Manual of Transportation Engineering Studies

Ta ble of Cont ents foreword

xxi

Acknowledgments

Chapter 1

Introduction

xxiii

1

1.0 Introduction 2.0

Purpose of the Manual

3.0 Organization

2 2

3.1 Organization of the Manual

2

3.2 Chapter Outline

3

4.0 General lips for Conducting Transportation Studies

3

4.1 Definitions

4

4.2 Developing a Study

4

4.3 Training the Data Collectors

s

4.4 Immediate Preparations for the Study

5

4.5 Conducting the Study ·

6

4.6 Data Collector Safety

7

4.7 Pitfalls of Data Collection

8

5.0 Summary

8

Chapter 2

Glossary of Terms

Chapter 3

Communicating Data to the Public

1.0 INTRODUOION

9 23 23

1.1 Objective of rhis Chapter

24

1.2 Guidance f(jr Other Readings

24

1.3 Evolution of Graphical Display of Data

24

1.4 Target Audiences

24

1.5 Chapter Organization

25

2.0 DESIGN OF GRAPHICS

25

2.1 Content-Driven, Not Software-Driven

25

2.2 Design Principles

26

2.3 Selecting a Graphical Display Method

26

2.4 Illustrative Examples

28

2.5 Engineering Drawings and Plans

32

3.0 WRITTEN REPORTS

34

3.1 Sections of a Report

35

3.2 Writing Style and Target Audience

35 Table of Contents • vii

3.3 Body of the Report

36

3.4 Use of Exhibits

37

3.5 Use of Appendices

37

4.0 PRESENTATION TECHNIQUES

37

I.

r I· ''·

4.1 Podium Presentations

38

4.2 Poster Presentations and Displays

40

4.3 Web site Design

41

5.0 SUMMARY

41

6.0 REFERENCES

42

6. 7 Literature References

42

6.2 Online Resources

42

Chapter 4

Volume Studies

1.0 INTRODUCTION 2.0

43 43

TYPES OF STUDIES

44

2.1 Intersection Counts

44

2.2 Area Counts

49

3.0 METHODS OF DATA COLLECTION

58

3. 7 Manual Observation

58

3.2 Automatic Counts

62

4.0 DATA REDUCTION AND ANALYSIS

67

4.1 Manual Counts

67


68

4.3 Count Periods

69

4.4 Volume Data Presentations

70

5.0 SUMMARY

74

6.0 REFERENCES

74

Chapter 5 1.0

Spot Speed Studies

77

INTRODUCTION

77

7. 7 Safety

78

1.2 Ttme-Mean Speed Versus Space-Mean Speed

78

7.3 General Speed Measurement Concepts

79

2.0 INDIVIDUAL VEHICLE SELECTION METHOD

79

2. 7 Introduction

79

2.2 Types of Studies

79

2.3 Data Collection Procedures 3.0 All-VEHICLE SAMPLING 3.1 Introduction

84

86

86

viii • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

3.2 Types of Studies 3.3 Data Collection Procedures . 4.0 DATA REDUCTION AND ANALYSIS 4.1 Data Reduction and Display

86 87 88 88

4.2 Descriptive Statistics

91

4.3 Inferential Statistics

92

5.0 SUMMARY

95

6.0

95

REFERENCES

Chapter 6 1.0

Intersection and Driveway Studies

INTRODUCTION

2.1 Equipment Needs

98

2.2 Personnel and Training Requirements

99

2.3 Field Procedures and Analysis

4.0

5.0

6.0

97

98

2.0 DELAY

3.0

97

99

QUEUE LENGTH

104

SATURATION FLOW AND LOST TIME

105

4.1 Equipment Needs ·

105

4.2 Personnel Training Requi rements

105


106

GAPS AND GAP ACCEPTANCE

109

5.1 Equipment Needs

109

5.2 Personnel Training Requirement

110


110

INTERSECTION SIGHT DISTANCE

112

6.1 Equipment Needs

112


113


113

7.0

SUMMARY

115

8.0

REFERENCES

115

Chapter 7

Traffic Control Device Studies

1.0 INTRODUCTION

117 117

1.1 Purpose

117

1.2 General Requirements

118

2.0 TCD STUDIES 2.1 Types of Studies 3.0 ESTABLISHING THE NEED FOR TRAFFIC CONTROL DEV ICES 3. 1 Traffic Signals

118 119 123 124

Table of Contents • lx

3.2 Signs 4.0 REMOVAL OF UNNECESSARY TRAFFIC CONTROL DEVICES 5.0 EFFECTIVENESS OF TRAFfiC CONTROL DEVICES

6.0

7.0

132 135 135

5.1 Road User Compliance Studies

136

5.2 Before-and-Aher Studies

136

5.3 Changes in Spot Speeds

136

5.4 Evaluating Safety Improvements

137

TCD CONDITION

137

6.1 Sign Retroreflectivity

137

6.2 Pavement Marking Retroreflectivity

139

6.3 Feedback from Citizens

140

SUMMARY

8.0 REFERENCES

140 140

8.1 Literature References

140


141

Chapter 8

Compliance w it h Traffic Control Devices

1.0 INTRODUCTION

143 143

J!

1.1 Purpose

144

~};

1.2 Applications

144

I

rt·

2.0 TYPES OF STUDIES

1:

145

2.1 Study Locations

145

2.2 Data Needs

145

2.3 Compliance Data 3.0 · DATA COLLECTION PROCEDURES

145 146

3.1 Personnel, Equipment and Training Requirements

146

3.2 Time and Duration of Study

147

3.3 Sample Size Requirements

147

3.4 Types of Compliance St udies

148


155

s_o

SUMMARY

156

6.0

REFERENCES

156


156


157

Chapter9

Travel-Time and Delay Studies

1.0 INTRODUCTION

159 1S9

1.1 Applications

160

1.2 Chapter Overview

160

2.0 TEST VEHICLE METHOD

160

x •

MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

2.1 Introduction 2.2 Data Collection Procedure

161

2.3 Data Reduction and Analysis

165

2.4 Volume Extension

166

3.0 OTHER TRAVEL·TIME STUDIES

171

3.1 Vehicle Observation ·

171

3.2 Vehicle Signature Matching M ethod

173

3.3 Platoon Matching Method

173

3.4 Probe Vehicle 4.0 REFERENCES

173 174

4. 1 Uterature References

174


174

4.3 Other Resources

174

Chapter 10 Freeway and Managed Lanes Studies 1.0

160

177

INTRODUCTION

177

1. 1 Chapter Objectives

177

1.2 General Freeway Facilities

178

1.3 Managed Lanes Facilities

179


182

2.1 Spot Evaluation

182

2.2 Segment Studies

187

2.3 Special Freeway Studies

190

3.0 DATA COLLECTION PROCEDURES

191

3.1 Facility Performance Data

191

3.2 Data Acquisition

191

3.3 Equipment Needs

192

3.4 Personnel Training Requirements

194

3.5 Field Procedures

194


195

4.1 Spot and Segment Evaluation

195

4.2 System ~onitoring

197

4.3 Managed Lane Measures

198

5.0 REFERENCES

Chapter 11

Simulation Studies

1.0 INTRODUCTION

199

201 201

1.1 . Purpose of this Chapter

201

1.2 Limitations of this Chapter

202

1.3 Types of Simulation Models

202 Table of Contents •

xi

1.4 When to Use Simulation

203

1.5 Definitions

205 206

2.0 lYPES OF STUDIES 2. 1 Sensitivity Analyses

206

2.2 Evaluating Alternatives

208

2.3 Predicting Behavior

210

2.4 Emergency Scenario Modeling

212

2.5 Safety Analyses

215

2.6 Environmental Studies

217


218 218

3.1 Model Setup

221

3.2 Types of Measures 3.3 Input Calibration

.

3.4 Output Validation

227

3.5 Procedure Sumr.nary

228

4.0 DATA REDUCTION AND ANALYS.IS

229

4.2 Determining the Required Number of Simulation Runs

229

4.3 Reporting Simufation Results

231

4.4 Documentation

232

4.5 Animation and Visualization

232 233

REFERENCES -

229

4.1 Concepts of Stochastic Variability

5.0 SUMMARY 6.0

223

234

"'··· ·

Chapter 12 Pedestrian and Bicyde Studies

237

1.0 INTRODUCTION

237


238

2.1 Volume Studies

238

2.2 Pedestrian Walking Speed Studies

246

2.3 . Gap Studies

246

3.0 PEDESTRIAN BEHAVIOR STUDIES

250


251

3.3 Data Collection Procedures

253


257

4.0 REFERENCES

vii •

250

3. 1 Introduction

259


259


261

4.3 Other Resources

261

. . "1111\1 f'IF

TR AN';p()RTATION FN<'iiNFFRIN(i STUDIES. 2ND EDITION

Chapter 13

Public Transportation Studies

' 1.0 INTRODUCTION

263 264

1.1 Purpose and Limitations of this Chapter

264

1.2 Transit Quality of Service

265

1.3 Interaction with Other Modes


265 265

2.1 Problem Identification

265

2.2 Transit Performance Measures

265

2.3 Transit Field Data

268

2.4 Use of Existing Data

271

2.5 Choice of Study Method

271


272

3.1 Manual Data Collection

273

3.2 Automatic Data Collection

281

3.3 On-Board Transit Surveys

283

3.4 Office Studies

285

4.0 STATISTICAL ANALYSIS 4.1 Sampling

.

285 285

4.2 Sample Sizes

286

4.3 Selecting a Sample

288

4.4 Estimating Mean Values and Proportions

288

5.0 REFERENCES

288


288


289

5.3 Other Resources

289

Chapter 14 Goods Movement Studies

291

1.0 INTRODUCTION

291


291

3.0

2.1 State of the Industry/Ex/sting Data Sources

293

2.2 Route Studies

298

2.3 Loading ~nd Unloading Studies

299

2.4 Vehicle Weight Studies

302

2.5 Hazardous Materials Studies

304

REFERENCES

305


305


307

3.3 Other Resources

307

T
Chapter 15 Inventories 1.0 INTRODUCTION

I

31 0

1.2 Purpo~e of the Inventory

310

2.0 STRUCTURE OF THE INVENTORY

l:i

309

1. 1 A Word of Caution

1.3 Choice of Data Elements

3.0

309

311 311

2.1 Means of Recording and Displaying Inventories

311

2.2 Inventory Location Referen ce Systems

316

2.3 Inventory Classification

316

2.4 Access to the Inventory

317

2.5 Storage

317

2.6 Retrieval

317

ESTABUSHING AN INVENTORY

317

3.1 Step 1: Determine the Purpose of the Inventory

317

3.2 Step 2: Select the Data Elements to Be Collected

318

3.3 Step 3: Select a Data Collection Technique

318

3.4 Step 4: Prepare a Data Collection Plan

318

3.5 Step 5: Collect the Data

320

3.6 Step 6: Construct the Database and Data Displays

320

4.0 MAINTAINING AN INVENTORY

320

4.1 Software

320

4.2 Frequency of Updates

321

5.0 REFERENCES

321

Chapter 16 Parking Studies

323

1.0 INTRODUCTION

324


324

2.1 Parking Inventory 2.2

Study Locations

325 325

2.3 Personnel and Equipment

325

2.4 Method

326

2.5 Parking Usage Studies

329

2.6 Accumulation and Generation Studies

330


332

3.1 License Plate Checks

332

3.2 Parking Interviews

33 6

3.3 Postcard/Flyer Interviews

337

3.4 Persona( Interviews

338

3.5 Parking Space Counting

339

xlv • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

4.0

DATA REDUCTION & ANALYSIS 4.1

. 5.0 6.0

Tabulation

339 339

SUMMARY

343

REFERENCES

345

6.1 Uterature References

345


345

6.3 Other Resources

345

Chapter 17 Traffic Collision Studies

347

1.0 INTRODUCTION

347

2.0

TRAFFIC SAFETY STUDIES

347

3.0

DATA COLLECTION PROCEDURES

348

3.1 Collision Reports

348

3.2 Sources of Data

352

3.3 Collision Data Reduction

354

3.4 Merging Collision and Other Data

355

3.5 Recognizing and Dealing with Problems in Collision Data 356 4.0 ANALYSIS OF COLLISION DA.TA

4. 1 Number of Collisions and Trends

357 358

4.2 Identifying Hazardous Locations- •Network Screening• 359

5.0

4.3 Selecting Countermeasures

369

4.4 Countermeasure Evaluation

379

REFERENCES

379

Chapter 18 Alternative Safety Studies

383

1.0

INTRODU CTION

383

2.0

ROAD SAFETY AUDITS

384

2.1 Introduction

384

2.2 Study Design

384


385


388

3.0 TRAFFlC CONFLICT STUDIES

4.0

390

3.1 Introduction

390

3.2 Study Design

390


394


405

ADVISORY SPEEDS

406

4.1 Introduction

406

4.2 Study Preparation

406

Table of Contents • ~

4.3 Data Reduction and Analysis.

5.0 REFERENCES

Chapter 19 Roadway lighting

409

412

413

1.0 INTRODUCTION

413


414

2.1 Existing Conditions

414

2.2 Before-and-After Analysis

415


415

3. 7 _Inventory of Existing Lighung

415

3.2 Roadway Grouping

415

3.3 Roadway. Pedestrian Walkway and Bikeway Gassirteations 419 3.4 Area Classifications

419

3.5 Collisions

420

3.6 Traffic Volumes ·

424

4.1 Control for Normal Variation

424

4.2 Use of Night Percentage

424

4.3 Ratio of Rates

424

4.4 Improvement Evaluation

425

5.0 REFERENCES

426


427

5.3 Other Resources

427

Chapter 20 Transportation Planning Data

429

1.0 INTRODUCTION

429


I

•I

430 430

2.1 Defining Study Areas

431

2.2 Inventories

432

2.3 Origin-Destination Surveys

437

2.4 Survey Partklpant Incentives

438

3.0 DATA COllECTION PROCEDURES

438

3.1 External Surveys

438

3.2 Internal Studies

446


i·

426

5.1 References

1. f ObjKtive of this Chapter

\\,

423


448

4.1 Presenting 0-D Data Results

448

4.2 Checking Survey Accuracy

448

·-·• - ""'""' m TIIA~I<;PnRTATlC)N ENGINEERING STUDIES, 2ND EDITION

5.0

4.3 Additional Sources of Data

450

REFERENCES

450

Chapter 21

Environmental Impacts of Transportation Projects

451

1.0 INTRODUCTION

451

2.0

HIGHWAY NOISE IMPACT STUDIES

452

2.1 Noise Impacts and Analysis

453

2.2 Determination of Existing Noise Levels

454

3.0

2.3 Noise Measurement Procedures

455

2.4 Prediction of Noise Impacts

456

AIR QUALITY IMPAt;T STUDIES

457

3. 7 Air Quality Impacts

458

3.2 Emission Rates

459

3.3 Air Pollutant Dispersion Models for Impact Analysis

459

3.4 Input Data Requirements

459

3.5 Construction Impacts

459

3.6 Air Quality Reports 4.0 REFERENCES

460 .•

4. 7 References

461 461

4.2 Online References

461

4.3 Other Resources

462

Chapter 22 Traffic Access and Impact Studies

463

1.0 INTRODUCTIO.N

463

7.7 Purpose of Studies 7.2

· Pr~paration

and Design of Studies

464 464

7.3 Need for Studies

464

1.4 Study Timing

465

1.5 StudyComponents

466


473

2.1 Site Traffic Forecasts

473

2.2 Nonslte traffic Forecasts

477


478

3.1 Total Traffic Estimates

479

3.2 Capacity

479

3.3 On-Site Circulation

480

3.4 Site Access and Off-Site Improvements ·

480

3.5 Transportation Demand Management

481

3.6 Residential Neighborhoods

481 Table of Contents • xvii

4.0 SUMMARY

4.1 Presentation

482 482

5.0 REFERENCES

483

Appendix A Experiment Design

485

1.0 INTRODUCTION

485

2.0 GENERAL CONCEPTS

486

2.1 Definitions

486

2.2 Objectives

486

2.3 Statisticallnference

486

2.4 Random Assignment

487

3.0 SIMPLE COMPARISONS

3.1 Unpaired Comparisons

4.0

487 487

3.2 Paired Comparisons

488

BEFORE-AND-AFTER EXPERIMENTS

489

4.1 Drawbacks to Before-and-After Experiments

489

4.2 Overcoming Before-and-After Dra wbacks

490

4.3 Analyzing a Before-and-After Experiment

492

4.4 Before and After With Control Experiments

492

4.5 Before and After with Comparison Experiments

494

5.0 FACTORIAL DESIGNS

494

6.0 REFERENCES

498

Appendix 8 Survey Design

499

1.0 · INTRODUCTION

499

1.1 Articulating Study Objectives

500

1.2 Why Survey?

500

2.0 METHODS

500

3.0 SAMPLE SELECTION

501

3.1 Random Sampling

501

3.2 Nonrandom Sampling

504

4.0 COMPOSING QUESTIONS

505

4.1 Number of Questions

505

4.2 Question Order

505

4.3 Types of Questions

506

4.4 Genera/Tips

507

5.0 EXAMPLE SURVEY FORM

508

6.0 PROTECTING RESPONDENTS

509

7.0 TRAINING INTERVIEWERS

510

xvlii • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

8.0 PRETESTS

510

9.0 SURVEY ADMINISTRATION

510

10.0 SOURCES OF ERROR

512

11 .0 REFERENCES

514

Appendix C Statistical Analyses

515

1.0 INTRODUCTION

515

2.0 DATA REDUCTION

516

2.1 Tables and Graphs

S16

2.2 Frequency Distribution

517

2.3 Time Series Distribution

S20

2.4 Spatial Distribution

521

3.0 DESCRIPTIVE STATISTICS

S22

3.1 Central Tendency

522

3.2 Variab/1/ty

526

4.0 STATISTICAL INFERENCE

4.1 Estimation

s.o

528 528

4.2 Reliability of the Sample

529

4.3 Significance Testing

532

4.4 Nonparamerric Tests

533

CALCULATION AIDS

536

6.0 REPORTING RESULTS

536

7.0 REFERENCES

538

Appendix D Supplemental Material on Communicating Data

539

1.0 INTRO.DUCTION

540


540

2.1 Section Overview

S40

2.2 Graphics Design

540

2.3 Tables

542

2.4 Types of Chart

543

2.5 Design Considerations

549

2.6 Use of Color

549

2.7 Graphics Checklist

550

3.0 WRmEN REPORTS

4.0

550


550

3.2 Organization of the Report

550


552

PRESENTATIONS

554 Table of Contents • ,...cix


554

4.2 Oral Presentations

555

4.3 Pvrpose and Scope

555

4.4 Organization of the Presentation

557

4.5 Answering Questions

557

4.6 Preparation and Planning Checklist

558

5.0 REFERENCES

559

Appendix E Useful Resources for Transportation Studies 1.0 DEVELOPING A TIME-STAMP MACRO

561

561

1.2 Coding

562

2.0 USEFUL DATA COllECTION FORMS

\i

i

561

1.1 Overview

.. """ '"' r"

TO"MCOf\OTI\Tif\M 'Mr.o•ot~QIMr. (TO '"It(

564

Foreword

T

his second edition of the MIZ'miAI ofTrrmsportatitm Engin«ring Studies (MTES) is an updated and expanded version to the 1994 MTES, which it3d.fwasa mision of the l976MIZ'miAiofTraffic EnginmingStvdUs, 4th Edition, by Box and Oppenlander. Some original ooncent of these earlier editions is maintained throughout this publication.

This edition includes a reorganiurion of the clupters into six pans of related transportation studies. Fout new chapters were added to this edition including a G/QSSJJry ofTn711S, a chapter on (;QmmunictZting Dlll4 to tht Public, a chapter on ~=ay and Managtd Lann and a chapter on SimulatWn Studies. These new clupters represent new and emerging trends in the transportation profession and are important additions to this publication. Work on this second edition of the Institute of Transportation Engineers (ITE) MTES began in 2008 with the confirmation of the project and author team. The team worked closdy with ITE staff and members of the Technical Advisory Committee to draft an outline of the revised manual. Individual chapter~uthorship responsibilities were based on areas of e:q>ertise and a detailed annotated bibliography was created to update references throughout the manual. The revised oudine and annotated bibliography were n:viewed and approved by ITE and the project Technical Advisory Committee before work on the final chapters oommenc:ed.

.

The ooncent of this handbook was finalized prior to the release of the 2009 Edition of the Manual on Uniform Traffic Control Dmm (MUfCD). Therefore, although all general references to the MUTCD have been updated to 2009, content with specific references to the 2003 MUTCD has been retained. Users are encouraged to consult the current edition of the MUTCD when making technical determinacions. A free copy of the 2009 MUTCD (PDF or HTML format) is available at htrp:f/mutcd.fhw.a..dot.gov. ITE, in oooperation with AASHTO and ATSSA publishes a hard oopy version available via the ITE bookstore at www.ire.org/bookscore.

This publication is .designed to aid rransportacion professionals and communities in studying their transportation JJWUICf, following procedures accepted by the profession. •

system in a suucrurM

The MTES should be used in conjunction with ITE's Traf!ic EngiMmng Handbook, Transportation Pilznning Handbook and other resources in order to prove most useful. As we release the second edition of the MTES, I hope that it will continue to be improved in later edicions. Your

coqunents and suggestions are earnesdy solicited as a means of making those improvements.

Bastian J. sduoeder, Editor c ......o •• , ......,.a

..

,..._,

Acknowledgments ••

I

I

I

I

I

I

I

I

I

I

I

I

I

I

o

0

I

I

I

I

I

0

o I

I

I

I

I

I

I

I

I

I

I

I

•

I

•

I

I

0

I

•

I

I

I

I

I

I

o

I

o

I

I

I

0

I

I

I

I

I

0

o

1

1

I

I

1

0

•

1

1

1

1

1

1

I

1

I

I

1

M

any hours of concerted effort wenr inro the development of this second edition of che Institute of Transportation Engineers (ITE) Manual ofTransportation Enginuring Studies (MTES), which would not have been possible withouc the dediatced and thorough work by the individuals listed below.

My responsibilities as editor were shared by co-edicqrs Christopher M. Cunningham, P.E. and Daniel ). Findley, P.E., who also served with me as authors of the overall manual, and as principal au~ors of individual chapters. Appreciation also goes to chapter co-authors Mr. Robert S. Foyle, P.E. and Dr. Joseph E. Hummer, P E. (M), whose professional expertise and technical insights made invaluable additions to the content of this manu al. As editor. I gceady appreciate the synergy and professionalism of ch.is core project team, which led co the productio n of a coherent and consistent publication. The authors would further like to thank a group of reviewers, who offered valuable reviews and feedback during chapter development including Mr. Loren D. Bloomberg (M), Dr. James A. Bonneson, P.E. (F), Dr. Darcy . M . Bullock, P.E. (M), Dr. Ronald Hughes and Dr. Nagui Rouphail. The authors also excend their thanks co the members of the Technical Advisory Coi:nmiccee ~ted below and ocher reviewers of draft chapters for the expert ise and bdpful i~ight they provided. lWp,b W. Boaz (M)

Philip H. Nicollama, T.E. (M)

Thomas W. Brockenbrough Jr., P.E., AICP (M)

Patrick]. O'Mara, P.E. (M)

MichadE.Kti

Eric C. Sben, P.E., PTP (M)

Gary L Heberr, P.E., PTOE (F)

Eric J. Tripi, P.E.; PTOE (M)

John M. Hemingway, P.Eng., PTOE (M)

Berty H. Tusrin,

Steven L. Jones Jr. (M)

Jan 0-:Yoss, P.Eng., PTOE (F)

P.E., PTOE (M)

Roberr J. Nairn (F) Since this document is an update of the 1994 first edition of the lTE MTES, all authors further acknowledge che thorough and det::illed work by the original author team, including Dr. H. Douglas Robertson, P.E. (Rtt.) (:J?l, Dr. Joseph E. Hummer, P.E. (M), Dr. Donna C. Nelson, P.E. (M), Dr. Ellis King and Marsha D. Anderson Bomar (::fl). The presenc effort to update this manual was greatly f.!cilitated by the exceptional quality of this first edition. The project ceam w'ould further like co thank the staff at the lnscirute for Transponarion ~earch and Edu:au on (lTRE) at North Carolina State University for their assistance in answe.ring d.iflicult technical questions, and for otP-er administrative suppon of project activities. Finally, the authors would like to express their graciwde co Lisa M. Fontana Tierney, P.E. (F) and Natalia V1a.sov Qf lTE for their patience, support and coordination effortS with the many people involved in this project.

Bastian}. Schroeder, (MS) Editor

(Lettm in parentheses jnJicate ITE member grad4: M-Mmsber, F-Fel/Qw, MS-Member Graduate Student.) Acknowledgments • XlC :5ii

___ .. -.

Chapter 1

Introduction OriginAl By: H. Doug/iu !Wbmson, Ph.D., P.E. joseph E. Hrnmner, Ph.D., P.E. DtmnA C Nelson, Ph.D., P.£. UptLzutl By: BastiAn]. Schroetkr, Ph.D. Cbrisfqpher M. Cun~injhmn. P.E. Dtmiel]. Fitulky, P.£. 1.0

Introduction

2.0 Purpose of the Manual

2

3.0 Organization

2

3.1 Organization of the Manual

2

3.1 Chapter Outline

3

4.0 General Tips for Conducting Transportation Studies

3

4.1 Definitions

4

4.1 Developing a Study

4

4.3 Training the Data Collectors

5

4.4 Immediate Preparations for the Study

5

4.5 Conducting the Study

6

4.6 Data Coliector Safety

7

4.7 Pitfalls of Data Collection

8

s.o Summary

8

1.0 INTRODUcriON ransportarion ~tems ue the h•ckbone of a nation's infrastructure. They provide a means foe people to travel to work, suppon home-based activities, enable vacation trips and serve as access to recreational and leiswe activities. Swface transportation networks further ~ntinue to cury a majority of freight movement and therefore directly affect economic prosperity. They enable rapid movement by fi.rst responders indudin~ fire, police and ambulance traffic, and represent strategic infrastructure for evacuation routing and martcrs of national secwity. The fields of transportation engineering and transponation planning have devised numerous methods for describing traffic How, forecasting tcavd demands and predicting future operations of transportation facilities. These methods are often focused on the automobile and truck modes, buc they also encompass nonmotorizcd modes of transponation, as well as transit. They encompass a range of transportation facility types, including rural roads, signalized anerial streets and limited access f.acilities. Many of these methods ue derived empirically and all require some level of field data as inputs.

T

The nation's transportation infiastrucrure also f.aces challenges. Problems of congestion, delay, puking. po!IHtiQp and safety are expected to dominate swface transportation foe the foreseeable future. Solutions to transportation problems Introduction • 1

and improvements to transportation facilities and services can only reasonably be developed after the magnitude, location and extent of the problems or the need for improvementS are well undemood. Such understanding comes from factual information gathered in an unbiased, objective manner, and analyzed to present a clear, concise picrurc of the narure of the problem.

2.0 PURPOSE OF THE MANUAL This manual is designed to guide and assist traffic engineers, transportarion planners, technicians and those ·assigned rraflic engineering responsibilities in the planning, design and execution of transportation studies in the field and in the office. It also serves as a reference for practicing engineers, researchers and engineering srudentS. The primary focus of the manual i.s on how to conduct transportation engineering studies in the field. The different chapters may address a specific data item of interest (for example, volumes or travel rime) or may cover a common classification of srudies (for example, simulation studies or planning studies). Each chapter introduces the type(s) of srudy that can be performed to obtain a specific data item, and describes data collection procedures. The discussion includes the types of equipment used, the personnel and level of training needed, the amount of data required, the procedures to follow and the techniques available to compile, reduce and analyze the data. A detailed discussion of applicacions is left to other sources. The focus is on planning the study, preparing for '6dd data coUection, executing the data collection plan and compiling, reducing and analy'l.ing the data. Guidelines for both oral and written presentation of study resultS are also offered.

3.0 ORGANIZATION 3.1 Organization of the Manual Each chapter in this manual is written so thar it largely stands alone without the need for extensive cross-referencing. However, to assure a con~e and efficient delivery of material some chapters will build on methods described in an earlier chapter. The chapters are grouped into parts of related topics as follows: • Introduction • Pan I: Spot Locations

• Part II: Segments and Networks • Part lll: Mulrimodal • Pan N: Asset Management • Pan V: Safety

• Part VI: Planning • Appendices Parts I through VI represent the heart of the material on data collection methods and are discussed briefly bdow. The Introduction features, in addition to this chapter, a glossary of terms used in the manual, and a general chapter on Communicating Data to tht Public that gives an overview of modern techniques for data presentation, visualization and public involvement. Information that i.s applicable to several cypes of studies is presented in the appendices. Such topics include general 5tudy design, questionnaire design, fundamental sraristical analysis and additional irtformarion on presentation techniques. An appendix containing cypical forms useful in transportation studies i.s also included.

PtZrt 1: Spot Locations Part I deals with basic srudies performed ar spot locations, including volume, speed and dday studies at intersections and driveways. In addition,~ section contains studies evaluating traffic control devices (TCO) and compliance. 2 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

· Part If: Segment and Network Data : Pan: II expands the basic conceptS described in the first pan ro segmenrs 2nd networks. It includes descriptions of speed. ·. travel time and delay srudies along corridors an.d networks, with reference to dara collection technologi es. This pact .,/Urther contains chapters on frj:eway studies and simulation studies. .

m;

Part Muhimod41 and Network Data Pan: III discusses alternate modes of uanspornrion, including a chapter on pedestrian and bicycle studies foUowed by a chapter on public tranSportation studies. Both chapters emphasize concepts of user perception to describe the qu:Jity of service of the transportation service. This parr concludes with a chapter on goods movement studies, an important area of uansporration receiving increased attention in the profession. Part IV: Assn Management Data Part IV contains two chapters on asset management studies. The chapter on inventories contains detailed discussion on automated data collection, including Global Positioning System (GPS) data and Geographical Information Syscem (GIS)-based data management. This pan: also contains the chapter on parking srudies.

Part Vt Safoty Data . Pan: V presents safety studies and srarrs with the chapter on collision studies and also contains a chapter on surrogate safety data. This part also contains the roadway lighting chapter, which focuses on the perceived and real impactS of lighting strategies, including crime and roadway safety data.

Part W: Planning Data Part VI presents uansportation planning data. The chcee chapters in this part describe general uansportation planning studies, environmental impactS of uanspoctation and craflic access and impact srudies.

3:2 Chapter Outline All the chapters generally follow the same outline to enable the reader to quickly navigate the material. This outline is followed wherever possible, but in some chapters the authors have had to deviate to ma.intain a proper flow of macerial, The general chapter outline is as follows: • Introduction: describes the pwpose of the chapter and comains general guidance for the described group of srudies. • Types of Snidies: outlines different types of studies that can be used to obtain a certain data clement and gives information on preparation and planning for a study. In some cases, the type of study is mentioned i~ an earlier chapter. • Data Collection Procedures: provides a specific methodoiogy for carrying out different types of studies, including equipment and personnel needs, observer locations and data collection technologies. • Data Reduction and Analysis: presenrs sample size calculations and common analysis steps for the collected data clemenc, including a discussion of data display and visualization techniques. • References: gives citations of sources used in the development of the chapter, and provides online and otb.er printed reso~ces th.a t may be helpful to the reader.

4.0 GENERAL TIPS FOR CONDUCTING TRANSPORTATION STUDIES This section contains general instructions and tips that are common to all the transportation engineering studies covered in this manual. These common elements include how to choose the right study (or none at all), ho-w to practice the data collection method, how to prepare on the day of the study, how to avoid mistakes duric::J.g the srudy and how to enhance data collector safety. The informacion i.n this section is viral for those witho&-l t experience in. transportation engineering srudies and should be reviewed occasionally by those with experience to make sure essentials are not being overlooked. The section startS with definitions of commonl:r used terms j.. l'l this manual. lntroductioo •

3

4.1 Definitions This book is a mwual of transportation studies. In this context, a study is defined as the detailed examination wd analysis of all or parts of a transportation system, supported by empirical data collection. The study statts with the identification and definicion of a transportation problem, followed by the design and execution of (field) data collection and the reduction and :malysis of the data in the office. A study Is typically performed to explore a specific aspect of or question about a transportation system, and study results arc usually written up in a ree,ort or similar document. Transportation studies arc oftentimes supported by dtzta colkction to empirically gather data in support of the study. Data collection traditionally is performed in the fie!~ using various study tuhniqu~ and titztll colkction equipment that arc described in more detail throughout this mwual. A Hudy technique describes the stepby-step procedure used for gathering data. Data collection equipment refers to a technological device that is specifically designed to measure one or more data elements (for example, volumes or speeds). Data collection for transportation studies can also be performed without gathering data in the field, but by extracting information from a central system (for example, the U.S. Census or other database), or by modeling the trmsportation system in a traffic simulation tool.

4.2 Developing a Study Transportation srudies can be divided into twO categories:

I

1. Studies carried out at regular intervals to monitor system performance ~d trends ri

volumes or crash statistics); wd

/ "WTTple, tiafli.c '

2. Studies to analyze specffic problems, usually in response to a concern or request. / ; . I .

Transportation studies serve to quantifY the extent of a problem, or to provide an estimate~ • 1e r applied solution strategy (before and after srudy), The rype of srudy is therefore uruqudy · I to analyst is trying to address. J:

C.

·,rmancc of an '!roblem the

: I

Transportation srudies are expensive and should not be conducted without considering tl{ /
I

• What is me purpose of the srudy and what measures are the desired outcome? • What analysis method will be used to solve the 'problem being faced? (Do not proceed until an analysis method is selected.) • What input dara are needed for the analysis method? • &e there acceptable values &om previous work that can be used as input <:bra? (If"yes~ for all inpua, do not use field <:bta collection.) • Are data available that can be mmipulated to become acceptable as input <:bta? For example, if ruming movement counts are needed, ace estimates from available link volume <:bta and nerwork geometry acceptable? (If "yes,• do not use field dat2 collection.)

• &e field srudy techniques available that will provide the input data needed? (If "no," do not use field data collection.) • Ace the time, money, personnel and other resources needed to conduct the field study available? (If"no," do not use field <:bta collection until the resources become available.) i"

\\ \

• Is there more dwl one 6dd study technique that will provide the needed input data with available resources? (If"yes,·. use the most cosr-efficient study technique.) 4 • MANUAL OF TRANSPORTATlON ENGINEERING STUDIES. 2ND EOffiON

, The question whether to conduct fidd dat:a collection and the choice of a particular study technique arc driven by the ~ needs of the question that must be answered and the analysis that is planned. Ax any time before or during the study, ! the engineer has the option of canceling or re-scheduling data collection if there is a change in the conditions that led ··fo the choice of study technique.

4.3 Training the Data Collectors After study sites, times, personnel and equipment arc sdccred and arranged, it is usuaUy necessary to train the data collectors in the study technique. Training and practice may nor be necessary for field personnel who have experience conducting the study of interest, bur it is essential for inccpericnccd personnd..Thc training session is typically scheduled for the day before or scvecal days before me fi.dd study is to be conducted, under conditions si.miLu to the most extreme expected during the study. For instance, if a day-long study is planned, training during the pw hours will be bene:fi.cial. Data collectors should practice under the direct supervision of me engineer for a short time, so that obvious mistakes can be corrected. The engineer and the data collector should also record data independencly for a short period of time; a comparison of those data will revo.lless obvious errors. The sire, time, p'ersonnd, cquipme.nr, memod and data collection form should be scrutinized by the engineer during and after the uaining session. The engi!lccr should also seek feedback and comments from the data collectors and answer any question. If extensive changes in the study technique are made as a result of a training session, it may be a good idea to schedule another session. to test mosc changes.

4.4 Immediate Preparations for the Study Too often, inefficient data collection or data collcccor discomfort results from inadequate preparation for the study. The checkllst in Exhibit 1-1 provides a means for data collectors to make sure that they have nor forgotten anything. Not.all of the items on the checklist are needed for every srudy. However, the best policy, espccia1ly for srudies at remote sites, is to take even marginal items from the chcckl..ist to the site. Unused items will not cause much aouble sitting in an automobile trunk, and the coruequences of forgetting an item may be large. Checklist item 1 is the mOst imporumt thing a data collector can do to avoid failure prior to the study: Equipment must be calibrated and checked from inpur, through storage, to output.

lntrodurtinn • 5

.-,

.c.

-~~:t~®f;l.,

" •

. .' __

e •. ·_..

o

-•.:-o~

:

. ...

...

~~:>'"',:r~:Y'4n~ ·;·.:'""'-~~·"'\·''tJ~('iY'""' ·H·.:::5Ji;~t""'l''

-~t ..····t.;:ri~~~--.h~~tl~.~~::i~~·;!...~&~;q*-~-~~-Jjt* .. jl~-~:

:~""f-F;:.,3

Before going to the site, do not forger

0

I.

!

To check che d2ta collection equipment. M3ke sure that ic records, scoces, and!or pcoduces oucpuc as you rcquirt. Make sure that the equipment is c:alib1'2.ted pcoperly.

0

2.

To label the equipment as needed (e.g., turning movemenr counters can be labeled with the approaches being watched).

0

3.

To bring the data collection equipment. Also. bring spares of equipment such as small, reliable stopwatches.

0

4.

A." acc=e watch set to the correct time.

0

5.

The correct siz.es of spa.te batre.ries for all the equipment.

0

6.

An abundance of forms. Also bring a clean copy of the form from which more copies can be made if needed.

Q

7.

Paper for caking notes.

Q

8.

Plenc:y of pens.

Q

9.

Clipboards or other writing surnces.

0

10.

A letter &om the landownet or bis/hec contact information (if private propercy will be used) and/or from a responsible agent of the highway aurhoric:y giving permission co collect data.

0

II.

A few business cards of the engineer supervising the study. The engineer's name and telephone number are ~etimes adequate.

Q

12.

A short, simple answer co the question, "Whac are you doing here?"

Q

13.

The telephone number where the supervising engineer can be reached on the day of the nudy, in case questions arise.

Q

14.

A 1112p showing the site or directions to the sire, if it is un&miliar.

0

15.

Folding chairs.

Q

16.

For a long study, a cooler or insulated conWrlec with beverages.

0

17.

A bar, sun visor, or sunglasses.

Q

18.

Sunburn protection.

0

19.

Ext1'2. cold-weather clothes, such as a sweatshirt and gloves.

Q

20.

Exu:~. warm-weather clothes, such as aT-shirt and shoru.

4.5 Conducting the Study On a study day, there are two main actions required of the engineer overseeing data collection. First, the epgineer must monitor the data collectors to make sure they are using agreed-upon procedures to assure quality control of the collected data. If necessary, the engineer may need to vmt the site to assure that data are being collected correctly and consistendy. Data should also be reviewed after the study for any unusual patterns that might indicate a compromise in data quality and integrity. Second, the engineer must be available by phone in case the data collectors need co confi:r. Any number of unusual occurrences can confront data collectors unhmili.ar with the analysis planned or unaware of alternate courses of action. Data collectors should be encouraged to call and place the responsibility for a decision with the engineer in charge of the project. Cellular phones are valuable in assuring char both the data collector and supervising engineer can be reached during the study. H owever, it should be emphasized to data collectors that personal conversations or text-messaging are distracting and should be discouraged during the study.

I·

Data collectors must typically arrive at least 15 minutes early at the site in order to assess conditions, distribute equipment, record crucial "header" informacion, assume positions and begin at the scheduled rime. The "header" informacion muse appear on each form or data record and include items such as (but not limited to) site name, date, time, observer name and weather condition. Sample forms included in Appendix E of this manual use a consistent header that assures all pertinent information is recorded. The same "header" informacion is also entered in decuonic count boards and is crucial to assure that the dataset can be uniquely idenci.6ed among all the informacion stored in the device, 6 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

, even if electronic count boards, computers, or other advanced data collection equipment is used. A hand h eld clipboard \is always a valuable asset to record any un usual events during the study or note general observations. Header information should be completed just prior to beginning the srudy. Data collectors waiting unril the end of che day to record header informacion on a stack of forms arc taking an unnecessary risk. Data collectot's should note any unusual occurrences in the transportation system that could affect the data being collected. Any deviations from accepted collection procedure should also be noted and probably should be cleared 6m by a responsible lead obser1fer or engineer. Data should always be recorded in ink to prevent fading, smudging and erasing. A clipboard assures that data collectors have a sturdy writing surface and that completed sheets are secured. Data collectors on pubUc or private property will often be asked, in cones ranging from poUte to threatening. •W'}lat this question chat will satisfY most members of the public without distracting too long from the data collection rask. A calm, professional approach and a referral to the supervising engineer are usually enough to defuse even very suspicious inquirers. Ic is always useful co carry the business cards of the supervisor whom people can call for questions. This is especially important if daa col· lectors are confronted by law enforcement personnel. are you doing here?" Data collectors should be taught a short, standard response to

If data collection is to occur on private property, it is recommended char the engineer overseeing data collection contaCt the property owner to inform them of the impending data collection, the nature and purpose of the data, the name of the data collection firm, the anticipated dates of data collection and the narure of the data collection material. Providitlg advanced notice to private property owners may a~id potencial conflicts in the field.

Local law enforcement should be made aware of data collection activities, especially if the study involves a laser speed "gun", or a permanent data collection unit tied to a sign post that may look suspicious to the untrained eye. Similar concerns may arise for any data collection chat uses video observations. Even ·standard" data collection (for cxample. counting traffic volumes from a parked car) may appear suspicious to some citizens and a quick phone call to the laW enforcement agency can prevent unnecessary complications. In addition, it may be beneficial to inform local government agencies about when and where data collection is planned and the method to be used, so local government staff can properly address any concerns raised by the public contacciflg the agency. If data collection involves the use of automated or video equipment that requires the equipment ro be a:crached or fixed to ai.sting ruuctures in the field, such as Ughr poles or sign posts, a permit may be required in adv:lflce from the local government agency. Consulting and informin.g that agency prior to data collection efforts can dctcnniP e if any pe.rmits are required.

4.6 Data Collector Safety The first responsibility of the data collector during the field srudy is t.omaintain his or her personal safety, the safety of the ocher data collectors and the safety of the traveling public. Traffic collisions are the primary safety threats duriP g rransportacion studies. To avoid traffic collisions and ocher safety hazards, data collectors should follow these cofl'3.monsensc defenses: • Follow employer's or agency's personnel safety policy, which often includes many of the points below. • Stay as far from the traveled way as possible. • Stay alert for 'errant vehicles. • Wear a fluorescent orange or fluorescenr yellow-green vest if working near the navded way. • Do nor interfere with existing traffic patterns and distract drivers as linle as possible. • Use standard TCDs, if applicable, to inform drivers of a closed lane, closed shoulder, activity near the traveled way, or other substantially changed driving conditions.

• If da~ collection is performed &om within a moving vehicle, a second data collector (other than the driver;> should perform all study activities, without distracting the driver. Introduction • II'

o

o

Data collectors should pay dose anention to the roadside environment, including holes, wires and potencial poisonous wildlife (snakes, ants) and plants (poison ivy). Data collection in cold climates or adverse weather requires adequate weatherproof clothing and ocher accessories, such as hats and gloves. Daa collectors should not underestimate the potential health impacts of cold and wee weather, especially when seated in a daa collection chair' for multiple hours.

Crime is also a threat to data collector safety during fidd studies. The: best defense: for data collc:ccors when criminal behavior chrc:atens is usually to abandon the study and leave the area. Safety from crime can be enhanced by the following measures: • Minimize nighttime: data collection. • Collect data in teams of at least £WO persons who remain in sight of each ocher at all times. Alert the local police when a data collection c:ffort is under way. o

Position data collection personnd in automobiles.

• Avoid the overt display of valuable equipment. • Keep personnel in the office aware: of the data collection schedule:. Other threats, .&om lightning to stray dogs, arise occasionally during srudies. A5 with threats from crime, the wisest strategy for many of these other chrc:ats is to abandon the study and leave the: area.

l

4.7 Pitfalls of Data Collection Fic:ld data collection requires careful planning, preparation and execution. The guidance provided in chis manual hdps analysts with these tasks, but is no substitution for personal experience with data collection. It is always recommended to initially team, or at least confer, with somebody who has some experience with a particular study to avoid common mistakes.

f' ! .

Data collection often requires coordinating efforts with several members of a research team, who collect different data items or may be: located in different locations. Data collection may also include the need to coordinate with equip· ment vendors or parmering municipalities for equipment installation or access to data sources. Considering these different contributors, the engineer needs to bring interpersonal skills to conimunicating with these partners and co supervise a pocentially large team of data collca:ors.

4 0 m cc cc

in fc: of

Data collection is subject to the uncertainty of weather and in some cases a project may actually require inclement conditions. Delays due to inclement weather or malfunction of equipment have to be anticipated. Consequ·encly, data collection efforts need to be well-scheduled and yet remain flexible to assure success. In addition to manual and small-scale data collection c:fforu, advances in technology and autonomous data collection methods may provide large quantities of data that can become unmanageable. Often, there is a need to prioritize data from available resources and be efficient in both the collection and analysis of data, given financial and time considerations. Without careful planning. the analyst runs the risk of collecting too much data (both in terms of data dements and guantity of data) that m;r.y result in inefficiency in data collection, as wd1 as in analysis time in the office.

wi

su

!• '

'

I I •

\:\~

\:

co

S.OSUMMARY

0 m ti< ol

Safe, efficient and c:ffective daa collection requires skill, artention to detail and common sense. The importance of "good" data cannot be overstated. Important conclusions arc: drawn from fidd daa that form the basis for decisions that a.ffcct the expenditure of large amounts of money and can have a significant effect on the safety of the public at large. Data collection demands the same levd of professionalism as any other task undertaken by an engineer or engi· neering technician.

th lx 6

8 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITlON

i

Chapter 2

'

.} .... ................. ......................... ........... ..... ...... .. .......... .... .

Glossary of Terms

ADT

Advisocy Sp«d Algoridun AMF

Acrolmt modifoa/Wn foC'tQr is based on quantitative results from safety re- I 17 search studies 'of various tratments indicating the expected percent reduction of crashes fOllowing the installation of a countermasuce. An AMP is based on more rigorous safety cva.luatioru than its similac f.u:tor, the crash reduction f.u:tor. This value is expressed as a dccim.al; an AMF of0.75 would man the expected crashes following a ~atment w-Ould be 75 percent of what would have been predicted had the' tratment not taken

Approach

All lanes of traffic moving coward an intersection or a midblock location I 7, 8 from one direction.

ATM

MvllnCed traffic 1Nln4gmu'11t, includes a variety of strategies to more I l 0 efficie.ndy manage traffic on existing hcilities, including managed lanes and ITS . .

.Art2irunent Ara

· meets or c:xceeds the U.S. Environ- I 21 standards used in the Oan Air Aa.

Average Day

A day representing traffic volumes normally and repatedly found at a loca- I 7, 8 cion. Where volumes are primarily influenced by employment, the average day is typically a weekday. When volumes are primarily influenced by entertainment or recreation, the aveta£C dav is wicallv a weekend

Ball-Bank Indicator

A tool used for determining the degree of dcB.ection of a vehicle through a I 18 horizontal curve. The deVice uses a curved rube 6lled with liquid. A weighted ball B.oats in the rube which determines the dcgru of deflection, which is based on a combination of effects related to superclevation, lateral acceleration and bodv roll. Glossary of Terms • 9

,~~~F""~ -_ -~ . :-· . ;.;- -~ _;,,.;:·:~.-:-.--:

'.

• _: ·~:t~!t~~~~-~\~~i;;~~;~~~~~~:-.~~~;7~l$-_-.~~ ~~;

Beacon

A highway traffic control device wirh one or more signal sections that oper- 7,8 aces in a flashing mode.

Benefit-Cost Ratio

A method for determining the cost-effectiveness of a particular countermea- 17 sute, where the benefit is based on the crash reduction and the cost is based on the countermeasure chosen.

Bicycle

A pedal-powered nonmotoriwi vehicle upon which_the human operator sits. 7,8, 12

Bicycle Lane

A portion of a roadway that has been designated by signs and pavemem 7, 8,12 markings for preferential or aclusive use by l:iicydists.

BRT

Bus rapid transit is a public transportation strategy where buses travel on 13 exclusive right of way or receive preferential treatment at signalized intersections, with the objective of reducing transit vehicle travel time and increasing reliability.

[CAD

Computn<-aidd tksign is a term describing software used for drafting engi- 3 neering design drawings.

Calibration

The process of changing the inputs or configuration of a device or system 4, 5, 11 to improve the accuracy and precision of the estimate. Calibration can refer to the process of adjusting data collection technology to match a known ground-truth. Calibration can also refer to changes in inputs to a traffic simulation model to assure the outputs match field observations.

Capacity

I

A transpotration facility's ability to accommodate a moving scream of people multiple or vehicles in a given time period.

Car-Following

An algorithm used in rnicrosimulation models to des~ibe the behavior of 11 one driver following another vehicle.

! .I

Causal Chain

The list of factors leading to the onset of a crash.

CBD

Cmtra/ businm district is the commercial and oftentimes also the geographic center of a city. The CBD may often have unique transportation systems incl11ding closely-spaced blocks and signals. It generally exhibits a high developmental density and often features a mix of differenr land uses.

Centerline Markings

The yellow pavement marking line(s) chat delineates the separation of traffic 7,8 lanes that have opposite directions of travel on a roadway.

Checker

A person performing a transit srudy while positioned either on-board a 13 transit vehicle or at a transit stop.

Chord

Any straight line measurement becween cwo points along a horiz.ontal curve, 18 typi~y taken from the centerline of the roadway.

Classi.6carion

A grouping of data into bins that describe different vehicle categories, such 4, 10 as passenger cars, trucks and buses.

Clearance Interval

The yellow and all-red intervals at a signalized intersection intended to 7,8 provide a safety buffer between green phases for conflicting movements.

Collision Diagram

Schematic, not-to-scale, graphical representation of the collision pattern ar 17 a particular intersection. This can also refer co a schematic of a single crash on a collision report.

Collision Report

The form used by the srate's department of motor vehicles (DMV) for 17 recording informacion about a collision.

Compliance

The lawful abid.ance of a regulated traffic control device.

Concurrent Flow HOY Lane

High occupancy vehicle (HOY) lane chat operates in the same direction as 7, 8, 10 the adjacent miud flow lanes, separated from the adjacent general pw-pose freeway lanes by a standard lane stripe, painted buffer, or barrier.

I

!

10 • MANUAL OF TRANSPORTATION ENGINEERING SnJDIES, 2ND EDITION

I I

17 .

8, 12

17

Condition Diagram

Confidence Level

of statistical confidence in an estimated parameter or I 5 stgnrncance. A 95 percent confidence level is common for most

Conflict

12, 18

Conuol Count

4

Cordon Coum

4, 13

Cosine Error

5

Count Expansion

n<"rtnnn<"tl

over a short time I 4, 12 volumes.

Countermeasure:

A measure or action taken to counter or offiet anocher action or series of l 17 causal faaors, such as a collision chain.

Covel'2ge Count

A short-term but continuous count of traffic performed over a period I 4, 13 of 24-72 hours at a particular ioa.cion. Coverage counts are multiplied by daily and seasonal adjustment factors from conuol counts to develop AADT estimates at che covera2:e-counr location. ·

Crash

wich anocher vehicle, animal, inanimate object, I 17

Crash Rate

The frequency of crashes per unit of distance, usually expressed as collisions I 17 per 100 million vehicle miles for a ucrUm:or collisions·per million entering vehicles for a mar. i

CRF

Crash rtductionj/U1Qr is similar to an AMF; however, che results arc based on l 17 any number and type of safety evaluations. Therefore, CRFs could be based on safety studies that are not rigorous in their method. Another difference is that a CRF is expressed as a percent; a CRF of 25 percent would mean that the expected crashes following a treatment would be 25 pctccm less than had the treatment not taken

Crash Severity Critical Gap

17 The threshold gap 'rime used to determine whether vehicles, pedestrians, I 6, 11 or bicyclists at a minor approach enter or cross the major aaffic saearo. In traffic flow theory, it is defined as the gap time where mi.nor·saeet traffic is equally lik.dy to accept or rejecr a gap. In the context of traffic simulation models it is often lised synonymously with the concept of a minimum gap where the minor-street vehicle will always accept gaps greater than the critical 2:11>. and will alwavs reiecr shorter ones.

Glossary of Terms • 1"'111

Crosswalk

The part of a roadway ac an intersection or at a midblock section that identifies the location for pedesuians to cross. The crosswalk is located within the lateral connections of the sidewalks .on opposite sides of the highway, typi. ' or otherwise marked to be visible to drivers and

DDHV .

Delay

The difference berween acrual travel time and the theorecical uavd time at I 6, 10 free-flow oonditions. Ttm4-in--qunu tklay (TIQD) is the difference be~ the time a vehicle joins the rear of a queue and the time the vehicle clears the intc.rsection. Control tklay is the component of delay that results when a control signal causes a lane group co reduce speed or to stop; it is measured by comparison with the uncontrolled cxindition. It is defined as the TIQD plus time losses due to decderation &om and acceleration to fiee-Bow speed. GrometrU Jelay is the component of delay that t'tStl.lts when geometric fewtts cause users to reduce their speed in negotiating a facility. Tn:tvd-timl! Jelay (TID) is the d.iffcrence berween the time a vehicle passes a point down.sttearn of the incerscction, where it has regained normal speed, and the time it would have passed thar point bad it been able to continue thro~ tbt intc.rsection at This includes all control and

Delineator

A retroreBectivc device mounted in a series on the roadway Slll'face or at the I 7, 8 side of the roadway to indicate the alignment of the roadway or a roadside hazard.

Dapand

The number of vehicles forecast to pass a point in a certain period of time. I 4 The acrual observed volume at that point can be less due to congestion mecering upstream traffic or a traffic control device metering the throughput at the

!

.. ;

!~,•

4

j:

time divided by the 14, 10 lane).

Density

~

"

5

7,8 i

II

rl ,.!I ;j

:l T i

Disuibution

11

DMI

Dist4n&e musurint i11JtrUmmt is a tool thar accurately records distances and ooordinarcs as a ocrson drives.

Doppler Effect

The physical principle behind speed measurements using a radar or Laser gun. I 5 The speed is inferred by measuring the fr~u~q $h.ift of a high frequency laser/radar wave as iris reflected &om an aooroachirut or recedin2 vehicle.

12 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDillON

0

DTA

Edge Line Markings

Dynamic trafJic a.uignmmt is the process of allooting traffic routes in a I 11 simulation model based on current traffic conditions. See Route.

or ycllow pavement marking lines that delineate the right or left I 7. 8 of a traveled

Engineering Judgment I The ~uation of avaib.ble penincnt information and the application ofl7, 8 appropriate principles, standards, guidance and practices as contained in the appropriate manual or other sources, for the purpose of deciding the or installation of a device. Engineering Srudy

The comprehensive analysis and evaluation of available perrinem informa- I 7, 8 cion, and the application of appropriate principles, standards, guidance and practices as contained in the appropriate manual or other sources, for the purpose of deciding upon the applicability, design, operation, or inscallation of a device.

EPDO

Equivalmtpropmy dam4g~ only is a method of adjusting cra5h frequencies or I 17 rates to rdlect greater cnsts o( injury of fual collisions. Using the KABCO scale, the equivalent ratio of &tal (I<) and severe collisions (ABC) are used to determine ifa sire is truly hazardous.

ERC

EtHJ1:714lion mpqnu curvt is a model that predica the rate of evacuation of l 11 people from an ~mergc:Qcy area (e.g., hurricane) as a function of time until the predicted event starts.

ET

Expms to,U lane is a managed lane freeway f.tcilicy that is open ro roll-paying I 10 vehicles only. ·

FHWA

multiple

Flow

The srudy of interactions between vehicles, drivers and infras~ccurc, with I 4 the aim of understanding and developing an optimal road network with efficient movement of tnffic and minimal traffic

Flow Rate

The number ofvehicles passing a pointin a period Je5s than aii hour expressed I 4, 10 in vehicles per hour. Operational analyses frequently usc a 15-min. flow rare, which is the number of vehicles observed in 15 min. multiplied by four, to obtain the hourly flow rate.

Gap

The available time in seconds between two successive vehicles at the same 17, 8, 12 point in space, measured &om the rear bumper of the lead vehicle to the &om bumper of the followin2 vehicle. AD. algorithm used in microsimulation models to describe the behavior of 1 11 a driver, pedestrian, or bicyclist entering or crossing a conBicting major traffic scream. 3

Gap-Acceptance

GIS GPLanes

all traffic without I 10

Glossary of Terms • 1J

~~;.¥

•••• "";~~ 0 . :;:~;J~k;-

.

~w~~~~k~~i~~: ~-~

GPS

Global Poritioning System is a system of sarellires, computers and receivers multiple chat is able co determine the latitude and longitude of a receiver on earth by calculating the time difference for signals from different satellites to reaeh the receiver.

I

Guide Sign

A sign char shows route designations, destinations, distances, ~rvices, points 7,8 of interesr, or other geographical, recreational, or cuirural informacion.

I

Highway Capacity Manual is aTransportation Research Board (TRB: see Iacer multiple term) document providing computational methodologies for estimating the operational performance of traffic Bow.

I

HCM

Headw:~.y

The time in seconds between the arrival of a common point on two succes- 6, 7, 8 sive vehicles (e.g. front bumper) co reach the same point in space. Headway also describes the time in minutes between successive transit vehicle arrivals.

I

HOT

High occupancy and toll facilicy that is open to both HOY and drivers willing 10 to pay a toll to travel on the facilicy.

!

HOY

High occupancy v~hick is a motor vehicle carcying more than one person, 10 including carpools, van pools and buses.

HOY Lane

Any preferential lane designated for exclusive use by high occupancy 7,8, 10 vehicles for all or part of a day, including a designated lane on a freeway, other highway, street, or independem roadway on a separate right of way.

HSM

Highway Safay ManUill is an AASHTO document providing compuracional 17 methodologies for estimating the safecy performance of a particular roadway or roadways.

Impact Score

A way to assign a weighting to different performance measures or questions 13 in a survey.

!

Incident

A crash, broken-down vehicle, or other event on a transportation facilicy 10, 11 that represents an unanticipated source of congestion.

1:

Interchange

A system of interconnecting roadways providing for traffic movement 7, 8, 10 between two or more highways that do not intersect at grade.

Interrater Reliabilicy

A test to assure that multiple independent observers reliably arrive at the 1, 4, 12 same result in a transporrarion srudy.

ITE

Inrtitu~ ofTmn.portati
Iteration

A replication or run of a simulation model starting with a different random II number seed. Each iteration populates the simulation model with a different set of stochastic base conditions that will affect the final estimated paramecers. Simulation studies require multiple runs to obrain an average estimate and a sense of the variabilicy in the estimate.

ITS

In~/ligtnt

Jay-walking

An unlawful crossing by a pedestrian, eirher in rhe form of crossing against 12 a "DON'T WALK" signal indication or an informal crossing at a midblock location.

f'

,. I

is an international association of trans- multiple portation professionals responsible for planning, designing, implementing, operating and maintaining the surface and ground transporracion systems of the world. ITE provides for th~ pcofessional development of members and others in meeting society's needs for safe, efficient and environmencally compatible transportation.

li

I I

transportation syrtems include the application of advanced 10 and emerging technologies in fields such as informacion processing, communications, conaol and elecuonics co surface cransporration needs.

I "•

MANUAL Of TRANSPORTATION ENGINEERING sruol!s, >No eomoN

~-··1l.J.~--·.;,. . - . ~~·:...,....... ';'i,l.\""i~ .

'

~

.

a': ·.·· · ,·· ~JJ~:t.:~ii3~':;~·~ .··~.-~-,;:.,~?;;/j:~~~'TJ':·;f~~t~~ ~-}'

KABCO

An acronym describing the method for expressing the severity of a collision, where K represents a "'killed• individual in a crash event. ABC represents three different levels of injury and 0 is a non injury collision.

Lag

lime elapsed bccwecn the arrival of a minor-street vehicle ready to move 6 into the major street and the arrival of the front bumper of the next vehicle in the major traffic stream. The distinction berween a gap and a lag is critical, because drivers may react differently to each.

Lane-Changing

An algorithm used in microsimulation modds to describe the behavior of a 11 driver switching into an adjacent lane of traffic.

Laser

A technology used to measure vehicle speeds. See Doppler Effict.

Level

A condition in an experiment that describes the scare of a particular treatmenr. Appendix A A rating of traffic performan.ce ranging from lecter grade A (excellent) to F 10, 12 (failing), stratified _by some performance measure observed iq the field or calculated from analysis methodologies.

Levels of Service

18

5

Loa.d

The number of passcngen in a tranSit vehicle.

13

Loop

See Smsor.

5

Macroscopic

A traffic model of aggregate vehicle How (not individual vehicles) based on 11 macroscopic traffic How principles.

Major Street

The street normally carrying the higher volume of vehicular traffic.

Managed Lanes

7. 8 A te;m encompassing alllimired access roadway &cilitics, including HOY, 10 HOT and Truck Only Toll (TOT) &cilitics. There are also some managed lane &cilities along arterials.

Mean

The average of a sample calculated by the sum of all observations divided by multiple the sample size.

Measwes of Effectiveness

Measures or resrs which rdloct the dcgR:e of attainment ofparticular objectives. multiple

Mesoscopic

A model of traffic flow that represents individual vehicle ml>vements based 11 on macroscopic traffic stream models (speed-How-density relationship).

Michigan U-Turn lntenection

A reconfiguration of a traditional intersection that functions by redirecting 4 through and lefr rurning traffic from the main and sjde street approaches to rum right, proceed to the nearby U-~ and then rerum to irs original course. It is a method to safely and efficiently manage high traffic volumes ar intersections with multiple approaches along a divided highway. Similar to a "Supersrreec lntenection."

Microscopic

A model of traffic flow that represents individual vehicle movements based 11 on vehicle-level behavioral algorithms such as car-foUowing, lane-changing, or gap-acceptance.

Middle Ordinate

The measured distance from a known chord to the centerline of the roa.dway. 18

Minor Street

The street normally carrying the lower volume of vehicular traffic.

Model

A computational approach for representing traffic How, which includes 11 tkw-ministic (equation-based) procedures found in the Highway CtJpaciJy MtJnual, as well as ftQchtJJtic (probabilistic) methods found in simulation.

I

7, 8

MOE

MellSU" ofEffictivenesr, sec Prrform4n« Mearu~.

Multiple Threat

A type of pedestrian-vehicle conHicr, where a yielding vehicle in the near 12 lane blocks the pedestrian's view of a vehicle in the fur-lane and vice versa.

---

i

..

'

11

-

Glossary of Terms • 1!!$

TraJ!ic CAntrol Dn~icts is an FHWA documenr specify- ! multiple ing requirements and guidelines for the we of signals, signs, markings and other traffic control devices in the United States.

MUTCD

I ManUitl on Uniform

NCHRP

I National Coop"atiw HighwllJ Rntarch Program is a TRB research I multiple program of pooled state funds that addresses broad-reaching issues in the transoort:uion field.

I A collection of streets, highways and intersections that define the physical I 11

Netwark

characteristics included within a study area or within study boundaries. This also refers to boundaries in a simulation · · NonattalnmentArea

I An area considered not to have met the U.S. Environmental Protection 121

The time a vehicle spends traveling over a roadway sensor expressed in I 4, 5. 10 seconds per vehicle or percent. Occupancy can be wed in combination with an assumed or known vehicle length to infer the speed of the vehicle. Also, occupancy can refer to the number of persons per vehicle as a measure of

Occupancy

I Origin-Dtstin~~wm. Starting and ending points of all aips in a tnnsporta-14

0-D

cion network are compiled into an 0-D maaix that is used for planning and analysis. Online Mapping Tools IA desktop application for Mac, PC. or Limn computers that allows you to I 3 navigate around the world from multiple views. These tools combine satellite photos and maps with a search engine to allow the user to search for directions and specific addresses or general locations and services. Multiple service oroviders exist that orovide similar online maooin£ tools.

IA speed at which a typical vehicle or the overall traffic operates. Operating I 5

Operating Speed

speed might be defined with speed values such as the average, pace, or 85th Optimization

The process of using a computer algorithm to obtain optimum configura- I 7, 8, 11 cion of sip timing or another traffic control scracegy. The optimum configuration is identified by minimiz.ing or maximizing an objective function such as total vehicle

0 nho-Rectification

The process of calibrating a video image for the pwpose of applying video I 4 image processing software for automated data collection from known (true) dimensions such as pavement markings, lane widths, or the height of the camera. reoresents more I 5 typically

Pace Speed

13

Paratransit Path-Based Count

A type of volume study that requires the analyst to trace a vehicle through I 4 an intersection such as a roundabout or

PCE

PIZSSmgtr-cJlr ~quivalmt is a measure of vehicle Bow used by the Highway 110 Capacity MmuuJ that converts heavy vehicles to an equivalent of passenger than 1.0) for calculation 4

PwHour ~c

•

~A ANI IAI ()F

TRAN<;P()RTATION ENGINEERING STUDIES. 2ND fOITION

Performance Measure I A data clement used to quantify a particular aspect of cransportarion service I 13 and used to describe its quality ofservice. Examples ofperformance measures include delay, travel time, or emissions. Permissive Turn

I Left rurns may be made during the same signal phase as the opposing through I 4 movement. There is no exclusive signal indication (prorected phase), and drivers must vidd to oncomiruz craffic and ' · ·

Platoon

I A group of vehicles or pedestrians traveling together as a 'group, either 17. 8, 12 voluntarily or involuntarily, because of traffic signal eoncrols, geometries, or other hcrors.

4

12, 13

Queue

multiple

Queue Length

7, 8, 10

Raised Reflective Pavement Marker

lUmp Metering

Reliability

10

Glossary of Terms • 17

~~ti:.~"".l''·~' . ..· ·" .... '''"''~. ;)';/' :·,.:. . .·

.

-,~~lmr~>'ll· ~ <.t't~~"lf~~~~!tt.~~-

·~~;

i~~~~

Replication

Used in experimental design to describe how many units are evaluated or Appendi x reseed for a particular treatment. For example, a study on a mfli.c-calming A installation may include four replications, if the same treatment is applied to four different incecsections. The term replication is also sometimes used co describe an itnatirm in a simulation study.

Reversible Lanes

One or more Lanes char can be operated to carry traffic in either direction of uavel, 10 found on ~y .&cilicits, arterW streets and at sporting/event complexes. Traffie is managed through ITS treatments, variable signing and movable barriers.

RightofWay

The permirring ofvehicles and/or pedestrians to. proceed in a lawful manner 7. 8 in preference to other vehicles or pedestrians by the display of sign or signal indications.

RSA

&ad S4[dJ audit is a formal safety evaluation of a furure or existing roadway 18 or alternative user facility by an independent audit team. A circular intersection with yidd control of all entering cnflic, channelized 4, 7,8 approaches and app.ropriace geometric curvarure, such chat rravel speeds on the circulatory roadway are typically less chan 50 km/h {30 mph).

Roundabout

Route

The specified course of a driver traveling through a simulated transportation 11 nerwork. Routes can be permanendy defined by the user (static) or can be determined through a path-optimization algorithm by the modd (dynamic).

Run

See ltnation.

Running Speed

A measure of the average speed over a segment, which is calculated as the dis- 9 tance travded divided by che running time.

Running Time

The rime a vehicle is acruilly in motion (or moving faster than a predesignated 9 speed) while traversing a given segment ofstreet or highway.

SPF

Safoy pcfomumajimctUm provides a prediction of the cxpeacd crash frequency 17 at a site for a base set of conditions (lane width, shoulder width, etc.) and is calculated for a segment using, typically, only AADT and segment length.

Sample

A representative subset of the total population used for data anal)'3is. Scatis- multiple tics (mean, st2ndard deviation) of the sample are used to approximate the true parameters of che population. A counr of ttaffic between two subareas performed by recording traffic flows 4' at crossing points berween che two regions. A screen-line count is typically performed along a narurai or human-made barrier wich limited crossing points, such as an urban freeway or a river.

Screen-Line Count

11

Sensitivity Anal~is

The process of varying input parameters in a stochastic modd to gauge che 11 effect of these changes on che predicted performance measures.

Sensor

A term describing various types of automated data collection equipment 5 used to obtain traffic operational data. Sensors include magnetic inductance loops, video-based detection methods and sensors using microwave or infrared technology.

Service Measure

A performance measure in public transportation studies used co define chc 13 transit levd of service.

SbarCd-Use Path

A facility out'Side the main traveled way and ph)'3ically separated from motor- 7,8, 12 ized vehicular ttaffi.c by an open space or barrier and either within the highway right of way or within an independent alignment. Shared-use pachs are also used by pedestrians (including skaters, users of manual and motorized whedchairs and joggers) and other authorized motorized and nonmocorized usets.

Sign

Any aal:lic control device chat is intended to communicate speci.6c informa- 7,8 cion to road users chrough a word or symbol legend.

18 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDffiON

I

I .

Signal Phase

Simulation

Speed

A software-based analysis approach that uses stochastic algorithms dacrib- I II ing behavior of motorized and nonmotorized users of the transportation ..,.,......·-uNrh the obiective of ouanrifvine: traffic o~rfnrm~nc~

4,5,9,10

Speed Limit

The maximum (or minimum) speed allowable on a section of roadway as I 5 atablishcd by law. Speed limits must be posted, as well as established by law to be

Speed Trap

A way to indirectly mosure the spot speed of vehicles by m~g the I 5 time it takes a vehicle to travel a known (short) distance, such as the distance ~cween cwo closely spaced magnetic inductance loops or cwo rransverse

Spot Map Spot Speed

SSAM

17

rs,

as measured at a point location or over a short 7, to the travel speeds measured over a segment or longer Appendix C 11

Statutory Speed

Steady-Scare

Superclcvacion

The cross-slope of a road (somecima referred to as camber or cross-slope). I 18 It is the difference in elevation between the cwo edges. A cross-slope which is not equal co zero results in a banked turn, allowing vehicles to traverse the turn at higher speeds than would otherwise be possible. Superelevarion is as a oercem or decimal such as 0.03 or 3

Glossary of Terms • 'I 9

Superscreet

A reconfiguration of a traditional intersection. A reconfiguration functions I 4 by redirecting through and left-turning traffic from the side stteet approach to tum right, proceed to the nearby U-turn and then return to i.ts original course. Left turns can be allowed from the mainline or made to U-turn; however, lefts from the mainline do not cause any progression issues. It is a method to safdy and efficiently manage high traffic volumes at interseCtions with multiple approaches along a divided highway. Similar to a "Michigan U-Turn lnursection."

Synchronization System

4, 12

the simulation analysis domain in space, levd of complexity I 11 duration. ·

Timt-mean speea is the basic f;lrithmttic mean of speed collected at a spot I 5. 9 location. It is simply calculated by summing the speeds of all individual vehicles crossine: over a point and dividine: bv the number of observations.

TMS

I Traffic managemmt center is a centralized data processing and observation 110

TMC

unit that monitors traffic flow and ITS technology deployed on a freeway or arterial network. Tolerance

I The user-defined allowable difference in a test of statistical significance. I 5

TOT

In a speed study, the tolerance may be the difference between before and ~e.r speed observations that is considered I Truck only toll is a f.l.ciliry that is intended exdusivdv to process heavv 110

Traffic

I Pedesttians, bicyclists, animals, vehicles, sueetcars and other conveyances, I 7, 8

Traffic Conflict

I Interactions between two or more vehicles or road users when one or more 118

vehicles but that

vehicles or road users take evasive action, such as braking or weaving, to avoid a collision. Traffic Conflict Study

I An alternative safety study often conducted in lieu of an actual collision I 18 study because it may be quicker or easier, or because there may be a lack of available collision data. In this case, a surrogate measure, such as brake lights or excessive dippine: of the front end of a vehicle, is used.

Traffic Conuol Device I A sign, signal, marking, or other device used to ree:ulate, warn, or £Uide I 7, 8

7, 8 11

Trajectoty

Treatment

-.A

-

A particular combination of levds (states) in an experiment that describes I Appendix A a condition being studied. An example of two studied ueatments would be two tYPes of sUmin£ used in a before and after

'- A!A"-1111\1 f'\C TDA~l(:Df"\OTATI/"\tJ Ch.tt:lt,ltE:QU,.If:. CTIII"'\It:C

'U.,II"\ CI"\IT1f'U.I

TSP

Trrmsit sigruJ priority is a signalize
Units

the traffic operacions are experienced by a driver, I 10, 12 transit

User Perception Validation

Vehicle

The comparison of estimates obtained from the simulation model (uutputs) I 11 with similar measures obtained from field data for a common location Every device in,

4, 5, 7,8

ed or dr.lwn exclusive or Vurual Earth

A Web-based, interactive geographical guide that allows users to search I 3 and observe maps and aerial photographs of specific areas, neighborhoods and points of interest on the earth's surf.u:c. The program is provided by Microsofr.

Visual Analytics

The science of analytical reasoning facilitated by interactive' visual interfaces I 3 such as traffic simulation

V LSUa.!iution

3,11

Volume

multiple

Warning Sign

7,8

Warrant

A rule describing threshold conditions to the engineer in evaluating the I 7, 8 potential safety and operational benefits of traffic contrOl devices based upon average or normal conditions. Warrants arc not a substiruce for engineering judgment. The fact that a warrant for a particular traffic conttol device is met is not conclusive justification for th~ installation of the device.

WIM

Wdgh-in-TMtion is used to determine if a ttuck is traveling within a reason- I 14 able range of the legal limits as adopted by a local goverruncnt. The data collected may include all or some of the following: gross vehicle weight, axle · ' and tandem axle

Glossary of Terms • 21

Chapter 3 o 0

o

0

0

o o

•

o

0

0

o

0

0

0

0

0

o 0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

o 0. 0

0

0

0

0.

o

0

0

•

•

0

o

0

o 0

0

o 0

o e

0

0

0

o

0

o 0

0

°

II

Communicating Data to the Public RobertS. Foyk, RE. BastiAn J Schroeder, Ph.D.

1.0 INTRODUCTION 1.1 Objective of This Chapter 1.2

Guidanc~

for.Other Readings

1.3 Evolution of Graphical Display of Data

23 24 24 24

1.4 Target Audiences

24

1.5 Chapter Organization

25


25

2.1 Content-Driven, Not Software-Driven

25

2.2 Design Principles

26

2.3 Selecting a Graphical Display-Method

26

2.4 Illustrative Examples

28

2.5 Engineering Drawings and Plans

32

3.0 WRITIEN REPORTS

34

3.1 Sections of a Report

35

3.2 Writing Style and Target Audience

35


36

3.4 Use of Exhibits

37


37

4.0 PRESENTATION TECHNIQUES

37

4.1 Podium Presentations

38

4.2 Poster Presentations and Displays

40

4.3 Web Site Design

41

5.0 SUMMARY

41

6.0 REFERENCES

42


42


42

1.0 INTRODUCTION b.is chapter gives an overview of modern techniques for data presentation and visualization for public involve ment in. \lllderstanding issues and bdping make decisions. With advances in computing technologies and >f'i--

T

sualization capabilities of modern software cools, new realms of dara display and ways to com.m,unicate·data c:4:i the public and decision-makers have become possible. Ax the same time, people's levels of expectations of reporu an. .d Communicating Data to the Public • 2 ..::::3

public meetings have become elevated, and they have come to rely on sophisticated presentation techniques and visualization to convey traffic engineering data. The effectiveness of this presentation trend is further enhanced by simulation and animation techniques, providing the audience with four-dimensional representation nf traffic operations. One question to be asked is, •Where does 'data' stop, and 'information' begin?" (Transpornuion Research Board: TRB, 2006). That is, when does the gathering of data and subsequent analyses everinW!y become informacion? For the purposes of this chapter, the term data will mean some compilation of raw data into refined groups with the intent of conveying information to the audience.

1.1 Objective of This Chapter The objective of chis dupter is to present to the reader basic concepts of the graphical display of data in the form of cables, charts, written reportS and public presentations.. This chapter presents the key principles and concepts of interest to most readers. Appendix 0 contains supplementary material discussing additional derail for graph and chart design, written reportS and presentation techniques.

1.2 Guidance for Other Readings This chapter explores and discusses the key concepts involved in communicating data in written and oral forms. The conceptS and guiding principles come from research on cognitive learning (for example, bow the brain interprets and comprehends both auditory and visual messages received). Recent publications by Tufte (2007) and Kosslyn (2007), for example, dearly demonstrate :l need for tnrtsportacion practitioners (and others) ro understand these principles to properly communicate data to the public and decision-makers, and thus achieve some consensus in d~ign concept or acceptance of a point of view, Ignoring these principles may mean your audience loses interest in your presentation, chara and graphs arc confusing or misleading, tables cannot be seen or understood, figures are distracting. or data are not uranged or grouped in understandable patterns. As practitioners, your job is to convince your audience that the data you have collected, analyzed and used to form your conclusions (or d.esign decisions) are logical and technically sound. If your audience cannot follow your reasoning. then you have failed to communicate effectively. The principles presented here will hdp overcome some typical problems that occur in both written and oral communication of data ro the public.

1·.3 Evolution of Graphical Display of Data There uc few compelling presentations that rdy soldy on the spoken word. Not char long ago, engineers often worked with ucbitectural firms to create hand-colored d.rawings of plans or buildings, including 30 modds, for review by the client and the public. MOst presentations today rely on visual, auditory and interactive features to capture and maintain audience attention. This can include both 30 and 40 visualization techniques. Prior to the advent of the personal computer in the 1980s, achieving compelling graphics also meant hiring a graphic designer or graphic design firm to create posters, figures,llyers, brochures and other materials for communicating with the public. With today's computers and software paclagcs, anyone with time co lc:am can create stunning graphics. And ycc, all this .8c:xibility may confuse the transportation practitioner who has not been trained in graphic design principles. .How chen can the practitioner rc:adily grasp and apply the major principles of graphic design that lead to better understanding of analysis results? This chapter covers those principles of graphic display creation to enable effi:aivc informacion acha.ngc and understanding. Simply having the rools does not mean you can automatically gcncruc a graphic to go in a report or presentation.

1.4 Target Audiences It is common knowledge: Both written and oral communication must be wgered to an intended tudiencc. Is yout presentl.cion for a public hearing on a proposed design, or a meeting of the town council on results from a parking 24 .. MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

study? Are you looking to inform or reach consensus? Are you uying to entertain or share knowledge? Is this h ighly technical information unfamiliar to the audience, or will the audience understand the terms being used wi thout definitions? Asc maps or design drawings needed? These are some of the questions you need to ask yourself when developing a report or presentation. Public involvement is both mandated and necessary for many engineering projects. The designer should ~ow how information will be wed; for example, small user group versus auditorium, and anticipated composition of the audience. A successful project will demand attention to these kinds of details (Kihl, 2008). Highly sensitive or controversial projects may require a public involvement officer to mediate any discussions to help achieve consensus on a preferred design. In some instances, the legal department of an agency may be called in to present a legal opinion on an issue or situation. Kosslyn (2007) defines three goals for presentation material: Goall: Connect with your audience. Goal 2: Direct artd hold attention. Goal 3: Promote understanding and memory. These three goals seem straightforward at first glance, but there arc countless examples of violations in many presentation materials. A closer look at these goals is provided in Section 2.3.

1.5 Chapter Organization This chapter presents three different ouders for communicating data co the public: graph and chan design; written reports; and presentations. Similar design principles apply to all three and there is clear overlap between these types of content delivery. Nonetheless, they are presented separately and sequentially to identify specific points of interest and special considerations with each presentation outlet.

2.0 DESIGN OF GRAPHICS 2.1 Content-Driven, Not Software-Driven Section 1.2 mentioned rwo authors who have received wide acclaim for their understanding of effective communication of refined data. The mistake often made by practitioners is using default formatting built into word processmg, spreadsheets and presentation packages while disregarding some core design principles. These principles are based on the overall principle that informacion be 'displayed in a content-driven manner and not in a software-driven manner. Just because you can easily create a bar chart in Excd to show traffic growth over time docs not mean the data should be presented as a bar chart. The more appropriate and effective chart is a line graph. Why? Because the temporal nature of the data supporu that format. And within that format, what principles help in understanding the line in the graph, and any symbols or colors used? Which labeling provides the best understanding? Further, because Excel can automatically create a legend docs not mean that legend aids your audience's understanding of the chart or graph. Often placing labeLs directly above a bar or on a line, or appropriately grouping data with titles, will be more effective iii enhancing an audience's comprehension . .For research reports, a data table may actually be more effective to convey complex content than a figure. A graph created automatically by a software •wizard~ is rypically limited to cwo or three dimensions, while a table can readily show multidimensional relationships. Another equally important point is not placing coo much informacion on one slide. An overabundance of words, pictures, clip art, etc., can be distracting and lead to viewer confusion. The designer should consider breaking up the information into two or more slides, and then concluding with a summary slide showing bow all the pieces connect. '

Communicating Data to the Public • 25

Think about the appropriate ways to share data with your audience. What measures of effectiveness (MOE) best relate to them? Engineers may expecr one rype, like average control delay, while rhe public may bener understand number of stops or reduction in average tr:~.vel speed or incrc:ased uavel time berween points A and B. Sec Appendix D for additional details 6n this topic.

2.2 Design Principles Tufte (from National Cooperative Highway Research Program [NCHRP] Project 08-36 and as adapted by Schroeer [TRB, 2006]) provides eight core design principles for inform:uion display: 1. Enforu wiu visual comparisons. Force answers to rh~ question "compared to what?" 2. Show thes~ comparisons

suu by silk, rather than sequentially, and especially not on separate pages.

3. Use smalJ multiple to facilitate comparistms. Show the same basic chart scve.ral times, with different data in each. 4. Show causality by linking varUtbks. For aample, do not just show vehicle miles cnvded (VMT) inc;reases over cime; show what might be linked m them. 5. The world ~ seek to undnstami is mullivariate, and so should our displays. Again, VMf does not increase by itsdf. what changes with it!because ofit? Show those as well in the same chart6. Int~grate word ami image. Almost all reporrs or presentations use both text and graphics. Make sur~ they work together as closely as possible. 7. Content is luy. If your numbers are boring. you have the wrong numbers. 8. Don't throw out data, or "dequancify" by (for example) changing numbers to "yes" or "no.» These principles get at the heart of any presented data-the designer should spend the time developing appropriate and compelling visual charts, graphs and figures to ensure audience comprehension.

2.3 Selecting a Graphical Display Method The above design principles by Tufte apply to any graphical display created from data.. Kosslyn (2007) has taken a slighdy different approach and created eight principles under his three goals from Section 1.4. The goals are restated, along with the eight general design principles as applied to visual graphs, cbarrs and figures. However, these principles apply equally to written reports.

Goall. Connect with Your Audience. Principle 1: The Principle of Relevance. Communication is most dfictiv~ when n~th" too much nor too link infrmnatitm is pmented . Principle 2: The Principle of Appropriate Knowledge. Communication requires prior knowledge of pminmt conctpts, j argon and symbols. Goal 2. Direct and Hold Attention. Principle 3: The Principle of Salience. Attmtion is drawn to large p"ctptibk differrnm. Principle 4: The Principle of Discriminability. Two propmW must diffir by a larg~ enough proportion or they wi/J not be distinguishable. Principle 5: The Principle of Puceptual Organization. Ptopk auzqmatica/Jy group &mmt:r into unil'S, ':'hich they thm amnd to ami rm:embtr. 26 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Goal 3. Promote Understa.ndiog and Memory. Principle 6: The Principle of Compatibility. A message is easiest to undmtand if itr form is compatible

with its meaning. Principle 7: The Principle of Informative Changes. Peopk e:xpm changes in propmies to carry informaJi4n. Principle 8: The Principle of Capacity Limitations. Peoplt have a limited capacity to ret:ain and to pro-

em information, and so will not understand a message iftoo much information m ust he retained or proctsud.

men

How do we take these principles and apply them to create compcUing data displays? First, the designer n~ds to know what kind of communication is relevant. Ifyour need is co communicare quantitative information, use graphs. Charrs, diagrams, maps, ehocographs and clip arc are used in communicating qualitative informacion. Exhibit 3-1 shows the overall decision levels associated with choosing an appropriate display (adapted from Kosslyn, 2007) along wich the intended P.urp?se(~).

Desired Communication

Qu.amiative

Grapbs

Bar Graphs (Vertical, Horizon tal, Saclced)

Maps

Qualitative

Photographs and c1ipan

Communicating Data to the Public •

e-1

2.4 Illustrative Examples Examples for these choices and design principles are illustrated in the following exhibits, with commentary following each exhibit.

Sl-10 A.M .. 1/12190

l

j Approach

~ Rlghttums

-Through

~ Leftturns

Grouped bar graphs can be vertical, as shown, or horizontal. Vertical is the most common. If titles for bars an: used and they are too long to fit above each bar, use a horizontal bar graph. In this gnph of traffic volumes at an intersection, the bars are shown with a third dimension into the paper. The reader naturally wants to discern what the third dimension represents, and in this case will waste time uying to figure out chat there is no additional informacion. Unless the third dimension represents a data dement, then present bars in two dimensions only. The.designer chooses the shading, color, or pattern within each bar. For color presentations or reproductions, colored bars are effective. However, Kosslyn (2007) emphasius char competing colors, such as red and blue, should not be plated next to each other. Because of the way our eyes process the wavdengths of these colors, a blurred transition is created where the colors join edges. The colors red and cream, or blue and cream, better complement each other and make it easier for the reader to process the visual information. The opposite diagonal cross-hatching for right rums and left rums in Exhibit 3.2 is acceptable, but using a different pattern for one of them might be more pleasing. Also, as each grouping represents one approach to the intersection, it makes more sense to orient the turning counts as left turns, through, and right rums, just as a driver would approach the intersection. Note: Uu ofcolor for shading or iJmtifying sptcific dat41114J not trans/.au ~tt ifth~ 1114teri4/ wilt be "P~~d in blaclt and white. Test a printed wmpk ofJI'Ur work for clarity bifrm using it in a public docummt. In general, data groupings using up to four data series are easiest for the reader to track and follow (Kosslyn, 2007). If the groupings total more than four series, consider breaking the data into rwo graphs. The reader should have no problem understanding the data in this ahibit. The horizontal lines at 50-interval points on the volume axis in Exhibit 3-2 hdp the reader see the relative values of the tops of each bar. Ifspecific values are most important, then these should be labded directly above the bar. Do not make the horizontal lines too chlck; they will become a distraction to comprehending the message sh.own by the bars. Also, as discussed above, the third dimension is difficult to process vi.!ually. Legends should be used when it is too complicated to place the information directly on the bars. Make sure the legend is described in the same order as the bar3 Ocft to right in~~ or clockwise for a pie graph).

· ., . . , ---- • .,,.""'n T •"~'"•"'••

'"'"rntrrn•••r rTllf""'U"'f"

.,~,n ~niTinM

9-10 A.M .• 1/12/90 500 450 400 -

J• ;

-~

· 350

~

250

~

300 200 1150

100 150 0

NB-ftrst Approach

~ Right turns

c=J

Through

~ Left turns

Exhibit 3-3 shows the same traffic count informacion presented as a stacked bar graph. In this case, the designer was more inrercmd in showing the differences in the roral approach volume than individual movement councs. Hence. the sta.eked bar graph clearly shows the relative differences in roral volume. There is some additional information robe gathered from the individual movement counts compared across each approach, bur the reader must really search out !his information, moSt easily seen for right rums bea.wc each bar component starts at the same lc:vd on the horizontal axis.

Motorcycl-. tnactors end others 2% Buses8%

Cera 3% •

\ ~

Utility vehicles 0!5% ~

Pic grapbs such as Exhibit 3-4 are hdpful for showing relative proportions of items. Colored sections of the pie, versw patterns, are useful in presentations, because patterns can become distracting: The reader must look at too many diagonal lines. In this graph, each pie slice is labded along with its proportion to the whole. This is more hdpful to the reader than trying to match slices up with a legend. The 3D view of the pie gives an added dimension to the standard 20 view that is not associated with any content. One 6.nai point is that a portion of the pie has been exploded away from the rest. It is hard to tell if this is trying to highlight the largest percentage of vehicle mix (uti.liry vehicles at 65 percent) or that the remaining 35 percent is comprised of four other vehicle types. Regardless, either an oral or wtit· ten aplanation of the graph should point out why a portion was exploded, as the reader is led ro believe the design aspect is associated with some content. • ·· · Communicating Data to tli~ Public • 29

35

30 25 Ill

~ ~

20

.~

s...

u..

16

10 10,000 vehicles

5 0

1910

1920

1930

1940

1950

1980

1970

1980

1990

The line graph in Exhibit 3-5 combines multiple dimensions of data in one visual display. A long time line for the X-axis indicates the designer is interested in looking at fatality rates (Y-axis) since the beginning of vehicle production. Line graphs arc especially useful when data suggest a continuous element. One line in the graph depicts the number of vehicles (I 0,000 vehicles), another depicts vehicle miles of uavd (I million VM1} and the third. line depicts population (100,000 population). Perhaps labeling the lines with •per~ before each label would aid viewer understanding of fatality rates per these three measures. While these lines arc weful, it seems some other relationships arc missing or not c:xplicicly shown in the graph, for c:xamplc, the sharp decline in rate per population during WWII. Also, why is there a steady declining rrend in the lower two lines when population rate is Huetuating from 1960 to 1985? Some of these related causes should be c:xplained.

Observed data July, 1989

60~ 50

:a Q.

40

:::>

30

.§

-o4> 4> Q.

(/)

20 10 0

0

50

Density, k (vpm)

30 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

100

150

XY plots like the one shown in Exhibit 3-6 are very helpful in discovering if a mathematical relationshi p exists between rwo variables, in this case speed and density. A plot without a fitted line becomes a scatter plot. The firred line, often · a straight regression line, helps define the relationship of the data because an equation for the line can be determined. With data points plotted, the lit of the line (R2) can be determined as well, providing a descriptor of how well the line represents the data points. If the line equation and R2 value are irnpocrant for the reader, these should be shown on the graph. One item missing is an explanation of what the triangles and squares represent. A legend is needed here to explain the clliference in these two elements. Also, data points best show up using plus(+), open circle (0), solid reiangle (A) and solid circle ( • ) (Kosslyn, 2007). Certainly other symbols, coloring and line hashing can be used, but care needs to be taken that the viewer can see the differences, especially if being viewed in an auditorium presentation, or if printed in a black and white publication.

b) Southbound Pedestrian Crossings ao-zo

•2o-40 D40-60

•.o ..o

aUI0-120 •tzo.-140 M14C>lAO •:uo.-uo • 11:9""2.00

••o-.~

-

ii--~ _! l

·&:: ,, ·I 1:5opm

1~,

l l.:SOpm

u:s oam 10!SO;arn

J!

-.so.rn &:SO.m

1

A

1

, ·

= -': .

·· a

4

S

; I_!-=-

C

I

D

7

L __ __ __l__t__ _ _ _-'.Jei._ ______j_L _ _ ___f_

.---

~f . '-' <>;

-51

lili

C51

e

""'' 4>i

='

g'1

C..i

.,! .., F=

ol :X:i

-. -.:

l

!i

~

II c

.,!:.;

!
b li

v

CQ

..:s

I

7:50 am

a

9

I I

=~ • ·c; ,::... 0

:;:;

"'-l

I

10 I I

Cl

; .:5!

-_,;

..g! f:ii

"""'

Exhibit 3-7 gives an example of a 3D graph of 6-hour pedestrian flows along.an arteri~ street (Schroeder ec al., 200 9). In this case, the legend ac the top of the graph, just under the title, uses color to depict a range of pedestrian volu£lles that correspond to plotted "contours" that add a third dimension to the graph. Often, these types of graphs are b e:st viewed in color as lower values can be darker colors and higher values can be lighter colors which catch the eye quicker. While there is a lot of information here, the value of this graph is :in having tht viewer quickly see that Pogue St. to Logan Ct. is the section with high pedestrian flows, and that Hoine St. and Chamberlain St. have the highest peak levels during lunch time (11:30 a.m.-1:30 p.m.). In addition co the dimensions oflocation, time period and pedestrian volume, the graph gives additional dimensions of road names, and the distinction between crosswalks (numbers) and rnidblock crossing locations {letters). The exhibit would be improved if the mid-block locations were S(a)ed to match the actual distance between intersections. However, this is one example where the Excel software package is limited in its abilities.

Communicating Data to the Public •

;::31

Density Contours (veh/ml/1 n}

.. 90.o-100.0

100.0

c8o.o-9o.o

90.0

70.0•80.0

80.0

::ii ::::.

i

• c

!!.

GZ60.0· 70.0

70.0

... 50.0-60.0

60.0

a4o.o-so.o

so.o

•30.0-40.0

40.0

• 20.0· 30.0

30.0

1010.0 -20.0

20.0

•0.0-10.0

10.0 0 .0 nmelnt erval

7

8

9

~

10 11 12 13 14 1S 16 17

18

1

Sepnent Number

The graph in Exhibit 3-8 depict.S &ceway congestion (de~ity, Y-a.xis) for 12, 15-min. time imerva.Ls (right side, Z.. axis} along 18 freeway segments (X-axis), some segmenrs being basic segments, some on ramps, some off ramps and some weaving sections. The view clearly shows the third dimension (time) going into the paper. Similar to the other 3D c:xhibit above, shaded ribbo~ are wed to emphasize different density interva.ls, where the vertical height at any point tepresenrs the density in that segment at that point in time. These kinds ofp2phs are atrerndy weful for viewing congestion along a highway section to determine bottleneck locations and the effects of queue buildup and dissipation (queue discltarge) in both time and space. In this case, a density of 45 vehicles per mile per lane (veh/mi/ In) corresponds to capacity and the Levds of Service •p• threshold as defined by the Highway Capacity ManwJ (TRB, 2000}. lt is directly evident from the graph that congestion and queuing are present during the analysis period, but the actual onset and extent of congestion would likely be better estimated in a •Bat" contour plot like the one shown in Exhibit 3-7.

2.5 Engineering Drawings and Plans Engineering drawings and plans f.ill into the same category as maps. MOSt of them are presented in 2D for viewing by the public and as part of a design project package of plans and specifications. H~r. there are certainly computeraided design packages that allow designers to create plans wing 3D topography and create maps in 3D. The question for the designer becomes what information is important to show co the public, especially at public hearings? Depending on the purpose of any meeting with the public and/or decision-makers, show only those pieces that are relevant to yow audience. Basically, show what is new compared co what currcndy exists, in as much decail as your audience needs co understand your mCSRge. This will generally include: • a site map; • road layout; . • rightofway; • adjacent property owners; • curb and gutter or shoulder; I

1. 1


• sidewalk: • parking; and pc:desuian access or possible: adjacent path. Construction plans would also include the: • cc:mc:rlinc: alignment; • drainage features; • storm water detention; • sediment conuol; construction phasing and work zone: safety; traffic co~uol (signs, signals and pavement markings); • vertical profile; • earthwork; • a North arrow; and a scale.

The: small section of an engino:ring plan in Exhibit 3-9 shows many of the: forures d=ibed above. Note: that by show· ing information rdc:vant to the road realignment and new stormwacer collecrion system (for cxarnpk. station informacion, catch basin nwnhc:rs, supcrdc:va.tion slope, c:tc.) along with uecs, the: drawing appears too cluttered for a public prescn12rion. Again, dtcidc what informacion your audience: needs, and the: outcome you want at me end of me meeting or present:~rion. This section of an acrual plan sheer is part of the consuuaion drawings and, as such, all the information is rdc:vant for: the intended putpase. Color ~ of existing~ new features would be: appropriate for a drawing presented at a public hearing. such as described in the nc:xt ahibiL Communicatinq Data to the Public •

33

...

0;

f&~~ ~-- -~--~

Source: Im~ coune:sy ofKimley-Horn and Associates, Inc.

Exhibit 3-10 is a computer rendering of a proposed streecscape redevelopment project along Hillsborough Sueet in Raleigh, NC, USA. Note that decailed information about cenreriine alignment curvarure, drainage consuuction elements, etc., would clutter the drawing. This additional dccail is not needed given the intent md argec audience of the graphic. This plan view drawing was created for a public presentation of the project; the entire layout required a 160-in. x 36-in. printing.

3.0 WRmEN REPORTS The results of mmy well-planned md wdl-cxecuted engineering srudies have been misunderstood, misconstrued, or disregarded ~cause of poorly written reporrs. Considerable resources are wasted ifthe srudy findings, conclusions md recomme,ndations are not clearly conveyed to chose charged with acting on the results. With careful attention to a few writing principles, reports of engineering srudies can be powerful tools co inform and aid the decision-making process. Regardless of the study's size or importance, the same fundamental principles apply (Ro~rrson ec a!., 1994). 1. Write co the level of the intended audience. 2. Use clear, concise language, not technical jargon. 3. Present findings and conclusions in a logical sequence.

4. Clearly show how the findings support the conclusions md recommen
5. Use figures (graphs) and cables ro portray the most important results. 6. Gee to the point. . 34 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

7. These principles are amplified in the sections r.hat follow. Remember, good engineering must be accompanied by good writing. Bad writing can trump good research . . See Appendix D for additional derails on this topic.

3.1 Sections of a Repor:t What's included in a transportation engineering report is determined by the report's purpose, length and complex.Jry. In general, shorter reportS have fewer components. The following is a list of componentS, any or all of which may be included in a given reporr: • Letter of transmittal • Tide page • Copyright notice • Dudaimcrs • Foreword or preface • Acknowledgments • Table of contents • List of tables (or exhibits;. if.using exhibits, then list of figures can be removed)

• List of ligures • Summary or executive summary • Body of the report • List of references or bibliography

• Appendix • Glossary • Index This sequence follows normal conventions bur may be altered to tic a given siruacion or local preferences.

3.2 Writing Style and Target Audience Adapting the language, style and presentation of a report to the intended audience is nor easy. The writer mUSt fi .rsr know the audience. This is true of both an audience you know and one about which you have little information ~e garding its knowledge and interest. Many transportation engineering reports are developed for engineers, planners, or public officials with some kno~l· edge of or experience with transportation issues. In these reports, less effort is required to define commonly used transportation engineering terms. If the report is prepared for the general public or a special-interest group, however, you may need to define terms that may be unfamiliar. You can do this either in the text or in a glossary. In all as.:=s, define acronyms the first time they appear in the reporr. The usual practice is to spell our the full name or t'CI!ll a.C'ld follow that with the acronym in parentheses (for example: Institute ofTransportacion Engineers (ITE)). You can th..en usc the acronym only throughout. Keep in mind that documents available for public comment should be written wi m additional effort made to define terminology and explain equations or data in terms the public can understand. Tb!s can save consi~crable time and effort by avoiding after-the-fact explanations. · · Communicating Data to the Fublk • ::;:35

A report should be artractive to the reader. lts appearance should be neat, the tone authoritative and professional (but not dry and overly formal) and the content accurate. It should be free of spelling and grammatical errors, which tend to distract the reader from the content. Break lengthy seccions by using headings and subheadings liberally and by inserting tables, figures, photographs, or lists to portray results or present data. Make the report a convenient size (for example, 8.5 by 11 in.); a standard page size is easier and cheaper to reproduce. Where possible; design tables and figures so they can be read without rurning the report sideways. However, rotated pages or even Z-folded inserts (11 by 17 in. pages that are folded to fit into an 8.5 by 11 in. report) may be necessary. Your audience needs to be able to read figures and tables in the report; thus, they need to be adequatdy sized. Font size and scyle are a personal preference. However, font size should be 11 or 12 point for the text portions. Using Times New Roman or something similar is suitable for a wide range of readers. Further, one and one-~alfline spacing is easier to read than single-line spacing. In general, double-line spacing does not add clariry.

3.3 Body of the Report The body is the heart of the report and is supported by all of the other components. It contains a series ofsections or chapters. In a transportation engineering report, the body offers answers to the following questions: 1. What was the objective or purpose of the study? 2. Why was the study necessary?

3. When, where and how was the study conducted?

4. What were the findings? 5. What conclusions were drawn? 6. What recommendations were made? 3.3.1. FDrmat ofthe &dy The body of a rypical uansportation study report generally includes:

• purpose or objectives of the study; • background (that is, what led to the study or why was the study needed); • scope of the study (that is, what limits were placed on the study);

• methods used; • data collected (for example, rype, arnowu, when, where, etc.); • analyses performed; • findings; • conclusions; and • recommendations. Other subjectS that may be appropriate for a tra..Osportation engineering study report include: • alternatives devdoped or camined; • sdecrion of alternatives, TCDs, or routes; • evaluation results; • cost analysis or financial impact; \ I

\.

36 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

• environmental and social impactS; • traffic impacr; and • implementation plans or recommendations.

3.4 Use of Exhibits Exhibits are figures and tables char hdp the reader understand the information presented iJl the report. They typically show results from analyses of data collected ror the study. Figures and tables shol!ld be id~tified early in the report, and should be clear and appropriately labeled. They can be more complex chan in a presentation because the reader has time to reSect on what he or she is being shown. Explanatory text should always accompany a figure or table. Using the word •exhibit• to label each figure and table will make it easier to number them sequencially. This provides a quick way for the reader to locate a specific exhibit, versus hunting through the report for Figure 12 which nlight come afrer Table 6~ ·· · ·


_

Detailed material that supports but is not essential to the body of the report should be placed in an appendix. Appropriate appendix materials include • supporting data; • detailed explanations of methodologies or procedures; derivations of formulas; • conversion factors;

• lists of symbols; • data collection formats; • data collection protocols; and • checklists. Appendices are an effective means of fully documenting and suppqrting the results of a study without cluttering the body of the report. The main report should remain an independent and free-standing document that provides sufficient decill co understand the study objecrive, findings and recommendations. Appendices are strictly supplementary and can be longer chan the succincr main report. In some cases, appendices may even be electronic or online documents that supplement the printed report. They are therefore readily availa.ble to the reader if needed, but resou.rces and paper are not spent on a document few readers will use.

4.0 PRESENTATION TECHNIQUES Effective oral presentations are not condensed versions of written reports, nor are they speeches. Each presentation should be prepared ror the specific conditions under which it is to be given and adapted to the presenter, the audience and the environment. The following five characreristics distinguish effective oral presentations from wrinen reports: I. Specific audience

2. Limited scope

3. Personal presentation Communitaling Data to the Public • 37

4. Need for instant understanding 5. Limited cime for presentation See Appendix 0 for addicional deta.il.s on this topic.

4.1 Podium Presentations If some portion of a written report is unclear, the reader may reread it, refer back or ahead, or even consult other sources. However, the spoken word only lasts the moment of presentation. The speaker must be careful to be as clear as possible at all times and pay particular attention to audience reaction. The presenter's voice, language and the use of transitions and summary Statements helps to engage the audience. Proper use of voice includes adequate projection and distinct enunciation to permit even those seared farthest away to hear and understand. The speaker should use pauses freely to break the Sow of ideas into meaningful thought units. Also, speaking forcefully and using a variety of infleccions avoids monotony and adds life and meaning. A speaker must be sensitive to the rate of delivery and must pace the remarks to enable understanding, variety and emphasis. Practice is essential to master these techniques. 4.1.1 Pitfalls ofPresentations The audience must understand the speaker's vocabulary. If the readers of a written repon encounter an unfamiliar word, they can refer to a dictionary. In an oral repon, the speaker must define unfamiliar terms for the listeners. Obviously, the problem varies with the situation, but the speaker should be sure the lis\eners understand the terminology. While a broad vocabulary permits clear, precise and colorful expression of though rs, the presenter should always keep audience comprehension in mind. Transitions and summaries help guide the listener through the devc:lopment of the presentation topic. After a section is completed, summarize the main points brieRy before continuing. The presenter should also illustrate the relationships among various sections by offering well thought-out transitions.

Organize and rehearse the presentation for the established cime limit. The average person speaks at a rate of I 00 to 150 words per minute. (A reader can comprehend up to 600 words per minute, but a listener cannot). Talk at a pace that holds the listener's attention. You can check your riming by practicing with audio, video, or an audience of coUeagues. Video is the most effective. It gives the presenter the opportunity to see how he/she is coming across and how to improve his/her presentation skills. Simply speaking aloud helps to check your organization; however, when practicing in an "empty room" people tend to speak much fasrer than when they're recording or speaking to a live audience. Rehc-.usal will hdp you achieve a relaxed, at-ease posture, and a smooth, confident delivery. Include only those pieces of information relevant to the presentation's purpose and what is appropriate for the intended audience. Data interesting co the engineer, but unnecessary for understanding and potentially confusing to the audience should be simplified or left out of the presentation. Conversely, if the infonnation is necessary for the presentation, make sure ic is clearly understandable and can be easily commurticated during the presentation. 4.1.2 Vuwd Aids tnul Hmulouts Some presentations are more effective if the audience has a copy of the presentation or if there are maps or other props used during the presentation. Examples include brochures or Syers on a design concept, plans for a project, the presentation slides in handout format and samples of products, devices, or equipment needed co perform a job. VISual aids and handouts should be developed and included if audience comprehension is co be improved, or if ic is imponanc to have audience members leave with something presentation-related in hand.

Handouts are c:speciaUy useful if complicated or detailed graphs and charts are used. The frequent use of projectors in presentations lends itself co "squeezing" a lot of content on the displayed slide. However, what may be clear on a computer screen can be harder to process when projected. A handout on 8.5 by 11 in. or even 11 by 17 in. paper can overcome the limitations of projected media. As Tufte (2007) points our, a well-designed handout or "super graphic" can also keep the audience engaged as they arc actively involved in exploring the presentation content. Prese.nters should not be afraid the audience might "read ahead" ifthey're given handours. On the contrary, a presenter should feel a sense of reward and accomplishment if the audience is interacting with his or her material. A handout stimularc:s curiosicy and ke~ps the audience engaged in the presentacion. 38 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

4.1.3 3D and 4D VtsUalization As stated before, public involvement and interest in transportation projects and expectations concernin g the delivery of content are increasing. The use of 30 modeling images and 40 visualization is growing as computer p rograms and \ equipment make them easier to create. These techniques can encompass "predict" the projection of data or ouccome of a project (that is, create a simulation or visualization of a future condition). One of the more recent applications of 3D modeling (XYZ dimensions in a static environment) is wi th srreecscape redesign projects. Before-and-after pictures can be shown to an audience, demonstrating what the street will look like once the project is completed. This technique provides an opportunity for immediate audience feedback on a design concept. The two exhibits below show the use of this technique. [This is the same srreetscape project along Hillsborough St. in Raleigh, NC as presented in Exhibit 3-1 0.]

Souree: Image courtesy of Kimley- Horn and Associates, Inc.

Source: Image courtesy of Kimley- Horn and Associates, Inc.

Exhibit 3-11 shows the existing view of the srreet lanes, cemer rum lane, sidewalk, ov<;.rhead urilicy lines, utility poles, buildings and traffic signals. Exhibit 3-12 is a computer-generated view of the new srreetscape with one lane, curbside parking, underground udlicy lines, a ne_w modern roundabout at the downstream intersection, no left turns at this lim intersection, a plan red median and new light poles and decorative banners. i By using this kind of visualization, the audience can see what is being planned and is in a better position to react ~d comment. By adding a time dimension to the visualization, proposed projects can be presented in "recorded fly-through" vie-ws, &om both an observer's and a driver's viewpoint. The following four images are screen shots from a ttaffic simulation model of the proposed project and give a static sense of what this could look like.

Communicating Data to the Publi< • :P9

The simulation and visualization in Exhibit 3-13 were created as pan of the Hillsborough St. streetscape project introduced previously. Additional details on creating these kinds of By-through views are provided in Chapter 11 on simulation studies. 4.1.4 Inter4ctWe Wib Fetzturu A new Web tool is emerging called data visualization (or visual analytics). This kind of Web applic;arion performs queries on databases and then displays the ~ula, often in a graphical map. Because of the graphical nature of the outcome displays, daCl visualization applications are often developed within a GIS program and architecture.

VJSUal analytics bas a strong •data mining~ component and the expectation of the discovery of n.:w knowledge or relationships within the data. For aample, suppose a traruportuion agency could query a county to locate high crash rates associateci with specific criteria, such as speeding vehicles or alignment curvature. The agency could then target resources for remedial measures or enforcement where the payback-reciuced crashes-would be most noticeable. The agency could print out a map showing these locations for usc in meetings to discuss possible actions.

4.2 Poster Presentations and Displays Displays and postery used for presentations, discussion s~ions, or vendor exhibits can incorporate many, ifnot all, of the dements described in chis chapter. Both 3D and 4D images and visualizarions can be shown.1icles, images, written material, graphs and figures must tell a story. be legible and not be confusing. The principles for creating graphs apply to these med4 as well. Vasualizarion techniques shown on a portable computer can demonstrate the presentation point, in a live setting. In same instances, the viewer could have an opportunity to interact with a program and perform a •cest drive" of the sofrw2re. Hands-on interaction with, for aample, a simulation program is often more important for a viewer than seeing a demonstration. 40 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

When creating a poster for a research conference or a public meeting, the presentation material should be printed in large font and simplifi~ to the essentials. The viewers will be uncomfortable and become unen~ed if they must · scand roo close to the display or if they have to read long paragraphs. Graphics and tables should be large, and simple, · and self-explanatory. Alternatively, short captions can be used to explain the display or illusuate irs key points. Poster presentations can be supplemented with handoutS, as discussed above, to provide additional information.

4.3 Web Site Design The Web is fast becoming the first place individuals search for information. A big project in a city might necessitate the creation of a separate Web page(s) devot~ to the project, along with a method of obtaining community feedb~ck as the project moves along irs various planning and design scages. By creating and maintaining a Web page on a big project, the public can get accurate information concerning the project without confusion, or misinformation th.ey might hear concerning the project design, location, or expected outcome. · Another way co use the Web is to post presentation materials prior co any public meetin~ or hearin~. This may be l.inked to a .news broadcast or other means ofletting the public know this information is aVallable for advance viewing. The project scope may justify a project hotline (phone number) and specific e-mail address for receiving commen~ as the project devdops, all noted on the project web page. This can be very effective for large projectS (for example, major new shopping center, brown6dd devdopment, high rise office and retail) and for pol.icy projectS (such as new light rail service versus express bus service).

S.OSUMMARY This chapter has explored some cognitive learning principles as they apply to developing graphs. Focus points for both oral presen.tations and written reportS or displays outlined the key components for describing and conveying technical information to the public. Some of the major poinrs discussed include: • know and connect with your target audience; • be judicious in selecting appropriate graphics for inclusion in a report or presentation; • pay attention to default gcaphic styles and question if the desired message is dear. If not, change or modify the gcaphic style; • pay attention to default coloring or shading and question~ this adds or distraets from the message. Know if the information will be produ~ in color or black and white and adjust accordingly; • direct the reader/viewer to the information of most importance in a graphic;

.i I

• promote understanding in all graphics with the intent of having the viewer remember specific information;

I

• 3D and 40 visualization techniques can enhance viewer understanding of choices for a complex project;

I

I

• pi:'Cpare repo~ with the same intensity and thoughtfulness .as a presentation; and

' a podium presentation. • always rehearse

•

In addition, using the Web can both inform the public and solicit irs fe~back on major projectS. The Web can also provide a mobile environment for conducting queries on complex databases, resulting in seeing physical and topographical relationships that could not be cap~ in any other environment. The reader is encourag~ to find additional details on these topics from the following publications.

C6ri'lri'li.ll'licating Data to the Public • 41

6.0 REFERENCES 6. 1 Literature References Editorial stafl' of th~ University of Chicago Peas. Tht Chicago Manual ofStyk, 13th ed. Chicago, IL: Univ~rsity of Chicago Pr~.l982.

Hodg~s.

J. C., M. E. Whircen, W. B. Horn~r. S. S. Webb and R. K Miller. Harbrau Colkge Handbook, 11th ed. San Di~go,

CA; Harcourt Br:ace jovanovkh, 1990.

!Ghl, M. and K. Kat. "Involving the Public in Project Evaluation-A Case Study in Arizona."ITE]oumal (F~bruary 2008): 18-23. Kosslyn, S.M. CLEAR atui to THE POINT, New York: Oxford University Pr~. 2007.

Kroner, M. G., J. W. Presley and D. C. Rigg. Prtntiu-Ha/1 WOrkbook for Wrirm, 4th ed. Englewood Cliffs, NJ: Prentice-Hall, 1985. R.obenson, H. D. and D. C. Nelson. Manual ofTnmportation Enginuring StutUes, Washingron, DC: Inscirure ofTranspomtion Engineers, 1994. Schroed~r, B.)., N. M. R.ouphail and B. A. Lehan. "Observational Srudy ofPedesuian Behavior Along a Signalized Urban Corridor.~ 88th Annual Meeting ofthe Traruportation Research &ard. Washingron, DC, 2009.

Transportation Res=ch Board. NCHRP Project 08-36, Task 51: Primer on Information Daign for Effi~ DOT Decision-Making. Washington, DC: Transportation Res=ch Board.•2006. Tufte, E. R. The Visual Display of~ntitlltive Information, 2nd ed. Cheshire, CT: Gr:apbics Press, 2007.

6.2 Online Resources "Meet the Author Poster Session Guidelines." Participant and Artnulu Guidn, TRB .Annual Meeting Web sire,

www.crb.org/AnnualMeeting2010/AM2010PanicipamGuides.aspL "Creating Effective Poster Prescncarioru.~ North Carolina State University Web sire, www.ncsu.edu/project:lposter:s/NewSire.


Chapter 4 o o •

t

c •

•

• • •

•

o •

o o o •

o •

• o •

• •

o •

o • o •

• •

o o o

t

o • •

•

o o

o o •

o •

o •

o •

o o •

o o o

o •

a •

o •

o

~

1

•

0

•

0

0

•

0

,

0

0

0

,

,

0

•

•

o

Volume Studies Original by:

H. Douglas Robertson, Ph.D., P.E. joseph E. Hummer, Ph.D., P.E. EJ;uJby: &stian]. Schroeder, Ph.D. 1.0

INTRODUCTION

2.0 lYPES OF STUDIES

44

2.1 Intersection Counts

44

2.2 Area Counts

49

3.0 METHODS OF DATA COLLECTION

4.0

43

58

3.1 Manual Observation

sa

3.2 Automatic Counti

62

DATA REDUCTION AND ANALYSIS

67

4.1 Manual Counts

67


68

4.3 Count Periods

69

4.4 Volume Data Presentations

70

5.0 SUMMARY

74

6.0 REFERENCES

74

1.0 INTRODUCTION Engineus often~ counts of the number ofvehicles, bicycles, or pedestrians pusing a poinr, entering an intersection, or using a particular facility such as a travd lane, crosswalk, or sidc:walk. Counts are usually samples ofactual vo]uro.es, although continuous councing is increasingly performed for certain sitW~cions or circumstances. Modern automaced count stations are found along signalized arterials and on freeway facilities, and are Standard features in combination with weigh-in-motion (WIM) stations and automated toll facilities. Sampling periods may range from a few minu. -ces to a month or more. Despite advances in modern data coUeccion technologies, the length of the sampling period :remains a function ofthe type of count being taken and the use co which the volume data will be put. lp this chapter the focus is on the common methods for counting tra8ic in the field; how volume data are sampled, expanded and analyz.ed; and how count prognms are established. Brief descriptions of specific studies are prese~t ed along with references containing more dttail. The chapter distinguishes and contrasts manual versus auromaced counts and real-time versus post-processed data, and discusses tradeoff's berween data accuracy and data coUection e£ficiency using examples. Special emphasis is given to data collection challenges at geometric configurations that reqll.i.re path-based vehicle counts, including roundabouts, and ocher intersections (for example, superst.ttets), whert soO'le or all traffic movements are not observed in isolation, bur at all times share lanes with ocher movements. Appendix:: E presents sUmmary forms suitable for copying.

Volume Studies • ..,:::;3

2.0 TYPES OF STUDIES Volume dau can princip~y be divided into intersection counts and a.rc2 counts. While intersection counts typically requite only a limited number of observers, area counts are generally ltWtc complex to plan and execute, and always require multiple observers. In both categories, traffic volume counting is not always a simple, straightforward wk. Some types of volume studies are complex and difficult to perform. They require special preparations and observer training. especi~y in me case of busy intersections or unusual geometric configurations. This section presents me most common count studies and identifies clullenges rhe data collector must coiJ.fronc in preparing for data collection.

2.1 Intersection Counts The most commonly counted location in a traffic system is me intersection. Intersection turning movement counts are common inputs for planning-level applications such as traffic impact analyses, as wdl as operational analyses of signalized arterial corridors. At a traditional intersection, each approach has up co four possible vehicle turning movements: U-rurn,left, dtrough and right (although in most srudies, U-rurns are included with left rurns). Many applications require a separate count of buses, passenger cars and rrucks. At a four-leg intersection, an observer could be faced with recording 36 separate data eleme.nts during each sampling period to record only the vehicular movements. In addition, intersection volume studies may include pedestrian and bicycle move.ments, which add complexity. Except for very light traffic conditions, intersection counts require multlplc observers. If many vehicle classes arc to be examined at a busy intersection with several simultaneous movements, each observer muse be able to record data for rwo or three lanes. Simplified methods of identifying vehicle classes arc sometimes desirable. For example, one could classil}r ~ motor vehicles with two to four tires as automobiles and ~ motor vehicles with six or more ti~ as trucks. The classification scheme must be well understood by all observers before the beginning of the count. 2.1.1 Uruip4JizeJ lnun«timu An unsignaliud inte.rsccrion is typically controlled by STOP or YIELD signs. In the U.S., the rwo-way and four-way

stop configurations are most common. In other countries, yield~ntrotled intersections are more common, where me minor movement is concrotled by a YIELD sign. Many European countries further feature unsigned intersections, where the right-band-rule governs traffic operations (drivers yield to me vehicle on their righr). Yield~ncroUed roundabout intersections are addressed in a separate section below. Before counting an unsignalized intersection it is important to understand the operations at the site, as well as me anticipated volume levels. Generally, a four-way stop or right-hand-rule intersection has low-enough traffic volumes that a single observer can comfortably count all movements, including heavy vehicles. Ar busier two-way nop or yieldcontrolled intersections, a second observer may be necessary, depending on me volume levels on the major approaches. 2.1.2 Sigrudiutl lnunectimu At a signalized inrc.rscction, opposing movements arc conrroUed by signal phasing and therefore move at scheduled and alternating times. Since all approaches do not have the right of way simultaneously, an observer may alternate counting movements in rwo directions (for example, eastbound and southbound) as the signal phase changes. Low-voluinc signalized intersections may therefore be counted by a single weU-traincd observer. However, since the majority of volume studies arc performed during peak flow periods, it is likely that multiple observers are necessary in most cases. Counts at larger and busier signalized intersections are complicated because one or more movements occur during each phase, because each signal cycle contains rwo or more phases and because the green time for one phase often is not equal to char of other, opposing phases. In order to not bias the count toward any parti~ set of movements, the count interval should be an even multiple of the signal cycle length. This effect is important to consider at short rime intervals (and long cycle lengths), but tends to average out if longer count intervals are uscd.lt is good practice to select count intervals that capture at least five cycles. Actuated signals fUrther complicate counci,ng, because both the cycle lengths and the green times vary from cycle co cycle. One rule of thumb is to select counting incc.rva!s that will include at least five cycles, using the maximum cycle length to determine the interval. Assuming the sigll21 concroller is responding to demand, counts should be representa~C of the demand, despite the variations in timing (Roess, Prassas and McShane 2004).

me

'·.'. I


Special challenges at signalized intersections are permissive turning movements that do not move consistently during their (permissive} green phase. Permissive right turns and right-turn-on-red movements are easy to miss among busy through movements and special attention is required.

2.1.2.1 Arrival Vmus Departure ~luma Intersection volume counts are usually recorded as vehicles cr~ the stop bar and enter the intersection. This point is chosen so that turning movements can be observed accurately. If the intersection becomes saturated (demand exceeds capacity), queues will develop that may require more than a single cycle to dissipate. When this occurs the deparrure counts do not reflect the demand volumes. In these cases, arrival volumes should be recorded. In this context, departure volumes are defined as traffic ctossing the stop bar at the intersection. Arrival volumes (or demand volumes) arc defined as any vehicle that approaches the intersection on the studied approach during the analysis time interval. In an oversarurated flow period, the departure counts will be less than arrival counts, since some vehicles are held in a queue at the signal. Consequently, arrival volumes can be approximated by relating the departure count to che number of vehicles in the queue. Arrival volumes are not easy to observe, since the queu.es arc constantly changing. and may extend beyond tlte line of vision of the obseiver. Additional observers arc normally needed to count queue lengths. while the primary observers tount departure volumes. For greateSt accuracy. the queue can be counted every cycle or may be counted at the end of each aggregation interval (for example, 15 min.). If only a single observer is available, a video camera can be used to capture depanurc volumes, while the data collector manually records vehicle queues. [n this case, it is important to find a synching point to be able to combine daca.sowces in the office. Exhibit 4-1 illustrates how to estimate arrival volumes by observing departures and queue lengths. One can calculate the arrival count for each interval by adding the net change in queue length to the observed departure count. Note from the example that while the total departure and arrival volumes arc the same, the distribution of volumes across counting intervals (or cycles) is different. This procedure es~ates arrival volume for the approach. Turning move~ents for these arrival volumes may be 'obtained by assuming that arrivals follow the same distribution of left, right and.through movements as the.departure volumes.

Tunc Period

Aaival VoltUDe (Vehicles)

Source: Roess, Pruas and McShane, 2004, p. 179.

Volume Studies • 45

2.1.3 Petknritm and Bicyck Counts Pedestrian counts are usually taken at intersection crosswalks, ar midblock crossings, or along sidewalks or walking paths. Pedestrian Aows are also common at modern roundabouts in urban or suburban settings. Similarly, bicycle counts can be imporcant for planning applications and operational an2lyses, especially in areas where bicycle usage for .recreational and commuting purposes arc common. Pedew-i~n md bicycle volume data are used for rraffic signal and crossw2lk warrant studies, for capacity analysis, in collision studies, for site impact analysis and in other planning applications.

For both of these nonmotorizcd modes of transportation it is important to understand behavioral patterns, since they can affect the volume counts. Pedestrians frequently cross outside of marked crossw2lk areas and away from intersections. The observed pedestrian count at a crosswalk is therefore oftentimes less than the acrual pedestrian demand volumes. Bicyclists may travel on the roadway with motorized traffic, or may decide to dismount and we the sidewalks for certain maneuvers. Chapter 12 describes procedures for performing pedestrian and bicycle volume counts. 2.1.4 PAth-Based Counts Some modern intersection configurations require the analyst to perform path-based counts. Counts at traditional intersections (signalized or unsignalized) are typically done at a point at which the vehicle path is uniquely defined (for aa.rnple, the stop bar for an exclusive lane at an intersection). However, several modern intersection configurations combine multiple movements into one or more shared lanes, and the count i.s a function of both origin and destination of the vehicle, that is, the vehicle path. Eumples of these intersections include modern roundabouts, supcrstrcets and Michigan U-Turn intersections. At these intersections, the volume from any one approach (origin) mixes with other traffic (from other origins) before exiting at a common destination. As a result, individual turning movements (for example, eastbound left tum) are never observed in isolation but are at all times mhed with one or more other \ movements.

2.1.4. 1 Motkrn Roundabouts At a single-lane roundabout, all entering t\ow (right-tum, through, left-turn and U-tum traffic) enters the roundabout through the same lane, mixes with circulating traffic and exits to different destinations. Using traditional count techniques, these data arc difficult to collect because an analyst has to visually keep track of the vehicle paths. furthermore, many automated data colleccion methods such as tube counters arc appropriate to measure overall approach demand, but cannot provide turning movement counr.s for shared lanes. The volume data challenge increases dramatically at multilane sites. Recent literature has identified rwo main approaches for counting volumes at roundabouts: 1) manual counts and 2) video-image processing counts with vehicle tracking. Manual countS at roundabouts are possible if t:.raffic volumes are low or if multiple observers are used. The cognitive task is more complicated from a regular intersection count, since the observer needs to remember the vehicle's origin when he or she tallies the count as the vehicle exirs. Even if an analyst were co focus on only a single approach, this method may result in multiple vehicles within the circle ac any given time. The potential for ector is grcaL To successfully perform a turning movement count, the observer(s) should be positioned at good vantage points. To enhance the accuracy of a manu21 roundabout count, video observations can be wed. Given an appropriate vantage point that caprures the entire roundabout, an analyst can count different approaches to the roundabout by watching the video repeatedly. A national study (R.odegc.rts et al., 2007) on roundabout operations has wed this approach, aided by the we of multiple cameras mounted on a telescoping pole in the center island of the circle. Others have estimated roundabout turning movementS wing multiple time-synchronized data collection units (list et 21., 2006). To further facilitare the manu21 count of roundabout volumes, manufacturers of data collection equipment are beginning to include roundabout analysis features UAMAR 2008). But even with erthanced data collection equipment and video observations, a manual roundabout count remains a challenge. Given the challenge of counting roundabout flows, recent technologies and research have increasingly focused on automated roundabout data collection using video-image processing techniques. This approach requires a good vantage point for video observations and further that the image is calibrated. Video-based roundabout counts are already being offered by some private companies and are expected to grow as roundabout insrallations become more frequent.

46 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

2. 1.4.2 MedUzn U-Tum and Supmtreet lnteructions ·,. Other intersection configurations that require path-based counts are the median U-turn and superstreet in tersection. i Both intersections restrict movements chat would rypically need to be controlled at the intersection by diverting driv·.ers to aU-turn opening on the mainline. This eliminates unnecessary signal phases at the main intersection, allowing signals to operate more efficiently because less time i.s dedicated to protecting other (now eliminated) movements. Exhibit 4-2 shows rypical configurations of these intersections along with movement displays. The triangle shapes in dicate proposed locations of video cameras or field observers. While these intersection configurations are still relatively rare in the U.S. at the time of this publication, research suggesting their operational benefits (Hummer et al., 2007) will likely foster additional installations in this and other countries. The median U-mrn allows through movements along the major and minor streets to operate within the normal intersection zone; however, ALL left turn movementS are required to utilize one of the U-turn openings to complete che movement. Many of these inte.rsections arc located in the state of Michigan, which has led to many people calling rh is intersection configuration a "Michigan Left."Thesuperstreet divertS minor left AND through movements by maki ng drive.rs access one of the U-turn openings. Typically, the main street movements are allowed to operate in a normal fashion, because the additional left turn signal phase does not hinder the opposing through movemenrs. This is because the two main street through movements act as one-way pairs on each side of the l"{ledian, allowin g ne:u perfect progression that is nor possible with two-way progression at a rypical intersection. However, at times it is not possible to allow left turn movements due to issues that may arise such as right-of-way restrictions. Counrs at these intersections are even more chall.enging than at a roundabout, because of the larger spatial extent of the intersection. Another key difference to the roundabout, is that not all counts are path-based. Luckily, the major movements at a superstreet (typically) operate like they would at a regular intersection. They can thus be counted using any of the manual or automated study methods discussed in this chapter. The challenge lies with the mipor street movements. Depending.on the volumes of these movements, one or multiple observers in the field may be able to perform a manual couot to capture t,hese Hows. At greater volumes, video observations may be used to allow for repeated viewing and detailed count of the different movementS. Researchers bave found that video cameras positioned at the locations shown in Exhibit 4-2 can be sufficient, although additional cameras may be necessary if visual occlusion through landscaping and signage is a concern. An overhead van rage point i.s generally helpful and preferable if available.

2.1:4.3 Sampling Method for Path-Basfd uuntJ . In the absence of video-based technologies and in an effort to make path-based counts more efficient, a sampli~g approach can be taken that develops an origin-destination (0-D) marrix of turning percentages. An 0-D rnatriJC is essentially a table that relates multiple origins of traffic in rows to multiple destinations in columns. 0-D matrices are commonly used in planning appl.ications discussed in more detail in Chapter 20. To develop an 0 -D matrix of turning percentages, the observer(s) samples the puning movement percentages for' '3. short period and then applies these percentages to a general appro~ch volume couot. Depending on the BuauacioOS in traffic, a 15-min. sample per approach may be sufficient, supplemented by a 2-hour approach count. To impro-ve the accuracy of the 0-D matrix, an average of multiple sampling periods can be used. If traffic volumes are high, a sampling rate can be used to record the path of every fifth vehicle (or greater). If rate sampling is used, the sampling period should be extended longer than five min. Because of differences in rime-of-day crave! patterns, different 0-D matrices should be estimated for the Ml, noo.J'l, and PM peak hour, or any other time period used in the analysis. Exhibit 4-3 illuscrates the computation. ln rhe tP'-ample, the average 15-min. 0-D macrix was obtained by manually counting approach turning movementS for.each approach. which were then used to estimate turning percentages. The approach demand was obtained by simply tallying the vehicles on the approach at a location upstream of the intersection. Using the oudined method, a roundaboi.J. t, for example, could be couoted by a simple observer, who samples the 0-D matrix at different times of the day. For increased efficiency, the approach volumes are then obtained by automatic data collection equipment as discussed lac-=:r in this chapter. It is important that while the 0-D macrix in Exhibit 4-3 was sampled for only 15 min., the approach volumes represent a count performed over the full analysis period.

Volume Studies • 4-7

,~....,.~,.$'d.i\'~

.-:"71&.·.r..JJ::~:.JJ,..

....~~""!";

~~-~(!ll~

a) ~

~-···········t····

~

····i·············)oo 0

C:* ~

.............!..···

·····%···· .........)-

......

l••••t··············>-

...

+

b)

~

1

C>>e '

at::) ......)o

- - - + V ehiclular Movement and Direction of Travel Pedestrian Movement and Direction of Travel Proposed Camera Position and View Angle

........... -==::J

~

The .sampling method is appliable to all intersections that require path-based counr:s. The 6dd observer lim estimates typical 0-D matrices with turning percentages for different approaches and different times of day. This can be done froQI. video or in the field. The 0-D matrix is chen applied to the demand flows at the four approaches to the intersection to estimate turning movementS. An alternative way co perfOrm a path-based count ac a large intersection such as a supe.rsuecc is co use a li
matching approach. Multiple observers can be positioned at the minor approaches and U-rum bays to record the last three digirs ofvehicle licenses plates. These data can later be analyzed in the office and converted to an 0-0 matrix, or full turning movement count. This approach is labor intensive, but is a reliable approach to obtain a detailed count at these intersections. Modem licease plate recognition technology shows a lot of promise that may gready assist pathbased count applications in the future.


···~

.·'

~ ··,

.. o, -

.

.

.,

.

. • :r.t1ft!:. · · ,

, ....,. •

- •

"'-"'. • :~.·--~ ·-·~ , , · -. ~' .

'

•... ·

•

A-.uage Is-min OlD Mattix Mowmcnt

Approach

U-Turn

Left North

U-Tum

15

Left

33

100% 0% 24% 48% 28%

ff7

100%

6 24

4% 14% 35% 47% 100%

Riglu Total U-Turn

Left

..

Through Right

60

Total

171

81

•

~~·'

~ ~·. ~¥-

..

~

·'

Mo~ment

Deman~

Demand

(~hides/hour)

(~hides/hour)

. 37 I ll

578

307 123 59 130

39 243 0 21 42 24

Left

·~;·

- '·>--

578

!56

Through

~-

6% 14% 64% 16%

Right

'

.,.....

- ' - -~ -

100%

. Through

U-Tum

West

6% 19% 53% 21%

75

Total

South

9 27 30 141

Total

East

Percent

Right

_,,._~

:.:-:-;~t="'\·· :·.~~;lt

. -·-

Appi'!Jach

Count

Through

.

958

615 154 958 0

331 •

678

80 160 91 331 24 95 238 321 678

2.1.5 Other Uncotwm~Unud lnteruction Design In addition to roundabouts, supczstteets and median U-rum intecscct:ions, there is a variety of other unconventional intersection designs that look and opcrue differently from conventional inte.~ons with twa orthogonal approaches and eight-phase signal control. Ex:amples of these include continuous Bo~ intersections; single-point urban interchanges and diverging diamond interchanges. These new intersection geometries i:nay appear different to the analyst :lt first encounter, but a volume study generally does not require path-based counts as discussed above. At these intersection types. each m~ment (at some point) can be observt:d in isobtion, when it does nor share pavement sur&ce with other movemenrs. A volume study at these other, unconventional intersections therefore is consistent with ones perfOrmed at Standard orthogonal intersections, where turning movement counrs can be tallied as vehicles cross a point such as the stop bar. However, due to unconventional geometry, data collectors do need to exhibit cautiop and ensure they fUlly understand Bow p:mems before beginning the count. If all movemenrs can be observed in isobtion (not sharing lanes with other m~menu), automated data collection methods can readily be used ·as discu.sscd b.ter in this chapter.

2.2 Area Counts In many applications, it is necessary to obtain count data for a bigger area in the tranSportation network. All states maintain ongoing count programs on state highways for planning. estimating vehicle miles of travd, tracking volume trends and conducting traffic engineering analyses. Another objective of these programs is to estimate the average annual daily ~c (AADT) at coverage-count locations. Cities and counties may also have similar progyams for roads and screeu in their jurisdictions. Area counts can further be classified into cordon counts, screen line coun~. control counts and coverage counts. • Volume Studies • 49

2.2.1 Cordon Counts Agencies make :1. cordon coum by encircling m area such as a central business district (CBD) or other major activity center with an imaginary boundary md counting vehicles and pedestrians at all of the points where srreets cross the cordon. Observers classify each vehicle by type, direction of travel and occupmcy and typically use 15- ro 60-min. intervals. The counts show the amount of traffic entering or leaving and ehable an estimation of the vehicle and person accumulations within the area. · Agencies use cordon counts moSt commonly as part of an 0-D survey as a basis for expanding imerview data. The counts are taken in conjunction with the interviews. 0-D studies are described in Chapter 20. Cordon counts may also be taken for trend analysis purposes. For this application, agencies count one weekday each year, during a month with an average daily traffic (ADn that is dose to the annual ADT. The counts are made at the same time each year. The cordon is established by following several guiding principles d.efinc:d by Roc:ss, Prassas and McShane (2004): • The: cordoned area must be large enough to define the full area of interest, yet small enough so that aca.unulation estimates will be useful for par~g and ocher traffic planning pmpose.s. • The cordon is established to cross all sucets and highways at midblock locations, to avoid the complexity of establishing whether turning vehicles are entering or leaving the cordoned area. • The cordon should be established to minimize the number of crossing points wherever p~ible. Natural or manmade barriers (such as rivers, cailroads, limited access highways and similar features) can be used as part of the cordon. • Cordon areas should have relatively uniform land use. Accumulation estimates arc used to eStimate ~tteet capaciry and parking needs. Large cordons encompassing different land-use activities will not be focused enough for these purposes. The objective of :1. cordon count is to capture 90-95 percent ofADT entering the cordoned area. The most heavily traveled roads are observed for a full 24-hour period, less heavily traveled roads for l().. hour periods, md minot roads for 9-12 hours. Alleys and very low-volume streets may be ignored, if the aggregate loss is less than 3-4 pcrcenc of the toral. Count stations on the cordon boundary are always located at midblock locations. Agencies can keep the number of stations to a minimum by taking advantage of natural or human-made barriers. Counts should be made on the same day. However, if the agency maintains a set of control stations, cordon counts made on different days can be adjusted using the control station data (discussed later in this chapter). Short counts covering only peak-hour flows should nor be used because the distribution of traffic at each crossing location is critical to the determination of accumulation. Counts of transit passengers should be available from the local transit agency or can be made using the methods presented in Chapter 13. Exhibit 4-4 shows typicalfidd and summary sheets for a cordon count. Full-size versions of these sheets for copying are available in Appendix E.

50 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDmON

~~~~~~~~~~~~~ . - ·"" - .. . --· ,· .. ---y:..,.,._. _ _...-, . .~

·~-

.:..p:.p~•w· ·-"--~

Passenger Vehicles, Trucks and MisceUaneous Vehicles Bound on

Traffic

Street From

W eather

To

Numbers of Passenger Vehicles Including Taxicabs

1/4 Hour Starting

Other Vehicle Excl. Bus &; Str. Can

Trucks

----Recorder

Date

Pedestrians Side of

Traffic

Street Hours From

Weather

To

l.obound

l/4 Hour Starting

Oulbound

- -- --

--Recorder

Da.te

Pedestrians Summary Sheet Pedestrians on

Date

Street

Weather

Street Street

i

Compiled by

Street

St

St i/4Hour Sw1lng

Side Ia

0...

Sick lo

0...

'IOoal Ia

o..

Side lo

:

o..

Side lo

0...

St Tooal lo

o..

Side Ia

Oal

Sick Ia

o••

St Tooal Ia

o..

Side lo

Oat

Side lo

o.,,

Total lo Oac

.

Total

A--ae ..... }loot

Contintted on next p4gt

Source: Box and Oppenlander, 1976, p. 40. Volume Studies • 5 _,

~~·1!--,~ Passenger Vehicles, Trucks and Misc:dlaneoos Vehicles Summary Sheet

Ihte

Traffic on

Weather

Compiled by

Sueer

Namlov lolooaad 1/4 Ho..,

.........&<'

Scardaa

Vehicla

Num.bu OutboUDd

ToQJ

ra-p Truda

TraDJit

MJ.c

ToQJ

lD .....

Vebldes

'l'racb

Tramir

MJ.c

o...

ToQJ

Total

Average Peak Hoa.r

' All Penon SWIUIW'f Sheet Date

Traffic on

Weather

Compiled by

J/4 Hoa.r

SanloJ

.......... v.blda

-..lahoaadv. Tracb

Tamil WUda

1\Wc VCiolda

S=r

,.......o..m.....dV"oa Toal

Wo.IJdac lahoood

PuocDp Ve!Uda

Tta
y-a ~

.....

Hoat

Source: Box and OppeDlander, 1976, p. 40.


Ttu.oia VCiolda

Mloc WUd.

Wolkios

Teal TocaiiD Owl>oa...t PluO..I

Vehicle accumulations within the cordon arc found by summing the entering and leaving counts at all COI,Int statiQns by time interval. The counts usually begin when the street system is at its lowest B.ow. Agencies can count the number of vehicles parked on-street and off-street to estimate the number of vehicles inside the cordon when the study begins. '.Exhibit 4-5 illustrates the procedure for computing accumulation. Exhibit 4-6 illustrates a method of displaying cordon count data summaries.

SoUJce: McShane and Roess, 1990, p. 103.

Soun:e: Grater Toronto Asea Cordon Count Program, 2006. Volume Studies • 53

2.2.2 Screen Line Counts Screen line counts are made to record travel from one area to another. While a cordon count is used to count all traffic entering and leaving an area (such as a central business district), a screen liQe count is used to capture traffic flows from one area to another area. The screen line is some form of natural or human-made barrier with a limited number of crossing points where volumes are counted. Examples include rivers, railroad lines, or urban freeways with a limited number of crossing points. Analysts use screen line counts to check and adjust the results of 0-D studies or to validate traffic distribution results of a rransponation planning study (see Chapter 20). They may also be used tO detect trends or long-term changes in land use, commercial activity and travel patterns. For screen line counts, hourly intervals for a 12- co 24-hour period are usually used on a weekday. Using several counting periods that arc I week or more apace will preclude any unusual conditions occurring on a given day from causing a bias in the data. Classification counts may also be desirable when conducting a screen line count. Upon completion of the count, the screen line crossings (hourly or total) are compared co the crossings predicted by uansporcation plan· ning srudies discussed in Chapter 20. The result of this comparison is then used to adjust the ~pottarion planning model chat predicts 0-D Bows. In other words, the screen line study results are used to calibrau chc planning modd. Exhibit 4-7 illustrates the layout of screen lines in a regional screen line count study. Another screen line example and display of traffic accumulation over rime is given in Chapter 20.

N

A LAgtnd _ .......,.~

c:J.__ ~

.......... ......

~ pr-..q AUMu.nt~•el•""""ar,an

eut

111

,,

a

Source: Southem California Association of Governments.

2.2.3 Omtrol.Counts

Daily and seasonal (monthly) volume variation patterns are established and monitored using control counts in an areawide program. Counts are made either continuously or petiodically throughout the year. The most useful counts are made at pcrmancm-count stations, which operate 24 hours a day, 365 days a year. Control-count stations supplement the data obtained from permanenc-count stations to obtain estimates of seasonal and monthly volume variations at addicionallocarions in the transportation network.

'.i

Control-count sracions are distributed across the cransporcation network and placed at strategic locations. They can be divided into major and minor concrol·couar stations, where the major Stations are placed at key corridors in the network. Typically, at major control-count srarions, a 7-day continuous count is performed during each month of the year. At minor control-count stations, a 5-
:j

S4 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDmON

;i I

I

generally twice as many minor as major conrrol-counr stations. Concrol counts in urban areas may be refe~red co as key counts. For every 20 to 25 coverage srations (discussed below}, there should be a permanent- or concrol..:oum station.

2.2.3.1 Growth Factors One very common use of comrol counrs is co esrimace yowth factors for traffic volumes. Engineers and planners are ofren asked to project traffic many years inco the future co determine if enough capacity exists on the roadway being designed or updated for future conditions. Count sca!ions at or near the roadway under consideratio n provide the analyse with a great resource for estimating rather chan merely guessing or making poor assumptions based on limited or no daca. The furwe traffic volume is calculated using an exponential growth race equation, shown bd ow. Vrutur•

=Vcu,.,..,.t(l +g)"

Equation 4- 1

where

vfi...m = future year projected traffic volume (vehicles/hour [veh./hr.]) V""""' = currenc year traffic volume count, typically peak hour (veh./hr.)

g

= average yearly growth factor, expressed as a decimal

n

= number of years projected out in the future

Typically, a running average of 5 or more years of traffic data is used from a control station; however, engineering judgrnem must be used to make good decisions about the exact time period that should be used for the growth rate. For instance, a small town that has rece.ncly begun heavily sprawling should consider using a smaller time frame to calculate the growth rate if sprawl is expected to happen for many years co come. To calculate the growth race from just two volume observations of volumes (current and future) chat were !aken n years apart, the growth factor gcan be computed directly by rearranging Equation 4-1 (all rimes as defined b~fore) : Equation 4-2

2.2.3.2 Daily and Seasonal Factors One of the primary uses of control counts is to develop daily and seasonal adjustmem factors thac can be used ~o adjust other counts in the region. Permanenc-count stations provide the most accurate source of data for compucirlg daily and seasonal adjustment factors. The first Step in such a comp,'utation is tO find the average volume for each d:aY of the week over the entire year. The average of this 7-day profile is' the ADT of a typical week. The daily adjusune.nc factor is found by dividing the ADT by the average volume for each day of the week. Exhibit 4-8 illustrates tbe cor:oputacion of daily adjustment factors.

Volume Studies • !$i05

TOTAL 10.000 vehicles ADT • 1.429 vch./day 5

Source: McShane and Roess, 1990, p. 100.

The computation ofseasonal or monthly variation factors follows a similar procedure. The ADT for each month is the monthly volume from the permanenr-count mcion divided by the number of days in the month. The AADT is then computed as the average of the 12 monthly ADTs. The monthly adjustment factors are obtained by dividing each monthly ADT by the AADT. Exhibit 4-9 illusttates the computation of monthly variation factors. Daily and seasonal factors can be computed in a similar way from control-count data. Since control counts are samples rather than continuous counts, the ri:wgin for error is greater. However, carefully planned control counts will produce reli-

able estimates. For further discussion, see Roess, Pcasw and McShane (2004).

Total Traffic (-.ehides)

TOTAL= 290,851 vehicles MDT • 290,8511365 • 797 Sou=: McShane and Roess, 1990, p. 100.

\ '

\


i2.2.3.3 P~ak Hour and DirtttionAL FattiJn ~nan operational anal}'3is of uaffic flow, the methodologies in the Highway Capacity Manual (TRB, 2000) are typ ikally based on a peak hour analysis wing the 15-min. peak hourly Bow rate. The input for these methodologies is }~eally based on observed 15-rnin. counts. But in the absence of su~ data, a. directional design-hour volume (DDHV) can be estimated from AADT a.s follows (TRB, 2000): DDHV=AADT• K• D

Equati on 4-3

where

DDHV z dire.ccional design-hour volume (veh./hr)

MDT = annual average daily tra.llic (veh./day) K

= proportion of AADT occurring in the peak direction and

D

; proportion of pQk hour traffic in the pole direction

The Highway CAptz&ity Ma~ual (HCM) offm typical values for K and 0 that can be used for anal)'3is. If an a.gency desites to develop its own estimates of these proportions, a conuol-count station can be wed, provided that da.ta. are available in hourly bins and separated by direction. The basic computational procedure mirrors the development of daily and monthly factors described above. HCM computations use the peak 15-rninute Bow rate expressed in vehicles per hour. It is obtained by dividing the DDHVby th.e peak hour factor. (PHF). The HCM and many agencies offer default values for the PHF chat can be used in the absence of detailed (15-iilln.). 6.dd data. If 15-rnin. volumes are available, the HCM methodologies directly we the highest 15-rninute flow rate (expressed in vehicles per hour). The PHF can be estimated from 15-rnin. volume observations a.s follows:

y

PHF=~

Equation 4-4

4*V, •.•m where: PHF

= Peak Hour Factor

V,....-

=

V, ..w

= Traffic volume in the peak 15 min. in vehicles

Volume in the peak hour (60 min.) in vehicles

The peak how ranges from a theoretical minimum of 0.25 (all peak hour vehicles arrive in the peak 15 min.) to 1.0, in which case the vehicles arc evenly distributed over the peak hour. ~mmended HCM default values arc 0.88 for rural areas and 0.92 for urban area.s (TRB, 2000) . . 2.2.4 Conr.p Collllb A coverage count is a relatively short-term but continuow count performed at one location over a period of 24 to 72 hours. To adjust a 24-hour coverage count for a given location to an estimate of AADT, the count is multiplied by the appropriate daily and monthly variation factors obtained from conuol-count stations. The factors wed should be ~m a permanent- or conuol-count station !ocation similar in geometry and t:raflic characreriscics co the location of the coverage count. Guidance on how to match up the: appropriate factors for a site are given in the literature (for cumple, Min-Tang ct. al, 2006).


3.0 METHODS OF DATA COLLECTION The rwo basic methods of counting traffic are manual observation and automatic counts. For the purpose of this chap· ter, manual observations will be defined as any count where individual vehicles are tallied by an analyse either during field observations or from video recordings. By conuast, automatic counts use automated technology to ~rform the count, reducing the analysts' taSks to set-up, calibration and analysis. It is important to note both rypes of counts can be performed in the field or during pose-processing in che office.

3.1 Manual Observation 3.1.1 PurposeanJ Application In manual count data collecrion, the analyst manually tallies each vehicle as it proceeds through the intersecrion or point of interest. The main advantage of a manual count is that it typically minimizes equipment cost and set-up time. An obser\rer can quickly be trained to perform a manual count in the field or from video. However, manual counts tend co become inefficient the longer the analyst remains in the field. Many types of counts require classifications that are obtained more easily and accurately with trained observers in a manual count. Examples include vehicle occupancy, pedesuians, intersection turning movements and vehicle classifications. While most of these measures can also be obtained by automatic means, the additional set-up effon is typically not time-efficient for shon-duration counts. The main reasons for conduCting manual counts are time and resources. Practical applications often require less than 10 hours of data at any given location, with most counts focusing on peak-hour conditions. Thus the'effort and expense co sec up and remove automated equipment is not justified. While there will always be a need for manual counting, advances in technology and the ability to use existing equipment for counts (for Cll:ample, loops or cameras at signalized intersections) are shifting the balance away from manual and towards automatic methods.

3.1.2 Equipment 3.1.2.1 Taliy Shun The traditionally simplest approach foe conducring manual counts is co record each observed vehicle with a tick mark on a prepared field form. Exhibit 4-10 shows a field sheet for a vehicle turning movement count. Pedesuian '!lld bicycle counts may require separate sheets if volumes of those modes are high. For many in[crsections low pedestrian and bicycle movements can be added manually on the existing form. The form allows for whatever classifications may be desired. A watch or stop~tch is required ro cue the observer to the desired count interval, and a new form is used at the stan of each interval. The raw counts are tallied, summarized, or keyed into a computer upon return £O the office. This method is low-cost and is easily adaprable to different geometries and count types. However, its application is less common today with the availability of clecaonic count boards and laptop computers, which are convenient to use and reduce analysis time.


VEHICLE TURNING MOVEMENT COUNT FOUR-APPROACH AELD SHEET Tlma

to

Olote

NIS S tr. .t

E IW Str . . t

pe~aeng.,.. c.,.., •t8tlonw-.;a0ns.

. p -

motorO)"Ciea, pldc·uP tr~.

D•y

weath er Obaerver

T - · ottwf' trudca. (R-=onl _,y' Kh~ bus u · S B : ottltU'. ~·a. IS).

p

~ p

. IT

I I

T

p

T

., I

I

p

I

,. I I

I

--

I

I I I

I I

I

~·

IT

I

I

I

p

I

I

........

..1.

I

r

I

I

I I

I I I I I

-· .

IT

- - -I T

IT

p

jp

I I

I

I I

I

I

I I

I

I

1

I

I

p

I

I

IT

II

I

I

I

I

I

I

lp_

,

p

+

I

;-.: I

!T

I I I

I I I

I

I

I

;' ~ I

I

p

I

I .I I

I I I

T

r

Source: Box and Oppenlander, 1976, p. 21.

3.1.2.2 Handlu/.J Count Boards Battery-operated, handheld electronic count boards are CUirently t&.e most common device to aid in the colleaio~ of rraffic count data They have all but replaced traditional mechanical count boards that used accumulating pus~ burton dc:Yiccs mounted on co a clipboard. The biggest drawbacks of mechanical count boards we.re that the analy-st had to keep track of time while performing the count, and that data aggregation and analysis had to be done sep ::a-ratdy. They may still have limited application in situations where the only needed data are overall tallies of even. -.;s (for example, the number of passengers boarcling a transit line). In most cases, the added benefits of electronic cou,e:l--.: boards justify their expense. Electronic count boards are compact, lightweight, handheld computers with di.fferem buttons allocated to differel-1 • movements at an intersection. They are much simpler in design and visual clisplay chan laptop computer.;, and ftaru.c-e. rugged casing and long battery life. Electronic count boards contain an internal clock that separates tho data by whacever interval is chosen, so no field forms are needed. Electronic count boards have an advantage over Wly sheeG and Volume Studies • ~

mechanical counting boards, since paper forms are more sensitive to weather (wind and rain), and are sometimes hard to keep organized for long studies. Mosc importantly, they preclude the need for manual data reduction and summary. Data may be trarufeued directly from the field to a computer in the office via modem or USB connection upon recurn from the field. In the analysis software the data are summarized and processed, and the results displayed in a sdecred presentation format. This eliminates the data reduction step required with tally sheets and mechanical count boards. Many elecrronic count boards ate capable of handling several types of common traffic studies, including turning movement, classification, gap, stop delay, sacucation Bow rate, stop sign delay, spot speed and travel-time studies. Foe agencies requiring more than occasional manual traffic counts, the decrronic count boaid or handheld computer is a cost-effective, labor-saving tool. Exhibit 4-11 illustrates a typical electronic count board. The window on the counter displays data, menus ofcommands and messages about the status of the counter. Most current electronic count boards provide a shift key for special functions such as recording particular vehicle classes, but these functions should be limited because of the added cognitive load of the observer.

~

..

·!:! ::

} ) i ··

Sotirce: Jamar Technologies Inc.

3.1.2.3 Laptop ComputnA battery-efficient laptop computer can be substituted for a handheld count board in many applications. Oftentimes, laptop computers ate already available to the analyst, making this an attractive alternative to new hatdware. Commonly available spreadsheet software can be used to record time stamps of different types of events using a macro routine. Appendix E contains discussion of how to code a time-stamp macro in Microsoft Excel~ and VISual Basic~. The benefit of using a macroenabled spreadsheet to collecr volumes (or other temporal event data) is that it can be customized to the specific needs of the user. Also, many commercially available count boards are ~cted to output aggregated data.. Alternatively, this approach allows the analyst to obtain individual time-stamped events. The downside is that some software coding and post-processing analysis are required. Consequently, this approach may be more interesting in research settings and other special applications.

3.1.2.4 Vuieo-Based Counts :;'

:•;·

;i;

'!

;,;·

:~~,,

.

Manual counts can be performed in real time in the field or in a post-processing operation from video observations in the office. The use of videos may be practical if video is already available from a previous effott, or if it is easy to obtain. For example, modern traffic management centers often have live video feeds from permanent field cameras (on freeways or signalized intersections) to a central office, where they can be recorded. For manual freeway counts this is frequently a safer approach than positioning an observer on a bridge srruccure or other location in close proximity to fast-moving traffic. Video-based counts can also min.i.tnizc staff requirements, if the same analyst can replay the video to count different movements. A well-chosen camera angle, ideally from some overhead vantage point, is critical to cnswe the video is useful. With adequate light conditions and a good vantage point, one camera can capcure all curning movements at a typical intersection. A digital clock in the video image is an excellent way to note the end of intervals. However, notations by the camera operator or the camera operator speaking into the microphone arc a,dequate substitutes for the clock.

Modem video-image processing software (discussed below) may auromatically provide volume data. Alternatively, observers can record their counts with a handhdd count board, tick macks on a tally sheet, or directly into a computer. For most applications, this way of obtaining manual counts from video effectively doubles the analysis, making it

\

\


Jess practical if video observaciom aren't already available. One benefit of video is that the observations can be easily error-checked by a second analyst or can be harvested for additional data op vehicle dassificacio.n, dday, or queues. If necessary, observers can view the video in slow motion for additional scrutiny. The ability to conduce other stud. ies with video, including the intersection and driveway studies discussed in Chapter 6, can lead to very efficient uses of the labor of fidd crews. Usually, other studies require a higher-quality video than a turning movement count, S
3.1..3 Perromui.Requ~J Manual traffic counting requires trained observers. They must be relieved periodically to avoid fatigue and degraded performance. Breaks of 10 to 15 min. should be sclleduled at lca.st every 2 hours. If the data collection perio d is ~ore than 8 hours, breaks of 30 to 45 min. should be allowed every 4 hours. The size c;>f the data coUcction ccam depends on the length of the counting period, the type of count being performed, the number of lanes or crosswalks being observed and the volume level of traffic. One observer can easily count turning movements at a four-way, low-volwne, signalized intersection with one-lane approaches as long as special classifications and/or vehicle occupancy are not required. As any or all of the foregoing variables increase, the complexity of the counting cask increases and additional observers will be needed. Duties may be divided among observers in various ways. At a signalized intersection, one observer may record the north and west approaches while the ot:h~ watches the south and cast approaches. In that way, only one ap proach is moving for each observer at any given time. Another way to divide duties is for one observer to record occupancy or certain classes of vehicles, while the other observer counts total volumes. At complex sites, individual lanes, crosswalks, or classifications may be assigned to individual observers. Also at complex sites, one observer may have the sole job of rdieving me other observers on a rotating schedule basis. 3.1.4 Field Procedure 3.1.4.1 PrtparatifJn An accurate and reliable manual traffic count begins in the office. A locally devdoped checklist is a valuable aid, even to experienced teams, co ensure that all preparations for the field study have been completed before the team arrives at the site. Exhibit 1-1 is a general checklist that can serve as the beginning of a locally devdoped checklist. Pre~ tions should start with a review of the purpose and type of count to be performed, the count period and time intervals required, and any information known about the site (such as geomeuic layout, volwne levels by time of day, signat timing, etc.):This information will hdp determine the type of eq~pment to be used, the fidd procedures to follow and the number of observers required. Online mapping and visualization tools may hdp identify good vantage points, but local knowledge or a site visit to the location is usually necessary.

The selection of equipment will dictate the types of data forms needed, if any. Header information should be filled in to the extent possible in the office, and the forms should be arranged in the order in which they will be used by each observer in the fidd. When using handhdd count boards it is very important that a naming convention for streets and orientation be agreed upon beforehand. Without such convention, it can be difficult to matcll up multiple output 6Jes in the analysis step. The checklist should also include equipment items, such as pens, batteries, stopwatchd"and blank media (video tapes, discs, Hash media, or hard drives), as appropriate. Having to rerum to the office to retrieve forgotten itemS may delay the start of the st1;1dy or cause it to be postponed. An inadequate number of forms could also invalidate the study and waste resources. Equipment must opcrare properly to ensure accurate counting. Good counting boards have firm keys that provide the observer with tactile and audible confirmation when a key has been pressed successfully. Units with "soft~ keys should be repaired or discarded. The lack of tactile or audible confirmation when a button has been pushed poses a challenge when using laptop computers over long count periods. A commercially available count board may thus be prefera~lc for long coi.wts. An office review of the procedures to be foUowcd and a check of the proper operation ofall c;quip111ent complete the preparation stage of the study. Volume Studies • 61

3.1.4.2 Obrerver Location Observers must position themselves where they can most clearly view the t~affic they are counting. Observers must avoid vanrage points blocked regularly by trucks, buses, parked cars, or other fearure.s. They should be located well away from the edge of the travel way, both as a personal safety precaution and to avoid di.macting drivers. A position above the level of the street and clc:ar of obstructions usually illOrds the best vanuge point If several observers are counting at the same site, it is helpful for them to maintain visual contllct with one another, and to be able ro communicate so as ro coordinate their activities. Given char observers are likely positioned on opposite corners of an intersection, rwo-way radios or cell phones are helpful. Protection from the elements is also an important consideration for rhe observer. Proper clothing ro suit prevailing weather conditions is critical. Safety vests should be worn if the observer is near traffic at any time. Observers may co unt from inside vehicles as long as their view is unobstructed. Sitting in an automobile is safer and more comfort· able during inclement weather than sitting outside. Observers should park the vehicle in a legal space that is close enough to _the intersection so they can sec the farthest lane of their assigned approaches. Parking on priVllte property is sometimes convenient, but observers musr first obtain the property owner's permission. While sitting outSide, observers may use chairs to prevent fatigue and umbrellas for protection from the sun, as long as these devices are not distracting to drivers. A sign indicating that a traffic count is under way usually satislies drivers' curiosity about observers they can view from the roadway. The analyst should identify count locations ahead of dalll collection or plan sufficient rime to scope' the site just before the count. In general, it is goo
3.1.4.3 Daut &cording The key ro successful traffic countS lies in keeping the data organiz.cd and labeled correctly. CountS may produce a large number of data forms, or deccronic files. Each form or file must be clearly labeled with such information as the count location, observer's name, rime of study and conditions under which the counts are made. The form itSelf should clearly indicate the movementS, classifications and tiroe intervals. Por count boards, it is important ro maintain a naming convention for files and movementS. The observer must concentrate his or her attention on accurately recording each count in the proper place or with the proper button. Special care must be taken with deccronic counting boards to ensure they are properly oriented to the geographic and geometric layout of the intersection. When rwo or more observers arc working together, time intervals musr be maintained and coordinated accurately. Observers should also look for and note on their forms, or in a log, any temporary traffic events, such as collisions or maintenance activities, that may lead to unusual traffic countS. For this reason, it is good practice to always carry a field notebook, even if using automated count equipment. It is also good practice to have the data colleccor llllly each data collection sheer before handing it to the analyst ..This practice improves accuracy and decreases missed countS.

3.2 Automatic Counts 3.2.1 Purpose mui ApplicAtion There are many applications in which counrs are needed for extended periods of time (days, weeks, or even months). The use of observers for such purposes would be cost prohibitive. Automatic counting provides the means of gathering large amountS of volume data at a reasonable expendirurc of time and resources. The traditional disadvantage of automatic countS relative to manual counts is equipment reliability, but research suggestS increasing accuracy of modem technology and hardware offer solutions. While with modem equipment, there is a lower chance that detection or recording equipment will fail during a given scudy, there is still a chance of malfunctions, vandalism, large vehicle interference, weather, or numerous other issues. AnalystS must prepare contingency plans co use in case of equipment failure and understand the limitlltions of the selected technology.

Modem technologies for automated counts can principally be divided into on-road technology and roadside technology. On-road technology includes pneumatic tubes, piezoelectric scrips and various forms of magnetic inductance technology. Roadside technology can utiliz.c video, radar, infrared, or lasc.r technology. Roadside technology can also be combined with in-W!hick technology which typically takes the form of deccronic roll transponders that can communicate with roadside readers. 62 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES. 2ND EDinON

Automatic count technologies are frequently used at intersections, where chey provide turning movement councs. . They are also increasingly applied ro freeway segments and tolled facilities to provide lane-by-lane infor mation on vol. umes and other data. Generally, all of the above technologies can automatically count vehicle classifications, and may in some cases provide additional output on vehicle speeds, headways, density and even travel time from one count location co the next by using vehicle identification technology (for example from toll transponders). 3.2.2 Equipment Boch in-road and roadside technologies typically consist of rwo basic components: a data recorder, and sensors to detect the presence of vehicles and/or pedestrians. Some equipment also has the capability of communi cating the collected data to a cenual facility for processing. Space does not permit a.discussion of che wide range of equipment available today. Besides, advancing technology is causing continuing change chat would quickly render such a discussion obsolete. This therefore fo cuses on basic characteristics of in-road .and roadside technologies, and tradeoff's co consider.

3.2.2. I In-!Wad Count Technology . In-road count technology is mounted directly to the travel lanes or in some cases is p<;rmanendy embedded in the pavement. It can take che form of petmanencly installed equipment used to perform a long-term con trol count: or portable equipment used to conduct a 24-72 hour coverage count. Agencies establish petmanenc-count stations whe,re they desire long-term, continuous counts (for example, 24 hours a day, 365 days a year). The volumes collected at chese stations are usually part of an areawide program co monitor traffic charactetistics and trends over time. Automatic count technology is also available in portable format for use in temporary applications. · Permanent traffic monitoring s~tions 3!e common for both signalized nerworks and freeway facilities. In signalized n~rworks, the vehicle detection technology (most commonly magnetic inductance loops or video detection technology) is routinely used to get traffic volwne data. Often, special system loops are installed for the sole purpose of trafii~ monicoring and data collection. On freeways, in-road detection technology is used to collect traffic volumes and other traffic parameters. More information is provided in Chaptet 10. For permanent data collection equipment, durability and reliability are of central importance.

In addition to permanendy installed equipment, me analyst has me choice between several portable or temporar}' options for data collection equipment. Most frequently, agencies use pneumatic rubes or magnetic inductance technology mounted directly on the pavement. Data are stored in built-in memory and can be downloaded via USB con· nection, or even wireless transmission. Battery life in modern equipment allows chese devices to record several weeks of data on a single charge, although applications this long are rare. Exhibit 4-12 shows a sdection of modem in-ro~d data collection equipment. ··


c) On-Road Magnetic Inductance

d) Sensys Detectors

:~

·;

Source: Images by a) Jamar Technologies Inc., b) City of Richmond, CA, c) Vaisala Inc., d) 2009 Sensys Networks Inc. ; •, ~:~

The aforementioned in-road technologies are limited in their ability to count pedestrians and bicycles. Further, special care and attention is needed when installing and removing these devices from lanes of moving traffic. Depending on the intersection configuration, these technologies may require a lot of equipment to capture lane-by-lane data. The additional cost and set-up time make these devices more applicable for longer duration counts. For short-term counts (up to 8-12 hours), a manual count is typically more cost-effective.

:•

3.2.2.2 Roadsitk Count Ttchrwlogy Many modem automatic count technologies can reliably detect vehicles from a roadside or overhead location. AdV211CCS in radar, laser, microwave, infrared and video-image processing technologies allow for automated and nondestrUctive volume measurements at intersections and in freeway applications. While some reports claim reliability concerns with some of these technologies (Federal Highway Administration (FHWA) 2006), it is expected these concerns will decrease over time. In fact, others cite good reliability in research applications (Banks 2008), suggesting the bigger concern with these technologies may lie with long-term applications and maintenance.

~:.

::

~~

:t! .j( ;!. }'

:~

Just as with in-road equipment, roadside count technologies can be permanent installations or can take the form of portable or temporary equipment. Permanent installations include video detection systems at signalized intersections. The use of permanent equipment based on other technologies (radar, microwave, etc.) is more common at freeway applications and tolled facilities. More detail on this is provided in Chapter 10. Portable roadside equipmentis available, but used less frequently than in-road equipment. Exhibit 4-13 shows a selection of modern roadside data collection equipment.

'.

q

::1

i

\

l \

(

\ ,


3.2.2.3 V'ut'eo !mag~ Processing Video-image processing systems. can automatically collect volume and other data from video. The analysis process typically involves computerized meas\u'em.ent of lighting changes in pixels on the video; however. the exact algorithms are proprietary and vary among different manufacturers of video-image processing technology and software.

-I J

! '

Videe>-based automated counts are u.sed at signalized intersections, where video detection cameras can double as data collection technology. The positions of these cameras are typically st:a.tic and the c:a.meras are c:a.librar:ed to record diffen;nt movements on one approach of the intersection. Because of sight-angle restrictions, multiple c:a.meras :a.re used if more than one :a.pproach to the intersection is evaluated. Once the video detection cainera is installed, the an:a.lyst predefines virtual daecton using computer softwa.re at locations on the video image where movements :a.re to be recorded. The data aggregation interval is :a.lso a software input. Once the vide<\ image is c:a.librared and configured, data are collected and aggregared in the specified time intervals. In statc-of-the-:a.rt signal,systems, individu:a.l intersections are interconnected with each other and a central proccsser in a traffic managemenc center (TMC), where volumes and other dat:a. arc being stored and an:a.lyz.ed. Video-image processing can also be applied to offiine video recor~ in the office. The ch:a.llenge for offiine video analysis is that the video detection softwa.re needs precise information of the c:a.mera loca.tion to a.cau:a.tdy process the video. In the c:a.libcation step, the an:a.lyst has to enter the camera height and rdative distance to a known reference point on the video. This process is known as orthortctifoJZtion or imagt ca/ilnution. Without calibration inform:a.tion, the video typically c:a.nnot be used for offiine image processing. Due to the levd of effort involved, oflline autom:a.ted video counts are likdy to be a highly inefficient way to collect simple count data. The set-up effort may be justified if :a.ddition:a.l data are collected or if the goal is to obtain path-b:a.sed countS.

need

In addition to the for c:a.librarion, all video-b:a.sed detection. and count technology are susceptible to camera movement'(wind), lighting changes (daylight, clouds) and occlusion by call vertical objects (trucks). The reader is referred to documentation provided by the equipment manufacturer to overcome these issues. 3.2.3 Pn'$0tmel Required The only pcrsonnd required for :a.utomatic countS are those needed to inscall and recover the equipment. Crew siz.cs of two or three are usually sufficient to deploy most portable counting equipment. Depending on the type of equipment, the inscall:a.tion crew may have to be in the a:a.vded ~y and it may be preferable and s:a.kst to temporarily close Lanes or inscall equipment during periods oflow traffic. The recording component can be handled by one person; however, one or rwo persons will be needed to install ro:a.d rubes or magnetic sensors, while an addition:a.l person w:atches for traffic. Roc.overy of the equipment can usually be performed by one or two persons. The inscallariofl of permanent counters with in-pavement sensors rn:a.y require a l:a.rger crew and the closure of travd lanes. Volume Studies • 65

3.2.4 Field Procedures 3.2.4.1 Preparation Field work should never be undertaken without proper preparation in the office. A locally prepared checklist is an invaluable aid even for the mOSt routine task. The purpose of the count will drive the type ofequipment to be used and the deployment procedures to be followed. All equipment should be checked to see that it is limctioning properly. An ample supply ofaccessories (such as nails, clamps, tape, adhesive, chains, locks, batteries) and all necessary tools should be on hand. 3.2.4.2 Selecting the Count Location Agencies decide in the office on which street or highway the count will be made and the general location (midblock or intersection) where the counters will be placed. These decisions depend on the type of scudy being performed. The exact location of the count recorders and sensors may be determined in the field. Often, the transportation analyst will order a count from a contractor or field crew. In these cases, rhe exacr counr location may be marked on a computer drawing to ensure proper installation. The following guidelines for the deployment of auroqtatic counters apply to both in-road and roadside technologies.

. .! ,.

• Do not place sensors across marked or unmarked parking lanes, where a parked vehicle could activate the sensor continuously. • Deploy sensors at right angles to the traffic How. • For directional counts, keep sufficient space between the sensor and the centerline of the roadway. • At intersections or near driveways, place sensors where double counting of corning vehicles can be :~:voided. • Record sensor placement by noting the physical location on a condition diagram sketch. • Use a test vehicle to ensure that bidirectional counters are recording the proper direction. • Avoid locations where frequent queuing occurs.

• Set the count interval to ensure that totals will occur on the hour to make the data compatible with other counts. • Note the time that counter operation begins. Additional guldeJ4les for in-road counters include: • Avoid placing sensors on pavement expansion joints, sharp pavement edges, or curves. • Fasten the sensor securely to the pavement with nails, clamps, tape and/or adhesives made specifically for this purpose. Loose sensors will prevent the collection of data and may pose a hazard to motorists and pedestrians. Locate the count recorder near a sign pOSt or tree and secure it with a lock and chain, or place it in a locked signal control cabinet to prevem vandalism. • Keep the cable or cube that conneCts the sensor to the recorder as short as possible. • Check the installation periodically to ensure it is in place and functioning properly. In cold-cl.imate areas, agenCies should check sensors whenever it snows to ensure snow plows have not removed the sensors from the road. ·

3.2.4.3lmtallationand Retrieval The primary concern during installation and retrieval operations is the safety of the field crew. The crew's vehicle should be clearly visible to traffic and should be parked away from the travded way. All crew members should wear reflective clothing at all times. Deployments and recoveries should be accomplished during periods of low traffic volume and good visibility. If nighttime operations are necessary, the crew should employ extra safety measures (such as lights, cones, warning signs). Anytime that crew members must enter the roadway, at least one crew member should have the sole duty of watching for traffic and warning the rest of the crew. Details of installation techniques vary and in many cases are product-specific. Information of this nature is generally available from the product manufacturer. In some eases, police assistance may be required to ensure the safety of the crew and the public. 66 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

4.0 DATA REDUCTION AND ANALYSIS Following collection, raw data must be placed in a form suitable for analysis. This reduction usually c onsists of con. verting rally marks to numbers, summarizing the data by calculating subtotals and totals, and arranging the dara in ·an appropriate format. The analysis may range &om a simple extraction of descriptive information ro a sophisticated statistical treatment, depending on the type of study being conducted.

4.1 Manual Counts Data collected by observers using tally marks or count boards must be reduced to a form suitable for analysis. Tally marks ace counted for each time interval and classification and the counts ace entered on summary she ers as shown in Exhibit 4-14. Modern count boards typically ace associated with a so&wace analysis package that can create sum~nary reports and graphical displays (Exhibit 4-15) of count data automatically. In addition to the counts f oe each movement, totals for each approach and the intersection as a whole are shown foe both peak hours.

Source: Roess, Pcassas and McShane, 2004, p. 179.

Volume Studies • e;1

_J)~ n

StiniOil

~

Peek Period~~ for JalleiMdlon AN 7:30 110.:8:30=._ __

LU 12:11 PM 8:00

110.:1.:..:.:15;:.___ __

110..:"'=._ __

WAIIMrCoDtlllloM AN~

w~

PM~

N

0 c-ar<•>

- - -- -·

tt

l

:l'

il'

City of Raleigll Pullllc Worb

Source: Cicy of Raleigh Transporurion Operations Divilion.

"i J,,

ii :•

:~


,

Automatic data collection equipment typically records data on built-in or removable memory. Depending on the product, cbe data need to be downloade.d in the office or may be mnsfc.md directly over modem, fiber, or wireless connection. As with modern count boards, automatic data recorders typically come with analym software that will automati.cally reduce and tabulate data at a user-specified aggregation inccrval.

;.,

:;•

...

;i": .,, ·;; ·ll lit

.:.

'~

.}

ri ,:P.

~

j

l

This computer software accomplishes the reductions and produces summaries of the data and desired calculated values (such as ADT, peak-hour factors and percent turns). These tools save analysrs time and often eliminate errors.

4.2.1 Qnnerlmg Axle Counts to Wlbick Counts Older automatic cotintcrs used to be driven by single-point sensors such as pneumatic road tubes, tape Switches, or loops. They therefore recorded axles, not vehicles, and the raw count had to be converted in the analysis step. While methods exist to convert axle counrs to vehicle counts, modern data collection equipment typically makes this step unnecessary. Modem pneumatic tubes arc delivered and inscallcd in pairs, and built-in logic automatically determines a v_chicle and classification count. Other equipment based on magnetic induction technology also gives a vehicle and c:Wsification count automatically. For guidance on how to convert axle to vehicle counrs the reader is referred to R.ocss, Prassas and McShane (2004) or Robertson cr al. (1994).

j

\J 'I

\l

I

68 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

: 4.3 Count Periods ! The counting period sdected

for a given location depends on the planned use of me data and me methods available 'for collecting the data. The count periods should be representative of me time of day, day of week, or month of year that is of interest in the study. For example, Mondays and Fridays are usually not typical weekdays. Engineers and planners rarely need turning movement counts, vehicle classifications, or pedestrian counts from nights, Sundays, or holidays. Saturday counts are sometimes needed for shopping areas. The count period should avoid special events and adverse weather unless the purpose is to study such phenomena. Count periods may range from an hour t o a year. Manual counts are usually for periods less than 1 day. Typical count periods for curnlng movements, sample counts; vehicle dassHicarions, pedestrians and bicycles include • 2 hours; peak period

• 4 hours; morning and afternoon peak periods

• 6 hours; morning. midday and afternoon peak periods

• 12 hours; daytime (for example, 7:00 a.m.-7:00p.m.) Count intervals are typically 5 or 15 tnln. For capacity analysis purposes, 15-min. ~unt intervals are arlequate, which is consistent with HCM methodologies. Electronic count boards provide totals automatically for any interval. Wim a computer program to analyze these counts, shorter intervals require no greater observer or analyst effort than longer intervals and provide more detailed results. The choice of an appropriate count interval depends primarily on the needs of the analysis; shorter intervals will result in more data. While memory storage is oflittle concern with modern computers, excessive data resulting from short time intervals can become unmanageable. Automatic counts are usually c:dcen· for a~mum of 24 hours. They may extend for 7 days, a month, or even a year. The interval most commonly used is 1 hour, al!hough smaller intervals may be desired for certain purposes. Smaller intervals r~uire greater computer storage space and reduction time, and the analyst should consider an appropriate file management structure to organize count data. Modern database management systems allow the analyse to store counts from multiple locations over long periods of time. The database can be queried for a particular study period. Many providers of automated count technologies also offer services for data management, including databases and query tools. 4.3.1 Sample Counts anJ Count Expansions All counts are samples. Even permanent-count sucions represent a sample ofspecific IOc:ations among many locations in a given area. Count periods are also samples of the overall long-term traffic How. Tune and resources do not permit me continuous counting of every intersection and unique roadway section on all aisring streets and highways. Con~. sequendy, ~pie counts are taken over shorter time periods at specific locations. These counts are men adjusted and/ or expanded to produce estimates of the expected traffic How at tha~ or similar locations.

Short counts may be expanded by use of a control station. Ifa number ofsample counts are needed in a relatively small area, analysts sdect one location representative of the area streets to be sampled. It is important that !he control seacion service me same type of Street, and variations of traffic being sampled, on me Orner streetS. The COntrol Station is counted continuously during the entice sampling period using the same count interval (for example, 15 tnln.) as on the sampled streets. The counts taken at the sampling location are called covmzgt counts. Bod! the coverage and control counts are taken at midblock to avoid the complexity of~ movements. Each link or street segment to be sampled should be counted at least once during the sampling period. The counts may be made manually or «'ith automatic counters. The control-count data establishes me volume variation partern for the entire sampling period. The partern is quantified by calculating, for the control-count data, me proportion of the tocal sampling period volume occurring during each count interval. Assuming that this pattern applies to all of the sampled locations in the study area, the full sampling period volume for a coverage-count location is obtained by dividing the sample count by the control-count proportion for the corresponding count interval. Exhibit 4-16 shows an illusttation of me procedure. In the example, a control station was used and four coverage counts were made over a 2-hour sampling period. The count interval was 15 min. beginning at each quarter hour. Roess, Prassas and McShane (2004) provide a more extensive ~ion of this count expansion procedure with examples. Volume Studies • 69

•f§l' .. ---.... - ;'~W~ ~ ..• ~ • .• ~- ·j./." -~~!:f'~lt~1f~~~~~t'"~':~1~-·~t"" .... ~,~~tt~1!e~~,~~~ ~ . . ' '!. ~ ~· -~ ! , • . .- :,.,....~~ ~-- ,,::~ .~_:.-,..~~- ...... ·~. ~~tl-•:: ~~- s .,... : .. 'f ~~;~~!·/ .... ~,.~-!.~·t-v ... -•. Control Station Data Cou.nt (vehicles)

Proportion of Period Total

7:00-7:1SAM

720

72016,864 .. 0.105

7:15-7:30AM

nG

776/6,864 a 0.113

7:30-7:45AM

837

837/6,864 ~ 0.!22

7:45-8:00AM

95 1

951/6,864 = 0.138

8:00-8:15AM

1,022

1,022/6,864 . 0.149

8:15-8:30AM

986

986/6,864 = 0.144

8:30-8:45AM

874

874/6,864 =0.127

8:45-9:00AM

698

698/6,864 • 0.102

Tune

!

TOTAL • 6,864 vehicles F.xparulon of Coverage CollDts Ttme

Count (vehicles)

1

7: 15-7:30AM

784

784/0.113. 6.938

2

7:45-8:00AM

1,192

1,19210.138 =8!638

3

8:15-8:30AM

863

863/0.144 . 5,993

4

8:45-9:00AM

532

532/0.102 = 5,216

Location

Proportion of Period Total

•

TOTAL • 6,864 velticles

4.4 Volume Data Presentations Traffic volwne data may be poruayed in a nwnber of ways. The selection of the presentation method depends on the planned usc of the data and the audience that will view it. Analysts most often dcpiet volume data in summary cables or in one of several graphical forms. Graphs and bar charcs arc suitable for illustrating traffic volwnes over time. Exhibit 4-17 provides examples of bar charcs showing monthly, daily and hourly traffic variations. Peak periods of How are readily disccrnable. Pic charcs are useful co show proportions of volwnes by type of vehicle. An intersection graphic summary, such as the one shown in Exhibit 4-15, presents a piaure of how much traffic Bows through the intersection during a specified period.


~

·~-

Dr.

~'r!Z. . ~.·--

~ 1)0

"' 160

~ j

140

t

120

!

130

~0

JIO

'o

120

1

100

•

90

~

110

l.

": 100

I

90

:>'

80

~

?0

.s

.Dt.ain St •JWirold St

Q

l lO

i

•

~

I I • I I • • I I ~F.l~f~~"'g:lJr.>!:l - ~ ~ < ~ - ~ < ~ ~ ~ 5

~

~

80 70

r--

rr--

-

f-

r-----

f

~ -

I ro

":' 0

l! !!i

~

... E ~

f-

f-

r--

-

t-

f-

,_

t-

fl ~

~

~

~' ~

~

.,~

l>o.IolWotk

MH.IholYeoc Mool.bl11'rallk Varildou

9.0 8.0 _ 7.0

t--

f-

Dalli1nllic Variatloos

DMainSt

-

• Railroad St

6.0

s.o 4.0

3.0 2.0 1.0 0.0

.. n. ~

- ...

~

~

·~a ..

.... ~ ~ ~ 'T

,.:.

:t

0

~

... ;;;"'

-6 - U..Ptriod

. ....

.. ..

"'~ "';;; ~... ,.:. "' d.~ - - - - -

~~

;:; ~ ... " li! ;:; ~ ... ~

Howty1nllic Volunats


An intersection Bow diagram (Exhibit 4-18) provides a similar picture. The width of the flow band indicates the level of traffic volume. Traffic flows may be depicted in the same way on a route map (Exhibit 4-19). The thlckness of tb.e bands provide,s a visualization of the relative volumes found on the•sn:eets of a Highway network. Another common means of presenting volume data for an area or system of streets is a traffic count map. Exhibit 4-20 shows a nup wich total volumes in vehicles per day (vpd) in both directions for a city street system. These traffic count maps are typically maintained by and available through the state department of cranspottacion.

Volume Studies • ,....,

29 ~~·

I ;\. 'Ii .

,, ·;

:r1:

seleCt riinet . • ·

F!o.ni ,.' _·.-106: -:-·oo~M."----::J--,_.

. · To . .j 05:45 PM .

:.....··.~·.:_ -

.... ...... ... ~

2~ -~:. .

Sourg:: Jama.r Technologies Inc.


\

'.

.·

::J

. .,. . ~

1·.. t!~ ··1

CITY OF RENTON

2008 TRAFFIC FLOW MAP DO'M1l4)1T I1F "--/IMUlll«l/I'UIIUC MR
8CAI£: ,. •

NOT£:

Source: The Oty of Rmton Wa.shingron, 2006.

IKTEIISTA'tt -

~000

IEHilD

NCI STAlE IIQ1lE 18'7 Nt£ AT HAlF

~ I1F 1-4011}

I

i

I

i

!

:i

I

S.OSUMMARY In this chapter the various methods used to perform traffic volume councs have been described. Issues rdated to data reduction and analysis and specific types of counting studies have been discussed. For further information and more details on traffic volume studies, refer to Roess, Prassas and McShane (2004).

6.0 REFERENCES Banks, J. H. Eval=rWn ofPorusbk Dara GJil«tion T«hno!IJgit:s: Fmal &pon. San Diego, CA University of California-Berkdcy; Partners for Advanced Transit and Highways (PATH), 2008. Bonsall, P. W., F. Gbari-Saremi, M. R. Tight and N. W. Marier. The Performance of Handheld Data-Capture Devices in Traffic and Transport Surveys, Traffi~ Engineering and Contro/29, No. 1 (1988).

Box, P. C. and J. C. Oppenlander. Manual ofTraffi~ Engineering Studia, 4th ed. Washington, DC: Institute ofTransponacion Engineers, 1976: 13. Dixon, M. P.-R. "Fidd Evaluation ofRoun
Federal Highway Administration. Virginia State Route 1 V"uko Deuction ~stem hrfoT7711t11u Assessmmt. FHWA. Washington, DC,_2006. Garber, N.J. and LA. Hoe!. Traf!U & Highway Engineering. Pacific Grove, CA Brooks/Cole, 2002.


Grecne-R.oesel, R. D. •Effectiveness of a Commerci:llly Available Automated Pedesu~n Counting Device in Urban Environmems: Comparison with Manual Counts. • Pr«udintl ofth~ 87th Annual '"'~~ring ofrhe Transporrmion kwzrch Board. W:uhington, DC: TRB. 2008. Hummer, J. S. "Recenr Supersrreet lmplemenracion and R=arch." Confirmu Procudings ofthe 3rd Urban Smn Sympotium. Seactle, WA. Washington, DC: TRB, 2007. JAMAR Technologies. Data Col.kction Equipmmt. Retrieved December 29, 2008, from JAMAR Technologies: ja manech.com/ index.htrnl. ·

Kyre, M. and M. Lines. "Development ofVideo-B:ued and Other Automated Traffic Data. CoUection Methods. '' Traffic Track(r Manual. Retrieved April2010. National ln.stitute for Advanced Transpormion Technology, Universicyofldaho, Moscow, ID2004. http://www.websl.uidaho.edu/nian/ruearch/Final_Reports/KLK203_N04-02.pdf. Li.sr.• G. F. •Identifying Vehicle T rajectorie.s and Turning Movements ar R.oundabqucs." 5siJ International Sympo•ium on Highway CapariiJ and Quality ofSn-ttu~. Yokohama, Japan, Japan Society ofTraffic Engineers (2006): 44_9-458. Min-Tang. L F.-F. "A.s.signmeoc ofSe:uonal Factor Categories to Urban Coverage Counr Srations Using a Fuzzy Decision Tree: ASCE]ourruzi ofTransportarion Enginuring, 132, No. 8: 654-662. · · RobertsOn, H. D. and D. C. Nelson. Manual ofTrr:nsportarion Enginuring Studi~, W:uhiogton, DC: ITE, 1994. R.odegerdts, L. B. c!. al. NCHRP Report 572: Roundabouts in the United States. Washington, DC: National Coop erative Highway Resea.rch Program, 2007.

Roe.ss, R. P., E. S. Prassas and W. R. McShane, Traffic Enginming, 3rd eci Upper Saddle River, NJ: Pearson Pren tice Hall, 2004. Sm:~.d.i, A. B. "Advantages of Using lnnova~ve Traffic Data Collection Techniques." Applkations ofAduanud TechnobJgi11 in

Transportation. The Ninth Inrernational Conference. Re5ron, VA; ASCE Publications, 2006.

Tarlw, A. and R. Lyles. ~/opmmt ofa Portabk Vuleo Dmcrion Syffnn for Caunring Turning Vthicla tU lntmmions.

Purd~

Universicy Repon FHWNTNIJTRP-2001/18. hnp://doc:s.lib.pwdue,eduljap/62/. West Lafayette, IN: Purdue Universicy, 2002. Transporration R=ach Board. "Data, Survey Methods, Traffic Monitoring. and A.s.set Mmagernent.• Transpor111rion &rarch &cord:jourruzi ofthe Transporrmion &much Board 1993 (2007). Zou, N. and J. Wang. "A Video-Based Method for Evaluacing Traffic Dara from De rectors." App/.Wltions ofAdvanced Ttchnologi6t in Transportation. The Ninth International Conference. Reston, VA; ASCE Publications (2006): 232-237.

Volume Studies o ..,. 5

Chapter 5

.................................................................................... j

Spot Speed Studies OrigiluJBy: H. DouglAs Robertson, Ph.D., P.E. EJiutlBy: BII#Um ] . Schroetkr, Ph,.D. 1.0

2.0

INTRODUCT19N·

77

1.1 Safety

78

1.2 Time-Mean Speed VetSus Space-Mean Speed

78

1.3 General Speed Measurement Concepts INDIVIDUAL VEHICLE SELECTION METHOD

79 79

2. 1 Introduction

79


79

2.3 Data Collection Procedures 3.0 . All-VEHICLE SAMPLING .

84

86

3.1 Introduction

86


86


87

4 .0 DATA REDUCTION AND ANALYSIS

4.1 Data Reduction and Display

88

88

4.2 Descriptive Statistics

91

4.3 lrtferential Statistics

92

S.O

SUMMARY

95

6.0

REFERENCES

95

1.0 INTRODUCTION Speed is an important measure for tta.ffic operations, because highway users relate speed to economics, safety, time, comfort and convc:n.iencc. Speed is a basic measure of tta.ffic perfOrmance. Thus spot speed data have a number of applications, including detennin.i.ng tta.ffic operation and control parameters, establishing highway design demencs, analyzing highway capacity, assessing highway safety, monitoring speed trends and measuring effectiveness of controls or programs. Spot speed studies are designed to measure speeds at specific locations under the traffic and envirorunental conditions pteVailing at the time of the srudy. This chapter is divided into two fundamentally d.i.lferent approaches to collecting vehicle speeds at spot locations. The first is the individwJ whick ~!mUm method where a subset of vehicles in the mHic stream are sampkd using predominatdy manual speed measurement techrUques. Altem2tivdy, the all-vehicle sampling method records (almost) aJJ whick spmls using automated in-road or roadside measurement equipment. While the first method is wgeted to short-t~ speed measurcmencs, the second is appropriate for system performance monitoring systems that rdy on. continuous estimation. lbis chapttc presents study design, methods and analysis techniques for both methods, following a discussion ofsome geoaa.l speed study concepts. ~nnt

C:noor4 <:t,•.4io<- •

,,

1.1 Safety As with all fidd studies, safety is the paramount consideration in conducting spot speed studies. The measurement of

speed involves having workers in the proximity of the roadway, whether it is ro install and recover in-road detection devices or to operate roadside data collection equipment. Workers must use care and vigilance at all times while: working near the roadway. Workers should park their vehicles off the traveled way. wear reflective vestS and act in a manncr that does not disuact motorists or influence their driving behavior. Workers should place and recover detection devices in the roadway under low-volume conditions. Standard lane closure procedures and warning devices should be used if extended time in the roadway is required. Study teams may need police assistance co direct traffic during the deployment and recovery of equipment.

1.2 Time-Mean Speed Versus Space-Mean Speed There are !Wo types of average speed measures that express the rate of movement or speed of a vehicle. Time-mean speed (TMS) refers to the basic llrirhmt!tic melln of speed collected at a spot location. It is simply calculated by summing the: speeds of all individual vehicles crossing over a point and dividing by the: number of observations. It is the intuitive measure of speed that is collected, for example, by an observer with a laser speed gun at the side of the road. Space-mean speed (SMS) is the averagt! speed of all vehicles occupying a segment and is defined as the total distance traveled over the total cravd time for all vehicles. It is the speed measure used in traffic flow theory speed-flow-density rdationships. Speeds can principally be obtained at a spot location (using, for example, a laser speed gun) or can be inferred from the travel time over a known distance. Both SMS and TMS can be calculated from either measurement approach, so it is imporrant to understand the distinction. Exhibit 5-l presents all four equations.

.

.

-

Approach

-

•Lm

. . ~

~

Typical Eqaipment

.. .

.

'~.

~~i~~,' \~'t~~~~~{>3f.· .' · " lF.lfli --~ ._:/~ !_'i_'1~.·-.-id.jt!.· -~.-, . ~- ·

<.. ; .. TMS .Jll

Variabla

Spot Location

Laser!R.adar, Short Speed Trap

v1 • Individual Speed n •II of vehicles

Segment

Stopwatch, GPS, Overhead Video

~ • Vcb. Travel Time d1 .. Distance Traveled n • II of vehicles L - Sl!&,ment ~

TMS

SMS 2)

Ll'.o(v,) n

-

Spot -

.I:r. n(!f)

=

1 = y----r iiL~•• v-

_(4)

_13 TMSseg

SMS.rpoc

n

.I:r•• cd,) nL SMS,.g = .I:f••(tt) = tr. o(t,)

Exhibit 5-1 shows four potencial ways to calculate speed, labeled (1) through (4). For the purpose of chis chapter, (l) is the m~t common application, as individual spot speed measurements are averaged to obtain TMS. However, for some applications, including freeway and arterial segment operations, SMS is needed and equation (4) is appropriate. In the absence of segment speed or cravd time data, equation (2) can be used to estimare SMS from spot speed measurements (Roess, McShane and Prassas, 2004). The remaining equation (3) is rarely used, since it is unrypical to use TMS in a ,segment evaluation. Since chis chapter focilses on spot speeds, the resulting measure is typically the spot time-mean speed of traffic (Equation 1). This measure is appropriate: to sample approach speeds to a signalized intersection, to measure the speeds in a horizontal curve, or to quantify che dfect of a traffic-al.ming treatment on vehicle speeds at that location. However, it is not an appropriate measure to use in segrnenr-based speed evaluation related to segment travel times (Chapter 9)

. 1.3 General Speed Measurement Concepts

i Spot speed data are collected by one of two general approaches: direct and indirect measurements. D irect m~1uure

. mmts of speed are made using permanent or handheld technology that apply the Doppler principle via, for example 'radar or laser measurement technology. If a direct measurement is used, the resulting metric is typically the TMS, although SMS can be esiimated as discussed above. Indirect measurements of spot speeds actually calcula.te speed from time measurements of a vehicle traveling a known (short) distance, such as the distance between two closely spaced magnetic induccance loops. By m.inimizing the length of the spud trap the difference between the indire ctly measured space speeds and the desired spot speeds is negligible. A!; discussed in Chapter 9, speeds can also be inferred from vehicle travel times ove.r a longer segment. With increasing segment length the difference between TMS and SMS can be significant as discussed in Roess, Prassas and McSh~e (2004) or ocher sources.

Two basic methods of data collection are rhe individual vehick sekction method and the a/1-vehick sampling method. Both merhods can use direct measurement or indirect measurement. Each is discussed below in terms of study pur" pose, location, time period, personnel and equipment, 'sample size and field procedures ..

2.0 INDIVIDUAL VEHICLE SELECTION METHOD 2.1 Introduction Analysts use the individual vehicle selection merhod when the study purpose can be satisfied with a relatively sm.all sample of spot speeds taken over relatively shorr time periods. The objectives of such studies are usually vecy specinc and limited in scope to certain ·types oflocacions, time periods and conditions. Examples of such applications include measuring rhe effectiveness of a TCD, spot checking the effect of speed enforcement, or establishing the location of a traffic sign.

2.2 Types of Studies 2.2.1 LocatWn, Tmu and Conditions ofthe Study The location, analysis time period and the roadway, traffic and weather con4!cions at/under which the study will be conducted are genec:illy determined by the study itsd£ The srudy's objective and scope 'dictate the specific location for collecting the data, the time of day and day of rhe week as well as the conditions under which rhe speed data are desired. If approach speeds to an intersection are the sample of interest, speed measurements should be taken upscreaJ'll. on the app~oach just before the point that traffic begins to decelera~e for a possible stop at the intersection. If speeds are being sampled as part of a nighttime accident study, the data sh9uld be collected during the hours o f darkness. If wet pavement is a factor of interest in the study, analyses should measure speeds when ir is raining. If the srudy rea.XD needs free-flow speeds, they should conduct the study during off-peak time periods. 2.2.2 Persrmnel atul LJui;[munt The individual vehicle sdection method may use a manual speed trap, but is generally carried out using the Doppl .er principle, direct measurement technique wirh radar, laser, or infrared technologies. A well-positioned overhead video camera can also be used to manually estimate speeds from known djstances on the video.

2.2.2.1 Doppler Measurements Tbe mosc commonly used device for directly measuring individual vehicle speeds is a radar or laser gun. This device may be handhdd, mounted in a vehicle, or mounted on a tripod; it is also commonly used by law-enforcemem age.t:"lcies. Exhibit 5-2 show an example laser speed measurement devices used in practice.

Spot Speed Studies • 7 'S

The general operating principle of both radar and laser devices is the Doppler Principle. The speed device transmits a beam of highfrequency wo.vcs toward a moving target vehicle. These waves reSect off the vehicle back to the unit. The change in frequency b<:tween the transmitted and reSected waves is proportional to the speed of the target vehicle, relative to the speed of the radar unit. lf the unit is fixed (not moving), it does not matter if the target vehicle is traveling toward the unit or away from it.

.!1 l

·~·

:l

if,,

.,,.

The srimtific difftrme~ b<:rween laser and radar technologies is telated to the wavelength of the wave beam, with radar wave$ gen· erally having a higher wo.velength and thus lower frequency. The strength of a radar signal decreases with increasing distances. Effective rangc;s vary from a few hundred feet to rwo miles (approxi· mately 100m to 3 km). The tranSmitted beam b<:comes wider as it uavds out from the unit, so it may cover more than one traffic lane. Most units allow for adjustments in range and beam width. Check the manufacturer's specilications when selecting the radar device to ensure it will meet your srudy requirements. Laser b<:ams :.• . ·~~ are more concentrated with a shorter wavelength and higher fre. So CiS ock b m/Dan'd LoiscU 1 quency. They generally have a greater range than radar, allow the uroe: t P oto.co e amlyst to directly aim ar a selected vehicle and arc less likely to pick up adjacent vehicles. Most modem bser devices record both the vehicle speed and relative distance to the observer, which can be useful in some applications to filrer vehicle speeds for a certain segment. . I

;

L 'IJ,

I

I ;

t

The practWJ applklllion of the two measurement technologies also di£fers. A radar measwement device generally transmits a continuous spectrum of wo.vcs thac can re8ect off multiple objeets. In fact, radar units used in police cruisen commonly give multiple speed estimates of vehicles directly in front of and behind the cruiser, as well as the speed of the fastest vehicle. The scatter resulting from the continuous radar signal can b<: picked up by in-vehicle radar detectors, and analysts must recognize the bias associated with driven slowing in advance of a radar measwement station. Many of these detectors are visible to the observer as the vehicle passes, which allows chose observations to be discounted.

' .

A laser beam is typically not continuous, but is triggered by the analyse (or police officer) after focusing on a vehicle. It therefore is le.ss likely to be picked up by drivers in advance, even if an in-vehicle laser detectOr is installed. Due co the ooncenuated ll2.turc of the laser beam, there is less signal scatter that could b<: picked up by upStream drivers approaching the measurement station. A laser device is therefore more likely to result in an unbiased estimate of speed. The accwacy of laser and radar units is affected by two errors: round off error and cosine angle error. Radar units typically display the measwed speed in digital form rounded down to the nearest whole unit of speed. For a.a.mple, a reading of 55 mph (88.5 km/h) would mean this estimate was actually b<:rwcen 55 and 56 mph (88.5 and 90.1 km/h). Laser unirs typically provide speeds with greater accuracy to one decinu1 point, but differences may still aist b<:rwccn manufacturers. The cosine angie error occurs b<:cause the angle of incidence of the beam to me navel direction of the target vehicle produces a reading on the unit that is less than the acrual speed. As shown in Exhibit 5-3, the measwement is a function of me ~inc of the incidence angle. While in law enforcement this error provides a margin in favor of the target vehicle, other applica?ons of spot speed data may require a concction to the reading to ensure accuracy.

iI

J\l I.


.P lan

Radar meter

v l.

Vsin a.

~~

·b.---·vcos a

Vector diagra-m

Radar measurement= Vcos a Sou=: Taylor, M.A. P. and Young, W., Traffic AnaJytis: Nnu T«hna/4gy anti Nnu So/JIIitnu, Hargrecn Publishing Company, North Mdbourne, Victoria., Australia, 1988, p. 155.

Eth.ibit 5-4 illu.suata the effect of the a»ine angle error on true speed. Because of the absolute nature of thae two error sources, the relative error decreasa as speed inaeasa. Some units have a built-in correction for angle error based on praet angla of incidence (Taylor and Young. 1988).

1.0 mpb. equals 1.6 km/b. Source: Lunenfdd and McDade, 1983.

A radar or laser unit is easily operated by one person. Many modern units have built-in data storage using USB or Hash media. Alcernatively, operators can write down the digical readings displayed on the unit or can record the readings verbally on a voice recorder for transfer Iacer co paper or computer for analysis. Even ifbuilt-in data storage is available, it is useful to take notes on paper or a voice recorder, since it is otherwise difficult co reconstruct erroneous measurements from the elecuonic dawet. For c::xample, a study focused on free-flow speed measurement may inadvertently include some vehicles in foUowing mode chat need co be excluded from the analysis. If traffic is heavy or the sampling strategy is complex, rwo observers may be needed: one to call out the readings for the vehicles of interest and one to record the speeds on an appropriate form.

2.2.2.2 Manual Spud Traps Spot speeds may be estimated by manually measuring the time ir rakes a vehicle to travel between rwo defined points on the roadway a known distance apart using a stopwatch. This technique, commonly referred to as a speed crap, is a "low-technology" approach ro speed data collection. With the ready availability of radar and laser units, as described in the preceding section, and the automatic speed craps discussed in the all-vehicle sampling section, manual speed traps arc seldom used. 2.2.2.3 Vuko Principles of time-lapse phorography have historically been used by researchers in performing speed srudies. Today, video recordings can be used ro obtain time stamps ofvehicles passi.ng known points, similar to using a manual speed trap. The usefulness of this method for research lies in the c::xpanded amount of information recorded on the video and the ability to analyze and reanalyze speed along with other types of data. The technique applies primarily when analysts require tocal informacion about the scene over relatively shore periods of time (Taylor et al., 1989). Effective use of video requires a video camcorder equipped with a character generator or similar device for recording a time swnp, and a variable-speed or frame-step playback unic. Without these fearures, the video record may not proVide the ability co extract reliable time swnps foe precise and accurate estimates of speeds. As with all video-based studies, an overhead vanrage point drastically increases the analysts' ability to reliably estimate speeds.

l ·!

I'

;}

i!

2.2.3 Sample-Size Jlequirmlenl.t Analyses who use individual vehicle selection muse collect a sufficient number of spot speed observations co allow statistical analysis of the srudy results. A minimum sample size can be determined for a desired degree of statistical accuracy by using Equation 5-1 to calculate the nu.mber of speeds to be measured, when mean speed is the statistic of interest. ·

N = (s·~f

if

where

'I ':;''

N = minimum number of measured speeds

<

K ., constant corresponding to the desired confidence level

~ ~

E "' permitted error or tolerance in the average speed estimate, mph

S

Equation 5-1

= estimated sample standard deviation, mph

<

.

1 '

LI' 'i

''~ .l'

Analyses can estimateS for this equation from previous speed srudies under similar conditions of srudy or from speed monitoring data at a nearby location. In the absence of these data, Exhibit 5-5 presents estimated values of average standard deviations (S) as a function of traffic area and highway rype.

q ''

!•

!; !'

11

'I

!I II

I!

.! ; I I

I· I

!I I I


- -·~----

-~.

..._.... ·.·

; ~l*~~.w,.._.,~ •" • • • e .

_

. . .. .... ·--~-. ~.

,.

--~

-~~-D-~~~-·:

.

·~-~-~~~~~~H~~~·.:~~-: :~

Average Standard Deviation Traffic Azea.s Rural lmermediate

Highway Type

km/h

Two-lane

5.3

8.5

Four-lane

4.2

6.8

Two-lane

5.3

8.5

Four-Jane

5.3 4.8

8.5

4.9 5.0

7.9

Two-lane Urban

mph

Four-lane Rounded value:

7.7 8.0

Source: Box and Oppenlander, 1976, p.80.

For the greatest accuracy, analysts can conduct the study, calculate the acrual standard deviation of the data and check to see if the sample size was adequate. If not, additional data would have to be collected under the same conditions as the first study. Another technique is to use a calculator to update continuously a running toea!, average speed and standard deviation. When the standard deviation becomes stable, an adequate sample size has been obtained. The confidence level expressed by constant Kis the probability that the difference becween the calculated mean speed from the sample and the true average speed at the study location is less than the permitted error. This concept is discussed in greater detail in Appendix C. ~e corresponding constant K values for selected confidence levels are shown in _Exhibit 5-6 and are valid for any sample size:s greater than I 00 measurements. For smaller sample sizes, the cot'~ re-sponding K value can be obtained from the t·di.mibution table given in Appendix C (Exhibit C-12).

.

Source: Box and Oppenlander, 1976, p.Bl.

The permitted error, E, reflects the precision required in estimating the mean speed. This parameter is an absoluc:c tol~rance and is expressed as plus and minus a specified value. Typical permitted errors range from :1: 1.0 to :t 5.0 mp.b (:1: 1.6 to :1: 8.0 kmlh). For example, the analyst may be interested in the necessary sample size for a speed study on a rural tw(}-lane higbwa::Y at a 95 percent confidence levd and a permitted error of :t 1.0 mph (1.6 km/h). In other words, the analyst want$ to show with 95 percent certainty that the true population average speed is within plus or minus 1 mph (1.6 kmfb.) of the observed .sample mean. The estimate standard deviation, S, from Exhibit 5-5 is 5.3 mph (8.5 kmlh) and th.~ constant Kis L96 per Exhibit S-6 (for .samples greater than 100 measurements). Applying Equation 5-1, the neces-. sary .sample size is: · ·- ·

Spot Speed Studies • 8S!I

2

N

• = ( 5.3 • 1.96) - =108 observanons 1

Once the sample of speeds has been collected, the anal~t should recompute the sample size requirement using the uue field-observed standard deviation, to ensure the sample is still sufficient. If the statistic of interest is some percentile speed, such as the 85th percentile, Equation 5-2 is appropriate for determining the sample size required. S 2K2 (2 + U2 ) N=

2£2

Equation 5-2

where N, S, K and E are as defined for Equation 5-1 and U is the constant corresponding to the desired percentile speed. Exhibit 5-7 presents constants corresponding to percentile speeds. As a general rule, the' minimum sample size should never be less than 30 measured spot speeds to satisfy underlying assumptions of the statistical disuibutions in Equations 5-5 and 5-6.

2.3 Data Collection Procedures 2.3.1 Radar!Laser Spot Speed Studies Successful spot speed data collection depends on how wdl two aspeccs of the srudy are conducted. The fi.rst issue involves the configuration of the site for data collection, and the second pertains to how individual vehicles are selected for measurement. Poor ueacment of these issues can adversely affect the accuracy of the measurements and/or bias the results.

2.3.1.1 Layout ofSite The positioning of the radar/laser unit is consuained by three considerations: • capabilities of the radar unit • minimizing the angle of incidence

I I

! .1

l

j

l

!; I

\I f,l

• concealing the unit &om the view of motorists The capabilities of radar and laser units vary considerably. Units muse be set up and operared in accordance with che manufacturer's specifications and instructions. As discussed earlier, the larger the angle of incidence between the radar beam and the direction of uavd of the tatget vehicle, the larger the cosine error. An angie ofless than 15 degrees keeps the error under 2 mph (3.2 km/b), but depending on the specified srudy tolerance, a smaller angle may be critical. Concealment of the radar unit and operators will prevent motorist disuaction (a safety concern) and reaction (a potential source of bias). The equipment and crew may be concealed by veg(!:ation or roadside structures, or they may simply be located out of view of target vehicles. Roadway maintenance vehicles, which motorists expect to see along the roadside, may be used ro conceal the unit and crew. This approach will not work in states where the police also use maintenance vehicles to conceal radar units. Concealment will not defeat radar detectors. Exhibit 5-8 shows two potential ways to posicion the radar unit on an overhead bridge or on the side of the road. Depending 84 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

. on the spatial position of the observer the angle of incidence and associated cosine error may occur in the horizon~ ; or vertical direction.

_·_____ --

~~~~ I F--~------- - · ~ ------ - ---

I

~~

Source: Speed Moniwring

Proctdurtd M1111U11£ FHWA. 1980.

2.3.1.2 &kaWn ofTarg~t Vthicks The guiding principle is to randomly select target vehicles chat represent the population of vehicles under study. Thus analysts must clearly define the study population (that is, free-llow vehicles, large trucks, platoon leaders, all vehicles, c;tc.). Once the population is defined, study teams can adopt a selection strategy to provide a random sample of that population. With all strategies, teams must exercise caution to avoid selecting vehicles with radar detectors if a radar measurement device is used. -. . Except for studies conducted under low-volume conditions, it will be d.iflicult to obtain a measurement of every vehicle. The availability of built-in electronic storage in modern data collection units has made sampling easier, but a sampling .scheme is likdy still necessary. Foe radar measurements, vehicles may mask other vehicles from the radar beam. For laser measurements, the line ofsight may be occluded or otherwise visually obstructed by traffic or roadside obj~ets. Large vehicles return a stronger radar signal than do small vehicles, thus overriding the smaller vehicle speed. Vehicles from the opposite direction or in a different lane from that under study may override the measurement of the target vehicle. These concerns ace lower with laser units. Observers have a natural tendency to record those vehicles chat "stand out" in' some way, such as fast vehicles, slow vehicles, trucks, or platoon leaders. A procedure that controls for this bias is co select every third, fifth, 1Oth, or nth vehicle. Take care, however, that the nth vehicle is not controlled by some external effect. Foe example, every lOth vehicle on 3!1 arterial Street could be a platoon leader if a coordinated mffic signal system establishes a regular pattern of traffic Bow.

2.3.1.3 Docummtation The final layout of the data collection site should be fully described in any report of speed data. Observers should make an accurate sketch of the site showing the number of lanes, the position of the measurement unit and the x., y and z dimensions as shown in Exhibit 5-8. The dimensions permit calculation of the angle of incidence so that a cosine error correction may be applied, ifdesired. Observers should record the start time, end time, any downtime and the conditions prevailing during the study. Photographs of the layout may also prove useful. 2.3.1.4 UdibratWn i The radar and laser manufacturer's recommended calibration tests should be made before the start and again at the end of data collection. The results should be included in any report of speed data. It is also wise to make an initial test in the office to ensure the radar unit is operable before traveling to the site. 2.3.2 Mznlllll SJHed Trap1 atul V"uleo As indicated in a preceding section, analysts seldom use manual speed traps, and video is generally less efficient chan direct speed measurement with a laser or radar unit..If video is used, general set-up considerations are consistent with the d.iscuss!on offered in Chapter 4. lf video is co be used for speed studies, it is further necessary to establish known distances in the video field of view to use for reference in measurement. The dashed lane striping provid~ ~ gooc,i ref.ccence in many cases, as do closely-spaced roadside fearures such as barriers, signs,,or trees. When no existing features arc usable, the analyst can retrofit the site using temporary markings or roadside cones, but care should be taken that the installation does not affect driver behavior. Spot Speed Studies • 85

For the study, the analyse measures the time it takes a vehicle co cravd the known distance between two points. This is most simply done using a stopwatch, bur can further be performed using automated equipment including handheld eleccronic count boards or a laptop computer. More information about rhe procedures for conducting spot speed studies using manual speed reaps or video is given in Roess, Prassas and McShane (2004).

3.0 ALL-VEHICLE SAMPLING 3.1 Introduction Analysts use this method when the purpose of the study requires or can be accommodated by measuring the spot speeds of all vehicles passing a point for a sample of time periods. The objectives of such studies are usually more general than studies using the individual vehicle selection method, but may also be specific and somewhat limited in scope with respect to certain types of locations, time periods .and conditions. Examples of such applications include monitoring speed trends, assessing highway safety, or establishing speed limits.

•' !

!,,!


:i .

3.2.1 Location, Time, and Conditions ofthe Study As with the individual vehicle selection, selecting the spot to take speed measurements; the lime period over which to collect the data; and the roadway, traffic and weather conditions under which the spot speed study will be conducted are generally derived by the objective and scope of the study. If average speeds on a section of freeway are the sample of interest, speed measurements should be taken at the midpoint of a typical section. If speeds are being sampled to determine the 85th percentile speed for use in establishing a speed limit, the data should be collected over one or more 24-hour periods on several rypical days. If free-How versus platoon speeds are key to addressing a research question, the collection technique must be capable of distinguishing these two groups from all other vehicles.

'~

I

I

'I. j .

3.2.2 Personnel and Equipment The all-vehicle sampling method uses automatic data collection equipment such as sensors placed in the travel lanes that serve as input devices to recorders located at the site. Data from the recorders may be downloaded with a computer in the field or direccly by modem or fiber connection to a computer in the office. Significant advances have been made in the use of both permanent and portable sensors, recorders and computers. Computers are capable of sensing different types of vehicles, recording travel times over traps, calculating speeds, classifying vehicle.s and storing large quamicies of data. These advances permit analysts to study large samples of vehicles over long time periods.

.

I, I . I .

·~ ~

i i

~I j

l

q

.·l I

3.2.2.1 Pmonnel One of the advantages of automatic speed data collection is chat personnel are needed only during installation and recovery of the data collection equipment. Once installed, the equipment can operate unattended for several days or may be installed permanently. Agencies normally use two or three people to deploy and recover sensors and recorders. Safety dictate.s that one person never attempt this task alone. During installation and recovery, the crew and its vehicle should be highly visible co motorists. The crew should use cones and warning signs. One person should watch for traffic anytime a crew member is in the roadway. Permanent, long-rerm installations are typically done by professional work crews and involve lane closures to ensure safety. 3.2.2.2 Equipment fqr Alltcmaud Speed Mearuremmts The most commonly used devices for measuring speed are in-road sensors in the form of pneumatic tubes, standard induction loops, or point loops. These devices are normally deployed in paits. Two measurement units are placed a short, measured distance apart to form a speed crap that measures the time it takes the vehicle to travel from one detector co the next. Agencies may place these sensors in saw cuts or bore holes in the pavement, sealed for protection from the environment, in the same manner that sensors are installed on approaches to signalized intersections. The devices used to collect automated speed data are generally the same that were introduced in Chapter 4 for automated volume studies. The technologies include permanent in-pavemem installations, and temporary on-road measurement technology. For most technology, the manufacrurer specifies installation requirements, including the distance berween' two sensors. These should be followed closely to ensure that the unit is well-calibrated. To measure

f ;

l


speeds on multilane roads, study teams must place sensors with separate inputS to the recorder in each. lane. Exhibit

5-9shows examples of installations for two automatic speed measurement installations. i

tT~:l-~'"l_q1lJ.111l_tf~f-~~·

b) In-Road Magnetic lnductan(e Sensing Station

3.2.3 SAmpk-Siu Requirmumts With the all-veh.icle sampling method, obtaining an adequate sample is sddom a problem because dep loyments are usually made for at least a 24-hour period. Sample-size requirements may be estimate4 in the same m anner as described earlier for the individual vehicle selection method. Analysts seeking samples of certain rypes of vehicles maY need to collect a large total sample to ensure the rypes of interest arc adequately represented. Since the automatic daca collection system captures every vehicle (except in the case of a malfunction), the important sampling issue analysts need to address is the time period when the data collected will be represcnrativc of the desired study conditions.

3.3 Data Collection Procedures 3.3.1 FielJ Procedures for ~tomatic Spot Speed Stutlies Successful spot speed srudies using automatic data collection equipment depend on the operational reliability of the sensors and recorders, the physical installation of the sensors and lead wiring and the calibration and quality ronrrol measures employed. External facrors that can affect data collection include the temperature, weather, level of rra.f· fie volume, mix of vehicle rypes and the environment (such as din, dust, or debris at the data collection sire). SruclY procedures begin in the office with coordination preparations and qper:uional checks of equipment. In rhe fidd, r,b..e principal tasks are deployment, calibration, recovery and documentation. 3.3.1.1 Ojfiu ~araJions . The first task is to coordinate all data collection activities with appropriate state and local officials, including cran. sponacion, rraflic and law enforcement agencies. These agencies need to be informed, and analystS need to cnst~tt the::h studies do not interfere with ongoing activities. The second cask is to brief the 6dd team on the data collection pla-n to ensure data arc collected at the proper place in the desired manner for the required time period. The rhird wk iovolves the team's preparations. All cools, supplies and equipment should be assembled and inspected. Checldisa m.,.._Y prove helpful. and the general checklist in Exhibit 1-1 may be a good starting point. Each piece of equipmcm mou.J. d be tested co see it is functioning properly. This precaution can prevent cosdy trips back to the office to replace missica.g supplies or malfunctioning equipment. · - ·· Spot Speed Studies • •

7

f. n

cl Sl

J

3.3.1.2 DqJ/oymmt Safery is the 6rst consideration during deployment. It is preferable to close each lane to traffic while work is under way. The sensors should be prepared on the roadside to minimize: the time each lane is closed. Workers then place each set of loops, rubes, or other equipment prescribed in the data collection plan and in accordance with manufacturer specifications. The proper spacing ofsensors determines the accuracy of data and the crew must be careful. to install them as dmcted. After placing, the lead wires are connected to the recorder and ihe sensors are checked for proper functioning. After any needed repairs arc made, the crew can secure the sensors to the pavement. The field crew should have some leeway to select the exact position to deploy the data collection system, to avoid broken pavement and to locate the recorder near some fixed object to which it can be secured. 3.3.1.3 QzljbrlltWn With the sensors and recorders in place, the next step is to check the accuracy of the equipment in measuring the counts and speeds of the traffic stream. A calibrated radar or laser gun is used to measure vehicle speeds, which arc compared to the speeds of those same vehicles monitored by the data collection equipment. If necessary the crew can adjust the recorder until the speeds are within a :t 1.0 mph (1.6 kmlh) tolerance. For temporary, in-road installations it is ~dvisable to check the functions and accuracy of the equipment at lease once during every 24-hour data collection period. Permanent sensor installations should als() be checked at regular intervals. 3.3.1.4 &cowry After the data collection period, the recorded data are cliecked for accuracy by computing ~e ratio of usable vehicle data to total vehicle data. To compute this ratio, compare the total vehicle count to the usable vehicle count. A usable vehicle is a vehicle with both a length and a speed. Sometimes a vehicle will be sensed by only one of the loops in the pair. This occurs when the vehicle path does not cros:s both sensors or when one of the sensors &il.s. Such vehicles are stored in the total vehicle count but not in the usable vehicle count. The ratio of usable to total should exceed 0.75. If this criterion is not met, the team should collect data during another period. · Crews recover data collection equipment by reversing the process they used to deploy it. Again, safety should be the principal concern. If the equipment is to be deployed at another site, the data should be transferred from the recorder either at the site or via modem or fiber connection co the office. If the site is being monitored for an extended period of time, the data should be •dumped" periodically.

..

3.3.1.5 Docummt4lion The fuu.llayout of the data collection sice should be fully described in any report of speed data. The crew should make an accurate sketch of the site, showing the number of lanes, the position of the sensors and the location of the recorders. The crew should record the start time, end time, any downtime and the conditions prevailing during the study. Equipment malfunctions and repairs and the results of calibration and accuracy checks should be recorded. .Photographs of the layout may also prove useful.

!

\

I,

'\I.


I'· !II

'.

l \

I: )

! \

t~

i!, !

H 1[i~ ~

A typical spot speed nudy analysis has three parts. Data rcduaion is the first pact; that is simply the arrangement of the measured speeds, or "raw data, inro a convenient tabular or gr.1phical form. The second part is the calculation and presentation of~pcive statistics which illustrate the collection ofspeed data by means of a few representative values or variables. The third part of a cypical analysis is statistical i.nfuenec, which permits the devdopment ofstatistical estima.tes and the testing ofStatistical hypotheses. Appendix C presents detailed information on field srudy analyses and uses many examples from spot speed data. Therefore. the disrussion in this section is brief, and the reader is referred to Appendix C for greater detail.

I

~ I

ij

l!

~

\

1

,,

4.1 Data Reduction and Display Both the individUal vehicle selection method and the all-vehicle sampling method produce measured values of individual vehicle speed. The speeds are usually arrayed in tabular form in the sequence they arc recorded. To illustrate, ·Exhibit 5-10 contains data from a speed srudy conducted in rural Vllginia that used the individual vehicle sdeccion method. Exhibit 5-11 shows a frequency distribution table for the data in which the speeds have been grouped into 16 classes with an interval of2 mph (3.2 km/h). Oass frequencies, the percentage of observations in each class and the cumulative percentage of all observations per class are also shown. 88 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

1.0 mph • 1.6 kmlh

Source: Gar!M:r and Hod, 2002.

Spot Speed Studies • 89

.

IL

.,~~ :

~i

i

I i;

l!; ;[ ~

! ~ i

ii >I

f.

. .

I'

!

t

:

!

11 i

I

1.0 mph a 1.6 km/h Souro:: Garber and Hod, 2002

Analysts can display speed data in several graphical forms. Exhibit 5-12 is a histogram of the classified frequency data in Exhibit 5-11. A frequency distribution curve may be plotted from the percentage of observations column, as shown in Exhibit 5-13. Similarly, analysts can form the S-shaped cumulative distribution curves shown in Exhibit 5-14 by plotting the cumulative percentages of all observations versus speed. Exhibit 5-13 also shows the pace of the distribution, which is defined as a 10 mph (16 kmlh) range that contains the highest frequency of measUied speeds in the observed distribution. Exhibit 5-14 further shows the median speed, which is defined as the midpoint of the speed distribution, and the 85th percentile speed of the distribution. The use of these metrics is discussed in the non section on descriptive statistics.

... _.,."' ~

1:

J" 8

.. l

0

,.,

,., VtHd•S~

..

......

"'

1.0 mph • 1.6 km/h

Data from automatic speed data recorders can be Source: Garber and Hod, 2002 presented in tabular form. The all-vehicle sampling method usually yields a large quantity of data, so that computer software is necessary for reduction and analysis. The equipment vendor will typically provide :m interface for downloading the data through a modem or Internet connection {Exhlbit 5-15). 90 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

I'

40

~g 000 20

00

0

0

OS

j

• 0

€

i

l

00

"'

:

60

:

SO

0

o..:.,..... ,.•• I

....

t .. ~

.

..

·~~----.--.~.~--~--.~.~--~--~~--_:~~.~.v.ru~

SJioled

c~l

1.0 mph • 1.6 km/h Source: Garber and Hoe!, 2002

•49m. .

i

)O

:

lO

: 0

I

!

00

•

0

·~~---L--~.-~--~.t..-~--~~;---L-~~

1.0 mph • 1.6 km/h Source: Garber and Hod, 2002

4.2 Descriptive Statistics Several variables chat describe a collection of speed observations may be calculated from both the raw and classified data. The most common variables include the mean, mode, median/ standard deviation, pace and percen tile. Some of these variables may be obtained directly from the frequency and cumulative distribution curves as shown in Exhibi cs 5-13 and 5-14. The calculations may also be made manually or through the use of computer software. Most of th.e software agencies use with data recorders is capable of producing these variables. The mean is a common measure co describe the central tendency of a distribution, but may be skewed if some very high or low speeds are part of the data sec. The median speed is typically less susceptible to outlier observations of very high or low spee4s and is often a better measure chan the average speed (unless oudiers are removed). The mode describes the most frequently observed speed, and is less useful, unless speeds ~e first aggregated into bins. The pace is defined as the 10-mph (approximatdy 16 km!h) range that contains the greatest percentage of observa.tions. It is a term unique to the fidd of traffic engineering and should not be confused with, for example, the wallting pace of a pedestrian, which is the inverse of walking speed. The standard dc:vUrion describes the dispersion of the speed data set. A small standard deviation means that speeds ar~ tightly grouped around the mean, while a large =dud devUtion susgests a wider distribution. The 85th percentile lpCCd is a commonly used statistic for design applications. Ic is often interpreted as the fustest spccxl by reasonable drivers (Roes. PraQS and McShane, 2004). Similarly, the 15th pcn=tile speed is the slowest speed by =sonable driversfor desipl pur-· poses. Both measures e:xclude =me and ouclier observations char would ~tin impractical design considerations. Mor~ details oo the calculations of these measures are given in Appendix C and in Roess, Prassas and McShane, 2004. Spot Speed Studies • 9•

4.3 Inferential Statistics AnalystS can draw inferences from testing these variables statistically. For example, analystS can determine if significant differences exist in mean spot speeds for different traffic or roadway conditions. They can also determine if the speed distributions for data samples from rwo different locations or time periods are the same or different. Appendix C provides additional details on statistical tes!S and describes how to perform the necessary Calculations. A standard application of a spot speed study is to test an increase or decrease in speed resulting from some uearment in a before-and-after test. The statistical inference procedure is as follows: 1. Spot speed studies are conducted before and after the change is made. Conditions for both studies should be as similar as possible.

2. The before and after means (i ~and i...) and standard deviations (s~ and s.p) are calculated independently. The before and after sample sizes Should be as calculated earlier in this chapter. The following test should only be used for sample sizes greater than 30 observations for each period. 3. The difference in means, idlff is calculated from the two estimated means. Xdllf

= Xafter -

ibefore

Equation 5-3

4. The standard error of the combined sample estimate of the difference in means is calculated: 2

2

SE = ~~ + S~ntor• NA/tttr N"''"'' 5

Equation 5-4

5. The hypothesis in this statistical test is that the distribution has a zero mean-cluu is, there is no difference between mean speed before and mean speed after the change. Xafter - Xbefore

=0

Equation 5-5

6. The test sarisric is calculared &om the observed~ and hypotbesized (0) mean diffm:nce, the standard error of the combined sample estimate of tbe difference in means (SE) and tbe t-value obtained from Exhibit 5-16 fur the userspecified significance level (~ or confidence level (1- ~ and the before and after study degltes of fi=lom 0. The before and alter samples are sufficient to establish that che difftrcnce in spcds is significant if.

-

-0

Equation 5-6

~~t SE

idlff - 0

It is important to note that a before and after study that yidds S E < t does not establish that the difference .in speeds is not significant. Rather this result means only that the sample sizes (N+ and N ~ were not large enough to establish the statistical significance of the ~ d.i.B.i:rcnce. As with any study. the praaiCif implications should always be considered since even a statistically significant difference in speeds rnxt not be meaningful if the observed difference idlfl is small. This is espocially true as modem aummated means of collecting speed data read.ily result in very high sample sizes, which reduces che srandard error of the estimate and increases the test statistic in Equation 5-6.

l

lI

~.

Sample calculations for the procedure done on a spreadsheet arc shown in Exhibit 5-17. 92 • MANUAL OF TRANSPORTATION ENGINEERING STUOIES, 2ND EDmON

.

!1Z'litlt~r.lf! Ous Boundaries

:-

0 • ••

.. . , . .

.

.. '

-

-

O...limits

a ... midvalues (u)

Jlreqlltncics after (f)

Before (f,)

28-29

2B.5

0

0

.30-31

30.)

l

0

32-33

32.5

2

0

34-35

34.5

14

5

36-37

36.5

7

9

38-39

38.5

20

7

40-41

40.5

38

25

42..(3

41.5

29

34

#-45

44.5

35

44

~7

-46.5

15

29

48-49

48.5

12

12

50-51

50.5

9

13

52-53

52.5

4

II

>4-SS

54.5

0

7

0

2

~~~

27.5

29. 5

31.5

33.5

35.5 37.5

39.5

41.5

43.5 '

45. 5

47.5

49.5

Sl.5 53.5

'

55 ..5

56-57

56.5

57..5 To..J

:E

Man· _ X=

!

Scaadatd Deviuiom •

f ,u,

~~

$.IL

f,•.' - ~i~:]~J• ~

~ b
Xdi§

SE =

l

2

N.

N•

.:!L.+i

-0

z- ----s£ 'Dil£ertru:e ill Jpttds wilh 95% one-ailed con.&Wtce inaMI, +- aitic:ol (s • 0.05, df~ 200 . 1.64).

1&6

198

.8clore

After

42.35

44.67

4.50

4.68

2.32

0.4685

.. b57
Spot Speed Studies • 93

,.,.

!. •

'' '

. I

I

1., ·I

,, ..

!:· ,,ij '.\, 1·

1 ., l

~

.~

l (

".

f

[:

~~ !

l

i !

!

i

I

.l . I

.l

f

I .. • il

,,i;

MANUAL OF TRANSPORTATION ENGINEEruNo sruoiES. 2ND eomoN

S.OSUMMARY In this chapter various methods used co perform traffic speed studies at spot locations have been described. The chapter distinguished between individual vehicle sampling mechods, where data collectors randomly select a subset of che vehicle stream using mosdy handhel~ or manual data collection equipment or video, and all-vehicle sampling, where automated speed data collection equipment records speeds for the entire traffic stream. The chapter further discussed is5ues related to data reduction and analysis, including before-and-after statistical tests for speed studies. For further information and more details on speed studies, refer to Roess, Prassas and McShane (2004).

6.0 REFERENCES Banks, J. H. EvalUAtion ofPortable Data Colkction uchnologier: Final R.tport. San Diego, CA: University of California-Berkeley; Partners for Advanced Transit and Highways (PATH), 2008.

·

Bonsall, P. W., F. Gbari-Saremi, M. R. Ttght and N. W. Marier. "The Performance of Handheld Dara-Caprure Devices in Traffic and Transport Surveys." Traffic Engineering and ContrOl29, No. I. Box, P. C and J. C. Oppenlander. MantuJ! ofTraffo Enginming Studies, 4th ed. Washington, DC: Institute ofTransportation Engineers: p. 13. Dixon, M. P.-R. "Field Evaluation of Roundabout Turning Movement Estimation Procedures.• ASCE journal ofTransportatiofi Engineering 133, No.2. Reston, VA; American Society of Civil Engineers. Federal Highway Administration. Virginia State Route 7 Vuleo Detection System Pnfonnance Astessmmt. Washington, DC: Federal Highway Admioisuacion, 2006. Garber, N.J. and L A. Hoe!. Traffic & Highway Engineering. Pacific Grove, CA: Brooks/Cole, 2002. Gates, T. S. "Comparison of Portable Speed Measurement Devices. • Transportation Retearrh Record: journal ofthe Transportatit? ?'l &search Board 1870 (2004): 139- 145. Grecne-Roesel, R. D. "Effectiveneness of a Commercially Available Automated Pedestrian Counting Device in Urban Environments: Comparison with Manual Couors." Procmiin~ ofthe 87th AnnUAl Muting ofthe Transportation Retearrh Board. Washington, DC: Transportation Research Board, 2008. Hummer, J. S. ~Recent Superrueet lmplementacion and Research." Confemue Proem/in~ ofthe 3rd Urban Street Sympptium. Seattle, WA. Washington, DC: Transportation Research Board, 2007. Spot Speed Studies • 91' 5

JAMAR Technologies. Data Collection Equipmnrt. Retrieved December 29, 2008. JAMAR Technologies: ja.martech.com/index. hun I. Kyte, M. wd M. Lines. "Development ofVideo-Ba.sed and Other Automared Traffic Data Collection Mcthock " Traffic Trrulur ManU4L Retrieved April 20I 0. Nacionallostirutc for Advanced Transporw:ion Technology, Un~ity of Idaho, Moscow, lD. 2004. www.wcbs1.uidaho.edulnianlr=rcb1Finai_Repons/Kl.K203_N04-02.pdf. Laser Atlanta. Spud Mtasumnent Equipment. Retrieved December 29, 2008, Laser Adanta: www.laseradanta.com.

List, G. F. "Identifying Vehicle Trajectories and Turning Movements at Roundabouu.• 5th lnremational Symposium on Highway Capacity wd Quality of Service. Yokohama, Japan: Japan Society ofTraffic Engineers (2006): 449-458. McCarthy. J. TnzjJU J!naipis Toolbt»e.

~ashington,

DC: Federal Highway Administration, 2005.

Min-Tang, L F.-F. "Assignmenr of Seasonal Faaor Categories to Urban Coverage Count Stations Using a Fuzzy Decision Tree.• ASCE]ou1714i ofTrrwport41ion Enginming 132, No.8: 654--662. Robertson., H. D. Manlllli ofTransportation Enginming SnuUn. Washington. DC: Institute ofTran.sporration Engineers, 1994. Rodegeru, L B. NCHRP Repon 572: RowuiAb~IS in tht Umttd Statts. Washington, DC: Transponation Rc.scuch Board, 2007.

R.oes.t, R. er al. TnzjJU Enginuring, 3rd ed. Upper Saddle River, NJ: Pearson Prentice Hall, 2004. Smadi, A. B. "Advantages of Using Innovative Traffic Data Collection Techniques.· Appliurums ofAAwn£td T«hrw/4gia in Transporwion. The Ninth International Confewtce. Reston, VA:. ASCE Publications, 2006. '

i~

Tarlto, A. and R. Lyles. Dtwlopment oftJ PmtJble Yliko fRt«tiqn System for Counting Turning Vthicks at lntmectio'!l. Purdue

f

Taylor, M.A. P. and W. Young. Trttjfic Analysis: Ntu1 Ttclmo{ggy aNI. Ntu1 Solutions. North Mdbourne, Viaoria, Australia: Hargrctn Publishing (1988): 150--157.

University Repon FHWA/IN/jTRP-2001/18 doa.lib.purdue.edu/jup/62/. West Lafayette, IN: Purdue University, 2002.

Transponacion Resea.ch Board. "Data, Survey Methods, Traffic Monitoring, and Asset Ma!ugcment.• Tramport41ion RestJZrch &cord.: ]~1714i oftht Trrwportlltion kurch Board 1993 (2007). Zou, N. and J. Wang. "A Video-Based Method for Evaluating Traffic Data from Derea:ors." .Applicatioru ofAJvanctd Ttch":"login in Transporwion. The Ninth International Conference. Reston, VA:. ASCE Publications (2006): 232- 237. ;~ .;

.;! .:

i

·I.. · l

[ ~

~

i

d \


Chapt er 6

............ .... .. ......... ................................................. ... .. ..

Intersection and Driveway Studies OriginAl by: Joseph E. Hummer, Ph.D., P.E. Edited by: Christopher M. Cunningham, MCE, P.E. 1.0 INTRODUCTION

97

2.0 DELAY

98

2. 7 Equipment Needs

98


99


104

4.0 SATURATION FLOW AND LOST TIME

10S

4.1 Equipment Needs . 4.2 Personnel Training Requirements 4.3 Field Procedures and Analysis

·i

99

3.0 QUEUE LENGTH

5.0 GAPS AND GAP ACCEPTANCE

! 6.0

10S 105 106 109


109

5.2 Personnel Training Requirement

110


110

INTERSECTION SIGHT DISTANCE

112


112


113


113

7.0 SUMMARY

115

8.0 REfERENCES

115

1.0 INTRODUCTION ntersection and driveway studies are among !he most common studies in transportation engineering. In particular, many agencies routinely count turning mOvem.ents and study intersection delay. Other intersection and driveway studies include queue length, saturation Bow and lost time, gap and gap-acceptance, and intersection sight distance studies. Analysts perform lhese studies less often, but find !hem useful for special locations or when calibrating basic traffic Bow relationships. Analysts use the results of intersection and driveway stwiics to determine what kind of traffic control devices (TCDs) arc warranted and to determine intersection capacity,.uaffic signal timing, site development impacts, ~e speeds, driveway locations and oilier important parameters.

I

The basic data needs from intersection and driveway studies have changed very little lhrough the years.' However, the equipment used to conduct the studies has changed greatly in recent years and is likdy to keep changing. In this Intersection and Driveway Studies • 97

r

chapter, the basic procedure for each srudy is described together with the equipment that is most commonly used by agencies in North America for dday; queue length; saturation How and lost rime; gap and gap acceptance; and intersection sight distance srudies. The procedures for intersection turning movement counrs were derailed in Chapter 4. Analysrs should note this chapter contains methods for calculating u:Ufic measures developed for U.S.-bascd srudies. Different methods developed in other countries were extensive, and therefore were outside the scope of this chapter. Therefore, if outside the U.S., the analyst should consult specific guidance on procedures for intersection and driveway srudies.

·~ ; '!'

2.0 DELAY :i'

:} ~

)·

'

j

ij . I

~i

!':

lmersecrion delay srudies are very common. Intersection delay data have many US<:S, including measurement of the quality of traffic Bow and evaluation of the need for traffic signals. Analysts can estimate intersection delay with equations or simulation modds. However, the inpurs co the equations and models can be extensive, and the results arc only approximations of actual traffic operations. Therefore, field srudies of delay arc often used at operating interSections for greater accuracy or to validate theoretical delay prediction. One of the major problems with interSection delay srudies is the definicion of delay. There are many types of delay, and using terms casually can lead to error. The following are among the most useful terms describing delay at intersections (TRB, 2000, FHWA Trt1jjic Signa/ Timing Manual) .

' I!

• 7irM-in-queue tklay (TIQD) is rhe difference berween rhe time a vehicle joins the rear of a queue and the time the vehicle dears the imersection. ,., ' .; . !

• Ctmtrol tklay is the component of delay that results when a conuol signal causes a lane group to reduce speed or to stop; it is measured by comparison with the uncontrolled condition. lt is defined as the TIQD plus time losses due to deceleration from and acceleration to free-Row speed.

• Gro=tric Delay is the component of delay that results when geometric features cause users to reduce their speed in negotiating a facility.

•

d i

i

'I ·

·

Travel-ti~ tklay (lTD) is the difference between the time 3. vehicle passes a point downstream of the intersection where it has regained normal speed and the rime it would have passed that point had it been able to continue through the intersection at its approach speed. This includes all conuol and geometric delay.

Analysts use control delay most often because it is the easiest to measure and because the Highway Capacity Manual (HCM) (TRB, 2000) bases its definition of intersection level of service on conuol delay, among other reasons.

I

i.;; . ;

2.1 Equipment Needs

.~ !,

Observers can collect delay data manually or by electronic means. Data sheetS are often helpful for collecting del2y data at intersections, and allow data to be collected manually while in the field. Ho~r, more often the use of electronic counting boards, such as those de~ibed in Chapter 4, are used to collect delay. Built-in software allows the user to quickly obtain outputs. If electronic counting boards arc used, the user needs to make sure they understand the output delay measure. For instance, many count boards collect •orne-in-queue" delay, which is closely related to co nuol delay bur is lower based on the deceleration and acceleration rates and the length of the queue discharging.

·j :

.!.,I

,,

:q

'I!

l

i:

Becoming more popular in recent years is the use of a laptop computer. Computer software can be manipulated by writing *code" to use the key pad with the internal clock, allowing time scamps of various events to be recorded in real time. This manual provides an example code in Appendix E. This form of data collection is typically more flexible, and spreadsheers can be developed to calculate various intersection parameters.

j

SCleral srudies have used video with a time stamp overlay to measure some types of delay. Video provides a permanent record of the srudy period that may be used for further review of the delay data or for other srudies. Video may also reduce the num.ber of field personnel needed for a delay srudy. However, video recordings frequently suffer from poor lighting conditions and vantage points. Long queues are especially difficult to capture. When possible, an overhead vantage point should be U¥d to prevent occlusion. Many times, surveillance or video detection cameras arc already

jI


i

. I I

II

used at intersections and can be adopted for data collection of delay from video requires much labor in the office.

pu~poses.

It should be noted rhar produci ng an esrimat-:

. For a delay study, it is often necessary to determine the free-flow speed (FFS) of vehicles prior ro the intersection or interesr. This is typically done using a radar or laser speed gun; however, the space-mean speed (SMS) ·could be used by determining the run rime between known points for a large number of vehicles. Chapter 5 provides guidance on rhe use of various speed devices and methods for determining the average speed.

2.2 Personnel and Training Requirements The number of personnel necessary for a delay srudy depends on the equipment used and the volume o f traffic at the intersection being studied. A single observer can accurately record the number of queued vehicles on two lanes with moderate length queues (up to 25 vehicles per lane) or on one lane with long queues (Reilly et al., 19 76). However, a single observer working with moderate or long queues needs an audio signal of the end of an interval that is discernible above background noise from traffic. If a device that provides a loud, periodic audio signal is nor available, observers can use a recording with a signal after every time increment. Observers must check the playing device before a srudy and several times during a srudy to ensure battery levels are sufficient. Where long queues exist or if no audio signal is possible, the srudy needs two observers. One observer can then watch the time, signal when an interval has ended and record the count ofstopped vehicles while the second observer concentrates exclusively on counting. Ar;.alysts need a rurning movement count, or a classification of vehicles entering the inte.rsection as having queued or not queued, to estimate control delay per vehicle. Therefore, additional observers to conduct those counts may be necessary. As mentioned earlier, video may be helpful if additional observers are not available. Observers should position thc:!llSelves near the right shoulder or on the eight sidewalk at the approximate midpoint of the maximum queue. Observers musr have a clear view of the lanes they are observing and should move around as necessary to avoid being blocked by buses or trucks. Observers can sit in automobiles in locations where blocked vision is not a problem. Park the automobile in a legal space well our of the uavellanes. Observers should CQunr vehicles that are stopped, or nearly stopped, as being queued. A good estimation for a queued vehicle is any vehicle that is moving less than approximately 3 mph (5 km/h) or is rwo-duee vehicle lengths from the vehicle that is queued in front of it. The situation when the signal has just turned green and the queue has just begun to discharge is the most difficult to record accurately. Observers should record at least 60 intervals. Estimates of delay during pc:af periods are most useful. The appropriate weekday (Tuesday, Wednesday, or Thursday) or weekend should be used for consistent traffic patterns with dle highest volumes. Delay estimates will vary widely within short times, especially when peak periods begin and end, so analysts should interpolate between time periods with exueme caution. Light rain should not affect traffic volumes. q r queuing behavior on commuter routes during peak periods, so obs~ers can study delay if they can keep their for01S dry. Do not eonduct delay srudies in weather that affects normal volumes or driving behavior. Observers need to note on each data form all the usual "header" information, including locations, times and weather conditions.

2.3 Field Procedures and Analysis 2.3.1 Time-in-Queue Delay Buehler et a!. (1976), argued that TIQD is more closely related to TID than stopped delay and is fairly easy to measure. They recommended a procedure for determining TIQD in which an observer records the number of queued v-ehicles after each 10- to 20-sec. interval. The observer notes the number of vehicles in the queue regardless of wheth. er the first few vehicles have started into the intersection. TIQD per vehicle is then calculated as shown in Equation 6-2. The 0.9 value is an adjustment &eeoc to account for errors in the sampling technique which typically cause overes-cimation ofTIQD. A typical rime interval is 10, 15, or 20 sec. long, the key being that it is not divisible into the cycle length (TR:I3. 2000). Buehler et al. note observers should not just concenuate on the back of the queue. Observers must accou~lt for vehicles ~at leave or join the middle of the queue, and not all queued vehicles will dear the intersection duric:lg a single green phase. Since TIQD is used as an input in the field-calculated control delay, an example calculation is provided in Exhibit 6-1 in the next section. Intersection and Driveway Studies • $1'9

j

{: I'

H

[i

t ~.

fi

l~ 1!it

rJ:i

r,: ~~

·:· .i '! ~

i

~

!!:

.. ;

!:

;

2.3.2 Control Deilry Control delay is the most commonly used performance measure for determining the effect of control devices on traf. fie flow. Control delay replaced stopped delay as the primary delay mea.suremrot used in the HCM in 2000 because it was determined to be a more accurate representation and was relatively easy to collect. The 2000 HCM provides a methodology for collecting control delay in the field, which uses queued vehicles as the primary data source. Although TIQD (d ) is used, additional delay is added using an accderacionldecelerarion correction (d.J Control delay (d) is calculatd'using Equation 6-1. The 2000 HCM provides an "Intersection Control Delay Worksheet" that lays out the framework for conducting this type of delay scudy. An example worksheet is provided in Exhibit 6-1, with a blank form provided in Appendix E for copying. This method is applicable to scenarios where queues do not reach more th.a n 20-25 vehicles per lane. If queues ace expected to be longer, or volume to capacity ratios are near 1.0, the analyst should be careful to continue the vehicle-in~ueue count past the arrival count period so that vehicles queued in two or more different cycles ace accounted for. As mentioned previously, acceleration and deceleration ace acco~ted for in the control delay calculation; however, they ace not measured directly because more sophisticated methods using additional equipment would be necessary. Therefore, the acceleration/deceleration delay (d) is an estimate, which was shown to be reasonably accurate. The equation for this estimate is shown in Equation 6-1, where the &action of vehicles stopping (FVS) is calculated as shown in the example worksheet (Exhibit 6-1) and the correction factor (CF) that can be found in Exhibit 6-2. AccelfDecel (dMt).,. FVS* CF

Equation 6- 1

TIQD (d.) is calculated using Equation 6-2,

TIQD (ci,q) = (Is •

i::) •

0.9

Equation 6-1

where

i,.

= cime interval between time in queue counts (sec)

~ IV

= sum of all vehicles-in queue counts (vehicles)

IV•

c

'f

total number of vehicles during the scudy period (vehicles)

0.9 is the empirical adjustment factor. The intersection control delay calculation is found by summing the accderationldecderation delay and TIQD, shown in Equation 6-3. d=d.. + dMI


Equation 6-3

.

~~ ,.--·;"' _·. ·.. . .. ' . ...

;

-

INTERSECTION CONTROL DELAY WORKSHEET o.n•rM1

Sl~·

lonalysl

lnllaectlon

Agency or Com pal>f Dole Perfonnod

Ala Type

CC "' tJetmr>nc li9 Other

C CBO

Judsdlctlm f'l,(

~11-Periad

---- -

1899

'-~)~sis Year

Input fnttllll,

Numlzral loles, N

z

~- ~. Fl'S (mllh)

40

Tolll veNclounMng. v,. SlopJ*I whlcles count. v..,0

,,

~C"~ ~-cl~C SU.-...ycountlnllrwl.l,(s)

Cyclt

-.-llumbot

llrra

1

'I 2 I

4t:)4

I

I

2

2m

I

-

'

"

I

I

5

•

_j

--;------:-···cl.:...

~-s - ~----.iiiii;,=JF"~-:-- 1

- - · - - - . · · - - · ···- -

......

Clock

letvlh. c (s)

!5.!10

,

I

~

·,

7

D f a l " l tel t z l z l o 1 2 61 121 t:> I 16 1 6 I o I o I 2 1 o 1 o 2 2

"''"

Total

I

6

($

?::

7

D

6

&

8

4

7

II

I

l!Y7

=

Toll! ..ehldes Ill . . - . Dt,. n ...IIHf_. psrvohlc:le. cl,q •

0

16

641 ""I "' I I ..

o. • ~) • 0.9

No. alwhldesstopplng per lam ad!~ • lil:c:IIIOeral comcllon lldot. CF (&. A1~

~

~Z!

u

~

...

•

~

I 41 .

o

I

6

,...,...,oreyc~es~., .

v.

FIICIIon of whlcla slopping, f\IS "

it'f

AcciiiO-' conc:lkln delay. "- • FVS •

~ 0.42

a

Calnl ~" · ct.. •t~o~

1.7 ff-2

• $

Source: High1INIJ Ozp4LjJy M411Vid 2000. Copyriglu, Nationa!Acadt.m.y of Sciences, Washington, DC. Exhibit A16-l , p. 16-89. Reproduced with permission of the Transporotion Research Board.

=

of approximatdy 30 vehicles per lane are typically unreliable. Source: Highway Capacity Manwd 2000. Copyright. Nation21 Academy of Sciences, Washington, DC. Exhibit Al6-2, p. 16-

• Vehicle-in-queue counts in

9 t. Reproduc;~ with ~riiiWion of the Transporucion ~ Soard.

Intersection ~,n nriVO\ A#!)H

<"t ..... :.... -

· --·

'

Before starcing the srudy, rhe observer should record any genc:ral information, especially the input par.imerers. These will be usc:d in many of the calculations to determine the control delay. FFS is determined as the unimpeded speed through the intersection if it were green for an extended period of time. This can be determined by driving th rough during the green indication when vehicles are not impeding Bow, or by recordi ng the speed at a midblock location away from the intersection under study. Ideally, rhe survey will begin at the start of a red phase with no overflowing queued vehicles from the previous cycle. If rhis is nor possible, the overflowing vehicles should be excluded from the analysis. Two observers should be used, with the following tasks for each observer: Observer 1:

!:

I. Keep track of the queues for each cycle in the survey period by observing the last vehicle in every lane that stops due to the control device. This includes queuing vehicles during the green phase.

il.'

·'=:1·

2. At some predetermined intc:rval {between I 0-20 sec.), observer should record the numbc:r of queued vehicles. Again, a repeated time intc:rval (using a stop warch or othc:r device) is necessary to do chis srudy and the time interval should not be divisible inro the cycle length.

-~I

r

~I

,"

3. Enter the vehicle-in-queue counts in the appropriate box in the worksheet. The survey period should be assumed prior to conducting the study. The clock rime should be recorded at the beginning of every fifth cycle.

.i '

l

ji ·

.1 , G;

4. At the end of the srudy period, the observer should continue following any queued vehicles prior to the end of the srudy period until they exit the intersection. '

)j'

I:

Ji

Observer 2:

1. During the entire survey period, conducts a continuous vehicle count and a count of vehicles stopping one or more rimes. Vehicles stopping multiple rimes should only be counted once.

f

,q

L

!:1,:. 9

it ;• I ,., l'

'I;

11 I'

;j "' i ~\

:i

H ' I

f•

ji li

ll

Once a control delay is calculated, the level of service {LOS) can be determined using the appropriate facility type (signalized or unsignal.ized ime~ection) . For signals, when control delays for the minor street are high, the analyst should be aware it may not mean there is really a problem. Since control delay is measured in seconds per vehicle, if the volumes are low and the cycle length is long, a vehicle or two may have to wait a significant amount of rime before receiving a green signal indication. As an alternative, the volume-co-capacity ratio (vic) would be a better measure of the available throughput (or lack of throughput) for the given movement. · 2.3.3 Geo11Utric Delay Roadway geometry can significandy alter the time ir cakes to negotiate an intersection or route. Geometric effects are caused by changes in horiwncal andlor vertical alignment. A good example of an intersectio'n treatment with significant geometric delay is a roundabout, which uses curvature at the exit and entry of each leg to control speeds of drive~ as they accept available gaps in the circulating lane. Other exampiC$ include a signifiC2Jlt incline at a leg ofa ~ignalized intersection, changing a standard pedestrian crossing co a z.ig-zag ccossing, or installing a median barrier co force left turning and through vehicles &om the side sueer co cake a right and U-tum ar a median opening downstream of the typical four-legged inre~ection (see supe~treet or Michigan U-Turn discussion in Chapter 4). Geometric delay is easy to estimate but impossible to measure directly in the field. The basic idea is to determine what the travel time for the original movement would have been, and compare that to the uavel time through the new design. The analyst needs to know the average FFS of drivers (or other mode) and two known points prior before and a.fcer the geometric fearure. The most important rip is that each of these points of interest is located where the vehicle is determined to be at FFS. If the geometric feature had not been installed, the travel time between the two known points can be determined by Equation 6-4. d 2 -d1

1!

G7'T1

liof'.

102 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDIT10N

"'

FFS

'I

l:Ii

iJ

Equation 6-4

where . GTT ,

I

assumed geometric travel rime had rhe geometric fearure nor been installed

d2

downsueam point of interest where the vehicle is driving at FFS

d,

upstream point of inu!rest prior to decelerating due to the geometric feature

Next, the analyst should record the travel rimes of a large sample (usually 100 or more) of unimpeded vehicles (or other modes) to determine the travel rime (GTI2) between the same two pointS of interest. It may be h elpful to use a test vehicle to emer the uaffic stream at random times and determining the time between the two points. The average of the actual individual uavel times is then used to determine the geomwic delay (g) by Equation 6-5. g = GTJ1 - GTJ1

Equation 6 -5

It is important to point out again that geometric delay DOES NOT include the effects of any con uol device or conflicting traffic. It merely measwes the delay that the geometric feature causes by comparing it to the "no change" condition.

2.3.4 Travei-Tnne Delay TTD is often used at intersections and along corridors where the analyst is trying to determine the effect of a concrol device and any geometric effects (if they exist), as well as any other factors affecting delay. TTD is somerimes coi,ned "total delay" for obvious reasons. For instance, using one of the examples from the previous section, the roundaboUt would be evaluated based on ihe effect _of the curvature in the roundabout (geometric delay) and the YIELD sign. Tjle important difference is that the analyst is no longer concerned about unimpeded conditions to learn only about geometric delay, but also the effect the YIELD sign has along with conflicting circulating traffic. It is possible that TTD is only a function of the control delay because there may be no geometric features that cause delay (for instance, a perfectly flat and straight section of road). Let us assume an analyst wants w determine the travel time effect of a new intersection design, the "superscrceC·" A superstreet design can take a few differem forms, but for simplicity's sake, assume that a standard four-legged ince:.r· section was converted to a superstreet by eliminating the side street left and thru movements and providing a median opening upstream of the intersection where drivers can U-turn. In addition; the mainline left turn movements weJ:e cesuicted and must also U-twn at the median opening. The superstreet would have the basic geometry as shown in Exhibit 6-3. For instance, a driver traveling northbound and wishing to travel straight through the intersection would first take a riglu onto the mainline, maneuver to the median opening and make a U-turn, followed by a right nu:n onto the side street. Likewise, a northbound driver wishing to turn ~eft onco the mainline would follow the same geo· eral direction; however, they would continue straight on the mainline after making the U-turn maneuver.

Intersection and Driveway Studies • 10 .:::3

I

If the geometric dday were utilized, it would be calculated by comparing any unimpeded rravd time of any single movement through the intersection to the similar unimpeded movemenr using the constructed geomerry in the fidd, each under free flowing conditions (no TCDs) with no conflicting traffic ar the same beginning and ending poinrs. However1 TID would be differenr because it would also include the effect of the TCD employed u this rype of inte.rsection. For the mainline thru movements, the unimpeded travel time is easy ro determine because the vehicle's FFS is assumed to be consranr through rhe signal. However, the FFS cannot be used for mainline thru movements, or any of rhe U-twn or left twn movements. Therefore, one of two merhods can be utilized.

I! \i .I

f'

lJ

Firsr, the travel time through any of the movemenrs could be collected using actual driver data through the corridor during low traffic volumes when the signalized inrersection movement under analysis has a green phase. However, this could be complicated. Another possible option is to use a simulation program (such as those discussed in Chapter 11) and assume the default (or field-a.librated) acceleration and deceleration rates. The analyst could apply volumes only to the movemenr of interests and derermine the unimpeded travd time. This would be TI1•

~I ;

; ',.

.i

Next, the actual travel time is derermincd by conducting a fidd or simulation study and tracking vehicles through the supersueet intersection at the same two points: the beginning and end of the section of interest. This is TT2• It may be helpful to use a test vehicle, sometimes called a • Heating car," to enter the traffic stream at random times and determining the time between the two points of interest. AgWt, a fairly Large sample should be used to gain confidence in the result. It is best to use 100 or more measwements for each movement; however, it may be impractical to collect this much data and engineering judgment should be used.

l:

The TTD is calculated using Equation 6-6.

~

,r

rrD- rr, -rr~

!I' I

Equation 6-6

where

•' !

1TD .. travel time delay through the intersection

t

1T1

= unimpeded travd time through the intersection

1T1

= acrual average travd time through the intersection

Another option for determining TTD chat is becoming more popular is the use of GPS devices to track vehicle trajecrories through intersections using a time-space diagram. Chapter 9 describes this technique, along wirh other technologies, that could be utilized. Last, TTD is used more often for corridor stutlies where the effecr of multiple coortlinatcd intersections is of interest. For instance, a corridor may have new timing plans or signal p)'wing that has been employed along the corridor. A travel time study through rhe corridor could be used to determine if the new coortlinuion plans bad a positive effecr. Again, these corridor srutlies are described in much more detail in Chapter 9.

3.0 QUEUE LENGTH Queue·length srudies have several important applications. Q~e length data can hdp determine the necc:ssary length of storage lanes or can provide a useful measure of traffic signal efficiency. Queue length srutlies for these pwposes are conducred by a method that is very similar to the manual method measuring conuol dday discussed above. Observers count the n.umber of vehicles in a standing or slowly moving queue at designated time inrervals. Observers can make notations in the fid.d, or can count from photographs or video. At signalized intersections, the observer's record counts at the start of the green interval and the end of rhe yellow interval. Counts at unsignali.zed intersections are usually made ar equal intervals of 30 seconds or 1 minute (ITE, 2009). An analyst can investigate the feasibility of a proposed driveway location on an intersection approach using a slighdy d,i.fferent queue length study method (ITE, 2009). In chis case, observers record rhe amount of time the queue blocks the proposed driveway location. Observers can use several sropwarches for multiple locations. One would typically condua this type of queue length srudy during the peak how of the driveway and/or the intersection approach. Dividing the blocbge time for a proposed driveway location or dedicated turn lane (right or left) by the total study time produces the percentage of time the location is blocked, which is a v~ useful measure. \

\


; Field investigations are the best method for determining the :tctual qyeue lengths at an :tpproach. Macroscopic models : are used frequently for signal riming and can often be wed co determine a queue's percentile {e.g. 50th, 90th, etc.). i However, macroscopic models are equ:ttion-based and therefore do not cake into account the actual s cor:tge bay "'.lengths. Instead, they :lSSume an infinite length for each lane group; therefore, simulation is usually employed co determine if a stor:tge bay is long enough or if :t proposed drivew.~.y would be blocked for a significant amount of time. Further informacion on simulation srudies can be found in Chapter 11.

4.0 SATURATION FLOW AND LOST TIME Saturation Bow and lost time are two of the basic building blocks of traffic engineering. Analysts we chcse m easures co time signals :tnd estimate intersection capacity. SaturatUm fonu is the number of vehicles that can p:lSS a give n point ·on a highway in a given period of time with no interruptions. In studies of intersections, analysts focus on the flow past the stop bar in a lane in an hour of uninterrupted green signal (also termed the •idealn saturation flow). Ltm timl is the unused po~on of the signal cycle. There are two signUicant components to lt>St time for each signal

phase:

• StArtup lost time occurs between the time the green signal begins wd the queue begins moving efficiently. • Ckaranct lost time occurs between the time the last vehicle crosses the stop bar and the next sign:l! phase begins. Many agencies usc standard constant values for saruration flow and lose time in analyses. However, saruration Bow and lost time vary significantly between intersections and between different times of day. To avoid errors caused by in:tppropriace usc of a stmdard value, some agencies measure saruration flow and lose time directly before performing o~er analyses. More often, agencies sample saturation Bow and lose time periodically at several sires in an area and calibrate their equations based 9n those samples. The procedures for measuring saruration flow and lost time are described in this section. The procedures are rel:ttivdy simple and one can use a variety of equipment to perform them.

4.1 Equipment Needs Saturation How f:lte and lose time srudies are usually conducted with a stopW?-tch, count board, or computer software with code written to utilize key strokes and the internal clock. Methods other than a stopwatch have several advantages, including greater accuracy, instmc creation of a computer lile (for the laptop) and creation of a permanent record that is available for other srudies (wing audio and video). For nonresearch srudies, analysts can rarely justifY the cxq;J. time and expense of these other methods, how=. The laptop and video methods require special computer programs to be written ·or acquired, such as the one provided in Appendix E-2. The laptop method requires observers to press certain keys when the founh vehicle crosses the stop bar and other keys when the seventh, eighth, ninth, or lOth vehicle crosses. The program records the times those keys are pressed and performs the calculations. The audiotape method requires an observer at the intersection to speak into a tape recorder when the ...dUdes of interest cross the stop bar (Shanteau, 1988). Back at the office, a data collector plays the tape at the same speed while pressing the appropriate computer key for each audio cue. A program similar co that on the laptop is needed to record the times and perform calculations.

'

The video inethod requir~ a clear van~ point and good light conditions. In the office, a technician must stop the video and record the time on the on-screen qock as the vehicles of interest cross the stop bar.

4.2 Personnel Training Requirements Personnel conducting saturation B.ow rate and lost time srudies will need co have good vantage points near the stop bars of the apEtoach being srudied, but also need to view approximatdy 200ft.. (60 m) upstream. Since data is collected near the roadway, a reflective vest is a must. If a stopwatch is used, the person collecting the nec:esspy data should

have good reflexes and understand the exact data collection methodology prior ro going in rhe field . Any errors, even small errors, could have a significant effect on the values (especially lost time).

:~

The hardest part :~.bout conducting one of these two studies is uying to be inconspicuous. Rardy is the analyst able to obcain a good field of view wichouc ~ing out of a vehicle and in full view of approaching vehicles.

:

i1:


ti.

J' '0:

~:

iii •H

d

!I

I'

·\I .

~!

I :!

~~ :; ·I

4.3.1 Satur11tWn Flow /Uzte E.xhibir 6-4 shows a form that is useful for collecting saturation flow observations. Appendix E contains 01. blank form (Exhibit E-12) suitable for copying. The observer starts the watch when the rear axle of the fourth vehicle, in a queue that had been stationary whi.le waiting for the green signal, crosses the stop bar. This is the point where an average queue of vehicles begins co keep consistent he01.dways. The observer stops the watch when the reu axle of the seventh, eighth, ninth, or 1Oth vehicle in the queue (whichever was the last vehicle in the stopped queue at the instant the signal turned green) crosses the stop bar. For example, suppose that the stopped queue is eight vehicles long at the instant the signal turns green. The observer would start the watch for vehicle four, stop the watch for vehicle eight, and enter the elapsed time in the "eighth vehicle• colurnt:~ of the form in Exhibit 6-4. The observer cannot record a measurement if the queue is less than seven vehicles long when the signal turns green because short queues provide unstable data. If the queue is more than 10 vehicles long. the obse.rver srops the watch 01.1 the lOth vehicle. Ten vehicles is a convenienc maximum that decreases the chance of error due to the effeccs of spillback or to vehicles stopping for the red signal. Observers must ignore vehicles joining the queue after the green signal appCO!.fS. One obse.rver records sarurarion flow ·dara for one lane at a time. Saturation flow rates estimated for a lane usually apply to adjacent lanes of the sap~e rype on the same approach. One observer with a cleu view of adjacent approaches can alternately record data from a lane on each if the approaches use different pans of the signal cycle.

J

I ~

'

f

J

'

i

.l

j'

.'!

,,,,

' t: I

!l , j lt J i

~I L

,.I

' 1.

. I.

r. n · 11:


:~ ·.~-.~~ ....

-,

I FJEJ.D SHEET- SAn.NtA nON JILOW' S1l.JDY

~ 8 oundTtl f'flcO~to: ·

Une! ---·ca.!'__.~~·clr:u;._.-A~"' · CLC.QO"'"'!i~~t...~'----:--:-T!mv. !.f::oof'D!!='~ Wf!!.\:t'.

Y=la/92.. ·Tiy:cs

ObMNera~ Qrado:'

1\1. H I

Qbs

2!o

Weatt•or.~C..""';I~p~.....:f~":;j=?;:-::"!f:o._•_ _

.

Lan.e Wicnh~

I :Z.tl

Areo;

.tw.bkV'&e.,.:,

"" ....h.

81h ..... ,.

8lh voh.

l ,Olh -h.

1r- :z.. 1/!.9::._

1l ·'2..

10.1 h~

.

10.1..

·~

-:r." .

. . /0

, • II-

.. ••

. -

w

--

20

-

S'· ~

·!!!i'

11·3 ' ( .'I-

.

~0

I

Ttl\ v~ . ·

.._,~h .

Mhvo h ,

lO th. Vol'\,

It-S" / 1 -1

nl

S-2-

•••

•

c .~

s~

I

I

I.

I

~

.c:;,

Cf·O /0.,

"" I

S.'&

::I

I~.~

I

I

I

I 0 -2-

:13

.

••

.

.

••

-

.

:M

~

••· ""·o ·~

- -- --

22

.

••

-

2AO

' 10

c;,..,

- -- --

11m• (McondSt' botw•on 4th vot\lcfo •nd -

-n I

5-3

12

.,.

:s l 20 I

., ,

'-·3

..

Ot:hor. -

22

_1,1-_4

·•

-

·
=-------- --- ---

Obo

T1m• (a~Cb) becwoen 4U. vehicle and ...

No.

.eoee

~'S:==t ~ =~u

Sa.,o.cnn•"

Looallon:

&/.'2-

sa

·

~ oo Column..,..,.

I

I 5? • 0

J I

....

I I I lf-S.tlI 4-7.0jiO'J.S a..

""

ld>

'aoo· tos•• N,mpuofos.m·~!"

................... -"""',_ " •· ~

~ ~

• If -

The factors that affect saruration flow rates are grade, lane width, :intersection location (CBD versus other), type of lane and presence of adjacenc packing lanes (TRB, 2000). Therefore, the analyst muse carefully select approaches to measure saturation How co ensure an unbiased result. Do not use a saruration flow estimate from a steep approach 1:0 analyz.e a fiat approach, for instance. Heavy vehicles also affect saruration flow rates, so observers should not reco .rd data if a heavy vehicle is in one of the first seven positions in the queue. If a heavy vehicle is in position 8, the obserVer can record the time becween the fourth and seventh vehicles, and so on. Also, do not record data during a signal ph~.se in which traffic How is interrupted by buses, by left-turning traffic waiting for opposing traffic to clear, or by rigb-tcurning traffic waiting for pedestrians to clear. Analysts can calculate interrupted saruration How from id eal sa~~~r.~ti on How by the methods of the HCM (2000). The procedure for studying saturation flow in an exclusive lefr·rurn or right-turn lane with a protected signal phase is the same as the basic procedure for a through lane. For agencies that have difficulty finding sites unaffected by the factors mentioned above, Roess, Prassas and McSha.$'le (2004) suggest a procedure for estimating ideal saruration flow from measurements at nonideal sites. The analyst c;:;;afi solve the saturation flow equation in the HCM for the ideal saruration How given the measured oonideal saruratiOD flow and the standard adjuscment factors for the nonideal conditions. Trme of day, wearher, and evenrs that affe ct driver populations or behavior also affect saruration How. Measure ideal saturation flows during peak hours, in cl..JY weather, and during times when no special events are affecting drivers. It may be difficult to collect saruracion flow daca during nonpeak hours in any case, due to small queues. · ·- · Intersection and Driveway Studies • 101

One can calculate desirable sample sizes for a saturation flow study from a standard sample size equation. Usually, analysts have some knowledge of the precision of the saturation flow estimate they desire. For instance, an analyst may not want the mean estimated saturation Bow rate to differ from the true saturation Bow rate by more than d vehicles per hour. The analyst can find the necessary sample sitt n by Equation 6-7.

n = (z·~J

Equation6-7

where n

= required sample size

Z

• constant from the standard normal disuibution corresponding to a certain confidence level (see Exhibit 6-5) = estimate of the standard deviation of the population of saturation flow rates

A I)'Pic:al value for sis 140 vehicles per how (ITE Technical Committee 5P-5, 1991). If the analyst is willing to use

this !)'Pica! standard deviation and wants an estimated mean saruration flow rate within 50 vehicles per hour of the true rate with 95 percent confidence, the analyst would have to observe n = (1.96(140/50))1 = 30 valid queues. A peak period at a moderately busy intersection usually produces at least 30 valid queues.

Ii. .

Once the data have been collected on the form in Exhibit 6-4, one can calculate the mean saturation flow rate using the equation on the bottom of that form. Basically, a mean saruration Bow rate is estimated by calculating an average numb.er of seconds consumed per vehicle (headway) and converting that into a number of vehicles per hour. For the sample data shown in Exhibit 6-4, mean saruration Bow in vehicles per hour, SF in Equation 6-8, is estimated from the equation on the bortom of the form as shown in Equation 6-8.

SF=

3600n

~+~+~+~+ 3 4 5 6

SF= ~;:

~ t• .

Equation 6-8

3600 • 31 57.0 + 45.1 + 47.0 + 109.8 3 4 5 6

I.

i::

.,,

! •·

:j .

.~

~

l

.i, I

I· T

l \!I• : ~

SF .. 1,925 veh1hr In Equation 6-8, n is the total number of observations, and a, b, c and dare the times in seconds between the fourth vehicle and 7th, 8th, 9th and lOth vehicles, respectively.

4.3.21Ast Tmu Lost time is more d.iflicult to study than saturation flow for several reasons. Fust, lost times are short, so accurate measurements require quick reflaes. Second, observers can measure clearance lost time only during completely saturated green phases. FuWly. many of the variables that affect .tarurarion Bow affect lost time, plus others, including signal head position and lens size. The analyst must be cacefu.l when applying a lost-time estimate from one lane to other l:mes, approa.ches, or inrersections. Observers record lost-time data with a stopwatch, l.aprop computer, audiotape and 108 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

a computer back at the office, or with video that has an on-screen clock. Since analysts need an estimate o( sa turation flow to compute startup lost time (as described below), they often gather data for the two studies simultaneo usly. ··The majoricy of uncertainty in lost-time studies is where to establish the reference point for timing. In othe r words, "Where is the vehicle considered to be in the intersection?" Previous studies have used: • the front or rear tires as they crossed the position that had been occupied by the front tires of the first veh.icle in the queue; • the stop bar; • the crosswalk line; • the extension of the curb line of the intersecting sueet; or • other points. Berry (1976) showed that .these different reference points dramatically affected the startup i~st-time estimate"{an almost 3 sec difference in some cases). For capacity analysis, Berry recommended recording the time when the from bumpers of the vehicles crossed the extension of the nearside curb line of the intersecting screet. The total lost time (tL) is the sum of the average start up (t) and clearance lost times (t), shown in Equation 6-9. IL = t,1 +lei

Equation 6-9

The data needed to compute Startup lost time (t) are the times when the green signal begins and when the third vehicle in a standing queue passes the ref~ence point. Most studies use the third vehicle because that is the last vehicle that commonly experiences any measurable lost rime. Analysts can compute startup lost time for a phase by computing the difference between the .two recorded times (green signal and third vehicle) and subcracting three times the average headway for the lane (in seconds per vehicle) found during the saturation flow srudy. For instance, an average headway value for a signalized intersection is approximately 1.9 sec per vehicle. If the time between the green signal starting and the third vehicle passing a known reference point (such as a near curb line) were 6.1 sec, the start-up lost time would be 0.4 sec [6.1-(3*1.9)]. As mentioned earlier, the headway values are determined using a stopwatch and a predetermined reference point (where the times between ftonr bumpers of consecutive vehicles over the reference point are recorded) with sufficient queues of four-1 0 vehicles in length. The result from the startup lost time calculation can be below zero for a particular phase. In that case, anal.ysts assume a Value of zero when calculating statistical parameters based on the results. · Observers measure clearance lost times (t) directly at the end of a sarurated green phase. This is a signal phase whcie heavy volumes (and consequently tight, ftee..flowing queues) were o:insistendy flowing during the entire allotted phase time. They record the time when the last vehicle through during the phase crosses the reference point and the time when. the signal rums green for the next phase. The difference between these times is the clearance lost time. Observers need to lind a location where they can observe both the reference point for timing and the signal indication for the next phase.

5.0 GAPS AND GAP ACCEPTANCE

5.1 Equipment'Needs G;tp and gap acceptance studies collect data using count boards, laptop computers, certain types of automatic vehicle detectors, audiotapes in combination with oomputers in the office (as described above for saturation flow studies), video, or stopwatches. With automated detectors, analysts must ensure that only the lanes of interest are being measured. Laptop computers and audiotapes with computers require special computer programs, such as the one provided in Appendix E-13. If voice recorders are used to collect data, it is advisable to use a headset or microphone which clips to a shirt {Bonsall et al., 1988). Acceptable headsets are inexpensive and quite useful.

Intersection and Driveway Studies • t09

5.2 Personnel Training Requirement !~

:jj " I

!'

Observers can collect gap data during weather rhar does not affect normal traffic volumes. Observers need good visibility ro the reference point but also need to be inconspicuous to avoid influencing driver behavior. It is advantageous for many reasons, including inconspicuousness, for the observer to sir in an auromobile during a gap acceptance study, Buses and trucks usuilly do nor block the view at unsignalized intersections, so usually observers can sir in an automobile.

•:l

~:

H i!

:ii 1! :j! ~

!i.

~.

1!. lij;h

,

~! I ~~ I

~· ~

5.3 Field Procedures and Analysis 5.3.1 Gaps Observers usually collect gap data using electronic counring boards or laptop computers with time-stamp-based coding. When a vehicle in the major traffic stream crosses a reference point at che intersection of interest, the observer presses a key and the device records the time elapsed since the lase time the key was pressed. Many electronic counting boards record gap data by grouping the gaps into "biru" with intervals of 2 sec. The resultS will then consist of the number of gaps between 0 and 2 sec, 2 and 4 sec, and so on. Two-sec intervals are crude but acceptable for most gap studies, but larger intervals are generally not useful. With no other data to collect simultaneously, one observer should have no problem collecting gap data for a multilane major street.

The size of gaps in a traffic stream depends on the traffic volume, speed on the major approach, grade on the side street (minor approach), number of lanes to cross and the median width. Because volumes change over any given day, an analyst must sample gaps during each period of interest that has a volume different from those of adjacent periods. The mean gap bas ~mly marginal meaning in analyses using gap data. Stuistics that describe the shape of the gap ·distribution, such as percentiles, are more useful.

5.3.2 Gap keeptame Gap acceptance studies are more difficult to conduct than gap scudies. A gap acceptance study scill requires data on the gaps presented in the major traffic stream. In addition, observers must categorize each data point as an accepted lag, a rejected lag. an untested gap (there was no minor street vehicle present), an accepted gap, or a rejected gap. The difference between lag and gap is critical because drivers react differently to each of them. Therefore, gap and lag are defined as: • Lag: The time elapsed between the arrival of a minor street vehicle ready to move into the major street and the arrival of the front bumper of the next vehicle in the major traffic stream. • Gap: The available time in seconds between cwo successive vehicles at the same point in space, measured from the rear bumper of the lead vehicle to the front bumper of the foUowing vehicle.

1: ~;

~;

;. ) '

j

.

·'·'

!

·I

f ~:

!~

H }i 11

':j! ~

Lags precede gaps because a gap is measured between two consecutive main street vehicles, whereas a lag is only concerned about the rime before the first main street vehicle arrives. Gap acceprance srudies are conducted at locations such as two-way stop controlled intersections or roundabouts co determine the critical gap (or minimum gap) for capacity calculaciom or for calibration of simulation models. · The simplest procedure for collecting gap acceptance clara with typical agency equipment requires an observer with a count board, laptop. PDA, or video. If a PDA or video is used in the fidd, a technician with a computer in the office would need to record the data into a computer so it can be easily manipulated during analysis (adapted &om Shanteau, 1988). The observer would strike a key indicating (or say "major" in the recording device) when a major street vehicle passed the reference point and strike another key indicating when a minor street vehicle appeared at the stop bar (not at the back of a queue). Between the appearances of major and minor vehicles, the observer would also key whether the gap or lag was "accepted," "rejected," or •untested" using other specified keys for the specific data entry. The observer should record "header" informacion at the beginning of the 6.le or speak into the recording device being used periodically during the session. If a recording device such as a PDA or video is used, the technician in the office must listen to the tape at the same speed at which it was recorded. A relatively simple program determines whether a gap or lag was observed, computes the gap or lag time and swnmarizes the distributions of accepted and rejected gaps and lags. The tape recorder must run u a comtant speed, the computer clock must be accurate and the technician must manually edit the data (removing discrepancies by replaying the tape) before the analysis. For this reason, many analystS feel more comfortable coUecring the data in real time, supplemented with a video camera to account for any 110 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDmON

. potential errors. In chis way, were the analyse unsure whether che data was collected correctly, it could b e error-checked ·. with the video camera record. ' Analysts may require more information from the gap acceptance srudy besides that described above. For example, the gap acceptance behavior of truck drivers may be ofincerest. Analysts may alter the audiotape proc edure described above to include such variations. A$ wich gap srudies, data collected in 2-sec bins are adequate for most gap acceptance srudies. Ramsey and Routledge (1 973) suggest that 2-sec bins requ.ire a sample of200 acceptances, and 1-sec bins require a sample of 5 00 acceptances (with a somewhat higher-quality result for the 1-sec bins). Observers can also collect gap acceptance clara with laptop computers at the intersection or with videotape that has an on-screen dock. At intersections with low volumes, two observers with a stopwatch and a form can usually collect gap acceptance data successfully.

5.3.3 Emmating the Criti&Al Gap DistribUIWn Many previous refereno:s have recommended the calculation of an approximate mean critical gap fro m a gap acceptance data sec. There are several problems with this type of analysis, including the face that the result ca n be inaccurate (see Hewitt, t985); and the result is nor very useful. By contrast, Hewitt showed the ana)ysis method of Ramsey and Routledge (1973) was considerably more accurate. The Ramsey and Routledge method uses the same d ata to eswnate efficiently the entire distribution ofcritical gaps. Additional advantages of the Ramsey and Routledge m ethod are chat it does not require any assumptions about the discribution of critical gaps in the driver population, as some analysis methods do, and that analysts can apply it to gap Qr lag data. Exhibit 6-6 shows that the method is easy to use by hand or to program on a computer spreadsheet.

Tht Rmnuy and &utkdgt mtthod hq;i1u with convmion oftht acctpranc~ + rtjtctions data abtwt f4 a ptruntagt and ml:rt1114t int4 column 1 ofTabk A, h~ltJw.

Table A

'

Colwan Number

1

2

3

4

5

Critical Gap (sec:}

0

2

4

6

8

30.0

0.0

0.0

0.0

0.0

J j

Accepted Gap (sec:) 1

25.0

35.7

0.0

0.0

0.0

22.5

32.1

50.0

0.0

0.0

7

20.0

28.6

44.4

88.9

0.0

9

2.5

3.6

5.6

11.1

100.0

100.0

100.0

100.0

100.0

100.0

3

5

TOTAL(%)

:

Tht pmmtaga in column 1 ofTabk A trprt1mt tht distrifxaiqn ofaccq>~dgaps that would he ohrervedifaU driwrs httd a critic.J gap ofO stconds. Ifall drivm httd criJiadgaps of2 stconds, tht first mtry in column 2 would ht 0% htcaust noru ofrhem would accq>r a 1-s« gap, the stt:ond mtry would he 25/(1()()..30).; 35.7%, t& third mtry would he 22.31(100-30) • 32.196, etc.

-

Intersection and Driv~way Stud ies • 11 -1

Table B contains the number of drivers with each criliol gap size that accept a gap of a given size. First, column S ofTable B is filled out with the number of accepted gaps from the raw data. Next, the first entry in column 1 can be made; since 5 acceptances of a 3-sec gap were recorded, all of those drivers mwt have had critiol gaps of2 sec. Next, the total of column I can be derived from this first entry and Table A; since 35.7% of all drivers with a critiol gap of 2 sec saw and accepted a gap of3 sec, the cow of column 1 is 5 x 100/35.7" 14.0. Then,-the rest of column I can be completed by wing the percentages from Table A on the total in column 1, and Table 8 looks like: TableB Cola.m.o Nanibu

I

Cricic:al Gap (sec)

I

2

I

l

4

I

3

4

s

6

8

Total 5.0

Acaplcd Gap (sec)

3

5.0

0

0

0

s

4.5

10.5

0

0

15.0

7

4.0

9.3

11.7

0

25.0

1.2

1.5

1.9

5.0

21.0

13.1

0.5

9 I

TOTAL

14.0

I

,;~Cf.;t-(;#$fi:~~;~{jf ~~ ·:·2si:~S"J:.~~·~· il%·-·-

:.)· -~ !". ,~'4Ii -~---:· 1~'"-~7:: ''· ,.,--'~~..~-~~·~\~ .:v-r'[ :~._{i~~---- "·{'' - ·~· ,.\·~ ··""'· "':·

-,u-·",~r.s.:?:.i!~r<'l.

1.9 50.0 -~'26'!!( ~-}~l~-,~~t, -I -_.. - ~00% ~&i ~...... -.,l · ' ' , "" ·to( 1 '.;,.··\.t.O.."'-:. :~ ~.~-i,._.,. . .'_ ·'. ...:f ·-~ ---.-, ii -7•. ~-...J~· -~· -3?~ -~~

.I .

;-": ··-" -. ~~.~~ .F.

. ..

~- ~·~r~r~>~\-,~ l;-.::~ -~-

. ._.-- ~..; :

+-·-~:~----

'The fim mtry in colum 2 is 0, siN:t drivers with criticalgapt of4 ~con4s Jo 11JJt ~etpt gapt qf3 ttron4s. The ttanui mJTy ~ bt tht tqta/ number of~ctpttdgaps of5 ~Nmds rmmu tht number ofJrivm with criticalgapt o/2 stN!nds who acctpttdgaps of 5 ttcon4s, of15- 4.5 e 10.5. Ntxt. tht tot41 ofcols.m 2 can be tltriwd just Ill tht total for cols.mn I Will dtriwd tmce 1111 mtry Wlll/moum:. 10.5 X 100/50 • 21. 0. The mrutinJn ofcofum11 2 is compfmJ UU tht rmuzinJn in coJum 1 Will DN:t tht total Will !miiiV1L Columnt 3 and 4 llTr complmdjust like cols.mn 2. TIN t
It should be noted that this method will not work if the proportions of accepted gaps do not increase. This should not be a problem since, theoretically, pedestrians and drivers should accept larger gaps more frequendy than smaller gaps. If this is problematic, it is probably due to error in data collection, or there is an insufficient sample size. A reference sprea~heet is provided in Appendix E-45 which is very hdpful when using this method to determine the critiol gap. ~- :

[: 6.0 INTERSECTION SIGHT DISTANCE

i

t

:t

i

I i

.t '

1

Sight distances on approaches are aitiol to safe interseetion operations. The American As;;ociation of Scm Highway and Transportation Officials (AASHTO) (2004) provides recommendations fur minimum sight distances ar in.tersections. Provisions should be made to account fur proper interseaion sight distance (lSD) continuously along each highway and meet so that sufficient time to srop is av:Wable. The ISD is very imporcmt at all interseetion approaches and is aitiol to intcnc:ction operation and safety. The provisions provided in AASIITO have many cli.H'amt scenarios fur olculating ISD; the most common are prcsenco:l in this chapcer. This scaion should be used as a tool fuc undctst:anding ISD and olcul.uing the necessary daea for cala•loring necessary sight distance requimnents. Ifaltcm.a.IM scenarios are necessary, the reader should ronsult AASIITO.

-1

''

i;~~

.

!'

j j

~

!.

\! !

6.1 Equipment Needs Intcrscaion sight distance distance studies are based primarily on vehicle speeds and distance measurements on the opposing ~y. Speeds are usually dettnnined using the FFS, using a laser or ~ speed gun. Alternatively, the posted 5pCCd plus 5 mph (8 kmlh) could be used as a quick substitute. Disancc can be mca.nutd using a mc:asuring whcd or tape. If a laser speed gun is available it an be wed to dctaminc dist:ancc between twO points by wing a refitaive device such as an old sign. A laser distance mcccr can also be purchased fur these types ofstudies. Laser operated devices are much more quick and accucate than measuring wheels and tapes and should be seriously considered if multiple stUdies are being conducted. 1 .. ,

a

~AAMI

lAI

l"'\~

TI)4MC:Pf"\ATAT1tit..l Ct..lt";t,_tCtOit\lt:

C'TIIf"\lt'~

~f\U"\

Cf"'rTir'ltd

; 6.2 Personnel and Training Requirement s )Intersection sight distance distance studies typically require two or more people to conduct. They are especially dan·gerous because they require that field personnel be in or near the roadway. Ic is imperative each person involved jn the study wear an orange reflective vest and look out for each other when working under uncertain conwrions. When using a laser speed gun, the device should be calibrated periodjcally to make sure it is operating correctly. DistanCe measurements should be completed 2S accurately 2S possible, recognizing that obstructions such 2S curbing or landscaping may be in che way of the !Jleasuring device, a good reason for purchasing a l2SCC measurement device.

6.3 Field Procedures and Analysis AASHTO provides recommendations for the minimum ISD requirements for various facilicies types, inclu ding no control, yield, stop and craHic signal control. Aieas near the intersection corners. should be dear of obstruCtions that may block the driver's view of conflicting vehicles. These zones, or areas, are known as d ear sight triangles, shown in Exhibit

6-7. The no-conaol or yield-control sight triangles are depicted in the upper diagram and depict the approach sight tri· angle. Stop control is showri in the lower diagram and depicts the departure sight triangle. Each is defined below. • Approach Sight Triangle: The uiangular area that should be dear of obstructions for any approaching vehicles so drivers can see potcntiaUy conflicting vehicles in sufficient time co stop before colliwng within the intersection. • Departure Sight Triangle: The uiangular area that should be clear of obstructions for any departing vehicles stopped at the intersection and trying to enter or cross the major road.

ce-land ll No control- yfeld control on min or road

Mefo rroed

_::;;

J

-

---...._

a;--- .

.

'

.

'11/)tq';;;---

I I I lllllnor roed I I I I I 1-'

t::

I I I

~

d,

. I

::sa:

.

--.... .... 0

Cl>

f&~"''~ .

d:z

I

· ca- m

Stop convc>~ o n minor road

Source: AASHTO, A Policy tm Gtomam Deign ofHigbwtzys and Smar, 1990.

_..,..-

jEF=

6.3.1 Yield Corztrol Analysts can check sight distances at intersections with yield signs using a simple, indirect procedure. Firsr, workers measure diHances a and bas shown in Exhibit 6-7. Then the analyst obtains the stopping sighr distance of the major road, d_. from irs design speed and Exhibit 6-8. Finally, rhe analyse solves for the sight distance on rhe minor srreer, d,, using Equation 6- 10.

~:

F li,.

d = ad. b d. -b

f

ii ~;

'

Equarion 6-10

If the speed corresponding to d• from Exhibit 6-8 is lower than the current design speed of the minor road, the.sight distance is inadequate. ·

~\I'

,,,~

!~

~i

1: lh : '

,,I

;I ' II

t'' Sourc~: Crearod

t

6.3.2 Stop Control The departure sight triangle is used because ir describes the majority of intersection uaffic control conditions installed in the Unired States that use stop control. The minimum recommended ISD for a stop controlled intersection can be cal®ated using Equation 6-11, below.

:!•

1!

ISO = 1.47 •

t

~.! ;

·p J I

I

'I'

I

91

.

by author using information from AASHTO's A Policy on Geommic Dnign ofHighwtiJS and Strem, 2004.

l;

j f

~

I

.' '

v...,;... • t1

Equation 6-11

Where:

ISD

= inrersectioo sight distance (length of the leg of sight triangle along the major road)

v...",

~ design speed of major road (mph)

tg

= time gap for the minor road vehicle to enter the major road(s)

Typical time gaps recommended by AASHTO for passenger cars, single-unit trucks and combination trucks are 7.5 sec, 9.5 sec, and 11.5 sec, respectively. This assumes a stopped vehicle rurning right or left onto a rwo-lane highway with no median and grades of3 percem or less. However, the time gap value for the minor road vehicle (tg} varies depending on the type of control used, the type of vehicle analyzed (passenger car, single-unir truck, combinarion truck), approach grades, the number of opposing lanes, and the type of movement (left, right, or thru moverneor). 114 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

;Because the material is discussed in detail in AASHTOs "Greenbook," it should be consulted for the exact rime gap icalcularion so the min imum recommended ISO can be determined (MSHTO, 2004). Sight distances sh ould be mea:sured in the field to compare against the minimum lSD values determined using the AASHTO "Green book. • Jf che }neasured sight distance is found to be lower than the recommended sight distance, :~.gencies should consider removing sight obstacles, reducing approach speeds, changing TCDs, or taking other action.s. Usually, rwo observers on foot using some form of distance measuring equipment record interseCtion sight distance data for each of the movemenrs. AASHTO recommends the observers use a height of3.5 ft. (I m) off the ground and step back 8 to 10 ft. (2 to 3m) from the stop bar on the minor approach for departing vehicles.

7.0 SUMMARY In this chapter, methods used to perform intersection and driveway studies were discussed at length. Topics including delay, queue length, saturation Bow and lost time, gap and gap acceptance studies and sighr distance studies were discussed. Specifically, various equipment needs md personnel requirements were given for each study, and field procedures used to perform the srudies were discussed in derail. For further information and derails on a specific studY type, the reader should refer co references in the specified section in question. ·

8.0 REFERENCES American Association of Srace Highway and Transportation Officials. A Policy on ~ommic Design ofHighways and Strms, Washington, DC: AASHTO, 2004. Berry. D. S. "Discussion: Rdatioruhip of,Signal Design to Discharge Headway, Approach Capacil)', and Delay," Transportation Rn_earch &cord: journal ofthe Tra!IJjJortation '&uarch Board 615 (1976). Bonsall, P. W., F. Gbahri-Sarcmi, M. R. Ttght and N. W. Marier. "The Performance of Handheld Dara-Caprure Devices in Traffic and Transport Surveys,• Traffic EngiMmng and Conrrol No. 1: 10.

Box, P. C., and J. Oppenlander. Manual ofTra.ffic Enginemng Srudier, 4th ed. Washington, DC: Institute of Transportation Engineers, 1976. Buchler, M. G., T. J. Hicks, and D. S. Berry. "Measuring Delay by Sampling Queue Backup," Tranportlltion Rnearrh Rmrd: journal ofthe Tramportation Rnearch Board 615 (1976). Federal Highway Administration. Traffic Signal Timing Manual. FHWA-hOP-08-024. Washington, DC: FHWA, 2008. Hewitt, R. H. "A Comparison Becween Some Methods of Measuring Critical G~ps,• Traffic EngiNmng and Omtroi 26, No. 1: 13-22. : Institute ofTrmsportarion Engineers TE Technical Committee 5P-5A, "An lnfomutional Report: Capacities ofTriple Left Tum Lanes," ITEjoul7llli, p. 37. Washington, DC: ITE. May 1995. Institme ofTransportation Engineers. Traffic Enginming Handbook, 6th Edition. Washington, DC: IT£, 2009. Ramsey, ]. B. H and I. W. Routledge. "A New Approach ro Analysis of Gap Acceptance T uncs. • Traffo Enginemng Conm/15, No.7 (1973).

Reilly, W. R., C. C. Gardner, and J. H. Kell. A Technique for Measurnnem ofDelay at lntermtiont, Vol. 3, User's Manual, H{WA-RD-76-137. Washington, DC: Federal Highway Administration, September, 1976. Roess, R. P., E. S. Prassa.s and W. R. McShane. Traffo EngiMmng, 3rd ed. Englewood Cliffs, NJ: Prentice Hall, 2004. Shanteau, R. M. "Using Cumulative Curves to Measure Saturation Flow and Lose Tune." ITE]ouma/58, No. 10 (October 1998). Teply, S. "Accuracy of Delay Surveys at Signalized Intersections,• Transportalion Rnearch &cord:jounud ofthe Tra!IJjJortarU!n Rnearch Board 1225 (1989). lnte~ection and Driveway Studies

• 11S!i

Teply, S. and G. D. Evaru. "Evaluarion of the Quality of Signal Progression by Dday Dlsrributioru." Transpor141ion &uarch &cord: Joumal ofthe Transportarion &search Boart/1225 (1989). Tran.spon:acion Research Board, Highway Czpacily Manwzl, Washington. DC: TRB, 2000.

.,. ·;i

i' I

I

Chapter 7

Traffic Control Device Studies OriginJ by: C N,/son. Ph.D., P.E.

Dontu~

EJiutlby: <:brimlpher M. c-nmglunn, MCE, P.E. 1.0 INTRODUCTION

117


118

2.0 TCD STUDIES

~.0

4.0 5.0

117

1.1 Purpose

118

2.1 JYpes of Studies

119

ESTABliSHING THE NEED FOR TRAFFIC CONTROl DEVICES

123

3.1 Traffic Signals

124

3.2 Signs

132

REMOVAl OF UNNECESSARY TRAFFIC CONTROl DEVICES

135

EFFECTIVENESS OF TRAFFIC CONTROL DEVICES

135

5.1 Road User Compliance Studies

136

5.2 Before-and-After Studies

136

5.3 Changes in Spot Speeds

136

5.4 Evafuating Safety Improvements

137

6.0 TCD CONDmON

137

6.1 $ign RetTOreflectivlty

137

6.2 Pavement Marking Retroreflectivlty

139

6.3 Feedback from Citizens

140

7.0

SUMMARY

140

8.0

REFERE NCES

140


140

8..2 Online Resources

141

1.0 INTRODUCTION

1.1 Purpose (MUTCD) defines trafJic control tkvic~s (TCDs) as •a11 signs, signals, markings, and other device$ used to regulate, warn, or guide traffic, placed on, over, or adjacent to a ruec:t, highway, pedestrian f:aci.ljcy, or bikeway by authority of a public agency having juri.!diccion• (FHWA. 2003). The: general purpose ofTCDs is to provide visual information to the road user. TCDs are used to. help ensure the safe, orderly and efficient movement of·all-types of traffic. -

T

h~ Manwd on Uniform Traffic Control Dmc~s

Traffic Control Device Studies • 117

Devices are classified inco groups that regulate, guide, or warn craflic. &gulaiQry tkvim inform the road user of regulations that are in force, instruct the road user to rake some action, prohibit or permit the road user from making certain maneuvers, or assign the right of way. Gttuu tkvicn typically are used to identify routes; provide rravder directions; delineate the roadway; and provide information on facilities, services, points of interest and political boundaries. "Wtming devic/!1 provide notice of unexpected conditions. They draw auention to the presence of geometric fearures with potemial hazards, major changes in roadway character, obstructions or other physical hazards in or near the roadway and areas where bazard.s may exist under certain conditions. They inform the motorist of regulatory controls ahead and advise drivers of appropriate actions.

I\

_,ll :

(' i ;!( i

In the United States, the MUTCD defines the basic principles that govern the design and use ofTCDs. The MUTCD presents TCD standards for streets and hlghways open to public uavel, rc:gardless of che cype. class, or govemmencal agency having jwisdic.tion. While the MUTCD is not a starute, it carries the power of a statute in ddining national standards. Many jurisdictions adopr the MUTCD without revision; others modify or eliminate SJXdhc designs, applications, or requitemen!S by state l~larive action . Frequendy, modifications rdiea mote stringent requitemen!S than the minimum expressed in the MUTCD. Equivalent state and local manu& that meet or exceed the MUTCDS minimum requiremen!S also carry the power of a starute (FHWA, 2003).

.I' .;' '

•

1t;

:;, : .

~I : ~· I

.fll.l '•

l~

·. ;! '

I

~. .

. I

Readers of this chapter should note that similar manuals exist in other counaies; however, the sheer volume of other studies and differences in each manual are not able to be presented as part of thls chapter. Therefote, if outside the United Scates, analyses should seek specific guidance on procedures for installing various TCDs. The content of this handbook was finalized prior to rhe release of the 2009 Edition of the ManUAl on Uniform Traffic Control Deuim (MUTCD). Therefore, although all general refeccnces to the MlJfCD have been updated to 2009, content with specific references to the 2003 MUTCD has been retained. Users are encouraged to consult the current edition of the MUTCD when making technical determinations. A free copy of the 2009 MUTCD (POF or HTML format) is available at http:l/mutcd.fhwa.dot.gov. ITE, in cooperation withAASHTO andATSSA publishes a hard copy version available via the lTE bookstore at www.ite.orgfbookstore.

.

I


• !, ,_

The MUI'CD setS out general requirements for the design, placement, operation and maintenance ofeffective TCDs (ITE, 2009).

it•

• D(sign: TCDs must be designed with the combination of physical features (size, color and shape) needed ro command attention and convey the correct message.

:.~o·~ ·

li t

• PIAcm~mt: Devices are placed to fall within the road user's cones of vision so the devices are able to command attention and allow time for driver response.

tl !

I

.i

,;,. ~·f ' j·

J,

~~

t ·

d' '

r~

• Operation: Devices must be employed in a way that meers traffic requirements in a uniform and ~nsistent manner, ful1i.Us a need, commands respect and allows time for response. • Maintmant:(: Devices must be maintained to retain Iegibiliry and visibility. Devices that are obsolete or are no longer needed should be removed. Application and operation oiTCDs should be uniform. Similar devices should be used for similar situations and in similar locations to minimize road users' confusion and gain their confidence. The inappropriate or overuse ofTCDs can lead to a number of problems including driver disregard, increased dday, excess fuel consumption, increased vehlcle emissions and increased crashes. Concrol devices should supplement each other by providing a meaningful message to motorists and should be designed and placed so they srand out from the environment..

J~ ~

g• I

~.

1

f

:\ .

~' ;

·:: ;i . 1; ~ ,'

.' l

2.0 TCD STUDIES TCDs are studied for a wide range of reasons. Typically srudies ate conducted to • support warrants for the inscallation or removal·ofTCDs • determine the effectiveness of existing TCDs

~· I

ri

J~ j

• assess the condition ofTCDs • assess ongoing maintenance and improvement programs 118 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

.TCD studies may be conducted at any location where excessive delays, excessive speed, or other traffic problems have '; been observed; where there have been citizen commentS or complaintS; or where analysis has indicated that rraffic !contwl will be needed to accommodate furure demand. Most state agencies have devdoped their own procedur
2.1 Types of Studies A wide variety of studies may be conducted m collect information regarding TCDs. The most common studies include roadway conditions, collision studies, volume studies, spot speed studies, delay studies, gap distribution and TCD inventories. The planning and implementation of some of these studies are described in other chapters. The informacion presented below focuses on the application of these studies to the study ofTCDs.

2.1.1 Roadway UJ111UtUm Diagram · . Most traffic studies require ·a_ thorough d10scription of the study sire. The extent and detail of the information nee4ed depends on the analysis to be performed. A conditions diagram and location plan show the derails o f the physical layout, including such features as intersection geometries, channeliv.tion, grades, sighr-
symbols. 2.1.3 Volll1.ile Studies Volwne data :ire required for most TCD studies. The specific volun?e data required and details of data collection a.£e determined by the purpose of the study. Traffic counts are an important factor in cv.aluating improvements and recocrJ.mendations. Vehicular volumes may be collected and grouped by movement or by approach. Studies may be limircd to specific time periods (such as peak hour), direction of travel, or geographic location. Studies may be conducted specifically to establish axle counts or vehicle cypes. Chapter 4 contains derailed information on the collection, reduCtion and presentation of volume data. When n~ int=tctions and roadways are being planned or where major construction projects will be implemented i .1:1 the near future, actual traffic volumes cannot be counted and therefore muse be estimated. The ITE publication Trij GmertltUJn OTE, 2008) may be hdpful in deriving trip estimates for this purpose. Additional methods for esrima~g future traffic volumes or projecting sample ~untS are described in Chapter 4 of this book, the Traffic Engintm?"$$ Handb()(J/t (TEH) OTE, 2009) and the Transport4tion Planning Handboo!t (fPH) (ITE, 2009). Roadway volun?-e projections arc generally expressed in rerrns of average daily traffic (ADT). Although peak-hour Bows can also be e5· timared, the hourly d.iscribution is not normally available. Consequently; minimum requirements based on estimate~ ADTs (EADTs) may be obtained from values given in Exhibir7-1. These values are based on the assumption that th-e eight highest hours will each exceed 6.25 percent of the ADT (equivalent to 500 vehicles per hour (vph)). This~ vary by agency (for example, Texas uses 5.6 percent of the ADT, which results in ADT values 11 percent higher tha..$1 those shown) (Kdl and Fullerton, 1982).

Traffic Control Device Studies • 11 ~

TRAI'FIC SIGNAL I~ ~n

URBAN

WARA~S

Estlm- Av...ge Deily Tn~ftlc-s- Now 21

' RURAL _ _ _ _ __

Minimum Requ lrwnenu

EADT

t. Minimum Vahieular Not Satlofled - - - - -

S.tlofied

Number of -

fOf' mo~ lng tr.fflc o n . -

street (toul of both

Vehlcf. . per dey on major

Vehlct• per day on hlgt>.-.,.,lume mlno<'·

--o-1

otreet epproech (one

direction only)

--o.ch 0# M IY\oOre .............. ...... .

Minor Street 1 ... ................" .. - ....... . 1 ................... ............ .. 2 or mOt·e ...... ............. ~•.

1 .........: .......................

2 or more .....................

11-'•Jor Street 1 ............................... ..

2 2

moe-• ......... ··-· ....... ..

Urt>a¥1 8,000 8.800 8.600 11,000

Rurel

Urban

Ru.-.1

15,1100 8,720 8,720 6,1100

2,400 2,400 3,200 3.200

1,880 1,880 2,240 2,240

2 . Interruption of Continuous Traffic

Vehlcl. . per day on major Satitfled

Not Slltllfled - - - - -

. - (toul of both

· appro~hesJ

direction only)

Number of Ia,.. for moving traffic on each · _,..._., . . MejcwStreet 1 .................................

Minor Street

Urban

1 .... ............................ .

2 ot more .................. . . 2or more ................... .

1 ............................... ..

12,000 14,400 14.400 12.000

1 .................................

2 or more .................... 2 M tnOre ................. ..... .......

3. Coll'blnatlon Slltitfied

Not S.tllfled - - - -

Vehicl• per day on hfvher...,.tume mlnDf' ocreet - - ( o n e

Rural ·&,400 10,080 10,0110 8,400

2 W•R"anta

Urban

Ru,..l

1,200 1,200 1,.800 1,.800

1,120 1,120

860 '

8150

2 WarrenU

Ng QM Mrrant at!lfltd but following -rrents fulfilled 110% or mora

·2 NOTE ; . 1. Left. turn movements from the m•Jor nreet may be lnctuded with minor nrwet volum.. If • seperete aignal ph... is 10 be provld«i for the feft..turn movement. 2 . To be uoed only for NEW INTE ASECTIONS or other locetlona where actual traffic volumes Cllnnot be counted.

So wee: Sr<~U ofCtzlifo,.,U, T1'lljfo Mmrwzl..

Exhibit 7-1 provides three possible signal warrants for new intersections where aaual traffic volumes are not known: 1.) Minimum vehicular, 2.) Interruption of continuous traffic ft.ow and 3.) Combination warrant. For instance, the minimwn vehicular warrant has four possible combinations of major and minor roads to chQO.Ic from. If the major and minor roads only have one lane each, and arc located in a rural area, the minimum nwnber of vehicles that must be present on each approach are 5,600 and 1,680.

Pedestrian volumes.may be recorded with vehicular volumes or on a sepa=e sheet. For signal stUdies, counts should be taken on each crosswalk during the same periods as the vehicular counts and al.so during hours of highest pedesaian volume. For other signal stUdies, pedestrian volumes crossing the major street should be sufficient. Where young or dderly persons need special consi
• Over 60 years

\I

'120 • MANUAL OF TRANSPORTATION ENGINFFRIN(, '\TIIniFc;

mr~

FmT1mo

fhe collection of pedestrian volume data is described in more detail in Chapter 12.

~1.4 Speed Studies ¥anY aspects of traffic control planning require speed distribution information. Speed disuibutions are commonly used to establish maximum and minimum speed limits; to determine the need for posting safe speeds at curves; to determine the proper location of regulatory, warning and guide signs; to establish the boundaries of no passing zones; and for the analysis ofspecial operational siruations (such as work zones and school areas). Spot speed studies arc made by measuring the individual speeds of a sample of vehicles passing a given point (spot) on a street or highway. These individual speeds arc used co estimate the speed distribution of the entire ttaffic stream ac the location under the conditions prevailing at the time of the srudy. The results of a spot speed srudy are shown in Exhibit 7-2. Spot speed srudies are usually conducted during off-peak average hours; however, the period during which speeds are measured depends on the purpose of the srudy. When spot speeds are needed to determlne the 85th percentile speed ofspecific roadway sections and/or intersection approaches, measurements for low-speed intersection approaches (I 5 co 25 mph [24 to 40 km/h}) should be made 150 to 200 ft. before the intersection. On high-speed approaches (50 to 65 mph [80 to 105 kmlh)), the checkpoint should be !oared 800 to 1,200 btfore the intersection (Kelland Fullerton, '1982). Spot speed studies are discussed further in Chapter 5. . .

:n

flW:dftifi?~l~ll:~t~

!

•J ~

43 58 46 47 45

;;j '

45

;l:

··l.,l ·. 'i! .

I' '

l'l I

57 48 51 44 46 53 52 36 57

•

'I' ! ~

.

~

I

I

;p'" ·,,~ ; u

~·

. IJ)I..

j; : I

!I'

Sam

II

Category (I)

Actual Samples (2)

Range

II

<35 kph 35.1-40 40.1-45 45.1- 50 50.1- 55 55.1-60 >60.1

6 12 12 17 13 7 8 75

35 40 45 50 55 60

11%

60+

II%

I

r~

...5 10

I

I

,1 '

""

lli

. L l

8 6 4 2

0

"'v M

~ ;6

M

......,"'0 'I'

"'. 0

~

It) 'I'

1

i

I 1:

Frequcocy Distribution 8% 24% 40% 61% 78% ·89% 100%

16 14 12

'~1'. i

63 38 37 44 49 36 43 59 47 37 47 32 54 64 53

Cumulati....:

Frequency Distribution 8% 16% 16% 21% 17%

Upper Limit of

Total

iji -

61 43 31 46 48 37 62 37 42 39 51 32 47 61 59 Samples (n) • 75

47.44 3,558

Mean

.

r·

~~ ~I

54 58 34 53 47 61 54 62 48 54 47 37 50 42 48

53 41 63 52 42 32 36 39 55 34 43 50 47 37 58

122 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

ttl

~

0 ttl

...~ It)

10

~

0

CD A

2.1.5 Delay Studies .TCD studies may also require information on the amount of delay encouncered by vehicles at si~nalized and unsigoal. ized incerseccions. Two mechods, stopped-time delay and the travel-time mechods, are commonly used co measure che 'delay at intersections. The stopped-time delay method consists of determining che amounr of time chat vehicles are actually stopped at che incerseccion. The amount of stopped rime can be determined using visual observation and a good data collection sheet, a count board set up to collecr stopped delays, or a sofrware based code using rhe internal clock of a computer (such as the one provided in Appendix E, Exhibit E-1).

Travel-time delay studies look at the amount of time lost by a vehicle through a roadway system by comparing che time a vehicle travels through a roadway system to the rime it would have taken if it had traveled at the desired speed wirh no delays. Three common studies are che test vehick, vehick observation and probe vehick methods. These methods can use various techniques for collecting data, from simple daca collection forms, count boards and softwar(·b3Sed programs using an internal clock to more complex methods used in probe studies that rely on transponders or cellular devices. Delay srudies are discussed furcher in Chapters 6 and 9. 2.1.6 Gap Distributions The ability of vehicles to enter a major scree~ from a side screet or driveway, and often the ability of pedestriaJJS. to cross at an unsignalized location, depends on the distribution of gaps in the traffic .stream. Ifgaps of adequate lwgth are infrequent, it may result in unacceptable delay for vehicles arrempting co enter the scream. Some road counters can record the gap distribution as well as the axle count, by time of day. Gap distribution can also be observed in the field manually by a person with cem..in eleccconic count boards or using a portable computer with che software-based code using the internal clock of the computer. Gap acceptance and gap studies are discussed further in Chapter 6.

21.7 TCD In~ . The establishment and maintenance ofTCD inventories is a necessary pan of day-to-day traffic operations. Chapcec l5 focuses on the planning ani! imple~entacion of inventories. TCD inventories for signs, markings and signals should contain, at a minimum, information on location, condition, TCD type, MUTCD or state code, message (for signs), mounting type and when the device was installed and last inspected. A TCD inventory can be conducted manually using a trained person or crew to inspect each device and record the data on its condition, locacion, if ic meets standards and if it serves its intended purpose. Manual data collection may work well for small jurisdictions; however, it is time conswning and expensive. Larger jurisdictions are using video to record and inventory datl everY few years. These dat2 may be stored in video libraries or computer bard drives for easy retrieval.

3.0 ESTABUSHING THE NEED FOR TRAFFIC CONTROL DEVICES TCDs are typically installed as they are warranted based on municipal, stare ancl/or fedecal guidelines. A warrant is a. set of criteria used to define the relative need for a pacticulac device and is intended co ensure roadway user safety and convenience. Warrants are often expressed as numerical requirements, such as the volume of vehicular or pedesuia.~ traffic. Some are, however, discussed as general policy statements rather than as absolute warrants. A warrant normallY includes a means of assigning priorities among several alternative choices. In che MUfCD (FHWA, 2003), two prin.ciples guide the devdopment and implementation of warrants: 1. char the most effective traflic conccol device is that which is the Least resaictive while still accomplishing the intended purpose; and

2. that driver response to the influences of a TCD has previo~ly been identilied by observation, fidd experi~ ence and laboratory test under a variety of traffic and driver conditions.

Th~ Ml.JI'CD presents warrants as a series of guidelines that should be used co help evaluate the situation ar hand. ~ not as absolute v..Iues. The satisfaction of a warrant does not guarantee that a TCD is needed. Similarly, failure tO fully sawfy a specific warrant is not positive proof that the device could not serve a useful purpose. The application o:£ warrants is effective only wb.en combined with knowledgeable engineering judgment. The MUTCD describe war-rants, design and placement criteria for a wide range ofTCDs. Warrants for signals and signs are summarized btlow-In addition, state and local jurisdictions may have developed their own warrants for applications not included in ch~ MUTCD, including loading zones and speed bumps.

Traffic Control Device Studies •

12~

3.1 Traffic Signals Traffic signals can be valuable tools for the control of conflicting modes of £ravel. Because they assign right of way, signals can profoundly influence traffic ftow (positively or negatively}. When jwtificd and properly designed, a craffic signal inscallation may (ITE, 1998): • effect orderly traffic movement; • reduce rhe frequency of certain rypes of crashes, especially right-angle collisions; • potentially increase the traffic handling capaciry, provided the proper conuol measures are wed and timing plans are updated regularly as land use changes or traffic flow increases take place; • provide for the continuous Bow of a platoon of t raffic through proper coordination at a definite speed along a given route; • allow other vehicles and pedestrians co cross a heavy traffic sucam; and • control traffic more economically than by manual methods. Traffic here is defined as vehicles, pedestrians, bicyclists, ridden or herded animals, succtcars, or any other mode wed either by itself or together for the purpose of travel on any roadway.

It should be noted that traffic signals do not always have a positive effect on roadway operations. A:n. unjustified, poorly designed, improperly operated, or poorly maintained traffic signal may result in increased collision frequencies, excessive delay, motorist disregard, decreased capaciry and circuitous travel by alternative routes. Experience has indicated thac although the imtallatiOP of signals may result in a decrease in the number of rightangle collisions, there may be increases in the number of rear-end collisions. Consequcncly, a th9rough study of uaffic and roadway conditions should precede the inscallation and the selcct.i on of signal control equipment.

.!

The MUTCD describes eight warrants for the installation of traffic signals. In many instances, state or local jurisdictions have supplemental warrants that should be consulted in addition to the MUTCD. The MUTCD contains the minimum rcco~endations for installation of a traffic signal. The MUTCD signalized intersection warrant requirements arc summarized below. Traffic signals should be considered only if one or more of the signal warrants are met, or if good engineering judgment suggests a signal should be wed to serve a weful purpose. The analyst sho uld keep in mind the MUTCD states "The satisfaction of a traffic signal warrant or warr:mts . shall not in itself require the installation of a uaffic control signal (FHWA, 2003):

'·

3.1.1 mammt 1t Eight-Hour V.hicul4r VtJlum• . Warrant 1 is satisfied if one of two possible vehicular volume "conditions" is met for an 8-hour average day. Exhibit 7-3 should be consulted for use with each applicable condition. • Condition A is intended for application when large volumes of conflicting t.ra.ffic intersection.

• Condition B is intended for application where Condition A is not satisfied and applies to operati ng conditions where the traffic volume on a major street is so heavy that uaffic on a minor intersecting meet suffers excessive delay or hazard in entering or crossing the major street.

.

:~

.I .' I ,!\

i!

arc present at an

To meet the requirements for this warrant, the vehicles per hour for Condition A orB in the 100 percent columns for major street and higher-volume minor suect approaches mwt be met. In applying the major and minor street volumes, the same 8-hour time period should be used. If the posted speed limit or 85th percentile speed exceeds 40 mph (64 kmlh}, or is located in an isolated communiry with a population less than 10,000, the 70 percenc columns for both conditions should be used in a similar manner. Alternatively, a combination Condition A antfB could be considered after all other remedial measures have been reviewed. This measure is similar to the single condition analysis; however, it only requires chat the 80 percent vehicles per hour measure apply to Conditions A and B. As before, in applying the major and minor street volumes, the same 8-hour time period should be used. If the posted speed limit or 85th percentile speed exceeds

1 'M

•

MANilA! O F TRANSPORTATION

ENGINEERING STUDIES, 2ND EDffiON

40 mph (64 km/h), or is located in an isolated community with population le.ss than 10,000, the 56 percent columns for both conditions should be used in a similar manner. '.Warrant 1 is satisfied if either Condition A or B is satisfied. In extreme cases, if A or B are not satisfied, then the Combination Warrant can be used.

~~~~~i(~.J~,.

~···

Condition A-Minimum Vehicular Volume

Number of lanes for moving traffic on each approach Minor Stnoat

Major Street 1 ••.... ..•.. ...••••. 2or more .•• 2or more ..• 1 .................

Vehicles per hour on major s treet (total of both approaches) 100%'

1 ....•....•••••... . 1 .•....... ....•... 2 or mora .. • 2 or more ....

80%.

70%.

56%4

600

480

350 420

280 336

600

480

420

336

!500

400

350

2 80

500 \ -400

Vehicles per hou r on higher-volume minor-street approach (one direction only) 100,-o• 80'o/o" 150 150 200 200

!2!!: 56o/ocl

120 . 105 120 105

84 84

160

140 112

160

140

112

Condition &-Interruption of Continuous Traffic

Number of lanes for moving traffic on each approach M ajor Street 1 ................. 2 or more ... . 2 or more ... 1 .................

· Minor Stree t 1 ................. 1 ................. 2 o r more ... 2 or more ....

• Bask: "*'~mum hOUr1y VOlume~

.

Vehicles par hour on major s treet (total of both epproaehea) 100%' 750 900 900 7!50

60%.

~

!16%.

600 720 720 800

!125 630 630 525

420 504 504 420

Vehicles per hour on higher-volume minor-siFeet approa ch (one direction only) 100%' 80%. 70%' 56%4 75 75 100 100

60 60

80 60

53 53 70 70

42 42 56 56

-1<1

• u Md IIOf combrtaoon Of Condliol\a A and 8 lltler adequate trlal ot ocn•r rernedtal meuurft. . • MayDa UHan tn• mr· SIJ. .t 70 1an1t1 Ot <40 ms:>t> or tn en loOiat
- <>
Source: 2003 Manual on Uniform TTlljfo Control Dtvicc. Table 4C-l. Sect.4C.02. Page 4C-3.

3.1.2 Warnint 2: Faur-Htn~r Vebicul.r YO~ : Warrant 2 is s~tisfied when each of any 4 hours of an average day the plotted points representing the vehicles per ltouc on the major street (total of both approaches), and the corresponding vehicles per hour on the higher-volume minor sueet approach (one direction only), both fall above the curve shown in Exhibit 7-4 for the c:xiscing combination of approach lanes. When the posted or 85th percentile speed of the major succt traffic exceeds 40 mph (64 km/h) or when the intersection lies within a built-up area of an isolated community having a population less than 10,000. the 4-hour volume requirement is satisfied when the plotted points referred to fall above the curve in Exhibit 7-5 for the existing combinacio~ of approach lanes.


n -If ·'

:c a.. > :c

I . ~.

<..:>

500

I: 1: i

tue5 wCC

400

I

I-

a..

300

~~

~3

200

a: w :c (!) :X:

100

.;:I If. I

lj•

a;O..

en<

.,,s

~§?

:,liI :

~

,:,r,!

·so 400

300

~

500

600

700

800

900

1000

1100

1200

1300

1400

MAJOR STREET- TOTAL OF BOTH APPROACHESVEHICLES PER HOUR (VPH)

I ' *

_I.I I '

•Note: 11 S vph appll8$ a s the lower threshold volume lor a minor-stree t approach with two or more lanes and 80 vph a pplies as the lower threshold volume lor a minor-street approach with one lane .

',f

'i~

I ll

Sowce: 2()()3 Mamuzl on Uniform Traffic Control Dwices. Figure 4C I. Sect. 4C.04. Page 4C5.

~~

~ 0

8~ ~~ ~~

I

!

300

200

z3

l

2 ~

~~

~

(!)

1:

r

!·I <.

~~ l

~:I

~~~I

!l l

~l

100

I

I

I

200

300

400

I ==---~ S?SI

~

500

600

700

Sowu: 2003 Manual tm Uniform Traffic Control Dnnets. F.gure 4C2. Sea. 4C.04. Page 4C-5.

,, I

~

:

'

.l j I !

900

•Note: 80 vph applies as the lower threshold volume for a minor-street approach with two or more lanes and eo vph applies as the lower threshold volume lor a minor-street approach with one lane.

~!I 1,

800

MAJOR STREET-TOTAL OF BOTH APPROACHESVEHICLES PER HOUR (VPH)

.. ' '

1=:~


1'000

· 3.1.3 Wan-ant3:Peak-Hour '; Warrant 3 is intended for application where traffic conditions are such that for a minimum of I hour of the day, the ',minor street traffic suffers undue delay in entering or crossing rhe major street. The warrant is intended to be applied 'to unusual cases where high volumes of traffic are occurring during a very limi red time period. These could include office complexes, manufacturing plants, industrial facilities, ere. The peak-hour warrant is satisfied when one of rhe cwo categories below is mer: • Category A is satisfied when all circumstances given below exist for 1 hour (any four consecutive IS-min. periods) of an average weekday. All rhree provisions must be met. 1. The total stop time delay experienced by the uaffic on one minor street approach (one direction only) controlled by a stop sign equals or exceeds 4 vehicle-hours for a one-lane approach and five vehiclehours for a cwo-lane approach tmJ

2. the volume on rhe same minor street approach (one direction only) equals or exceeds 100 vph for one movi~g lane of traffic or 150 vph for cwo ?loving lanes and

3. the total entering volume during the hour equals or exceeds 800 vph for intersections with four (or more) app roaches or 650 vph for intersections with three approaches. • Category B is satisfied when the plotted point representing the vehicles per hour on the major street (total of both approaches) and the corresponding vehicles per hour of the higher volume minor street approach (one direction only) for 1 hour (any four consecutive 15-min. periods) of an average day falu above the curve in Exhibit 7-6 for the existing combination of approach lanes. Al ternatively, if the posted speed limit or 85th percentile speed of major srreet traffic exceeds 40 mph (65 kmlh). or if the intersection lies within a'built-up ¥ea of an isolated community having a population less than I 0 ,000, Exhibit 7-7 should be used in place of Exhibit 7-6. It should be noted that the peak-hour warrant is not accepted by all states, so checking with the proper authority or reference is iroporranr if considering a signal based on this warrant alone. ·

iE

eoo

>

~

eoo

i~

400

5~

300

~~

200

~ ~

' 150 '100

100

400

600

600

700

600

QOO

1000

1100

1200

1300 1400

1 eGO

1600

1700

1600

MAJOR STREET-TOTAL OF BOTH APPROACHES: VEHICLES PER HOUR (VPH) "No t e: 16 0 vph applies as !he lower thre shold volume for a minor-street approach with two or l"(lore lanes ancl100 vph appnas as the lower threshold volume for a minor-street approach with one lane.

Source: 2003 Manual on Uniform Traffic ContrrJI Devim. Figure 4C-3. Sect. 4C.06. Page 4C-7.

Ttaffic Control Device Studies • 12!:. 7

~ :X:

0

ii ~~

200

~ ffi

100

:e-'

:X:

l

I

300

400

I. I

I

C)

=r-

I

I T

I

I:~~o

s:

500

600

700

800

goo

tooo

1100

1200

1soo

MAJOR STREET-TOTAL OF BOTH APPROACHESVEHICLES PER HOUR (VPH) •Note: 100 vph applie s as the lower threshold volume for a minor-street approach with two or more lanes and 76 vph applies as the lower threshold volume for a minor-street approach with one lane.

Source: 2003 Mmua/ 1111 Uniform Tmfo Control Dwices. FJgUrc 4C-4. Sect. 4C.06. Page 4C.7.

3.1.4 W4rrmtt 4: Pet/utrUm Ycllume Warrant 4 is intended for use when t.raffic volumes along che major rueet are so heavy pedesrrians experience excessive delay in crossing. A traffic control signal could be installed at an intersection or midblock crossing if the two conditions bdow are met. • Condition A is Rtisfied when pedestrian volumes exceed 100 or more during any 4 hours or 190 pederuians during 1 hour tmd .• Condition B is Rtisfied when there are fewer than 60 gaps per how of adequate length to allow pedestrians to cross safely when Condition A is satisfied. When a median refuge is available for pedestrians, the · condition applies to each direction separatdy. In special cases where che crossing speed of pedestrians is less than 4 ft./sec (1.2 mlsec), the pedestrian volume criteria could be reduced by as mucb as 50 percenL Note: Many other counuies require signaliucion for the •physicalfy challenged," typically nored as those that are disabled or visually impaired. Analysts should make run: that all Americans with Disabilities Act (ADA) or country guiddines are followed to promote me aossin~ and limit the possibility of furure judicial action.

3.1.5 WDTmtt 5: School Crouing Warrant 5 may be considered as a special case of the pedestrian warrant. The warrant is satisfied for an established school crossing when a gap study shows the number of adequate gaps in the tra.ffic stream dwing the period when children are using the crossing is less than the number of minutes in th.e RIDe period. The ITE repon A Program for School Crossing Prot«tion (ITE, l9n) describes the procedure fur collection of data for chis warranL The pedestrian group size and gap study consi.su of two .6dd studies: (I) pedestrian group sizes and (2) vehicular ~ sizes. Data for vehicular gaps can be collected and analyzed using the forms in Exhibits 7-8 and 7-9. The procedure is described in the following section. The adequate gap time bcrweeo vehicles on the major roadway (G) is estimated using the equation:

128 • MANUAL Of TRANSPORTATION ENGINEERING STUOIF~ 7Nrl mmnN

!, G=R

+ W/3.5 +2(N ·I}

Equation 7-1

!

\ where G = adequate gap time, sec

R

z

reaction time, sec

W

=

width ofstreet, ft.

N

= number of rows of pedestrians crossing in the 85th percentile group

Equation 7-1 uses a walking time based on 3.5 fps. The walking speed may be changed to more accurately reflect observed conditions. R is the perception reaction time: the time required to look both ways, make a decision and commence the walk. A commonly used value is 3 sec. The term 2(N - 1) represents the pedesuian platoon time. Children are assumed to cross a street in rows of five with an interval of2 sec btr.veen each row (ITE, 1972). Next, the total survey time (7) of the vehicular gap survey is converted from minutes to seconds:

T

=

total survey time * 60

Finally, the percentage of the study time during which the gaps are of adequate size (D) is calculated. D=

(T-t)

-r-•too

where t is the total time of all gaps equal or greater than G. The data form in Exhibit 7-9 provides a convenient format for studying pedestrian group size. In the fidd, the number ofgroups of pedesuians is recorded by group size. Because schoolchildren are assumed to cross in groups of five, group sizes are incremented by five. The number of adequate traffic gaps is obtained from Exhibit 7-8 (gap-size study forms). Warrant 4 is satisfied when the number of gaps in the traffic stream during the time children are using the crossing is less than the number of minutes in the same time period.

.·. ,,-I•J•'··'"-'-'"-""" ''•'";b;t~,;,.- ·''!';;t~J®,j,iJ£-<-• .'~·-~~;-. -~~~~~~)-~~>~f~);-_;;1~~- - ::1~~~;):-~<~,·- .~ :-_..:.:,. ~·~i'(:-

~~~~~~

·- :• e. • _ ~·

+_.

~----. •

c _• ~-

__

.. __

·.¥ ••, _ , , - · • • • • • • - •

.

Survey Dace Location

Crosswalk across

End of Survey (to nearest minute)

Number of lanes 'N'

Scarr of Survey (co nearest minute)

Roadway Width 'W

Toul Survey time (minutes)

Adequate gap rime 'G'

Gap Size Seconds 8 9 10

Number ofG"' s Tally Total

Multiply by ~ Size

11

12 13 14 15 116 117 18 19 20 21 22 23 1

ComJ>..uu.rion.s G=

R + (W/3.5) + 2(N-1)

GG

_ _ _sc:c

Ta

to121 survey timex60

Ta

_ _ _sec

o.

r(T-1)rn x 1oo

D·

\

_ _ _%

24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40

41 42 43 Totals Source: KdJ and Fullerton, 1982. 130 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

I=

total rime of all gaps equal to or greater than G

~~~!J:tJ:'ii'Ofi,'-i;!fu'm,..-~~f. ·ift:t.r·~ ~!--'J~·."·~eJ"'~··''o ~..::il.1.:'·:;t:.iti".·, 4. .=- · . ... ~~.. . ·_• ~tifl _• ~·! • '/-. • __ ~~,, ~~~~~-::-vl~$~w'n~-~~"'&~; . ... -.~-:-;~i'~'t· ~~- ~:-- t_~_..-:!'i;...:·

Sul'Veydate Crosswalk surveyed Crosswalk across Divided Roadway?

Yes

No

Curb co curb disance NWilbcr of groups Group SW:

Number of rows

5 or fewer

1

6to 10

2

15

3

16 tO 20

11

4

21 to 25 26 to30

5 6

31 ro 35

7

(0

36 to40

8

41 tO 45

9

46 to 50

10.. ,

Tally

Total

Cumulative

Comp utations

Tune period studied Number of adeqtute uaf6c gaps Nuinbcr of minurcs in che same period Is che warrant satisfied?

Sour= ITE. A Programfor School Crossing Pro~ction: A R.tcommnukd Practice, 3rd Edition, 1971.

3.1.6 WArrant 6: Coorrlin4ted Signal System Warrant 6 expresses the desirability of holding traffic in compact platoons, even wheh they would not otherwise be necessary. It is satisfied when: 1. On a one-way street or a street that has predominantly unidirectional traffic, the adjacent signals arc so apart they do not provide the necessary degree of vehicle platooning and speed conuoL

fa.r

2. On a cwo-way street, adjacent signals do not provide the necessary degree of platooning and speed conrro 1 and the proposed and adjacent signals could constitute a progressive signal system. According to this warrant, the installation of a signal should not be considered where the resulting signal spacio.g would be less than 1,000 ft. (300 m).

3. 1.7 WArrant 7: Crath Experience Warrant 7 is intended to provide safety considerations for the in.stallation of a traffic signal and is predicated on rtduc:ing the frequency and severity of collisions. It is satisfied when all of the foUowing criteria are met. 1. An adequate trial of less restrictive remedies with satisfactory observance and enforcement has failed to

reduce the collision frequency. 2.

Five or more reported coUisions of types susceptible to correction by traffic signal control have occwrtd within a 12-month period, each crash involving pe.rsonal injury or property damage exceeding the reportable crash thresholds in the state or local jurisdiction.

3. There existS a volume of vehicular traffic not less than 80 percent of the requirements speci.6ed in wanant6 1 and 4: Traffic Control Device Studies • 13-1

3.1.8 ~mmt 8: RDadway Network Warrant 8 is intended to encourage concentration and organization of traffic Bow along a roadway. It is satisfied when two or more conllicting major routes meet one or both of the following conditions: • Condition A is satisfied when the intersection has a total c:x.isting or projected entering volume of at least 1,000 vehicles per hour during the peak hour of a cypical weekday. This condition further requires that the 5-year projected traffic volumes meet the requirements of warrants 1, 2 and 3 during an average weekday. • Condition B is satisfied when the intersection has a total c:x.isting or projected entering volume of at lease 1,000 vehicles per hour for each of any 5 hours of a non-normal business day (Saturday or Sunday). A major route, as defined in this warrant, should have one or more of the following characteristics: • The route is part of a street system that serves as the principal network for through uaffic flow; or the route includes rural or suburban highwa~ outside, entering, or traversing a cicy; or • the route is referenced as a major route on an official plan such as a thoroughbre plan.

3.2 Signs The MUTCD prescribes standards for the uaffic signing within the right of way of all classes of public highways. Traffic signs &11 into three broad functional classifications according to use. These include regulatory signs, warning signs and guide or informational signs. ~ch is defined as follows: ' • Regulacocy signs give notice of traffic laws and regulations. • Warning signs give notice of a situation that might not be readily apparent. • Guide signs show route designations, destinations, directions, disunces, services, points of interest and other geographical, recrcacion.al, or culrural informacion. Signs should be used only where warranted by facts and field srudies. Regulatory and warning signs should be used conservatively because these signs, if used in excess, rend to lose their effectiveness. Route signs and directional signs should be used frequendy because they promote reasonably safe and efficient operations by !=ping road users informed of their location. Signs arc essential where special regulations apply at specified places or :u specific rimes only, or where haz.a.rds uc not self-evident (lTE, 200 1). The MUTCD contains the complete, current warrants for a wide range of reguktory, warning and guide signs (FHWA. 2003). The summaries below arc presented for convenience only and do not represent the entirety of available signagc options. The MUTCD or the equivalent local manual must be consulted for explicit requirements for specific signs. 3.2.1 IUgrJtmn, s;xru Regulatory signs inform road users of traffic laws and regulations. They arc to be installed near the location .at which the traffic regulation applies and with adequate visibility to obtain compliance. Rl:croreHectivicy or illumination sca.nduds should be followed to allow similar day and nighttime viewing ofshape and color. Regulatory signs uc used in many applications; however, this section focuses on stop sign, mulciway stop control, yield control and speed limit warcanu.

Stop sign wamtn~ at an intersection require an examination of collision diagrams and knowledge of operating conditions at the site. A stop sign may be warranted at an intersection where one or more of the following conditions exi!t based on engineering judgment. I. inrecseccion of a less important road with a main road, where application of the normal right-of-way rule

is Wlduly hazardous;

2. street entering a through highway or street;

3. unsignaliz.cd intersection in a signaliud area; and/or

4. unsignaliu:d intersection where a combination of high speed, restricted view and serious collision records indicate a need for control by the stop sign. Stop signs cannot be erected at intersections where traffic control signals are present, and they should nor be installed for the sole pwpose of controlling the speeds of the motoristS.

Multiway slfJp control sign warrant.i incorporate numeric:al criteria for collision occurrence, vehicular volumes and approach speeds. Four-way or all-way stop sign installations can be used as a safety measure at some locations where the volume on the intersecting roads is approxirnatdy equal and the following conditions have been established. I. Where traffic signals are warranted and urgently needed, the multiway stop conuol is an interim measure that can be installed quicldy to conrrol traffic while arrangementS are being made for the traffic signal

installation.

2. The:re is a.collision problem; as indicated by five or more reponed t~hes in a 12-mortth period of a rype susceptible to correction by a multiway stop installation. Crashes would likdy include such types as righ~ and left-turn collisions or right-angle collisions. 3. Minimum traffic volumes must meet the following criteria: a.

The total vehicular volume: entering the intersection from all approaches must average at least 300 vehicles/hour for any 8 hours of an average day; and

b. the combined vehicular and pedestrian volume from the: minor street or highway must average at least 200 units/hour for ~ ~arne 8 hours, with an average delay to minor street vehicular traffic of at least 30 sec per vehicle during die: maximum hour; but c. when the: 85th pc:rcencile approach speed of the major street traffic exceeds 40 mph (65 km/h), the minimum vehicular volume warrant is 70 percent of the foregoing requirements. Alternative criteria could be considered if they exist, including: l. problematic left-turn confliCts;

2. vehicle}pedesrrian confliCts where high concentrations of pedestrianS exist;

3. sight-distance problems at the site causing the driver to take inherent risk negotiating the intersection unless the other approach is required to stop; and , ·

4.

two residential roads inrcrsect, each having similar design tnd operating characteristics, where mulciway stop control would improve the ch.aracreri.stics of the intersection.

Yield sign w
2. A merging maneuver in the entering. roadway docs not provide adequate accderacion geometry and/or sight distance for merging maneuvers. 3. Within an intersection with a divided highway, where a stop sign is present at the entrance to the first roadway and further control is necessary at the entrance to the second roadway, and where the median width bcrween the two roadways exceeds 3~ ft.

4. M intersections where a special problem exists and where an engineering scudy indicates the problem to be susceptible to correction by the use of the yield sign.

·· ·

··

·


Typically, yield sign applications take place at two-way srop-concrolled intersections or channelized turn lane applications; however, it should be nored yield signs should also be used at the entering approaches of all modern roundabouts, which are becoming more common across the coumry. Spud limit rign warrantr $hould only be wed after an engineering study has been conducted based on established engineering practice, regulation, or as adopted by the aurhoriz.M agency. The MUTCD recommends posted speed limits should adhere to the following guidelines. I. Speed limits should be posted in multiples of5 mph (10 km/h).

2. State and local agencies should re-evaluate nonstatutory speed limits every 5 years if roadways have undergone significant changes in land use or the roadway. 3. When a speed limit is posted, it should be within 5 mph (1 0 krnlh) of the 85th percentile speed of freeBowing conditions. 4. Other options exist that can be considered when establishing the posted speed limit. These include: roadway characteristics, shoulder condition, grade, alignment and sight distance; 5. the 'pace' speed; 6. development along the roadway; 7. parking practices; 8. pedestrian movement and activity; and 9. 12-month crash ~perience. 3.2.2 Warning Signs Warning signs give notice of a situation that might nor be readily apparent. Care should be taken to use warning signs sparingly and only when necessary so compliance is nor degraded. If warning signs are used to provide informacion during seasonal events, they should be removed or covered when the condition no longer exists. Typical applications of warning signs include:

• changes in horizonral or vertical alignment; cross-section changes such as narrow bridge or low clearance; • roadway surface conditions such as grades, drops, or bumps; • advance traffic control such as "STOP/YIELD AHEAD" or "SPEED REDUcnON;" • traffic How including merging, adding, or end of traffic lanes; • change in advisory speed; • weather conditions such as "BRIDGE ICES BEFORE ROAD;"and • nonvehicular warnings including deer, canle, horse crossings, or other &lremative safety considerations. Placement of warning signs is very critical (ITE, 2001). Placement is detamined by the time necessary for PIEV: Perception, Identification (undemanding), Emotion (decision-making) and Volition (execution of decision). Depending on the type of sign, PIEV can vary widely. Placing signs coo Jar in advance of the condition could cause d.rivm ro foq;et the warning because of other driving clistracrions; that issue is compounded in urban environments. PIEV requires the posted or 85th percentile speed be used for dctcnniniDg the ideal placement. The application of various PIEV times i.s highly dependent on the application; the!ttore, the engineer should consult the MUfCD for further guidance on the placement ofeach warning sign.

3.23 GuUU Signs Adequate signing at intersections to show route directions and destinations includes advance ootico-of..route junctions and rums, directional route markings at the intersection and destination signs showing the names of important cities 134 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

, and towns with directions and distances. The MUTCD provides very basic instruction on the types of guide signs; ',how chey can be assembled; background and lettering colors/siu:s; and limiced information on location of very specific 'signs such as "distance" signs. The engineer should consulr rhe MUTCD directly for this basic informacion on rhc: 'guide sign in question.

4.0 REMOVAL OF UNNECESSARY TRAFFIC CONTROL DEVICES Although che MUTCD is generally dear on the warrants co install various signs, particularly mulciway scop signs, local citizen or political pressure frequencly results in the placement of unnecessary signs. Changes in local conditions may render current signs, markings, or other TCDs unnecessary, redundant, or unduly restrictive. Removal of sorne signs (STOP, YIELD) may require legal action at the agency level. Other signs, such as parking and regulawry signs, may also require legal approval prior to change or removal. Current practice indicates sign removals are i nfrequent unless a MUTCD standard has changed and there is a need to conform co new uniform standards. Even in these cases, conformance through removal is a slow process. The MUTCD does give guidance on the removal of uaffic signals provided changes in craflic patterns eliminate che need for the signal (FHWA, 2003). Typically, the signal is replaced by an alternative TCD; however, it is possible che TCD is removed entirdy. An engineering study would need to be conducted to determine if the traffic control signal is no longer justified. If properly justified, removal of a signal may be accomplished through the foUowing seeps: • Determine the appropriate traffic control measure to be used foUowing the removal of the signal. • Remove sight-distance obstructions. • Notify the public of the removal_srudy through appropriate means, such as signage at the intersection seating •Traffic Signal Under Study for Removal." Make sure the sign is visible to all road users. • Cover (or place in flash) the signal heads for a minimum of90 days and install the alternative TCD. • Conduct appropriate engineering studies (safety, operation, ere.) to determine if rhe signal is no longer needed. If removal of the signal takes place, leave the poles and cables in place in rhe event further study indicates the signal should be installed again. If necessary, a more thorough procedure to justify the removal of unneeded traffic signals is described in the repo .rt User Guid~ for the &moval ofNot Nmkd Traffic Signals (JHK, 1980).

5.0 EFFECTIVENESS OF TRAFFIC CONTROL DEVI(ES T he effectiveness of signs and markings or the need for additional control can be measured in a variety of ways. Among rhe methods char can be used to determine the effectiveness of traffic control devices are the following: • roaduser compliance studies; • before-and-after srudies of traffic characteristics, collision records, and enforcement records; • measurements of the device's performance based on its condition; • analysis of complaints or comments &om ciriu:ns; and • inappropriate use of a TCD. Driver and pedestrian studies can provide an objective, quantitative evaluation of existing TCDs and give guidance fc> r corrective action. Measures of efficiency are usually described by changes in type; frequency and duration of crallic de:lays; spot or cravd speeds; and increased compliance of drivers or pedestrians to traflic regulations and control devices;; Safety evaluations are provided by changes in type and frequency of traffic coUisions or conflict. However, other mea.sures of traffic flow and/or safety conditions can be selected as decision-malcing variables in a before-~d-afte£ study. ·

Traffic Control Device Studies • 13~

5.1 Road User Compliance Studies Studies are commonly conducted to evaluate driver, pedestrian, or bicyclist compliance with a specific TCD or regulation. The noncompliance problem appears to be concentrated in specific situations and/or with specific TCDs. Typical problem areas include: • exceeding advisory speeds and/or posted speed limits; • not stopping at stop signs; • not stopping at right-tum-on-red (RTOR) locations; • violating the red signal; • violating active railroad (RR) grade crossing signals; • violating left-turn-lane signals; and • traveling too fast for conditions. Studies of road-user compliance with stop signs, traffic signals, no-turn restrictions and RTOR provisions are discussed specifically in Chapter 8. The techniques presented can be adapted to study compliance o.f most traffic regulations and control devices, for example, parking regulations, advisory signs, speed limits and parking regulations.

5.2 Before-and-After Studies Before-and-after analyses arc commonly used to evaluate the effectiveness of highway or traffic improvements. The criteria for evaluation may be economic, or based on measures of the efficiency and safety of traffic flow through the improved roadway section or intersection. Economic assessments are often expressed as the dollar value of the benefits to the road users, the adjacent property and the general public as well as the actual costs for making the necessary improvements. These economic evaluations are generally computed on an annual basis. A before-and-after study should be planned as an integral part of the evaluation process for any significant traffic improvement. The design of beforeand-after studies is described in Appendix A. Common sources of mistakes made in comparing before-and-after dara arc (McShane and Roess, 1990): • poor choice of time periods for before-and-after data coUcction; • inadequate or noncomparable data; • f.illure to allow for a stabilizing period for the public to adjust to the change; • f.illwe to take into account other changes that may .affect the situation; • lack of control data to account for traffic trends; • f.illure to account for exposure; and • evaluating a change as significant when in reality the change is due to chance variation. When data arc collected for a before-after comparison of traffic movement characteristics such as speed, volume and delay, time periods should be selected to minimize the influence on the results. For example, if before the installation of a speed zone a speed study was performed in the early afternoons of several weekdays, the after study should be made on similar weekday afternoons. Observations should be spread over a period of days wherever possible. There is less chance of selecting an abnormal period of time if the data for several days are averaged.

5.3 Changes in Spot Speeds Spot speed data may be collected in several ways. Commonly used methods include individual and all- (or almost all-) vehicle selection. The individual vehicle method takes a sample of the vehicles traveling along the roadway segment using either 1'21:: a

lAAPJ.IIAl

r\~ T~AM
C"'-lf:IMCCDit...Jt: CTI Ir"\ICC ")Mn Cr'\lTtntd

, dirca measurements with count tubes, radar or laser speed guns, or calculating the rime a vd:ticle travels over some shon \ (predetermined) distance. TCD studies usually use individual vd:ticle methods because a relatively small sample ofspeeds is ' enough to infer if the device had any dfca-positive or ncgative-<>n speeds. Otaprer 5 lays out the basic &amewodc for 'conducting a spot speed study and should be consulred if the TCD is expected ro have some effect on speed.

5.4 Evaluating Safety Improvements TCD effectiveness is often evaluated in light of improvements in safety after a device has been installed. In designing befon:-and-afrer studies, the period of time considered before and after the improvement must be long enough to observe changes in collision occurn:nce. For most locations, after periods greater than 1 year are selected. Results of before-a~d after studies of safety improvements may be confounded by factors including changes in cxposun: (such as traffic volumes) and nearby changes that may affect collision occum:nce. Evaluating the dfc:aivmess of a single improvement may be further lwnpered by the fact that roadway improvements may be done in packages. For example, the installation of signals may be accompanied by geometric changes in the intersection and/or improvements on intersection approac!tes. It is best to include data in full-year increments before and after the improvement. This will help avoid any d.il'ference that might exist because ofseasonal fluctuation in traffic patterns. For instance, if a 6-month period is compared immediately preceding and following an improvement that was made in March, the before situation reflects winter conditions, while the after period reHcccs summer and autumn traffic conditions. If winter crashes were more frequent than summer crashes in the locality concerned, then the after phase may include a n:duccion in collisions not necessarily attributable to the traffic engineering change. If collision statistics are available for only 6 to 8 months prior to the change, it would be better to select an after period for the same months in the following year. It is generally acceptable to compare periods 0 f s~eral years' before and after; however, if a longer time period is used, it is especially important to take changing conditions into acco.unt (such as trends in car sizes, 55 mph [88.5 kmlh] speed limit). Tabular comparisons of collision fiequency should be supplemented by collision diagrams to emphasiu the relative frequencies of different types of crashes before and after a specific installation. For instance, after the installation of multiway scop signs, collision diagrams might reveal a decrease in right-angle collisions, but an increase in rear-end collisions. Crashes can be separated by a severity classification in making comparisons (see Exhibit 17-4). Collisions may be tabulated as shown in Exhibit 17-5 or included on a collision diagram such as the .one shown in Exhibit 17-9.

6.0 TCD CONDITION Day-to-day traffic operations involve an awareness of the effectiveness of the TCDs placed in the field. Srudies must be performed and inventories maintained to determine: · where the devices are and their condition; • how effectively the devices arc performing; and • the maintenance needs of the devices. Many ti~es this is done through an asset inventory program such as those discussed in Chapter 15.

6.1 Sign Retroreflectivity It is periodically desirable to measure the rctroreflectivity ofsigns' markings. The MUTCD provides limited guidance on maintaining minimum reuoreHectiviry levels for signs using various maintenance methods, as well as tables containing minimum rerroreBectivity thresholds and descriptions for various sign types (FHWA, 2003). This guidance is based on information published in FHWA's MaintAining Trr~ffic Sign &tror#ctiiJity, which should be consulted for further information as necessary (FHWA. 2007). Exhibit 7-10 provides minimum retroreHectivity thresholds for various sign ~ that should be maintained. The MtiTCD should be consulted for changes in thresholds over time..


~~~~wt~·

.'<.

t€1$~~:$-.~. Sheeting Typo ASTM 04956·04)

Sign Color .

White on GrGQn

Black on Yellow or B l ack on OranQe White on Red Black on White (!) (!) Q)

.

(!)

I

Beaded Sheeting I W':G0!:7 W';G2: Y';O'

v·:o·

Prismatic Sheeting

II Ill, IV VI VII, VIII, IX X Ill W':G2: 151 W';G2:25 W:t250:G2:25 W2:120;Glt15 Y2:50;02:50 Y
Additional Criteria :Nerhead

Ground-mount a Q)

(!) (!)

-

The minimum maintained retrorellecllvlty l evels shown In lhls !Able are In units of cdllxlm' measured al an observation angle ol 0.2• and an entrance angle of -4,0°. For texl and line symbol signs measurng at least 1200 m m (48 In) and for an sizes ot bold symbol signs For text and line symbol signs m easuring len lhan 1200 mm (48ln) Mlnlf!'lum Sign Contrast RaUo 2: 3:1 (while retroreflecUvlty .. red retroreftecllvlty) This sheetlng type should not be used for this color lOr this application •

Source: 2003 Manual on Uniform Traffic Control Devices.

Sign re~ordlectivicy standards will be increasingly important to highway agencies because each agency will be required to implement a sign assessment or management program by January 2012, with sign compliance by January 2015. Thi.s section covers assessment and management methods used to evaluate when signs should be replaced, and includes visual inspection, taking actual reuordlectivicy measurements, service life, fixed time period and a control sample. Other methods can be used as long as they are documented in an engineering srudy and meet the thresholds provided in the M'UTCD. The: only signs that may be: excluded from recroreflectlyicy standards are parking; standing, walking, hitchhiking, crossing, adopt-a-highway, any sign with a blue or brown ~ackground and bikeway signs intended for exclusive use by bicycliSts or pedestrians. 6.1.1 Assessment Muhods The actual assessment of traflic signs can be done by visual or measured methods. VISual inspection is the preferred method because it can be done much faster and is generally easier. Visual assessment is conducted by a trained inspector from a moving vehicle at posted speeds. The inspection should only be done during nighttime: conditions with low-beam headlights. The inspection could be done in a number of ways; however, the typical method of assessment requires the inspectOr to use a set of signs that meet minintum threshold requirements for rettoreflectivicy as a base for nuking decisions about 6dd measured signs. These signs arc: viewed prior to conducting a nighttime fidd srudy as a "calibration" for . each inspector. These calibration signs could be insralled in a maintenance yard or at known locations along a corridor where various signs are located. Trpically, the inspector uses the amount of time he or she can read the sign at the posted speed as the threshold for replacing the sign.

An alternate method of assessment is to measure acrual retroreflectivicy using a retrorefiecromerer. This method is more time consuming; however, the advantages are that assessments can be conducted during daytime or nighttime conditions and the actual measurement is more exact than a visual inspection. A direct comparison is made to minimum retroreflectivicy levels for various sign types noted in the MI.JI'CD. Trpically, a portable rettoreflectometer is used to conduct the srudy, such as the one shown in Exhibit 7-11. Newer methods are cunencly under devdopment to try to collect actual measurements from a moving vehicle. At the time of this publication, there was limited success using this method of measurement; however, fururc efforts may have more success. 6.1.2 Mantlgemmt Methods In many cases, it is easier for an agency to have a management plan in place for replacing signs. In this case, e:~.ch individual sign is not assessed, so signs are frequently replaced before minimum rctroreflecrivity measurements arc reached. Three methods for managing sign maintenance are brieRy covered in this section..

Source: DELTA.


: FllSr, signs ace often replaced based on their e)Cpected service life, usually indicated by the sheeting warra my period, rnea·~ surements from field-insralled signs over some period of time, or an alternate method. This method re
6.2 Pavement Marking Retroreflectivity More reccndy, guidance on minimum levels of retrorellecriviry for pavement madcing have come to light. Although nor specifically mandated by FHWA, this guidance is useful to highway agencies looking to make m eir roads safer during nighttime conditions. FHWA has provided these recommendations through a simple table shown in fJ(hibit 7-12. The values represented in Exhibit 7-12 are minimum rerrordiectivity levels with ·a nd without raised reflective pavement markers and roads with fully marked and centerline only marking configurarions. ~J.

_,.i!~·

_1 [-;l.,

·~tt~~

......., ..

~:!:mfl~~~~~~~".<"··~·-~1 ·- ~ -~ !\.-~~~~.. ,'li.:,'·t~:...*·r~~:pg, .h~ 1s

R.o,.dway 1\.IAl'ldn& Coull:ua"Udon

."

:

_

·_

Without R.R.Pl\:ls !5 50 milh

55--45 ml/!a

Fully n:uukcd roadways (w.ith <:c:ntcr line, 40 lftlle lines. and/or edacline, u needed). Roadways "'"ith center lines ooly I 90 • Applies to both yellow and white pavcmtnt ruarlciogs. 1 m.i.lb. - 1.61 la:oJh

Source: PHWA-HRT-07-059, Table 11, page 33.

'\Yhb ~ 70

JD1Ib

RRP:\l.s

60

90

40

2 50

S7S

50

-

Pavement markings can be evaluated using similar methods discussed in Section 6.1. However, when conducting ~ assessment of a pavement marlcing, the equipment used is different. In many cases, a pavemenr rerroreflectomcret' 15 wed such as the one shown in Exhibit 7-13. lr is important to take multiple readings along a shon segment of rhe marking and average the readings to account for variation in paint thickness, as well as the proponions of beads ::;1ft~

Source: DELTA.

Source: Gamma Scientific LLC.

Traffic Control Device Studies • 13 ~

d~ents th:n may be wed in addirion

to the paint. Becoming more popular are reuoreBectomerers which are attached co vehicles coUecring asset inventory daca at highway speeds. An example is shown in Exhibit 17-14. This method is gaining acceptance as measurements are becoming increasingly accurate.

6.3 Feedback from Citizens Roadway wers will often inform agency personnel chat certain traffic conuol devices are damaged, out of order, knocked down, or needed. While some comments from the public are very negative and charged with emocion and . others are very calm statements of filet, all are valuable sources of information on how the system works. Employees of meet and highway organizations, police and other governmencal employees whose duties require that they uavd on the roadways should be encouraged to repon any damaged or obscured signs ac the first opponunicy. Many agencies have found a system for receiving. recording and responding to complaints. Such a system helps ensure prompt conective accion is cahn when needed and allows for prioritizing other activities. Also, by recording complai.nts, a permanent record is established that can be reviewed from time to time. These records can form a basis for changing work emphasis areas, reallocating resources, or beginning a ttaining program.

7.0SUMMARY [riscallacion ofTCDs is a common task, especially as it relates co traflic signals and signs. However, TCDs should not be inStalled if they are not warranted. The MUTCD is a signi1icant resource that provides a foundation of warrants applicable to the installation of such devices, especially signals and signs. IfTCDs are inscalled char are not wah-anted or have nor operated as imcndcd, guidance for removing the devices is provided in this chapter. However, the amount of work to remove a TCD further necessitates the need to think through whether. it should have been ~ed in the first place. New or controversial TCDs often need studying to determine if the device a.ctually worked as intended. That subject is covered in this chapter, with references co the appropriate chapter. Many times, TCDs need to be ftequendy inaintained to provide the performance intended when they were inscalled. Users of this chapter wishing to obtain relevant TCD-related information should consult the MUTCD or ITE's Traffic C1111trol Devic~s Handbook. Other resources are provided in the reference Section of this chapter.

8.0 REFERENCES 8.1 Literature References Box, P. C. and J. Oppenlander. Manlllli oJTrriffic Engirumng Studia, 4th ed. Washington, DC: lnstirute ofTranspo~tion Engineers, 1976. Fcden1 Highway Admi.nistntion. MlllwtJ on Uniform TtttjJU unml DMas. Washington, DC: FHWA, 2003.

Federal Highway Admini.!cration. UpJtztes to &s~=h on Rrcommmdtd Minirnwm Lewis for Pa~~m~mt Mar/ring &tro+aivity tQ Mm Driwr Night Vt.fibiJjt.J Nmis. Washington, DC: FHWA, 2007. Institute ofTransporcation Engineers. A Prognmt for Sch«Jl Crotring Prot«tiQn: A Rr~J Pruniu, 3rd ed. Washington:

DC: JTE, 1972. In.stirute ofTransporati
In.stirute ofTransporation Engineers. Trip Gmmztjon. Washington: DC: ITE. 1988.

1411 • MANIIAI OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

, Institute ofTransporcuion Engineers. Trip Gmerarion, 8th Edition: An ITE Informational Report. Washington, DC; ITE, \2008.

'-JHK and Alsociates. A Uur GuUkforth~ &moval ofNot Nmud Traffic Sigllais, FHWA-IP-80-1 Z. Washington, DC: U.S. Depanmcnr ofTraruportacion, Federal Highway Administration, 1980. McShane, Wand R. RDess. Traffic EngitUmng, Englewood Oiffi, NJ: Prentice Hall, 1990.

8.2 Online Resources Federal Highway Administration. "Maintaining Traffic Sign RcuorcBcctivity." Washington, DC: Federal Highway Administration, 2007. fbwa.dot.gov/reuo. kccsscd July, 2009. Florida Department ofTra.nsporacion. Opmuions-MIZilual on Uniform Trrz.ffo StudUs. Department ofTra.nspon:ation, Tallahassee, FL, 2000. www.douate.B.usffrallicOpcrations/Opcratioru./Studies/MUTS/MUTS.shttn.


Chapter 8 o o o • o o

'I

o • o •

0

o <> ·> o

t.

I

o 0

0

C

•

o 0

<

'

0

•

o o 0

10

•

o o,. 0

<> o

o o o 0

<1

o o o • o o o o o 0 •

•

o o 0

o o o o o o o o 0

0

o 0

'•

•

0

0

0

O

0 0 0

o 0

•

Compliance with Traffic Control Devices Original by: Donna C. Nelson, Ph.D., P.E. Edited by:

Christopher M. Cunningham, MCE, P.E. 1.0

2.0

3.0

INTRODUCTION

143

1.1 Purpose

144

1.2 Applications

144

TYPES OF STUDIES

145

2.1 Study Locations

145

2.2 Data Needs

145

2.3 Compliance Data

145


146

3.1 Personnel, Equipment and Training Requirements

146

3.2 Time and Duration of Study

147

3.3 Sample Size Requirements

147

3.4 Types of Compliance Studies

148

4.0


155

~

WMMA~

1~

6.0

REFERENCES

156


156


157

1.0 INTRODUqiON ser disregard ofTraffic Control Devices (TCOs) is of increasing concern to !:hose involved in traffic engU!cering and roadway safety. The level of public olxdience is one test of the effectiveness of regulations and devices • the adequacy of publicity and education dealing with these controls and the level of enforcement (Humme.s::'• 2009). This chapter focuses on the srudy of driver and pedestrian compliance with TCOs. The speciJic srud.ics addressed in this chapter include driver compliance wit:h stop signs, traffic signals, no-turn restrictions and right tur.O. .s on red. While these studies focus on compliance at intersections, the techniques presented can be adapted to &rud.y compliance with most traffic regulations and control devices at a wide variety of locations. Bicyclist and pedesaia.J:"'=l compliance with control devices and regulations may readily be measured by these field procedures as well. Additiona-l background information can be found in FHWA Publication RD-89-103 Motorist Compliance with Slllndar~ Tr4f£.t::: Control Devices (Pierrucha, Opiela, Knoblauch and Cringler, 1989). -

U

Compliance with Traffic Control

Devices •

1~

1.1 Purpose Road-wer compliance studies are conducted to evaluate driver, pedestrian, or bicyclist conformity with a specific TCD or regulation. An engineer conducting an analysis that indicated unacceptable compliance with a regulation or device would rethink whether the device was necessary, determine if an additional improv~ment were needed (made larger, brighter, or supplemented with another device), consider enforcing the regulation through added police enforcement, or explore ways to better enforce the regulation, such as automated enforcement (Hummer, 2009).

1.2 Applications There are two basic methods for conducting a compliance study on traffic regulations or TCDs: a fidd study or a controlled study in a lab setting (ITE, 2001). The most desirable study is conducted in the fidd at any roadway or intersection location for which information is desired. However, it may be more feasible or desirable to conduct the study through a questionnaire using pictUres or video in a lab setting. The fidd study is typically considered a better option beca.use the actual observed reaction to TCDs is recorded. The Manwd on Uniform Traffic Control Droica (MUTCD) defines traffic control tkvica as "all signs, signals, markings, and other devices used co regulate, warn, or guide traffic, placed on, over, or adjacent to a street, highway, pedestrian &cility, or bikeway by authority of a public agency having jurisdiction" (FHWA, 2009). Compliance studies may be conducted to: • evaluate the effectiveness of traffic control devices; • develop edua.rion:U progwhs for drivers, schoolchildren and the general public; • determine critical locations for sdecrive enforcement efforts; and

• analyze the effectiveness of traffic improvements through before-and-after studies. A violation could occur for any number of reasons, including a deliberate violation, difficulty in detecting the TCD due to poor placement or visual overload, the TCD is not legible or needs maintenance, or the Violation is a traffic law (for example, speeding) (ITE, 2001). AASHTO survey respondentS indicated the mOst common TCD compliance issues were 55 mph speed limits (65 percent), TCD miswe (48 percent), driver perception of need (33 percent), signal timing (30 percent), TCD misunderstanding (26 percent), lack of respect for TCDs (24 percent) and stop signs (22 percent) (Hicks, 1985). Police commonly cite violations ofTCDs on traffic collision reporcs. Although the extent of the problem is not entirely known, manyTCD violations can be attributed to facrors other than the driver's ability to comply, such as poor placement or visual overload, the TCD is not legible or needs maintenance, or poor design/uniformity issues. A survey conducted by Piettucba et a!. found that the major compliance issues appear to be concentrated in specific siruatioru and/or with specific TCDs (Pierrucha, Opida, Knoblauch and Cringler, 1989). These include: • exceeding advisory speeds and/or posted speed limitS; • not stopping at stop signs; • not stopping at ~t tum on red (RTOR) locations; • violating the .red signal; • violating active railroad grade crossing signals; • violating left-tum lane signals; and • traveling too rut for conditions. Although not included in this list, compliance Studies could also be conducted on other devices identified in the MUfCD, such as temporary TCDs (like those wed in work zones), pedestrian crossings, pavement markings and craflic control for school areas (Hicks, 2005).

\2.0 TYPES OF STUDIES ;Compliance studies arc performed to determine the proportion of roadway users (motorists, bicyclists and pedestri~) that conform with regubdons and regulatory'TCDs. Prior to any analysis, the analyst should have a keen sense of the study location, its surrounding attributes, weather conditions and time of day (week or month). Iris ~pecial1y imporranr to have a clear definition of the compliance issue and how one intends to record it. ·

2.1 Study Locations Compliance studies may be conducted at any location where there is some indication a compliance problem exists. ~ noted e:utier, compliance srudies are typically conducted on driver behavior such as exceeding advisory speeds, and obeying traffic signal indications and regulatory and warning signs. However, with the increase in bicycle and pedestrian traffic, points of conflict berween drivers, bicyclists and pedestrians can make it necessary ro evaluate TCO complia.n~ mues of oilier (nonmotori~d) road users. Srudi~ may be initiated from citiz.cn complaints, observance of a specific traffic problem•. the existence of high accident locations, or an analysis of a new TCD device.

2.2

Oat~

Needs

The data required for compliance srudies ue of rwo types: 1. an inventory and sire description; and 2. compliance data. The sire diag.run and description are v~ important. During the data analysis, a good description of the sire can provide important informacion in interpreting the srudy rc:sults. Inventory and site information can be recorded on a condition diagram as shown in Exhibir 17-10. The site diagram should be approximately to scale and include all the physical fearures of the srudy site {such as sidewal.ks, crosswalks, bicycle lanes, vegetation, driveways, embankments, signs, traffic signals, markings, roadway shoulders, abutting land uses, bus stops and other characteristics or conditions that may affect visibility and/or sight distances). All streets or highways should be labeled by their official names and/or numbers. The diagram should indicate estimated measurements for roadway widths, shoulder widths and lane widths. A checklist of items commonly required for a complete site description is shown in Exhibit 17-11. All of chese items may not be required for every study. ·

2.3 Compl·iance Data Data collected for different rypes ofstudies of driver compliance arc shown in Exhibit 8-1. These should be modified for specific applications. Conditions of compliance and noncompliance must be weU~efined before the srudy begins. Even with wcll~efined compliance conditions, obse.rvacions and decisions can be difficult at best. Lack of clearly defined, well-understood definitions can result in inconsistent and unreliable data. For aample, four categories of driver action arc commonly used for stop sign compliance studies: 3. not stopping; :

4. practically stopped; 5. stopped by traffic (forced stop); and 6. voluntary full stop. Categories 1 and 2 represent total and putial noncompliance; categories 3 and 4 represent compliance. The choice berwecn categories 1 md 2 is at the discretion of the observer and may produce confusion and inconsistencies among data collected ?Y different observers.

]~l.~~~~~~~~!/®i:~!(-:,f[~if~~~~ ':!-.. :::•.. ~:~··:r-?f~'";:-t;~t'1c'{#:5 :·~ Running the ml indication

Violations of turn resuictioru

Turning movement

Turning movcmenr

Driver action

TyP" of traffic conrrol

Vehicle r:ypc

Vehicle type

Cycle lengrh

Temporal restrictions

Speed limit

Approach volume

Approach volume

location

Peak hour Location Improper rigbc tum on red

Stop sign compliance

Queuing conditions

Turning movemem

Driver a.ctionfT~ of stop

Queuing conditions

Location

Driver AccionfTypc of stop

Vehicle type

Vehicle type

Cross street volume

Cross SUCCI volume

Approach volume

Approach volume

Pedestrian paths ·

Pedestrian paths

Pedestrian signals

Location

3.0 DATA COLLECTION PROCEDURES 3.1 Personnel, Equipment and Training Requirements Because compliance srudies are "yes/no" data, they are most often carried out using manual methods. Typically, observers record road-user behavior using rick marks on paper forms or handheld mechanical counters. In some circumseances, alternative manual data collection methods such as electronic count boards or computer-based m.acros can be used. An example of how to code a spreadsheet rime-scamp macro is shown in Appendix E-1. These alternatives are typically used in applications where additional data may be necessary, such as a rtllearch study. An ex.a.mple would be a pedestrian-walk-signal compliance srudy, where the basic compliance question contains only data at the crosswalk; however, the analyst may also want to COllliider jaywalking pedestrians crossing at midblock locations and/or know at what signal phase indicarion rhe pedestrian crossed (Schroeder, Rouphail and Lehan, 2008). Buttons on eleruonic count boa,rds or keys on a laptop computer can be set to record specific events. One or more observers are used depending on the traffic demand and the complexity of data needed. Because the objective is to count road users violating traffic regulatiollli, che observers must be as unobcrusive as possible. For this reason, video cameras may be useful for che collection of some types of compliance data and offer the major advantage of being inconspicuous. Video offers other advantagtll over manual methods, such as the ability co observe che sire during the same time period multiple cirntll, validation of compliance between multiple data collectors (virtually eliminating errors in coding through crain.ing), use as a clara log that can be used at a later dace and multiple camera angles, which can be synchronized through a split-screen device. Data, however, must still be reduced from the video images and entered on paper forms or into a computer program ror analysis, so use of video does have the disadvan· tage of adding rime to the analysis process. If video is used, the analyst can expect to spend a minimum of twice the normal time gathering necessary data.


~.2

Time and Duration of Study

!he time and conditions under which compliance srudies u e conducted can affect resulu. These studies ue usually ~rformed in good weather and under "normal" uaffic conditions (when there ue no circumsrances presc:nr chat rnay affect the results). However, it is possible the TCD Wider investigation may be intended for use in an ab n ormal event, such as various pavement markings developed for nigh crime use in rainy weather. Samples ue taken tO cover all applicable periods of the day unless it is imperative that a specific time period or weather evenr be taking place during the analysis. Data are commonly CQilected in 5- or 15-min. intervals over a period that allows the collec tion ofdata for more than the minimum sample size as described below. Data should be collected under the conditions during which the problem has been observed or is likely to be most evident. For example, traffic delays and crashes generally occur more ofren during peak traffic periods; therefore, traffic data should be collected during these periods. Excessive speeds on some streets may occur only when tra ffic volumes are light to moderate. Compliance with TCDs in school zones is generally of interest during school hours , specificallY when srudenrs arc traveling to and from school. Compliance studies may be performed in off-peak periods to provide a comparative analysis of the violation problem. If results arc to be comparable for before-and-after analysis, simila r conditions must exist during both periods of data colleccion.

3.3 Sample Size Requirements Compliance with a traffic regulation is essentially a "yes/ no• matter; therefore, Equation 8·1 can be used to calcu!m: che minimum sample size for each observance study si re. pqKz N =Ez -Eq uauon . 8- 1 where N

a

minimum sample siz.e

p ,. proportion of drivers or pedesuians that observe the traffic regulation

q ,. proportion of drivers or pedesuians that do not observe the traffic regulation K = constant corresponding to the desired confidence level

E ~ permitted error in the proportion estimate of compliance The p and q values can be estimated by a preliminary study. If more:than one compliance issue is being determine d from the daraset (such as in Example 8-1), the higher compliance estimate should be considered for the entire sample The usc of0.5 forp and q provides the most conservative estimate possible of the sample size required. It is, in esstace. saying that 50 percent of the vehicles (or pedcsuians) fail to observe the traffic regulation. The sample size require: d. decreases as the value of p decreases or increases from 0.5. The sum of p and q should always be equal to 1.0. The: consrant K depen ds on the desired level of conlidence. Exhibit 8-1 presents some precalculated values of K bastdo ~ confidence intervals that represent from one to three standard deviations from the mean. A value of 1.96 for Kis o&e~ selected for a confidence level of95 percent, which corresponds tO one chance in 10 that the proportion of violatioo.S found would occur by;chance alone. The permitted error E is based on the precision that is required for this estimue:A common value for the permitted error is 0.05 or 0.10.

Compliance with Traffic Control Devices • 117

·\~\.
•• a_te-\..\.(O.'~ .. "'·

ft~.-.:~~~~:-:-"~~~.~~·~""4

Y•"

Coo.rtant, K

Coolidenc:e ~I (%)

1.00

68.3

1.50

86.3

1.64

90.0

1.96 2.58

95.0

99.0

EXAMPLE 8-I: Sample Siu Calculations with Change in Proportions

What is the difference in sample sizes necessary if stop sign violations are assumed to be 20 percent and 50 percent.

So/IliUm: Assume a 20 percent violation rate (p • 0.8, q =0.2), a 5 percent permined error and K a 1.96:

N

{0.8)(0.2)(1.96) • {0.05)2

2

•

Equauon 8-1

N-246 If the violation rate is assumed to be 50 percent (p = q = 0.5), then

N. (0.5)(0.5)(1.96)

2

(0.05) 2 N-385

Approximately 140 more vehicles are necessary for the sw-vey if the violation race were an equal 50/50 split. Exhibit 8-3 shows conservative (p = q • 0.5) sample size requirements for permined errors of 5 percent and 10 percent, with a confidence level of90 percem or 95 percent. For a more dcta.i.led explanation of K. sec Appendix C. Box(1984) suggests samples of 100 are ofu:n adoquaa: to indiact: compliance with TCDs, except when violations are rare.

3.4 Types_of Compliance Studies Typicalfidd sheets and general data collection procedures for several studies are shown in this section, followed by a discussion of other compliance studies that could be adapted to the methods discussed in this chapter. Exhibit 17-10 shows a sample site condition diagram that corresponds to a checklist in Exhibit 17-11 and is used to record the site description and inventory for sign2lized and unsignalized intersections, respectively. Exhibits 8-4 through 8-8 are used to record and analyze road-user compliance datAl. These forms should be modified as needed for individual sites and studies. The compliance of each road user to the craffic regulation is recorded by placing a cally mark in the appropriate space on the fidd sheet. The observations are continued until the desired sample size has been obtAiincd.

., ...

~<•:

Driver Observance of Stop Signs Field Sheet

location ____________________________________________________

~--

nme

to

Weather-----------------------------Non-Stopping

Practically Stopped · 0 to 3 mph

•.

Stopped by Traffic

I

I Voluntary Full Stop

"'0

?i ..; vi z

.

Right

-

Straight

Left

Straight Voluntary Full Stop

Right

Stopped by Traffic

Practically Stopped • 0 to 3 mph

Non-Stopping

"'0

?i ..;

I

vi z

I

Pate

Recorder _ _ _ _ _ _ _ _ _ _ _ _ _ _ __

Sou=: Box and Oppenlandcr, 1976, p. 184.

Swp-sign c-ompl.imrce dau are recorded on the form in Exhibit 8-4. For a four-way stop controlled intersecti.on, it will be necessary co use two forms, one for the north-south direction and one for the east-west direction. An observer for this study sb.ou1d be positioned to see the study vehicles as they arrive :1.t the stop sign and the aoss traffic on the through street. Dau are recorded by movemeac for each driver action: full stop, almost stopped, forced stop and no stop. A foil stop is defined as a "complete cessation of movement, however brief." Nearly tf(Jpped is defined most commonly as • ~ 3 mph (5 kmlh). • Afo~ui lfQP occurs when the motorist is cequ.Ued co stop because of conflict with cross traffic and pederuiansi no stop is commonly defined as"> 3 mph (5 km!h)." If more than one observer is used, m~ sure all observers are categorizing vehicles the same way. Cars and trucks (or large vehicles) can be differentiated by cmering tick marks in the cells for car or truck. Compliance with Traffic Control Devices • 149

·~~~~~r~i~~~~t~~¥::.~~--:f1E~t~~-~i~:~~t~~l:~±-::·~ ~uon

_______________________________________________________________________________

Oiceccion ofTm·el

Ttme Dace

Area Type.___________________________________

ro _ _ _ Weather ----------------------------------------------Observer Left

Right and Through

Total

Lefc

Rlgbt and Through

To cal

c A R

s

T R

u

c

K

s

Total Cars

To cal Trucks Total Noces:

Source: Adapted from Mttt4rist Onnpli4nu with St4ndttrd Traffic Cmrrol Dn~ices, FHWA-RD-89-103.

No-f4t-tum mnpllanu daca may be recorded on the form in Exhibit 8-5. The observer is positioned to observe the study vehicles as they approach the no-left-turn location. The nwnber of vehicles passing the study site that make an illegal left rum are counted along with the total number of vehicles. Through and right-turning cars and trucks may also be recorded.

150 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITION

:~ ~~ frfcl@r'[.'fd!)if~ -.:-.7-·.-.::-: ~ .-.-. ---~--~· - -"~

Driver Observance of Traffic Signals Field Sheet

lo~tion

nme

_______________________________________________________________________ to Weather ________________________________ N.S. E. - · W. ,,, V H

,., - ·

z

1: 1:~

leu3!S padwnr

1.!g

~ c .,3

pa~

"'Q.

~

~

.."'

Q.

~!!!.

U 013J9 Jaijt MDU&A

5'

. .."

....

~

~

~

~

"

Cl

;;

~ Ual~J9

....... .

.... ·- ...-· ~

-· ...

N. s. €. w. on

I

I

I

~"

I j

t

Green

.

10

~

~

c

..

.,3

a.

"' "!!!.

~-

..~

Yellow after Green

0

"'

ID 0.

~

.""' ~

"'

~1r

Red

Jumped Signal

·~

I"<§:

1

&i

:;:::: i

z R~rder

Date-- --------------------

_______________ _____ ____ _______


The drivtr compUan.« with traJfo: rignals study is particularly concerned with the response of drivers approaching the sign~ during the clearance interval and red phase. Dara may be recorded using the form shown in Exhibit~- The form Nsspace to record drivtt actions on four incersection approaches. D!Mr behavior is rallied by direction of crave! as the vdllde le:r.'t! lb.e: inccrseaion. The observed beh:Mor is related ro the traffic signal indication seen by the c:lrMr when the vehicle enterS the imej:"scaion. Entry is usually defined as crossing the near-side curb line. Tbe signal indications arc listed on the d.2,ra form ~ "green," "yellow after green, • "red" and "jumped signal. • The forms have space ro cecord driver actions. Care must be take~ to consider peraption-r=:tion times in classifying violations of craffic signals. Signal compl.iance studies are appropri~c.O:::: during both p~ and off-peak hours. .

Compliance with Traffic Control Devices • 1St

"'

.

,.:-

...

:

·--

Location Direction ofTravd Tune Date

~ ·~··! . !~.---~·lim.:~~~:.

·.

. . • -.,.. -·-·. .....rs

...,.;

"-'

-;~~J!U1.(·~~ ·~>G·-.;1' ·

1-::· ~-

-_~.~~ -vl!...

!"

-~~~ ..1:. -;,~:

Ast2Typc Weather Observer

to

Indication

Action

Cars

Trucks

Total Can

Total Trucks

Turned on G or Y

Stopped on R. Waited for G, Turned on G I

On Green lkhind a W~ter (above) Turned onG Attempted to rum on R, rumed onG Full Srop

' !

Stopped by Cross Traffic On Rd No Queue Stopped by Pedesuian Crossing

No Stop

Full Stop On Red Queue Stopped by Cross Traflic Stopped by Pede$trian Crossing Toal!

~'-------------------------------------------------------------

Source: Adapted from Motorist CompliAnce with S1"4111i4rtl T.mffic Control Devka. FHWA-RD-89-103.

Right-tum-on·mJ..aftu-ftOJJ compliAn« and data on the proper yielding ofsuch drivers ro pedestrians and other vchicub.r traffic can be ·oondua:cd using the fOrm shown in Exhibit 8-7. The measure of violation is the ob.ser-I slowing or brake light application of through aaffic, or interference with pedestrians. Right cum on red lll2Y be identified by sevaal oonditions: • failing to stop oompletdy at the stop line or before aossing the crosswalks a.t an urban incetsection if pcde$trlans a.ce present; • interfering with a pede$trian in either crosswalk being traversed; or • causing sudden slowing of a vehicle on the cross street which has the green indication. I:

\ ,

152 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDrTION

The observer should be able to see the study vehicles as they arrive at the traffic signal and the cross traffic o n the through street. Observe eVery right-turning vehicle that passes through the intersection during the study. Begin a new data sheet every 15 min., recording the following observations. • When the signal is green, record the number of study vehlcles that: • turned on green or yeUow; • stopped on red, waited for green before turning; • stopped on red behlnd a vehicle waiting for the green indication before turning right (rhe turn was executed on green); and • attempted ro turn on ted, signal rurned to green before turn was c:xecured (the turn was completed on green). • When the sisnaJ is red: Determine if srudy vehlcle arrived as a single vehicle or as part of a queue waiting at the signal. • Determine if the vehlcle made a full stop (a brief cessation of movement), was stopped by vehlcular or pedestrian cross traffic, or did nor stop at all before entering the intersection.

Pedestri an Observance of Traffic Signals rield Sheet

Location

Tome

Wuthet

to

Pedestrians crou£na of

SL on the (N.S.f.W.) - Olrtctlon

St. In

Cre>mdStnfiht St•pped from Curb on

(cronw1llc)

Slde

Toto!

~ ~

11!~ ~f

ilL

;

"1~ <:roued o;''""''lv I'

I

111 l ~f!H Tol>l

oor.

Recordtr


Pedestrian compliance with traffic 1ignals is recorded on the field sheet shown in Exhibit 8-8. Pedestrian behavior is noted as the signal indication on which the person steps from the curb onto the pavem~nt of the inrersecrion. Signal indications are observed for that intersection a.pproach which the pedesuian is crossing and include ·~en," "yellow" and "red." If signal indications arc separately provided for pedestrians, the appropriate indications arc "walk or countdown," "Hashing don't walk"and "steady don't walk." The field sheet is designed to check on crosswalks with a provision for recording diagonal crossings. If a bicycle compliance study is not conducted separately, the compliance of bicyclistS can also be recorded by using the lcner "B" for bicycle riders and •p• for pedestrians. It is important the analyst undetStand that any compliance srudy could be conducted using similar methods and data oollection sheers. For instance, a local jurisdiaion may want to check speed oompliance along a certain roadway segment with a 45 mph (72 km/h) posted speed. A threshold could be used to determine the compliance of drivers on the roadway. Alternatively, a city ll)ay want to study pedestrian compliance ar traffic signals as shown above. However, there may be an


issue, such as jaywalking, that needs to lx srudied as weU. This additional compliance daca could easily be added co the data shttt. The point here is that the methodology and data collection method is the same for any cype of comp liance issl.>e th< .agency wishes to study. In effecc, all that is needed is whether drivers were compliant or not, which is simply "yes/no" data.

4.0 DATA REDUCTION AND ANALYSIS After the field study has been completed, the tally marks are summed to give subtoral and total values. Totals and subtotals are developed for • each direction of movement; • each intersection approach; • each compliance category; • tot:2l number of road use.rs complying; and • tot:2l number of road users not complying. Compliance characteristics are summarized as totals for each compliance category. These subtouls and totals are divided by the sample size to yidd proportions or percentages. This procedure allows results from other compliance srudies (wi ch different sample sizes} to be compared directly. The standard error for proportions is calculated with the same equation used to calculate sample sizes, except the acrual proportions are known. Equation 8-2 is provided below. E

~ "'rn-

·

Equation 8-2

where E is rhe standard error, n is the actual sample size and p, q and K, as previously defined in Section 3.3. The K values corresponding to the desired levels of confidence are provided in Exhibit 8-2. Since compliance will be measured as a percent, the standard error musr also be converted to a percent by multiplying by 100. If standard error abour the mean does not include zero, the analyst says the findings were significant ar the levd of confidence used in the study to determine the "k-value." Although summary statistics fot compliance studies are usually expressed as a proportion or percentage values, other descriptors of central tendency can be developed. Data analysis and statistical summaries are descnbed in deta.i.l in Appendix C.

EXAMPLE 8-2: Analysis of Stop Sign Compliance Driver compliance with a scop sign was ch~cked on a coll~ccor street in an urban area. Based on the calculation i.P Example 8-1, the analyst collected a sample of 246 vehicles. The responses are recorded below. Voluntary Full Srop : 135 Stopped by Traffic: 25 Praccically Stopped • 50 Nonstopping " 36 Total Observed : 246 Solution: The percentages of driver compliance arc computed by dividing each group by the sample size and mulcip(ying by 100. The sundard errors are calculated using Equation 8-2 with a 95 percent confidence interval The: manual calculation for "Voluntary full stop• is shown below.

Voluntary Full.Stop Compliance: (135/246) x 100 • 54.9 percent 2 Standard Error, E • ( •/o.ss • o.4S" ... 1.96 ) x 100 ~ 0.0622 x 100 = 6.2 percent Compliance with Traffic Control Devices • 155

Therefore, the analyst could state that he or she was 95 percent confident that drivers came to a voluntary full stop at stop signs 54.9 :1: 6.2 percent. Because a "k·value" of 1.96 was chosen {based on a 95 percent confidence level), the analyst could say he or she was confident the findings were Jignificant at the 95 percent confidence level because the range does not include zero (48.7 !> x;,;: 61.1). The summary Statistics for each of the four compliance rates are shown bdow. Voluntary Full Stop:

54.9 percent :1: 6.2 percent

Stopped by Traffic:


Practically Stopped:


Nonstopping:


The perctntage of practically stopped, stopped by traffic and voluntary full stop vehicles are typically summed to represent the percentage of traffic "yielding.• which is considered safe in most jurisdictions. Therefore, the total percentage of drivers compliant with stop signs could be determined as 85.4 percent :1: 4.4 percent. Most jurisdictions would likdy agree that an 85 percent compliance rate would be acceptable; however, no refermce is a".-ailable on typical stop sign compliance rates.

S.OSUMMARY Disregard for TCDs is a serious problem that is frcquendy addressed by transportation analyst. Compliano! with such devices is one way to measure the effectiveness of a device intenclcd to enforce laws and regulations, education and publicity programs, or law enforcement efforrs aimed at making drivers more compliant with laws. Many types of common compliance srudies were addressed in the chapter pertaining to drivelS and pedestrians, many of which could be adapted for less-&equencly conducted studies. Users of this chapter wishing to find additional information related to TCD compliance can consult FHWA Publication RD-89-103 Mowrist Complilznc~ with St4ndarri Traffic Control DMces (Piecrucha, Opida, Knoblauch and Cringler, 1989).

6.0 REFERENCES 6.1 literature References Box, P. C. Transportation and Tr~ EngiMmng Handbook. "Traffic Studies.• Washington, DC: !Jutitute of Transportation Engineers, 1984: pp. 546-547. Box, P. C. and J. Oppenlander. Manual ofTraffic EnginJ:mng Studies, 4ch ed. Washington, DC: !Jutitute ofTransporcation Engineers, 1976. Federal Highway Administration. Manual on Uniform Traffic Omtrol Dnices. Washington, DC: FHWA, 2009. Hicks, T. Traffic CDntrol Dntke CDmplimue-Summary ofAASHTO At::Wns. AASHTO Highway Subcommittee on Traffic Engineering. Seattle, WA: American Association of State Highway and Transportation Officials, 1985. Hummer, J. Trttf!U &ginuring Hantlboolr, 6th ed. "Ciu.pter 8: Traffic Engineering Srudies." Washington, DC: Institute of Transportation Engineen, 2009: pp. 8(}-107. Institute ofTransportation Engineers, Traffo Omml Dnias Handhoolr. Washington, DC: ITE. 2009. Pietrueha, M., K. Opiela, R. Knoblauch and K. Cring!cr. Mowrirt CDmplimue with St41ldarJ Trrzffic Omtrol Dntices. Washington, DC: Federal Highway Administration, 1989.

.... .. · - · - · -• • •••-••·• •n•••,..

'""T'

•~•rr

""ta.U"\

c"~Tln~l

6.2 Online Resources Federal Highway Administration. Manll41 on Uniform Trrzffic Control Devict$. Washington, DC: FHWA. 2009. . hup://mutcd.fhwa.dot.gov.

Flotida Department ofTransportation. Operations-Manll41 on Uniform Traffic Studies. Tallahassee, FL: Depmment of Transporcation, 2000. hr:tp:l/dot.state.B.us/TrafficOperuions/Operatioru/Srudies/MliTS/MUTS.shtm.

I

i ~

Compliance with Traffic Control Devices • 157

..

:: •

s::

}>

z ~

r-

0

-n

..... ~

zVl (3

""};' :::1

0

z m z

Cl

~

m

"' z

Cl

VI -I

c

0

;:;:;

·"' "' z

0

m

0

3

~

Cha pter 9

....... . ... .. .... . .... ......... ..... . ... ........................ ... ....... .. . ............ . ~

~

Travel-Time and Delay Studies Original by: H. Douglas Robertson, Ph.D., P.E. Edited by:

DanielJ Findley, P.E. 1.0

2.0

INTROOUaiON

159

1.1 Applications

160

1.2 Chapter Ovetview

160

TEST VEHICLE METHOD

160

2. 1 Introduction

160

2.2 Data Collection Procedure

161

2.3 Data Reduction ancl Analysis

165

2.4 Volume Extension

166

3.0 OTHER TRAVEL TIME STUDIES

171

3.1 Vehicle Observation

171

3.2 Vehicle Signature Matching Method

173

3.3 Platoon Matching Method

173

3.4 Probe Vehicle

173

4.0 REFERENCES

174


174


174

4.3 Other Resources

174

1.0 INTRODUCTION ravel time and delay are two of the principal measures ofhigh"A"aY system performance wed by traffic engince:_.:-s, planners and analym. Vehicle speed is directly related ro uavel rime and delay and is also used ro evaluate u-af-6-c and highw..y systems. There are two types of average speed: rime-mean speed (TMS) (or mean spot speed) ~~ space-mean speed (SMS) (or mean travel speed). Measuring TMS is described in Chapter 5. SMS is covered in tt:;a.J.S chapter by examining the mean uavcl time in relation to the segment length.

T

Travel time varies inversely with travel speed. A travel-time study provides data on the amount of time it takes cot!"~ verse a section of sueet or highway. These data, combined with the length of the section under study, produce m~ uavcl speed. Travel-time and delay studies are conducted when the sources and amounts of delay occurring wich..i- .n the section are also noted. This chapter treats the measurement of delay along a roadway segment. Intersection dd~Y studies are addressed in Chapter 6.

Travel-Time and Delay Studies • 15 ~

1.1 Applications Engineers and planners usc data from travel-time and delay studies in a number of casks, including: • determining the efficiency of a route with respect to its ability to carry traffic relative to other routes · through the use ofsufficiency ratings or congestion indices; · system performance measurements; • providing input to capacity analyses of roadway segments; • identifying problem locations as indicated by delay; • evaluating the effectiveness of traffic operation improvements; • providing input to transportation planning models, trip assignment models and route-diversion models; • providing input to economic analyses of alternatives; generating rravel-time contour maps; • providing input to studies that evaluate trends in efficiency and level of service over time; .and • calibrating and validating simulation models.

1.2 Chapter Overview Travel-time and delay studies may be conducted wing the foUowing methods. • Test vehicle !

• Vehicle observation • Probe vehicle The first requires the analyst to perform measurements while in a moving vehicle in the traffic scream, while the others methods do not. The choice of method depends on the purpose of the study; the type of roadway segment under srudy; the length of the segment; the time of day of interest; and the personnel, equipment and resources available. The mOSt common methods (the test vehicle methods) are presented in detail, whereas the others are qescribed only briefiy. Appendix E contains data forms that are suitable for copying.

2.0 TEST VEHICLE METHOD 2.1 Introduction The test vehicle method requires the analyst to usc a vehicle to make runs on the roadway segment of interest. A typical srudy involves a driver and an observer driving from one end of the corridor to the other while recording times at prcdece~ed points of interest. In addition to uavd times, the analyst has the option of collecting addiriorutl data during rhe field runs including rraffic volumes and detailed information about the effects and sources of delays. Instructions on the driving technique should be dear, particularly if the data collection requires multiple drivers. The test vehicle met}:!od allows the analyst to determine the following attributes along the study route:

• rravel time; • running time; •

type. location, duration and cause of rraffic delays;

• distance traveled; · ',

'I

j

.., (

•

SMS ; ~d

• hourly volume. Agencies usually study travel time ~d delay during the peak hours in the directions of heaviest traffic Row. It may also be desirable to compare travel times, speeds ~d delays berween peak ~d off-peak periods or berween sets of other conditions. Some of these other conditions include ideal versus adverse weather ~d commuter versuS special-eyent traffic. During the peak period, speeds will probably vary, which may make it necessary to conduct separate traveltime and delay studies for differe.nt portions of the peak period.

2.2 Data Collection Procedure The data are recorded as the test vehicle traverses the study route. From these data, uavel time, SMS and running speed may be calculated. This method is applicable to ~y type of route, but is most widely used on arterial streets with at-grade intersections. The recommended minimum total length of route to be studied is 1 mile. Engineers may conduct areawide travel time ~d delay surveys on the major anerials leading to ~d fro111 the centtal business disukt (CBD) ~d d isplay the results as time-contow maps. . · Before test runs begin, observers select the starr point, end point ~d comrol point locations along the route whe re they will record rime mcaswes. On arterial ~d oPter types ofsurface sueets, these locations are usually at major signali:z.ed intersections or other easily identifiable control points. The choice of the ncar curb, far cwb, or center of the intersection :as the control point should be consistent throughout the study route. On freeways, control point locations~ include milepost markers, bridge overpasses, bridge underpasses, exit ramps, cn~ce ramps, or other ea,sily identifiable points. The choice of the bridge edge or center and ramp location (such as the gore) as the cont rol point should be consistent throughout the srupy route. 2.21 Dritting Teclmiqua

The driver of the test vehicle proceeds along the study route in accor~ce with one of the following techniques: • Avaage.Car Technique: test vehicle travels according to the driver's judgment of the average speed of the traffic sueam. • Floating-Car Technique: driver '"floats"' with the traffic by attempting to safely pass as many vehicles as pass the test vehicle. • Maximum-Car Technique: test vehicle is driven at the posted speed limit unless impeded by actual traffic conditions or safety considerations. The selection of test vehicle technique is based on the purpose of the study and the study team's judgment of the technique that best reflects the traffic stream being investigated. Most study reams prefer the ~verage~ technique. However, the average~ and Boating-car techniques function similarly in real world traffic situations. The travel lane should be consciously chosen, especially since the outside lane may be affected by slow-turning vehicles, parking maneuvers, or transit stops. In freeway studies, travel lane might be analyzed as ~ variable. 2.22 Penomuland Equipmmt The test vehicle method requires a test vehicle and the m= to record time and distance. These ~ be recorded manually oc automatically. A voice recorde.r can be useful to take nores of queues or other incidents, particularly in heavy traffic situations. Also, ~still or video camera can be useful to take photographs or videos of unusual events that will help explain the data resulrs. Manual daea collection requires~ driver and observer/recorder, two stopwatches and data collection forms. The srudy ~ be conducted with only one stopwatch. The distances between conuol points and the length of the total route may be obtained from ~ccurate, drawn-co-scale plans or maps. or from the vehicle odometer. Automatic data collection equipment i.s ~vailable to aid with data collection. Global Positioning System (GPS) unirs can be used to measure the test vehicle's position and speed along the corridor. The term GPS is used in gener_al in thi.s chapter to describe the standard technology which can be applied or can be augmented through diffcrcnoa1 methods

or ocher advancements in the technology. Numerous varieties of sofTWare programs for use wirh GPS units or lapcop/ handheld computers arc available to perform travel-time and delay srudies. These prooucu can caprure fearures including speed, distance, stopped delay and number of stops. A stopwatch or other manual equipmem should be used on at lease one run to error-check the data collection instruments. 2.2.3 Stzmple Siu Requirements The purpose of the uavel-time and delay srudy dictates the site of sample (in terms of number of test runs) required. The following arc suggested ranges of permitted errors in the estimate of the mean travel speed related to study purpose.

• transportation planning and highway needs srudies: t 3.0 to± 5.0 mph(± 4.8 to± 8.0 km/h). • traffic operations, crend analysis and economic evaluations: ± 2.0 tO ± 4.0 mph (± 3.2 to ± 6.5 km/h). • before-and-after srudies: ± 1.0 co ± 3.0 mph(± 1.6 to± 4.8 km/h). Engineers can apply these suggestions to similar studies as well. Exhibit 9-1 suggests numbers of test runs to perform travel-time and delay studies (Quiroga and Bullock, 1998). These values have a confidence level of 99.73 percent, 95 percent, 85 percent, or 75 percent and arc based on the difference between minimum and maximum speed measured in the inicial study. Multiple confidence levels arc presented to provide the analyse with the option to select the Je..·cl consistent with their specific needs. Upon completion of the inicial test runs, calculate the difference bcrween minimum and maximum speeds of the test runs (11), as shown in Equation 9-1.

-

.

..

1-1

1•1

R = maxv1 -minv1

F.q~on9-l

where m

= sample size of the initial study

v1

= running speed associated with travel run i

Select the approximate minimum sample size from Exhibit 9-1 fur the computed difference in minimum and maximum speeds and the desired error permitted(&). If the required number of test runs is greater than the initial number of test runs, make additional runs under the same conditions as the initial runs ro complete the sample required. Determine a sample size for each direction of travel and for each set of traffic and/or environmental conditions of interest. For ~ample, assume that five runs were made to obtain a difference in maximum and minimum running speeds of 10 mph (16 kmlh) and that the desired permitted error is± 3 mph (4.8 kmlh). Exhibit 9-1 indicates that 8 is the minimum number of runs required, assuming a con1idence level of95 percent. Therefore, three additional runs would have co be made to achieve the required sample site. However, if the additional runs result in a larger difference, additional runs wiU be required in most instances. The obscrver(s) must be sensirlve to changes in the traffic or environmental conditions. The sample number of runs represents a single sec of conditions. For example, speeds will probably vary during a peak pcrioo. Therefore, it may be necessary to conduct separate travel-time and delay studies for different portions of the peak perioo. Srudying speed changes during the peak period can be useful and important for agencies. Also, signal progression can introduce travel-time variability during data collection. Signal timing along the corridor should be understoO

.: ~;- ,~

-. ~:·~ ""$)'~-mi~Jtu.r.:O'i:~!lll~~ ·: ,~'::"- :·.

·:il:W~~itf4tcWdi{!f£&t~~:.~.::-- · ·

Minimum Sampi~ Si:te n for Specified Permined Error &

ii

Confidence le~l: 99.73%

Confidence leve.l: 95%

{mph)

1 mph

2mph

3mph

4 mph

5mph

1 m ph

2mph

3 mph

4 m ph

I

6

5

4

4

4

4

3

3

2

9

6

5

5

4

6

3

3

13

8

6

5

5

8

4 5

3 3

4

4

4 5

17

9

7

6

6

10

6

5

21

11

8

7

6

12

7

5

4 4

6

26

13

9

8

7

15

8

6

7

32

15

10

8

7

18

9

6

8

37

17

12

9

10

7

43

19

13

10

24

11

8

so

21

14

11

8 9 9

21

9 10

27

12

8

II

57

24

IS

12

10

31

13

12

64 72

26

17

13

11

34

13

29

18

14

11

38

14

80

32

20

15

12

43

15

89

34

21

16

13

47

20

-

50 .

17

71

2S

68

30 39.

21

.

27

21

30

.

89

so

34

26

5mph 3 3 3 4 4

4 5 5 5

9

5 5 6 6 7 7

6

15

10

8

6

16

11

8

7

18

11

12

9 9

7

19 27

17

12

10

99

36

22

15

12

-

47

27

19

15

6'

I I I

8 !

Minimum Sample Siz.e n for Specifi~ Pcmaitted Error 6

R.

Coo.6dence level: 75%

Coo.lideoce level• 85% 2mpb 3mpb 4mph

5mpb

1 mph

2mph

(mph)

1 mph

3 mph

4mp h

1

3

3

2

2

2

3

2

2

2

2

2

4

3

3

3

3

4

3

3

2

2

3

6

4

3

3

3

5

3

3

3

2

4

7

4

4

3

3

,6

4

3

3

3

5 6

9

5

4

3

3

i ]

4

3

3

3

5mpb

10

6

4

4

3

8

5

4

3

3

7 . 8

12

6

5

4

4

9

5

4

3

14

7

5

4

II

6

4

4

3 3

9

16

8

6

4 5 5

4

12

6

5

4

3

14

7

5

5

4 5

15

7 8

5 6 6

4 4 5

4 4 4 4 4 5

10

18

9

11

20

9

6 7

12

23

10

7

6

5

17

13

25

7

6

9

28

8

6

5 5

19

14

11 12

20

9

5

6

5

7

5

15

30

13

9

7

6

22

20

45

18

11

9

7

33

9 10 14

7

6

25

62 . 81

24

IS 18

11 13

9

44

18

11

8

7

10

58

22

14

10

. 8

30

30

-

Source: Quiroga, C. A. a.nd D. Bullock (1998). •Determination of Sample Sizes for Travd Ttme Scudies." lTE]ournal on k 'Web. Institute ofTrao.spon:ation Engineers. Washington, DC. Travel-Time and Delay Studies • 16 ..::3

.

-•

-~--· .

. l\..11':-'--

lOlL,.'""~·.

--

:J.· .

.,_

.

.. ~

. --- ·-

';

.

~:$ ~~~~-.

:;-. _}t~wit7~ft~~~

Travel-Tune and Delay Study Test Vehicle Method

Date

Weather

Route

Dira:cion

Trip Stalled at

At

Trip Ended at

. . .:

..,11.

it:IO • •

Trip#

n.caoonl

(Mqc)

(Loarion)

(Milaa<)

At

I

Control Points Location

Sto}l-' or Slows lime

Location

Type

Delay

I

(second.)

Trip Length:

Trip Tune:

Running Tune:

Stopped T L111C!

ITravel Speed: I Running Sp=i:

Symbols or Delay cause: S-Traffic Sigoa.ls, SS-Stop Sign, LT-Lcft Turns, PK-Parked Cars, DP-Double Parking. T-Gcneral, Pcd-Pedcstrians, BP-Bus Pa.s.scngcr Loa.ding or Unloading · CommeniS

Ob=vu

If the rehearsal runs go smoothly, count them as test runs. One round trip (one run in each direction) through the test section constitutes a test run. Test runs should begin promptly at the beginning of the desired study period so as to complete the required sample of runs before conditions along the route change. If the required sample cannot be obtained in a single period, complete the remaining runs on another day under the same conditions or add additional teSt vehicle teams to the peale period. As the test vehicle passes the begin point, the driver starts the first stopwatch. The test vehicle proceeds through the srudy route according to the driving technique selected. The observer records time readings from the first stopwatch as the vehicle passes each control point. When the test vehicle stops or is forced to rravd slowly (that is, 5 mph [8 kmlh] or less), the observer uses the second stopwatch to measure the amount of delay and noteS the location, duration and cause of each delay on the field sheet. Alternatively, the study could be conducted with only one stopwatch. The observer would stan the stopwatch at the beginning point and note the start and end times of any delays during the run.

As the test vehicle passes the end point of the srudy route, the driver reads the first stopwatch, and the obscrvc:r notes the total time of the run on the field sheet. The rest vehicle then rerurns to the begin point, the driver and observ~r reset the stopwatches, the observer prepares a new fidd sheet and the next run begins. This procedure is 164 • MANUAL OF TRANSPORTAnON ENGINEERING STIJniF<; ?Nn Ff'liT11'1N

repeated until the required number of sample runs is reached or until conditions that could affect the study change. If the reverse direction is also being studied, the same procedure governs the return trip, with the data recorded on .~ separate field sheet. A laptop computer software program with a rime-stamp procedure can be used to replace stopwatches. An o bserver using a laptop can reduce the additional tasks required of the.driver and capture the same information with predefined keyscrokes to represent locations or delays.

22.5 A*tomAiir Da14 Colkaion, Position the test vehicle a shon distance upscream of the begin point. Turn on and initialize the data recording equipment. GPS units and accompanying software packages automate many of the steps listed below relating to a distance measuring instrument (DMO. which could be more automated with a software package for automatic data reduction and analysis. For specifications and proper application of various automated software programs and associated devices, refer to the user manual. Test vehicle runs should be conducted under incident-free conditions along a representative lane. For two-lane cross-sections, use the second lane from median; for three-lane cross-sections, usc the middle Ian~. The number of test vehicles needed depends on the run duration (particularly important for congestion situations) and the desired run interval. For example, if the duration of the.run is 40 min. and the desired interval is 15 min., three test vehicles are necessary. · Calibrate DMls before arriving at the study site. Train the driver (and observer, if used) to operate the automati c equipment before staning a study. Safecy is a primary consideration in conducting this type of study with a driver only. Make a dey run and enter into the data recorder the location of the beginning, ending and control points for the route under study. Begin the test runs prompdy at the start of the desired study period, and complete the required number of runs before conditions change. If the required sample cannot be obtained in a single peri~d, complete the remaining runs on another day under the same conditions.

As. !he test vehicle passes the begin poirit, the driver activates the data recorder. The test vehicle proceeds through the study route according to the; driving technique selected. As each control point is passed, the driver activates the appropriate switch (this provides redundancy because the swcing point can be used to derive any control point along the route). When the test vehicle stops or is forced co travel slowly (the DMI will automatically sense chis), the driver pr= the appropriate button to enter the cause of each delay. The DMI automatically records the location and duration of the delay. Upon reaching the end point, the driver activates the appropriate switch to note that fact. The test vehicle then returns to the begin point, the driver resets the data recorder and the next run begins. This procedure is repeated until the required number of sample runs is reached or until conditions that could affect the study change. Some runs might need to be discarded if unusual events affect the run-for example, a passing emergency vehicle requires vehicles to pull over to the side of the road, a collision, or other atypical' circumstance.

2.3 Data Reduction and Ana lysis To evaluate travel-time daca, analysts conven the time and distance measures to space-mean speed. The average range in running speed is computed, as described above, to help determine the sample size required. The aavd speed for each individual test vehicle run is calculated using the following equation:

s = 3600 E.T where

S = travel speed, mph D • length of srudy route or section, miles

T

= travel time, sec

Equation 9-2

The equation for SMS or mean travel speed of all travel time runs is 3600ND S,w£=~

Equation 9-3

where space-mean speed or mean cravd speed, mph

SA;'.!:

:

D

"' length of srudy route or section, miles

LT

=

N

"' number of test runs

sum of ~vel times for all rest runs, sec

Use Equations 9-2 and 9-3 co calculate running speed per run and mean running speed of all runs, except use running speed in place of crave! speed and running cime in place of uavel rime. Mean speed may be calculated for each section of the study route in addition to the mean speed of the total route. Field notes can be used to justifY the removal of inconsistent values from the analysis due to an unusual siruation. Since the srudy team measures delay directly, it can summariz.e the delays of each type (refer to Ch~pter 6 for detailed informacion on rypes of delay) for each section and for each run. Total delay (the d.ilkrence between the recorded crave! time and the free-Bow travel time) is a useful and popular delay calculation. Mean delays are calculated by dividing the sucns of delays by the number of runs. Each of the resulting speed and delay measures may be ueated to statistical analyses, such as those described in Appendix C. Graphical summaries are useful in displaying the various travel-speed and delay measures. Exhibit 9-3 is an example of an average speed graph. For areawide studies, time contours overlaid on a map of the city's streets (as shown in Exhibit 9-4) offer a practical graphical display of travel times. Exhibits 9-5 and 9-6 present examples of typical travel time study output..Computer sofrware that accompanies automatic data collecrion equipment offers a powerful range of analysis and data summary capabilities.

•L•T~,-JW~m.tao

l~~~.-~~

-nt 45r---------------------------------A-oSpoedll-lntarMCIIono Feny RA>ed-nd (A.II Pule)

I 30

..j

l

f1s

"'

I I

I

I

A::

~-

0 ~------~----r---~--~--------~--~ 2000 4000 ~00 8000

lllol.once(looq

-Lht:

•·Con~oM~e~"-· 2· v...oy5 f · TI)
·-!liM.

3·

Clallt- 4 . --

2.4 Volume Extension The test vehicle method can also be used to collect volume along the corridor if desired. The estimates are obtained by measuring uavel time, opposing traffic, overtaking uaffic and passed traffic. Analysts can apply this c:xtension of the test vehicle method only to routes with controlled access or negligible driveway and cross-street traffic, such as freeways or uniform sections of arterial streets with 1 mile or more berween signals. On freeways, these locations are at interchanges that permit a fast turnaround. On arterial and other types of surface streets, they are usually at major intersections where U-turns are legal or where a reasonably quick turnaround is possible. The method is applicable only on two-way routes where opposing traffic is visible at all times. The vehicle extension requires a means to record the traffic volume. On multilane facilities with heavy traffic, srudy teams may use videotape to record the runs. The team will then reduce the data in the office. Manual data collection requires a driver and observer/recorder, handheld counters and data collection forcns. Exhibic 9-7 shows a sample field sheet. The test vehicle proceeds through the study section in the direction opposing the direction of traffic being srudied. The observer counts and records the opposing traffic mer during the trip. The test vehicle rums around and uavels the same section in the opposite direction. The observer counts the number of vehicles the test vehicle passes and the number of vehiclc:j it overtakes. 166 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITlON

(I)

I

0

'*-(fwtl ~

I

~

I

t:l:l

0

~

A.

0

1!1

'"0 '"0

g

~'i

~

g0..

l

0

"G

sf

'-J

9'

!

"?

;;

8

~

.......... (fHI,

g

B

i

I I I

II I I

I

I

I

i I

I

!

I

/I

IN l/!

I

·o ·

0

i

I

;I

0

1!1

i

j

t.-J! [J

~l i I . I ~:r-rr-~ ~ ~~~ 1[ , if [ . il I

lj f iz

a s 8

~;

Il '

.

X

t..

,.

•

-' ..

a·

..

I

I

.

,..

-~\-

·~jl>;.'>··

•'i'H~ .

..

...

··• .

··:;,

~ ~":t.

'

·,

··~- .

. ... • .f'""

0

&• •

.

.

. . . ·"" ~"'

<·

.

~

...

~

l' '

-

·· -~ 0 l .

·• 'tt ·-"'

•

•

. .... .

~

. <

• •

•

;,.,

>

.o ~. ~ .

Tra~l-Tunc and Delay Study

Test Vchlcle Method- Volu.me Extension Weather

Dare

Route End Point

Start Point

Vehicles Stan Tune

Run

Fmish T.unc

Travel Tune

Vehicles Met

<>vena.king

Vchlcles Passed

· Bound 1 2

3 4

I

5 6

7 8

9 10

Total Average Stan Tunc

Run

FtnbhTune

TnvdTune

Vchlclcs Met

Bound 1

2

3

4 5 6

7 8

9 10 Total Average Comments Obse~r


Vehicles Owrtak.ing

Vchlcles Passed

, Both directions may be studied simulra': neously. ln this situation, one observer f counts the vehicles passed and overta.k··c:n, while a second observer counts the vehicles traveling in the opposite direction. The team must ensure the counts are propc.rly recorded on the handheld counters. If traflic is heavy in the opposing direccion, it may be necessary to weight the count with a multiplier factor (for example, one count equals five vehicles). Studies on freeways with more than three lanes in each direction are difficult and may require an additional observer. Safety is a primary consideration in conducting these studies. V'JSual obstructions from trucks and a wide median can compromise the accuracy. If traffic is heavy, usc additional observers so the driver can concentrate: on the driving wk.

EndPoint

~

I I I I I I I

I I I I I I

t

:~

I

I

I I

N

.

I Start Point

Automatic data collection can utilize specially designed data-recording hardware to perform travel-time and delay studies using the volume: extension (Robertson and Courage, 1973). Handheld or laptop computers can be configured and software developed to serve this purpose. Because of safety concerns, automation would not eliminate the need for an obsecver/recorder.l,he procedure is the same as for manual data collection, except the observer uses a handheld or laptop computer to rerord time: and traffic count data. The procedure requires the same number of personnel. Automated data reduction saves time and money during the analysis.

2.4.1 W,lume Extmsitm Dal4 Arudysis tmd Summ4ry ofRuults Consider the: diagram in Exhibit 9-8. The subscripts nand s refer to th.e direction the test vehicle is traveling when the item is measured. The rest vehicle StartS at point A and travels south to B. The observer counts all the: vehicles met from the opposite direction (M). The time of the southbound trip (2) is recorded, the driver turns asound at B, proceeds north to A and records the time of the northbound trip (T). This constitutes a round·trip or one run. The observer co una any vehicle that is passed by the test vehicle {P) and any vehicle that passes the test vehicle (0) during the northbound trip.

2.4. 1.1 Hourly VolutM . During this round trip, the test vehicle is essentially measuring the n!=ber of vehicles that will pass the starting point, A, in the rime it takes the rest vehicle to make a round trip berween'A and B. The vehicles met (M) will pass point A before the test vehicle can return from B to A Those vdlldes overtaking the test vehicle (0) minus those passed by the test vehicle (P) will compensate for the test vehicle not traveling at exactly the average speed of the northbound traffic. Therefore, the volume past point A, in the northbound direction, in the time it takes the rest vehicle to make a round trip, isM, • 0.- P•. The formula for northbound volume:, then, is V,.

M,+ 0,. - P,.) =60 ( T,. + T, :

Equation 9-4

where ~

• volume per hour, northbound

M, • opposing count of vehicles met when the test v~hicle was traveling south 0. = number of vehicles overtaking the test vehicle as it traveled north Travel-Time and Delay Studies • 169

P. = number of vehicles passed by the test vehicle as it traveled north

T. = travel time when traveling north, min. T, =travel time when traveling south, min. The formula for southbound volume is the same except all subscripts are reversed. If the test vehicle turned instanta· neously at B, the volume count would be exact. However, some rime is lost during turning which introduces a small error, so the analyst should consider the result of Equation 9-4 an estimate. Since a single run may not be representative of average conditions, several runs are made and the analyst uses an average (Pignataro, 1973). · ~&lliimH~T.~~b'-"'"'"(~~""' ·· '~a:~~.,#~~~~·~ ··;.--~~~""'· 0 :"~~,~~~~~~"·r.@l&Jf~.~ . ~ . o . .• •.: ~ . • ' ~ G r!_D.ffiit_~ v ~~:.: - .• ~~;.~ ~~~l ..~N~ ·:~-·:i;6~:"~~:~~~~J~:·.~~·

..

!

Northbound Trips

T. (min)

M.

...

p

0

IN

2.65

85

1

0

2N

2.70

83

3

2

3N

2.35

77

0

2

4N

3.00

85

2

0 I

5N

2.42

90

1

6N

2.54

84

2

I

Tow

15.66

504

9

6

Average

2.6!

84.0

1.5

1.0

Southbound Trips

T (min)

M

0

P,

IS

2.33

112

2

0

2$

2.30

113

0

2

3$

2.71

119

0

0

45

2.16

120

I

I

5S

2.54

105

0

2

6S

2.48

100

0

I

Total

14.52

669

3

6

A-verage

2.42

111.5

0.5

1.0

Sou.rce: Adapted from Pignara.ro, 1973, Table 7-4.

2.4.1.2 Avn-agf Trav(l1i17U!.

The average travel rime for one..
r.a.. = r.-

eoco"-'")) v.

Equation 9-5

where:

T.... =average travel time for all north traffic, min. and the other variables are as defined for equation 9-4. As with equation 9-4, the southbound volume is the same except all subscripts are reversed. The quantity (0. • P) represents a correction factor co account for the faa the test vehlcle may not have travded ar the average speed. This correction factor is not necesury ifthe average-car or floating· car driving technique is a.pplied correctly.


,.

2. 4.1.3 Spact-Mean Spud. Analysts can calculate the SMS for one directional Aow using 60d sn--Tnaoe -

Equation 9-6

where S. is the SMS northbound, in mph; and dis the length of test section, in miles.

EXAMPLE9-I Exhibit 9-9 shows data for a 1.25-mile (2.01-km) length of rest section. From Equation 9-4, the ho urly volumes in the northbound and southbound directions are, respectively, Vn = 60 (

V. = 60 (

111.5 + 1.5 - 1.0) = 1336 vehicles per hour 2.61 + 2 .42

84.0 + 0.5 - 1.0) = 996 vehicles per hour . + . 2 42 2 61

From Equation (9-5), the average travd times for the northbound and southbound directions ace, respectively, 1.0)) r,.,.... = 2.61- ( 60 (1.51336 = 2.59 min

T,,_ =

2.42- (

60 (0.5- 1.0)) = 2.45 min. 996

From Equation 9-6, the space-mean speeds for the northbound and southbound directions are, respectively, . 60(1.25)

s,. = --r.s9" =29.0 mph

s, =

60(1.25)

~=

30.6 mph ( 49.2lon/h)

3.0 OTHER TRAVEL TIME STUDIES 3.1 Vehicle Observation Vehicle observation methods are technologies that are employed by the study team, which will sdect which vehicles will be observed, and noninrrusively study them. The license plate, intervi~ and wireless technology methods ~e pan of this group of methods. 3.1.1 License PlAte Method

This method produces travel time only, from which average travel speed may be calculated once the distance betw~en observation poincs is measured. A test vehicle is not required. Observers/recorders posicion themselves at the emra.oce and ait to the test section and at other major intersections or points·ofinterest along the route. There are four methods that can be implemeiued for license plate marching: manual, portable computer, video with manual transcription, or video with chacoa.ccer recognition (Turner, Eisele, Bem and Holdener, 1998). For manual license place marching, as vehicles pass the observers at each location, the observers record the last three or four digits of the license tag along with the time from a stopwatch, with pen and paper or on an audio tape. Using; a portable computer, an observer can record license plate numbers accompanied with an automatic time stamp. Lice fl.=' e plate numbers can also be collected using video. The video can be post-processed using manual uansaipcion; tP..i.s involves an analyst in the office watching che video and recording the license plate numbers. Character recogniriO ..J:I

.· Travel-Time and Delay Studies • 17 --1

software could be implcmcmcd on the video
3 .1.2lntnvintl Method Selected individuals who are wiUing to cooperate may provide a satisfactory sample from which to obtain travel times and delays without the use of a test vehicle or observers. These persons are asked to record their start and end times for designated routes. They also record the times and durations of delay. This is a variacion on the average-car method, except that in place of a single test vehicle there arc multiple test vehicles. Employees wbo drive on the job, truck drivers and raxi drivers often make good subjects. This method is useful when a large amount of
: 3.2 Vehicle Signature Matching Method JVehicle signature marching is aCC(lmplished by matching vehicle signatures at consecutive points on a roadway; this .;allows for travd time calculations (Turner, Eisele, Benz and Holdener, 1998). Inductance loop detectors can provide vehicle signatures by comparing detuning curves produced by individual vehicles. Alternatively, laser sensors can be used to develop a threc:-dimensional profile of each vehicle. Weigh-in-motion (WIM) sensors can produce individual vehicle signatures based on axle weight and conliguration. Video technology can be implemented to prOvide vehicle signatures based on attributes of individual vehicles. These technologies can be employed to develop travel times through vehicle signarure matching.

3.3 Platoon Matching Method Similar to vehicle signature matching, platoons of vehicles can also be marched by compari.ng vehicle platoon signaru.res at consecutive points on a roadway; this allows for travel-time calculations (Turner, Eisele, Benz and Holdener, 1998). Ultrasonic detectors can be used to determine. unique characteristics of vehicle platoons, particular:ly large vciUdes in a platoon.

3.3.1 Dmct Obserr~ation Method Observers at an devated vantage point can measure crave! time di.rectly becween cwo points a known distance apart. The method requi.res good visibility and is not suitable for Sections greater tlw! .5 mile (.8 km) in length. 3.3.2 AeriAl Surf~eY$ Aerial survey;s can be conducted to determine traffic parameters, including flow (Turne.r, Eisde, Benz and Holdener, 1998). ObseiVarion equipment can include observation balloons, fixed-wing ai.rcrafr, helicopters, satdlites, or unmanned aerial vehicles (J.JAVs}. Traffic density and traffic times can be computed using successive images captured fr
3.4 Probe Vehicle Probe vehicle methods a.re technologies, employed by the study team, which a.re sclecred/pcrmitted by drivers or fleet operators. Each of the resulting speed and delay measures may be treated to statistical analyses, such as those described in Appendix C. ~~l~~MeJT~ap~

Signpost-based uansponder systems a.re typically used by transit agencies for monitoring the srarus of the fleet vehicles (Tumer, Eisele, Benz and Holdener, 1998). This technology operat~ by Beet vehicles receiving timNramped identificarion data from signpost mounted decrronic uansmirtcrs. The ideqtification data is transmitted to a central dispatch ficility which can monitor schedule adherence. 3.4.2 AW TrtmiJH»>IIn't Electronic toll-road tags a.re a primary application of automatic vehicle identification (AVI) uan;sponders. These uansponden can also be used for .real-time traffic monitoring. incident management, travder information and performance measuring (Turner, Eisele, Benz and Holdener, 1998). T1me of arrival at each booth is .recorded on each tag. The toll-road rags provide data coUecrprs with a method to uniqu~ identify vehicles and measure the uavel times over a facility between toll booths. Knowing the distance between booths will allow the calculation of overall ttavd

speeds. 3.4..3 GrounJ.btuetJ llAJio Nllfliguion Ground-ba.sed radio navigation wes a receiving antenna netWork and demonic uan;sponders within vehicles (Turner, Eisele, Beru. and Holdener, 1998). Transit agencies and companies with large fleers are co=only we this technology. A cenual office computer can request vehicle location information; this results in a transmission from the vehicle to the radio towers to the central office. The location information can be determined by triangulation from multiple towers.

Travel·Time and Delay Studies • 173

3.4.4 Cellular Telephom Probe Method Cellular telephones can be used as a probe in a travel-time study with multiple approaches. Drivers can be recruited to use their cellular telephones to report their progression along a corridor relative co designated checkpoin ts (Turner, Eisele, lkm and Holdener, 1998). Although less accurate than others, this method can supplement the uavel-time clara with vi$ual confirmation of flow conditions or collisions. Another approach for the cellular-r~lephone-probe method is to recrui t drivers tO allow their cellular phones to be cracked (Krygsman and Schmitz, 2005). 3.4.5 GPS GPS technology is a probe technique char utilizes satellites orbiting the earth to transmit positional data to equipped vehicles. This data can be used tO determine the location, direction and speed of a vehicle; therefore, travel-time srudies can use the technology (Turner, Eisele, Ben~ and Holdener, 1998). The data can be stored onboard the vehicle and either retrieved at the conclusion of the srudy or transmitted to a cenrral office iftwo-way communication is available. Integrating GPS technology into a geographical informacion system (GIS) is useful for analysis (Quiroga and Bullock, 1999).

4.0 REFERENCES 4.1 Literature References Box, P. C. ind ). C, Oppenlander. Manual ofTrrzffic Enginming Studies, 4th ed. Washington, DC: lnsricuce ofTransporution Engineers, 1976: pp. 93-105. JAMAR Sales Company, Inc. PC-Trawl Sampit &portJ. Ivyland, PA: JAMAR, 1990. Krygsman, S. and P. Schmitz. "The Use ofCcllphone Technology in Activity and Travel Data CoUection." 24m Southern African Transport Conference. Pretoria, South Africa: Document Transformation Technologies, 2005: pp. 696-705. Pignataro, L. Traffic Enginming Theory and Prrutice. Englewood Cliffs, NJ: Prentice Hall, 1973. Quiroga, C. A. and D. Bullock. "Determination of Sample Sizes for Travd Tune Studies.• IT£ foumal 68, No. 8 (Augusr 1998): 92-98. Quiroga, C. A. and D. Bullock. "Measuring Control Delay ar Signalized Intersections. ASCEJoumal ofTransportation Enginuring 125, No.4: 27 1-280. Roberrson, H. D. and K. G. Courage. "Traflic Signal Srud.ies Using l Digiral Tape Recorder TnzfJU &searrh &pon, June 1973. T:unoff, P. J. et al. The Continuing Evolution ofTravel Tune Data Information Coll«tion and Procwing. Washington, DC: Transportation Rcse:arch Bo:ud, 2008. Turner, S.M., W. L. Eisele, R. J. Benz lnd D.). Holdener. Tnzw/ Time Data Co/kaion Handhook. Report No. FHWAPL-98-035. Washington, DC: Federal Highway Adminstrarion, 1998.

4.2 Online Resources (Availabl~ as of January 5, 201 0) Turner, S.M.. W. L. Eisele, R. J. lknz and D. J. Holdener. Traud Tii'IU Dara Colkaion Hmulbook. Report No. FHWAPL-98-035. Washington, DC: Federal Highway Administration. www.fhwa.doc.gov/ohim!starc.pd£

4.3 Other Resources Cambridge Systematics, Inc., Dowling Associates, Inc., System Mecrics Group, Inc. and Texu Transportation lnstirute. CorrEjfomw Pnfomumt:e Measum for TraiN!/ TtmL Delay. VatUuion, and &liability. Washington, DC: Transportation Research Boaxd, 2008. Eisele, W. L. and J. A. Crawford. Guidebook for Mobility Monitoring in Sma/J ta Medium-Sized CommunititS: A How- To Guide. Report 0-5571-Pl. CoUege Station, TX: Texas Transportation lnsitute, Texas A&M Univccity System, 2007.


; Eisele, W. L., J. A. Crawford and R. L. Stensrud. Mtasum, Mttbods, and Applications ofa Mobility Monitoring Prowsfor SfflalL ; to Mtdium-Siud Commrmitit1. CoUege Stacion, TX: Te:u.s Transportation lnstirute, Texas A&.\1 Univc:rsiry System, 2007.

·f.iscJe. W. L. and L

Rileu. "Travel-Time Estimates Obtained from lnteHigent Tra
Guindon, M. E. "2002 Congestion Management Report: Travel Time and Delay Study. • Ninth National Conference on Transportation Planning for Small and Medium-Si:r.ed Communities. Colorado Springs, CO: Transportation Research Board, 2004. Guo, B. and A. Poling. "Geographic Information Synems/Giob;al Positioning Systems Design for Network Tr..vd Time SrudY·»

Transportation Research & cord· journal oftht Transportation Restart:h Board 1497 (1995). Mortimer, W "Moving Vehicle Method of Estimating Volumes and Speeds.• Transportation Rntarrh Record: journal ofrht

Transportation !?=arc~ Board 156 (1957).

.

Roess, R. et al. Traffic Enginmin~ 3rd ed. Upper Saddle River, NJ: Pearson Prentice Hall, 2004. Srinivasan, K. and P. Jovanis. "Determinacion of Number of Probe Vehicles Required for Reliable Travel Tune Measurement in Urban Nerwork." Tnznsport4tion Rntarr:h &cord: journal ofthe Tranrport.uion Rnearr:h Board 1537 (1996). Transportarion Research Board. "Traffic and Urban Dara. • Transportation Restarr:h Record: Journal oftht Transporration ResetlfCh

Board 1945 (2006). Turner, S. "Advanced Techniques f!?r Travel Time Data Collection." Transportation Research Record: journal ofthe Transporratiof"

Resurrh&ardl55I ( 1996).

·

W~g, z_ "Using Floating Cars to Measure Travel Tune Delay: How Accurate is the Method?" Transp0rt4tion Research Record: journal ofthe Tnznsport4tion &search Board 1870 (2004).

Travel-Time and Delay Studies • 175

Chapter 10

\

................................. ................................................. .

Freew~y and Managed Lanes Studies By: BtUtilm &lrroeJn; Ph.D. Cbristt~pher M. Cunningham, P.£ 1.0 INTRODUCTION

177

1. 7 Chapter Objectives

177

1.2 General Free~y" Facilities

178

1.3 Managed Lanes Facilities

179


182

2.1 Spot Evaluation

182

2.2 Segment Studies

187

2.3 Special Freeway Studies

190

3.0 DATA COLLECTION PRoCEDURES

191

3. 7 Facility Performance Data

191

3.2 Data Acquisition

191

3.3 Equipment Needs

192

· 3.4 Personnel Training Requirements

194

3.5 Field Procedures

194


195

4.1 Spot and Segment Evaluation

195

4.2 Synem Monitoring

197 I

4.3 Managed Lane Measures 5.0 REFERENCES

i 198

199

1.0 INTRODUCTION

1.1 Chapter Objectives he objective of \:his chapter is to provide an overview of data collection on freeway facilities and modem managed lane facilities on freeways, including high-occupancy vehicle (HOY), high-occupancy and toll (HOT), express toll (ET) and truck-only toll (TP'nlanes. Each of the studies described in this chapter is intended co measure hca"Y volumes af traffic moving ;.c b.igh speeds. With a.dvance.s in modem ;.utornated data collection equipment it is rare these types of data uc collected in a manual &.sb..ion. Thm:forc, the m;.jority of this chapter is devoted co means of automated data collection using either permmendy installed or portable data collection equipment. The use of manual data collection rnay be appropriate for special studies in ramp and weaving ucas, as well as cransicions to and from a managed lane facility. Guidance is proyided as nece.ssuy.

T

The chapte; is-organized into four parts. The introduction section provides a general overview and de6.ni;iQAS of the. different facility types. Ncn, the types of studies and data items of interest for freeways ue presented; followed by a Freeway and Managed Lanes Studies • 177

discussion of the data collection procedures necessary co obtain che data. The chapcer concludes with a section on data reduction and analysis of freeway facilities. The data collection procedures on general purpose (GP) freeway facilities and on managed lanes (ML) are concepcually similar. In most parts of the chapter both are therefore discussed concurrently and differences are emphasized where necessary.

1.2 General Freeway Facilities The Highway CApacity Manual (HCM) (TRB 2000) ddines a freeway facility as a sequence ofindividual.&eeway segments that encompasses basic freeway segments, &ceway merge and diverge sections (ramps) and freeway weaving segments. In weaving segmentS, successive on- and off-ramps cause additio.nal friction, and may therefore operate at lower capacities than a basic segment with a similar aoss--seaion. All freeway segments should be built in accordance with freeway design standards outlined by the American Association of State Highway and Transportation Officials (AASHTO) (2004), with high design speeds, limited horizontal and vertical curves, limited grades, full access control and multiple lanes of traffic in each di.rea:ion that are divided by a landscaped median or barrier. Freeways rep=r the highest functional class of surface transportation facilities and typically carry large portions of the total traffic demand in a region at high spcds. Exhibit 10- 1 shows an eighc-lane freeway section of Interstate I-40 in North Carolina (four IanC:S each direction). According to HCM theory, the capacity of a freeway segment is 2,400 equivalent passenge.r-cars per hour per lane (pdh!ln), where eqwvalent passenger-cars (pc) is an aggregation of both passenger cars and heavy vehicles (which are counted as more than 1.0 pc). The fou.r-lane directional section of freeway shown in the exhibit therefore caa readily carry in excess of9,000 vehicles per hour. ·


,Due co high volumes and operacing speeds on freeway facilities, data collection on freeways frequencly involves auicomated data collection technology, although nonautomaced methods exist. Freeways represent the backbone of chc fransporcacion infrastructure, and as such, it is common to lind permanently inscalled sensing equipment co continuously monitor the traffic operations. Section 2 of this chapter highlights the most commonly collected data items on freeways. It further emphasizes some special freeway studies related co incident management or freeway. work zones.

1.3 Managed lanes Facilities Many modem freeway systems ue more than just a collection of links. With ever-increasing land and constrUction prices, along with the sign.ificanc incr~es in vehicle-miles traveled (VMT) compared co available paved freeway bn.es in the previous decades, the focus of the transpoccation profession has largely shifted away from new construction. a~d cowards maximizing the efficiency of aisting infrastrucrurc (Neudorff ec al., 2003). Consequently, agencies invest significant resources in improving and developing new strategies for freeway supply and demand manageme nt. AJvancl!d traffic managnnmt (ATM) is a term used in the Uniceq States to combine many strategies chat .aim co maxintize .t he efficiency of the ociscing infrastructure. In the United Kingdom, the teem •Managed Mocorways" is used co groUP a variety of strategies that llctively modify operations on a freeway co improve rraflic Row. ML facilities are a form of ATM on freeways, in which agencies strive to maximize the use of existing capacity on faciliries thro ugh deJlland mwagement strategies. ML facilities are usually installed to treat problems such as: • excessive peak-period demands in one or both directions; • a safety problem due co high congestion; • needed separation of Ol)e or more vehicle types in one or more lanes; • excessive ramp demands that aff~t mainline operations; frequent incidentS and the need to quickly mitigate their effect; • excessive vehicle ezcissions that degrade the air quality of a region; and • limited roadway construction funding. ML facilities cw include strategies such as HOY, HOT, ET and TOT lanes. All of these restrict access to one o:ctloJ"e lanes based on certain ~ceria. ATM can further include other demand mwagement scracegies, such as ramp nete.t"ing. In the following sections, these mwagemenc strategies ace discussed in more detail. 1.3.1 Ma~ged Lane 'JYpes . · HOY, HOT, ET and TOT lanes are ML treatments geared cowards pr!oviding more efficient use of the available ~ciCY" of a freeway. An HOY !we may require different vehicle occupancy lCvds, expressed as two or mote (2+), three OtJOO.-e (3+), or even four or more (4+) persons per vehicle. HOY facilities consisr ofone or more dedicated lwes that give1rio$ity co HOV vehicles, which is intended to increase the actual person trips along a freeway segment without increas~ c.Pe number of vehicles using the facility (Kuhn et al., 2005). HOV lanes aim co reduce vehicle congestion by redu~ t:Pe number of vehicles on the highway relative to the numbec of pecsons carried. In other words, HOY lwes maint.ae ~c same (or greacec) person-movements while reducing vehicle-movements. HOY users are provided with lwes dur 1pe~ ate at lowec congestion levels and thus higher speeds. They therefore (should) have lowec travel times associated wh ~ average trip compared co the general purpose (GP) lanes, which gives"w incentive for their use.

HOT lane facilities ue similar co HOV lane facilities, in that they allow HOV vehicles to travel on separate heS · However, they further include pricing strategies chat allow vehicles not meeting HOV oa;upancy rcquiremeq; ~ ~ purchase access to the lanes thtough electronic rolling. This strategy typically uses a time-of-day or dynamic pt:i.~~ strategy that is directly tied to the demand wdcapacityin the HOT lwes. HOT lwes aim co (paccly) 611 excess~~ ity of undcrutilizcd HOV lanes as illuscrated in Exhibit 10-2.

Freeway and Managed lanes StudiM • 'I '!!!!!fiiii3

:rc..,-

Z:J:c. .s C..podry:Tiae to B1cJ> Oc._acy Toll J..oDOs (from ColorD.SO VDJu. Eqlftss x..a-s :Feaolbillty S~: 1 - 11199) teO'!• 90'/o

""'" i ..,. "'" ,.,..

I a

j

-·

""~

2<1'.

toe•

.,... s.oo Alof

""

A3.l

7:'10

8:GO

H)O

Alr4

~t

A!ol

IGJ:IO U;OO 12m 1-GI AM .A.W: ftl ftL

S!OO PM

..... ""'"" c...-

laii!OVVo- • -... Source: FHWA mm~IIJ Opmuions Handbook.

2::a:l ftt

-4.'00 nt

.SA» n&

6IDO ft.S

?M

&.<0

P3r.(

.I'.W.

I

Report FHWA-OP~4-003. Figure 9-4.

Exhibit 10-2 makes evident HOY lanes can be undcrutilized, and improved utilization can be achieved through dynamic pricing. This is also helpful from a public perception point of view since HOY lanes are often perceived to carry few vehicle-ttips, even though the acrual person-ttips in the managed lane inay be equal to or greater than the GP lanes (due to higher in-vehicle occupancy). Finally, the revenue generated from HOT lanes supports maintenance: activities and construction of furure ML facilities, which is an important factor as agencies try to identify new funding sources for frc:c:way construction. The capaciry and operations of any tolled faciliry are critically linked to the means of toll collection, which can range from manual payment to an attendee, to semiautomared coin collection, to fully automated electronic toll collection. · HOY/HOT lanes are wually constructed wing one of three possible methods, described below. The dlfferent con· figurations generally vary in appearance and capaciry, and are listed here since they may affect data collection practices. • Concurrent Flow Lane: a freeway lane, typically the innermost lane, which is designated for HOY/HOT we only. This lane is not physically separated from the other lanes, but is instead designated through the use of pavement markings or a painted buffer.

_,_---

• Contraflow Lane: peak-direction use of an off-peak direction freeway lane. Traffic is separared from adjacent mainline traffic in the off-peak direction by some form of barrier, including movahle concrete barriers, pylons, or cones. · • &dasi:ft HOV/HOT Fa.cility: HOV lanes that are separated from All GP lanes (both directions) wing ~physical barrier. These lanes are cxclwivdy used for HOY/HOT during all or a portion of the day. Exdwive lanes can be operarc:d as two-directional or reversible. Reversible lanes typically operate With the inbound and outbound traffic patterns associated with commuting to a CBD.

One of the key differences between HOY and HOT lanes in terms of data collection is that HOT lanes are almost always fuUy instrumented to monitor traffic operations, while some older HOY facilities may not have broad (or any) sensor coverage. However, it is anticipated that most ML facilities will become increasingly instrumenrc:d in the furure. ET and TOT lanes are exclusive toU-only facilities for passenger cars and heavy vehicles, respectively. For purpose of data collection they are similar to HOT facilities, and will therefore not be discussed separately. One dif. ference worth emphaSizing is that toll-only facilities are typically les$ likdy to involve vehicle occupancy studies, which is common for both HOY and HOT &cilities.

Kuhn ec al. (2005) provides more detail on all ML &ciliry types. More detailed informa.tion on HOY and HOT lanes can be found in NCHRP Report 414: HOV Syrtmas Manwzl (Taas Transportation Institute, 1998) and A GuiM for

HOT lAne !Xvewpmmt (Perez and Sciara, 2003). UA .. tl IAI f'\C

TOI\~I('Of"\OTI\Tif"'M

CMf.IMC,OIMf': (TIIf'IIC<: ")P\111

~f"'ICTit"'lf\1

Source: FHWA Frrn~ay OpmztiolfS H~lt. Report FHWA-OP-04-003. Figure 8-8.

1.3.2 Reversible Lmu Rcvasible lanes a.re employed to accommodate heavy peak hour demands, especially as traffic goes to and from a CBD. Rcvenible lanes a.re typically utilized when there is a significant difference in directional ua.flic volumes (in the order of a 70/30 percent split}": Ofctn, HOY/HOT lanes a.re utilized co increase person throughput in the peak direction. The reversible lanes are locared between the two directions of ccaffic wing a raised barrier. Access points to reversible lanes can be at-grade or through grade-separated ramps, but are commonly limited in frequency to reduce friction. Barrier gates and signage are used co close rhe faciliry and switch to the opposite direction when appropriate. Typically, this happens at predetermined times of day when uaflic conditions a.re known to be highly congested. Exhibit 10-3, ofVa.rginia's I-395, shows a good example of rhe ~ional peaks that reversible lanes help micigare. . ... .

1.3.3 RAmp Mneritrg · . Ramp metering is an effective suategy for improving traffic flow along the aWnllne by controlling the rate of vehicles entering from the on-ramp into the mainline traffic. This improves the ovcraiJ capaciry along the &ccway. cspecially during congested conditions. Platoons of vehicles attempting to enter the mainline traffic stream can cawe a loc.gf turbulence during high demand periods, because gaps in traffic an; less &equent. The friction crcared &om merging craflic can slo"w the mainline considerably, especially in the outside :ttavd lanes. The metering rate is typically set as a function of the upstream freeway mainline volumes and the capaciry of rhe freeway section downstream of the ramp. Since the &ccway mainline carries a higha &action of overall traffic, the on-ramp vehicles are delayed to minimize the overall delay of the system. In some cases, the introduced on-ramp delay can be capped by the available queue storage on the on-ramp. The metaing strategy is employed by means of traffic signalization and signing wing three increasingly complex methods: I) a pretimed meter co provide a consistent pattern of vehicles ~cering the traffic ~cam; 2) time-of-day mere{ing plans along with logic to determine the most optimal metering pattern based on sensor information related to the dcnsiry of traffic along the freeway; or 3) more complex ramp metering concepts wing sensors inst:allcd along the freeway mainline to detect available gaps in traflic, which are then wed by vehicles exiting the ramp. An cxampk of a ramp meter is shown in Exhibit 10-4. The RAmp MAiwgmJmt tmJ C6ntrol H.anJboolt Oacobson et al., 2006) should be wed as a primary U.S. reference if ramp metering is being considered for implementation. The reader could furtha refer to guidance produced by the European Ramp Metaing Project (EURAMP, 2007).

Freeway and Managed Lanes Studies • 181

Source FHWA Frrm~4J Opn-4/iJJns Htmt.lboolt.

~P9r:t

FHWA-OP-04-003. Figure 7-2.

2.0 TYPES OF STUDIES Freeways can be analyzed in various ways, but most studies intend to describe the general flow characreril>tics of traffic or to determine the types of vehicles (classification) that are flowing in the traffic stream. Many types of data arc available through automated means using various detectOrs installed along the corridor. Some freeway srudies are conducted at spot locations to collect data such as speed, occupancy, classification and gaps or headways. Other srudies focus on freeway segments and gather data on the average travd time of vehicles, queuing patterns, or the average segment density. Freeway studies may also focus on in-vehicle occupancy (to estirrutte person throughput), or on collisions (to perform a safety evaluation). More specialized freeway studies include those dealing with freeway work zones, or those evaluating evacuation strategies on the freeway network. This section briefly describes each of th.e smdies that could be conducted. Section 2.1 presents fundamental principles · of freeway studies at spot locations and section 2.2 highlights data items of interest on segments. Both sections are equally applicable to general freeway facilities and MLs. Any differences between the two facility types are highlighted. Section 2.3 then discusses more specialized freeway studies.

2.1 Spot Evaluation The majority of freeway data can be measured at spot locations, including spccd-flow-densicy parameters, gap distribution and vehicle classifications. With modern automated data collection equipment, ir is increasingly common that large sections of both urban and rural freeways are equipped with permanent sensors that provide agencies with large amounts of real-time and archived freeway operational dara. The data available on these sensors is often not used in ways that could provide additional information to analysts wishing to conduct certain types of studies. Section 3.0 provides details on the type of equipment needed to perform these studies and gives e:xamples of large-scale strategic sensor deplO)'Ill.ent on freeway networks. 2. 1.1 FbJ.u, Traffic flow is defined as the number of vehicles crossing over a given point in a certain time interval. Vehicle flows are typically expressed in terms of vehicles per hour (vph). The HCM uses 15-min. flow races expressed in vph as the fundamental input in freeway operations analyses. If only hourly volumes arc available for input they have to be converted to 15-min. Bows by use of a peak hour factor (see Chapter 4). Freeway analyses are commonly performed for one direction of cravd at a time, which is consistent with H CM freeway analysis methodologies (TRB, 2000). It may also be of interest to separate traffic flow by lane, as the distribution of traffic across multiple lanes is typically not uniform. Analysts also tequlre estimates or measurements ofHows on merge and diverge sections to serve as input in HCM methods. 182 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Freeway Entering

RampEntering

::c:

Freeway-toFreeway Flow

Ramp-to-Ramp Flow

)

Freeway Exiting

-Ramp - Exiting

Flows at spot locations arc readily collected using automated detectors or can be obtained &om manual counts. Automated sensors can readily be installed on the freeway mainline or on- and off-ramps to the facility. More challenging are studies on freeway weaving segments. A weaving segment describes a situation where an on-ramp is closely followed by a downsueam off-ramp, and the two are connected by an auxiliary lane. The auxiliary lane is used by both movements to "weave• in and out of ~e main uaffic stream. For freeway weaving segments, the flow pattern is nor fully described by counts on the mainline, on- and off-=p; an additional estimate of weaving behavior is needed. This is most easily done by performing a supplemental manual count to estimate the ramp-to-ramp flow. With that value known, the on-ramp-to-freeway flow can be estimart::d by subtracting the ramp-to-ramp flow from the on-ramp demand. The freeway-to-off-ramp flow is calculated b y subuacting the ramp-to-ramp flow from the off-ramp demand. The remaining freeway-to-freeway Bow is similarlY estimated by arithmetic. Exhibit 10-5 shows a typical weaving diagram for illustrative purpose with four movement::s: freeway-to-freeway flow, freeway-to-ramp Bow, ramp-to-ramp How and ramp-to-freeway Bow. 2.1.2Speed Average vehicle speeds at a spot location can be obtained from a "speed uap~ using two dosdy-spaccd detectors or by other automated data collection equipment as presented in Chapter 5. The speed used in the theoretical ~y uaffic Row relationships is always the space-mean speed (SMS). As
Freeway and Managed Lanes Studies • 18.::;:::1

80

Mmllh

60

55 ml/h

li! ~50

i

,,

40

.::;:.~.,'

&;

20

, .,."'

.,."

,

'~~,~~,. .,." ,:'' .L>

~ ~~

.-

~" .. f:~~"~ .,. ,

1--

10

-

, , ,

-i"Mn::::::::::::::: 1750

~

lOS A

~ 30

f

7 ' - - - /_~a-1450 , , , R C':

10 mi/h 65 milh

70

~

,

fae:f!P!..Spdl _ffs.=_7~!l!!.

, .,.

.. ., ,.. .. -- , -- -- -

, ,

-

,

n

,

,

--- -

~<-,'

~ __ .,..

-~---

-

--

,.,.. "'·~-

-:..~-

0

400

0

800

1200

1600

2000

2400

Flow Rate (pclhlln)

Sou=: Higlrwtzy Czpacil) M1111UAI 2000. Copyright, National Academy of Sciences, Wa.shington, DC. Exhibit 23·3, p.l3·5. Reproduced with petmission of the Tn.n.sporuuoo Rcscarch Board.

70

i~

-a

J-50 ~40 >30 f20

0

<

ObserwdSpeeds

10 0

0

400

800 1200 Traffic flow

1600

2000

2400

Exhibit 10-7 ~evident pans of the field-observed data match the theoretical HCM speed-flow cdatiooship quire wdl. The fidd-observcd values that ace shown signilicandy below the speed-flow curve likdy cornspond to unstable or congested flow periods. The 6dd-obscrvcd data further show only one data point~ a. flow ofl,l 00 pdmilln. Ptcfcra.bly, more data would need to be coUected a.t this loca.rion to validate that capacity is in fact 2,400 passenger ca.rs, per mile, per la.ne (pelmi/ln).

2.1.3 lkcupamy Density, the third component of the fundamental flow relationship, is idca.lly meuured aver a segment of roadway as described below. However, with the presence of permanent traffic detectors, it is often easier to estimate the density from the average vehicle occupancy on a detector. Detector occupancy is different from person occupancy within a vehicle. It is defin.cd as "the proportion of time that a deteCtor is occupied by a vehicle in a defined time perioda (Rocss ct al., 2004). Density can be estimated &om deteetor occupancy using the following equation (and assuming an ~vc:rage vehicle le~, L). 1 tiA

a

"Alt.to..IIIAI ('\1: TQ AMC:Ql'"'IAT/\Tl(')M C:,.,lr.lliC:COH-IIf': (TI lf"'HC(' 1Mf"'\ Cr"\ITtf"\._1

1

5280.0

!

L,+Ld

'0=-I

Equation 10-l

~ere D .. average density (vehicle/mile/lane (vehlmillnl)

0 = avetage detector occupancy (percent, expressed 2S a decimal) L. • average vehicle length (ft.)

L,~ c detectorlength (ft.) The detector length, L,~ is feiatcd to the actual physical length of the detector; however, depending on the detector sensitivity, the actual detected area may be longer than the physical detector. This is especially true for newer "point" sensors, whose physical length is very small. Typically, the equipment manufacturer has pr«alibtated the occupancy measurement to provide an estimate of density in the data output. To illustrate Equation 10-1, consider a 6-ft. detector that reportS an average occupancy value of 0.1SO foe a vehicle stream with average vehicle length of 26 ft. The resulting density is given by: D=

5280.0.150 + = 24.75 veh/mL/ln 26 6

The densiry is then related th,rough the theoretical speed-How rdationship in the HCM to a LOS. Sina: the HCM is b2Sed on passenger-car equivalents, the ~cnsity needs to be adjusted for the pcesena= of heavy vehicles in the traffic ~. Assuming the observed measurement does not contain any trucks and following the radial density thresholds in Exhibit 10-6, a density of24.8 vehlmi!ln results in LOS•C (greater than 18 and less than 26). 2.1.4~

Traffic sensors can also be wed to estimate: the distribution of vchide classes within the traffic meam; by binning vehicles by their length. If the speed of the vehicle is known (for example, &om cwo dosdy spaced loops), then detector occupancy can be used to infer a vehicle length. Heavy vdUde traffic is a cenaal component in the evaluation of ficeway operations and a dassilication srud.y is very common. A heavy vehicle conswnes more capacity than a p=ger car. In HCM analyses, the percentage of heavy vehicles is ~re converted to a passengercar-equivalent greater than 1.0 as a function of terrain (foe example, one erucic lll2)' be analyzed as cwo passenger cars). Especially on steep upgrades and downgrWs, heavy vehicles alfea: traffic operaoons due to slowa speeds, either ~use their accderation powu is limited, or because they need to slow down and shift to a lowec gear for safety reasons. In addition, heavy vehicles have a more severe impact on deteriorating prn:ment conditions than do passenger cars. Therefore, a classification count can be important roth in temlS of pavcnent design and for pavement mainrenana: schedules. An example for results &om a freeway classification study on an urban freeway is given in Exhibit 10-8.

~ooo.---------------------------,

25000 20000

E 1sooo

i ,1;

10000

5000

.IS'

§

----------

5-Axle TraciDr-Semitraller Combination

lt

I 400 200 0

__ -~!! ~n!l!elJ'!!_! I,ru~ __ ... ... __ _

Mon

Tues

Wed

Thurs

Fri

Sat

Sun

Day of Week

Source: Highway Ctplll:ity MllnWZi 2000. Copyright. National ~=y

of Sciences, Washington, DC. Exhibit 8-5. p. 8-5. Reproduced witb permis.!ion of tbe Transpomtion ~ Boud. ~roo\"'~" ~nn ~11~n;lncui

I ~nP~ C\t11rfiP4\. • 1A5

,,-.,.

;r;,·,·~~··'·"

i

50

Legend

40

- - - Lane 1 Right Lane -------------· Lane 2 - - --Lane 3

30 20 10 0

---·------

0

5

10

15

20

Time Headway (s)

Source: Highwll] Capacity MIZIIuai 2000. Copyrighr, National Academy of Sciences, Washington, DC. Exhibir 8-28, p. 8-26. Reproduced with permission of the Transponation Research Board. ·

For HCM-cype analyses, all types of heavy vehicles are commonly aggregated imo a percentage of heavy vd\icles relative to the overall traffic B.ow. For ocher applications, including planning-levd studies and pavement-mwagement programs, the exact classilicarion type may be of interest. Modern detectors and associated software have built-in algorithms that convert axle patterns to bins of heavy vehicles. Manual classifications can be performed foliowing conventions in the AASHTO (2004) •Green Book" or Chapter 14.

2.1.5 Gaps and Hetulways In-road traffic detectors further give estimates of headways and gaps on a per-lane basis. Headway is defined as the rime between two successive vehicle arrivals at a fixed location (a detector) measured from the same point on the vehicles (for e-xample, front bumper to from bumper). A gap is defined as the time berween a vehicle departing the detector (rear bumper) and che arrival of the next vehicle (front bumper). The difference between the headway and gap measured between two vehicles is the occupancy time of the first vehicle on the detector. Gap and headway times are related tO the relative distance between vehicles as a function of the vehicle speed. A driver who foliows a vehicle at a 2-sec headway will keep a greater disrance at higher than at lower speeds. Presumably, gap times are more descriptive of driver behavior than g-.tp distances, since they are constant over a range of observed speeds, while distances thange.

A headway study can be useful in traffic flow theory applications to evaluate the arrival distribution of vehicles, which can, for .e-xample, be used as input in simulation models (see Chapter 11). A srudy of gaps in freeway traffic can be useful for analysis of ramp metering Strategies, where on-ramp demand is released as a function of gaps in the outside freeway lane(s). A freeway gap study can also give insight into driver-following behavior, where short gaps (at high speed) are evidence of an aggressive (and risky) driver behavior. Both headway and gap studies are more special.iz.ed than speed, .flow and occupancy studies. While detectors can readily be used to obtain chese data, they are less fre... quencly used in nonacademic applications. An example of results &om a freeway headway study is shown in Exhibit 10-9. The exhibit clearly distinguishes headways on the right lane (large variability) and the twO left lanes, where headways are grouped very tighcly.

2.1.6 Onnp/Umee Compliance studies are typically conducted at or ncar intersections with regard ro various TCDs. However, compliance can be studied at any location where disregard for a traffic regulation or law may take place. The concept of compliance studies is simple since they essentially answer a "yes/no" question. However, the acrual data coUeccion can be challenging on busy high-speed freeways, especially if the intent is co identify violators in an HOY lane. HOYcompliance studies require the analyst to estimate the number of persons within a vehicle, which can.ooc be done 186 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

·through automated sensors. Instead, roadside or overhead video recordings or manual observations may be needed :sample driver behavior.

to

Other compliance studies on freeways are more straightforward. Violators on tolled facilities arc identified automatically if che vehicle doesn't communicate with che toll tag reader. Similarly, violations of no-truck restricted lanes are obtained automatically from classification data. Chapter 8, "Compliance with Traffic Control Devices,M provides useful material regarding sample size calculations and relevant study methods. 2.1.7 Person Occupancy llnd Bus/Transit In some cases, a freeway study may include data collection on special vehicle classes, including high-occupancy artd transit vehicles. Most performance measures for freeways are based on vph mettics, which tend to underestimate che benefits of HOY and transit modes on the total person throughput on a facility. Person-based measure of cffcctivcsness (MOE) are recommended for these modes, because the multioccupancy vehicle presumably replaces several (drive-alone) vehicles. To escimau the person throughput on a transportation f.aciliry, a study of vehicle occupancy needs co be conducted. ·

Obtaining data on ~chide Occupancy for aa.osit vehicles is most easily obtained from the tranSit agency and usipg data collection methodologies laid out in Chapter 13 or TRB Transit CapaciJy and; Q}lality of&rviu Manual (TRJ3, 2003). This approa<;h is likely to be more efficient and result in greater accuracy data than any transit occu pancy study in the lidd. The person occupancy of HOY vehicles on the ocher hand, needs co be performed manually in the field, and as discussed in section 2.1.6, is challenging. !>. manual occupancy study is both time and labor intensive, but is mote reliable and feasible than automated methods that rely on video image processing. Person occupancy is best measured in transition zones, such as ramps or access points co an ML, since vehicle speeds are expected to be lower. It is u.nrealistic to collect data on every vel:ricle apd instead, analysts should employ a random sampling scheme by sdecti.ng every n"' vehicle. The value of n depends on·the volume, Bow rod speed of traffic, and the ability to observe and record occupancy. A constant sampling frequency (for example, 11• 10) asSures an unbiased study. The reader should refer co Appendix C for details on random sampling and statistical tests of significance.

2.2 Segment Studies While che fundamental types of freeway data can be collected ac spot locations, some m.~ures are more appropriately defined on the segment level. They include direct measurements of segment density, average travel times and vchicl e queues resulting from congestion. 2.2.1 DensiJy. . The discussion above highlighted that density can be approximated from detector occupancy. Buc it can also be measured directly, by dividing the number of vehicles in a segment by the segment length (vehicles per mile). Snapshot estimates of segment density can be obtained fcom overhead and aerial photography; however, this approach resultS in added data collection expenses. If aerial photography is used co estimate densiry, a photo is shot at regular intervals (for example, every half-hour) during the peak period from a plane, helicopter, or unmanned aerial vehicle (UAV) The analyst then counts the number of vehicles on a known segment length {for c:umple, between cwo interchanges) co obtain an average estimate of density for that time period {Dowling, 2007).

Alternatively, density ~ be inferred from occupancy measurements at spot locations as discussed above. Densiry is further related to the traffic stream parameters of 8ow {v, in vehicles per hour per lane) and SMS (s, in miles per hour) through the fundamental equation

d=~ s

Equation 10-:Z.

Density can therefore be computed from measurements of the other cwo parameters (Dowling, 2007). In the HCM. density is the measure used co define LOS for freeway operations as was shown in Exhibit I 0-6. A visual representation. of different deruitycondicions and the corresponding LOS caregory is shown in Exhibit 10-10.

Freeway and Managed Lanes Studies • 187

Sour= Hjtfnuay CapflriiJ Mllmi4L Copyright, N~tional Academy of Sciences, Washington, DC. Exhibia 13-5 chrough 1310, pages 3-8 chrough 3-10. ~roduo:d wich pennlssion of che Transpon:uion RA:scarc:b Bo:ud.

2.2.2 TrtSHl Trme anJ DJAy Travd times are becoming an increasingly popular perfonnance measure for frecm.y operations for twO major reasons. First, they :uguably relate more directly to the usee-perceived quality of service of the transportation facility thm the "engineering• concept of d.e.nsity. Second, they are useful in making routing decisions on ML facilities and ue used to set the relative pricing siiUcrure of tolled lanes relative to GP lanes. The measurement of travel time has been discussed in decail in Chapter 9, including an overview of state-of-the ut technologies such u toll-transponders, wireless technology (such u ceU phones or bluecooth devices) and GPS technology Wt can be U.sed to obtain travel-time da~ Through the application of these technologic$, the availability of travd-time data bu increased dramatically, and i.s therefore used more frequently in practice t~y. The concepts of cnvd time also enter into the measwes of travel-time reliability u i.s discussed in Section 4.0. Travd time i.s further used to estimue the dday on a facility, by ~ the free-Bow travd time from the 6ddobserved cravcl-time ~verage. The free-Bow travd time on a freeway can either be measured during off-peak conditions, or~ be approximated br. dividing the segment length by the free-flow speed. Even though the HCM does not use travel 488

-

l..tAft.JIIAI 1"'\C TDAt.I~DnDTATif"\AI C~lf':I~ICCOU,Ir" I'"'TI lf"''ll'"l'"

.._Ur" '""''""'"'''

time delay (TIT>), it is a useful performance measure. Agencies commonly use delay to estimate a congestion use r cost by . multiplying the total delay (in vehicle hours) by me average vehicle ocrupancy (in persons per vchicle),and men by an eco. nomic estimate of me value of time (in dollars per person-how). The resulting user cost (in dollars) gives a rough estimate · of me economic impact of uaffic congestion, and similarly, the benefits obtained from congestion mitigation strategies. 2.2.3Qunus Queuing studies are among the most important study types for freeways; however, they are also some of the most difficult studies to conduct. Queues are a key performance meuic repoHed from the analysis of freeway facilities in the HCM, as well as simulation-based studies. They are also commonly reported by the media to alert drivers of the spatial extent of freeway congestion.

Freeway queues are difficult to study due to their sometimes long and variable spacial extent and a lack of reliable automated means of measurement. There are three basic methods for estimating freeway queues in the field. 1. Th.e analyst can estimate me spatial extent of queues from speed drops on permanent detectOr stations. The problem with this method is the deployment of mffic sensors is not continuous, md even on very advanced systems, half..mile spacing is common. The agency therefore only has "spotty" data on vehicle queues and further has to assume that a certain speed or density (occupancy) threshold in fuct corresponds to a queued siruation. 2. The analyst can direaly examine queues in me fidd or from video using visual observation at fixed vantage points. Since the spatial extent of queues varies, me use of preinstalled freeway monitoring video cameras.is typically more.rd.iable than observations from a roadside or bridge observer location. Many modem freeway TMCs have access to a broad deployment of video cameras that can be used to monitor queues from within the TMC. The collection ofqueuing data'in this fashion is not automatic and requires some judgment about when a vehicle is considered to be in a queued state. The precise location of the back-<>f-queue can also be difficult to see depending on ·the video angle. Chapter 6 offered a methodology for performing a queuing srudy for the approach to a signalized intersection. The same general methodology applies here, with the caveat that the qurues are much larger in both spatial extent and temporal duration. For this reason it is acceptable to increase the analysis intervalli:om 1 to 5 min. and use the distribution of 5-min. back-of-queues to describe the queuing system.

3. The analyst can record queues from within a vehicle traveling in the opposite direction of the congested freeway facility (typically driving slowly along the shoulder with hazard lights on). Peak how traffic patterns are typically di.rectional and conditions in the opposite direction are therefore less congested. Given fast freeway speeds and the cognitive task involved, the ai:lalyst should not be the one operating the vehicle and should not disttaet the vehicle operator from the driving task. Queue lengths can be related to mileposts at the side of the freeway or through waypoints in GPS equipment. This fOrm of queue srudy can also easily be combined with a travel-time srudy of the congestod facility on the return trip. The selection of an appropriate queue study methodology ultimately ilepends on detector coverage, the presence of overhead video cameras and the ability to see the spatial menc of the queue &om video or the opposite ttavd din:ction. Sample size is also ~ important consideration, since travding co and from the 6dd can be timcxonsuming. Another important point in relarion to freeway queuing is that queuing patterns are highly sensitive to the source ofcoop cion. For cx.ample, a queue may be caused by a 6=1 boaleneck where demand =ds the (constant) capacity. In this case the queue will grow upstream of the bottleneck location. As demand drops, the quelic will begin to clear from the back towards the bottlene4: location. On the other hand, if the queue is caused by an incident such as a collision or cemporary lane dos~ the segmtnc demand may be constant, while the capacity changed. The reduced capacity will once again cause a queue that grows upstream of the bottleneck location. However, once the incident is cleared and assuming constant demand, the queue will noW' dear &om the front, which may cause some data collection challenges. In another caveat, a free" way queue may be caused by a moving congested source such as a slow-moving heavy vehicle or a military convoy. In these cases the low-capacity bottleneck is mobile and the resulting queues •roll• with the congtstion source. The analyst needs to be aware of the particular queuing paaems in the studied segment and needs to adapt the data collection strategy accordingly, possibly utilizing a second moving vehicle in the opposite direction of aavd collecting similar data at set time intervals. Queue srudies are also performed in rdation to ramp metering strategies, to evaluate the queuing patterns on the metered ramp. These are conceptually similar to queue studies at intersections as discussed in Chapter 6. Ftnally. queue studies are common in simulation applications presented in Chapter 11. •

2.2.4Safay Fre~vay operations arc substa.ntially affected by crashes and other incidentS. A significant amount of con~tion in the United Stares is the result of incidents resulting in huge annual productivity losses, in addition to the potential loss of human life and personal injury (PB Famdyne, 2000). Safety sru.dies on instrumented freeways are aided by availability of cmtera systems. While these systems typically are not used to record continuous observations (and therefore won't have a record of the incident), they do allow the agency to quickly identifY the location of a crash. This is especially hdpful in reducing the amount of incident clearance time, which can significandy affect the operations of a freeway facility. In tenns of user-cost elfeas on nonrecurring freeway congestion, reducing incident response time is a key target for many agencies. More generally, safety studies are used on freeways to evaluate specific design features (barrier type, median width, cuCve radii) or policy strategies, such as reduced speed limit, truck restrictions, or enhanced enforcement. Chapters 17 and 18 provide details on collision stUdies and alternate safety sru.dies that are also applicable to freeways.

me

2.3 Spedal Freeway Studies 2.3.1 huUienl Mmuzgemmr Traffic incidents are randomly oc.curring events that .result in a reduction in me avWable capacity of a roadm.y. Incidents include such issues as collisions, disabled vdlldes and spilled goods. Traffic incident rna.nagernenc is defined as "me systematic, planned, and coordinated use of human, .i.nscirutional, mechanical, and technical resources to reduce the duration and impact of t:raflic incidents, and improve the safety of motorists, crash viaims, and t:raflic incident responders" (Kuhn et al., 2005). Proper use of avWable tools results in reducing me time to detect and verify a t:raflic incident, carrying out the appropriate response and dearing the incident quickly and safely so capacity is rescored. FHWA's Trrrffic /ru:idmJ Managemmr Handbook should be consulted for further information (PB Farradyne, 2000). Studies performed to evaluate incident management include an assessment ofincident response and clearance times, and how either can be reduced through improved incident management. Analysts may further study the average queuing and vehicle dday impacts associated with incidents and the ~t strat~ as di.scussod in Section 2.2. 2.3.2 Work Zones . Freeway operational performance and incidents are also often of interest in work zones. The reduced frequency of new freeway consuuction, the associated emphasis on increasing capacity on aisting facil.ities and a gmeralty deteriorating cranspor· ration .i.n&asaucrure all .result in inaeases in freeway construction projeers. R.eoognizing the impact of freeway work ~nes on the travding public, FHWA •Safety and Mobility Rule• (FHWA. 2004) requires agencies receiving federal funding to better predict, ma.t12gC and evaluate the user C06ts associated wim work zones. Among omer things, this requires agencies to carefully assess the expected effects of work zones and to monitor mem during construction.

Work zone stUdies are gmeralty not difli:rcnt from other freeway sru.dies, as me same pcral measures are of interest: volumes, speeds, rravd rimes and queues. An important difference h~ is the potential lack of automated data sources, as sensors are often 1m1
2.3.3 ~ &nurin In recent yeats, much national attention has been devoted to c:rnergency evacuation strategies in response to narural disasters (hurricanes) or th.tea1:s to national security. Whatever the scenario, transportation engineers and planners have devdoped evacuation t:raflic plans that commonly involve special lane rcversal.s on interstates with lanes in both directions moving traffic away from the disaster area. While these reversal facilities typically fall into the freeway functional class, the aitical congestion points are 1oca.a:d at me transition points to and &om the freeway lanes (Kuhn er al., 2005). Oftentimes, Law enforcement has to employ special temporary traffic control Strategies to direct drivers to the evacuation &cility. As a result ofthis cemporuy aaffic control and c:xpectcdly high demands, the capacity of the transition points is expcctedly much lower than the evacuation freeway capacity and should be the focus of any sru.dy of evacuation scenarios. Since data coUeaion during a na.rural disaster or in a national case of emcrgcncy is unlikely to be performed manually, permanendy installed sensors are the best source ofdata. .Alternativdy, many evacuation strategies and scenarios are tested using simulation computer models; Chapter 11 provides more detailed discussion of those. 190 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

3.0 DATA COLLEcriON PROCEDURES . Freeway facility performance monitoring and evaluation typically includes two basicCltegories of daca: facility use (voliJlllCS• .speed, compliance) and facility performance, which includes both op=.cional (uavel cimes, delay. ecc.) and safety daca (collisions). Facility performance evaluation further includes before-and-after scudies on implementation of various freeway measures to improve efficiency as well as incident management. Data can be collected through a varicc:y of m~ induding automatic or manual methods, many of which can be done continuously along a facilicy or through vehicle sampling methods (Kuhn et al., 2005}.

This chapter addresses data collected on two types of &ecways: GP and ML facilities. While chc traffic operations on both facility types are generally similar, the dara coUeccion praaices differ in two ways. FllSdy. ML facilities are often outfitted with more sophistiC!ted data coUeccion equipment, allowing fOr more automated and more detailed srudies. In addition, ML facilities have additional and unique performance measures associared with them that are clifferent &om general freeWay facilities. These measures are rdared to vehicle occupancy, intelligent cransporcation systems (ITS) straregies and real-ciJne ~eed measurement for dynamic pricing in HOT lanes.

3. 1 Facil ity Performance Data The operational characteristics of freeways are described through the speed-flow-density relationship derived &om traffic flow theory. [nits most basic form, the flow of vehicles q (vehlhour) is the produa of the SMS v (miles/hour) and the density of traffic It (vehicles/mile). As c:raffic flow on a fi:eeway increases, the resulting density on the srudied segment inae~es as well, while the speed of traffic begins to drop. Exhibit 10-6 showed the theoretical speed-flow-density relationships as adopted from theory. while Exhibit 10-7 showe
This fundamental rdationship is-the underlying basis of freeway analysis methodologies in the HCM (TRI3, 2000) and als 0 finds apptiC!cion in simulation analysis :IS diScussed in Chapter ll. Speod, flow and density are key performance measu.te5 on freeways and are therefore common measures included in freeway srudies. Today they are most commonly measured using in-road or roadside sensors; however, it is somerimes necessary co collea data manually or chrough nonpcrmanc:Pt means. Since density is difficult to measure in the fidd, detector ocrupancy is a common substitution. Detcct:ors are also capable of providing analysts with an estimate of gaps in c:raffic, which may be important in some circumstances.

In addition co the above fundamental measures, freeway studies are often interested in the composition of the c:raffic ~ and the classi.fi.Cltion of vehicles. The presence of heavy vehicles directly influences the op=.tioos of chc traffic sueafll• especially in mountainous terrain. The vast majority of detectors are capable of providing.vehicle classilicarion ifthis dlJ. ta dement is necessary. Fmally, the safcc:y of motorists is a primary concern to engineers. Safety can be comP.romlsed in many ways. For insranC~: during congested conditions, c:raffic often becomes unpredictable as ~des begin to accelerate and decelerate em.ticallY· Facility improvementS such as HOV/Har lanes or reversible lanes could alleviate a large portion of the u:ailic collisioll.l tb.:lJ.C cause craffic to act unpredictably. Since many of the MI. aeaonents are installed co prevent unnecessary aaffic coUisiooS • sa6:ty srudies should-be condUcted and updated frequendy co determine the long-term sa6:ty dfcctiveness of the facili t:Y· Chapter 17 provid.-more information on how to ~cifically conduct safety studies.

3.2 Data Acquisition Performance da~are primarily collected using a continuous monitoring station; however, it is sometimes neccssu-Y to collect data using manual methods or nonpermanent data collection points. Continuous monitoring refers to tb..O:: use of detection to observe traflic conditions over some period of time for use in everyday freeway operations or traHi C. studies. Monico~ craffic through an automared method is typically more prevalent than nonpermanent or manual mccb-ods because craffic operational and planning studies are done frequendy to observe craffic congestion along the major sc~ of a state highway.ystem. Typical srudies or applications include traffic volume counts, dassificarion of vehicles, ocrupanc:Y studies to desaibe traffic Row during a typical day, ttaVd rime along a corridor, etc. Ar. times, it is necessary to coUea freeway data at specific IOC!tions that do not have continuous monitoring stations. Typically. these alternative methods would be use
against data collected by automated means, or along segments that do not provide consistent and accurate data using an automated format. For instance, it may be important to collect data at twO closely spaced interchanges where the weaving segment is very shon. The weaving narure of traffic ar this locarion would nor lend i!Sdf to automated dara collection methods because vchides would fRquendy be double-counted while changing lanes, not to mention that traffic may be traveling coo slowly during rush hour through one or more of the lanes. Lastly, it is also po55ible thar an agency might not have mearu or funds to c:Ollcct data in an automated fashion, and uses more rudimentary methods to updare outdated data or as a spot check for assumed growth patterns. Typically, data collected in this manner are for traffic counts, speeds and classification. In addition, ~en safety studies are conducted, it is nea:ssazy tO collect collision dara for the segment of intcn:st. Typically. the sate department of cranspon:arion or local municipality can provide the ncccssary data; however, it is po55ible thar the . analysts will neal to oollect the data on their own. Segment analysis methods would likdy need ro be employed using specific mileposts along the segment. Chapter 17 should be consulted for sample siz.e calculations and further reading on how t
"

l..asdy, it is often ncccssary to conduct surveys of the users of a new freeway or ML &cility to detennine the perceived effectiveness. This is especially useful when comparing aaual quantitative results to more qualitative results. Information on : }iow..to condua surveys can be found in Appendix B, •survey Design.•

3.3 Equipment Needs ~

Freeway monitoring is typically done through the use of a permanent count station. Typically. a data collection location takes the form of a small roadside station called an auromatic traffic recorder (!UR) or is part of a larger weigh-in-motion (WlM) station. are wually collectrd 24 bows per day, 365 days per~· ·

':Thea

An ATR will usually consist of a cabinet cont:aining a traffic recording unit which stores dara from one or more roadside or in-road detictors as described in Chapters 4 and 5. Many times, the data from an ATR is collected at a centralized 19C3(ion via aninternet connection on an as-needed basis. The detection type and daca collected could cake many forms depending on the type of detection used. Detection at ATRs usually uti.l.iza miccowavi:, point detectors, video detection, or surveillance cameras. Permanent data recording stations are also integrated in WlM stations. WIMs (Exhibit 10-11) are primarily used ro enfo1ce heavy vehicle axle weighcs; however, many times othc:r forms of detection are installed at the adjacent mainline &ciliry to acquire other forms of data. AI. ATR and WlM stations, dara can be downloaded in a raw form, but in many cases sofiware will.automarically reduce and tahulare dara in the format the user .rtqu.estS. . There is an array of detection devices available for collecting many types oftraffic data. Most are described in Chapters 4 and . 5, and include inductance loops, point detectors, video, infrared. miccowave, radar, acoustic and semipermanent deteaors thar can be removed for use ar othc:r sites. Alccmatively, auromated dara collection equipment, iri the form ofemissions and pavement monitoring. is becoming ioaeasingfy popular along freeways. The most common use of dara &om continuous '. monitoring sWions is traffic counts for det.cnnining ~annual daily traffic (AADT), typical daily traffic ~lurnc patrems. seasonality f2aors and to estimate vehicle miles cravdlcd (VMT). Each detection device has trade-offi; for instance, - 5ome are less acrurare or do not perform certain types of detection, some require increased maintenance, others allow more flexibility to change detection locarion, while others are more intrusive or cequire special mounting angles so vehicles do not · visually occlude adjacent detection mnes. For continuous moniroring purposes. the detection types used are more likdy ro be mounted 01 installed in a predetc:rmined locarion ro allow for more stabk daca cx:dlcction. For more information on the · tt.ide-ofl's between various detectors, users shoukl consult the FHWA Trrzffic lkt«tor HIINiboo!t (Klein et al., 2006). Although ma.!!r of the same types of data collected. by automattd means can be collected by manual or nonpermanent .methods, the.cquipment used is much more restricted. Manual data collection can be done by handheld count boards or . by using alaprap with a time-based maao such as the one described in Appendix E (Exhibits E-1 and E2). Often, a video camera is used to record a specific rime period of events so many types of dara can be collected in the office at a later time. , Nonpermanent methods of data collection commonly use detection through pneumatic tubes or movable point detectors communicating ro a roadside data collection unit. These are described in detail in Oupter 4, •volume Srudies." A small number of~encies and unMrsities havt a£lluircd equipment for video dctcaion tba.c can C3$ily be mm~ tO any necessary site. H~. these methods are typically used for research purposes since dal3 may be needed ar a specific location where continuous monitoring stations are not located. Exhibit 10-12 shows an example of equipment and a schematic of fidd deployment ar a freeway work moe.

·· j

·•

.,, I

<.

•

3.4 Personnel Training Requirements Aucomated freeway data collection is typically conducted by a sraff of individuals at the state levd; however, the research community frequcndy uses data provided by these agencies as well. Personnel are trained by vendors or internal staff members on various sofrware and/or hardware used with a multitude of detection devices. It is important that there are trained personnel on staff who understand how to use the devices so the necessary data can be collected accurately and reliably. Most continuous count stations are set up to collect data in an automated fashion w d arc only polled for d...ta daily, weekly, or on an as-needed basis. One common example is an ATR that uses a series of point detectors in each of the lanes of a freeway (Exhibit 10-13). This could be used to collect vehicle traffic counts, speeds, classification wd other details by lane. The personnel responsible for collet;t:ion of this d...ta must be diligent in organil.ing and storing data in a consistent and orderly manner so it can be identified quickly. The maghirude of data obtained by automated methods deployed over the transportation system can be large and the analyst needs to be capable of managing such quantities of information. If automated data collection is not possible, field crews can be used to collect field data manually, or for setting portable detection stations. These crews are typically staffed by the same department, or even the same staff members as those running continuous monitoring stations. Manual data collection methods in the field arc easy to conduct; however, they should be done well outside the travel way for safety reasons. Typically. an overhead vantage point is best; therefore, an interchange overpass or ramp is commonly used. When setting up nonpermanent detection stations, it is sometimes necessary to be in or ncar the roadway to install equipment. In this case, it is best to install equipment using temporary traffic control during periods of low traffic volume. For freeway installations, one or more lanes of traffic may need to be closed temporarily, which requires a more significant effort and potentially several road crews. While installing any sensors in or around the roadway, at least one person should be designated to pay attention to motorists. Reflective ves13 should always be worn by all personnel anywhere ncar the roadway.

3.5 Field Procedures Continuous monitoring requires Little-to-no 6eld preparation; however, collecting data along high speed roadways is a very dangerous task when using nonpermanent or manual methods. Proper preparation is vital to safely set up and collect data collection equipment. When these nonpermanent data collection methods are used, it may be best to set up a controlled test prior to acrually installing the equipment in the field. This is important to reduce the amount of exposure in the travel way. not to mention it will save time in the 6dd and make sure all equipment is operating properly. No.matter how rudimentary it may se~. a checklist is a valuable tool that should be developed to cover all tasks necessary to set up any equipment. Once rhc equipment is checked, plenty of time should be allowed to study the site through a field visit. Many times a supervisor provides a rough idea for placement of traffic detection devices on an aerial map. Although rhe placement of dcteaors may be installed cxaaly as noted in the rough sketch, it is a good idea to draw a condition diagram such as the one in Exhibit 17-10, oudining the basic features of the sire, where the dctc:ctors are positioned and distances to known points of interest. Chapter 4 provides additional guidelines for deploying nonpermanent dereaion devices along any type of roadway. When using manual methods for studying compliance or gathering data on person trips along Ml f:a.cilitics such as HOY/ HOT lanes, a good vantage point is critical to observing the actual number of occupants in each vehicle. Many times, video 194 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

is a useful method for obtaining such a vantage poinc;, it is also much safer to gather the data in the office chan nor the ·., roadway. Surveillance videos should be considered if observation cameras are already installed along the roadside. ' During the installation of ci:lta collection equipment on freeways (including nonincrusive roadside units), data collectors should ensure activities do nor interfere with traffic operaLions nor cause significant distraction to drivers. This sort of capacity-reducing "friction" from driver distraction is common for freeway work wnes, wh ere additional measures are taken (such as advanced signing, barrier placement) to minimize the impact on the uaveling public. For freeway data collection activities, the data collectors should be sensitive to these issues, and minimiu interference with traffic operations. This aspect is particularly important, since the focus of data CQIJection is cypically to measure "standard" operating conditions, without the preseno: of additional driver distractions. Lasdy, evaluating the effectiveness of various freeway strategies should not be con~ideced a one-rime accivicy, but parr of a periodic review of the component or strategy employed, as well as the overaiJ system performance. This is especially true just after a freeway management strategy is fust employed. It is recommended formal evaluations of a new facility take place several times during the first year of operation. EvaluationS at 1 month, 6 months and 1 year ~.fter insrall:uion would be ideal.

4.0 DATA REDUCTION AND ANALYSIS The analysis of freeway data are generally similar to the evaluation process of other facilities. Raw dara are aggrega~ed to a common analysis interval (rypically 15-min.), descriptive statistics and statistical testS are performed foUow1ng methods described in Appendix C and tables and charts are created acCQrding to principles oudined in Cbapcer 3 and Appendix D. This section therefore presents only the aspects of freeway data analysis that are different from che standard approaches presented. else:-vhere.

4.1 Spot and Segment Evaluation A significant portion of freeway swdies is focused on evaluations of one specific spot or segment location. Due to the automated narure of most freeway data, large amounts can often be gathered quickly and easily. As a result, ma.JlY freeway data sets are continuous, rather than being focused on a particular peak-hour interval. Consequently; data c~ be displayed graphically for extended periods of time. Exhibit 10-14 shows an example of a graphical represenraci0° of freeway volumes (also separated by cars and uucks) for a 24-hour period.

5-Minute Volumes over 24-hour period 800

li.

i i

700

600

+---------------------------------------~~--------------

500

+-----------------~------------------~

:\/.!¥.

400

1 J

I

J

•AIITraffl<

a c ars

300 +---------------~

•Trucks

200

!

100

I.

a£

0

0

1

2

3

4

.5 •6

7

8

9

10 11 12 13 14 15 16 17 18 19 20 21 22 2.3

Tlme of Day (O•OO- 24: 00 houra)

Freeway and Managed Lanes Studies • 19~

5-Minute Truck Volumes by Lane - 3-Spm 3S 30

l :

lS

1 l 20

lllane4

.E

j

lS

.I

10

II lane 3

~ 'S

Dlane 2 Blanel

E

~

s 0

.. .... 9.

0

....

t1 t1 "' ...

.,

... t1

.... r:'!

"'

.. ...."' .. "' .. .... .. .... .. .,.. .. ...... .. "'.. .. .. .. .,.. 0

'!'!

'!'!

0 ~

.... ~

0

'!'!

.... ~

.;;

9.

0

t1

0

t1

t1

t1

.~

'!'!

0

'!':

....

'!':

~

'!'!

....~

lime ofDI'((':OOpm• S10Gpn)

With the availability of lane-by-lane sensors, it is further possible tO evaluate traffic patterns by lane as is shown in Exhibit 10-15. The Exhibit shows the observed 5-rnln. counts of heavy vehicles per lane over a two-hour period.

In addition to craffic volumes, analysts arc frequently interested in dau. of freeway flow and the associated SMS. The paired speed-flow obsclvations are commonly displayed in a graph as was shown in Exhibit 10-7. The performance of an ovctall facility over time and space can be represented in a 3-dimensional contour plot. Exhibit 10-1~ shows the predicted freeway density and SMS over 18 analysis segments and 10 time periods. The graphs in the Exhibit were generated from the FREEYAL compuu.cional engine for the HCM 2000 freeway facilities chapter (fRB, 2000), but could easily be created for acrualfidd dau. as well.

__

.,...._~

-e......nCON4Il-l

...

·u

....... ·-......... ... ·....... ......... ....... ·.....

.......

..... ......... .... ..... ....

" •

,.,.

~-

I

II

· ~

..

,,,~

\ '

•""

-

u•~IIIAI n~ TIIAN~PI'1RTAT11'1N

FNCiiNFFRING STUDIES. 2ND EDITION

--

IAAuuu.,u.,uu'

--

~ 4.2

System Monitoring

JFor &ec:way systems with broad and continuous sensor coverage, comprehensive freeway performance monitoring ~rograms are employed to trade system performance (Ncudorff ct al., 2003). For this purpose, FHWA defines a series of system-wide performance measures that are derived &om various field data sources. The performance mctrics are conccptw.lly divided into 1111Jbili1J avtrages and rruabiliiJ mtasum. The following definitions (in italics) arc given by FHWA. . Average Mobility Measures (Definitions &om Neudorff et al., 2003)

Tram-time index - This a rttlio ofITIIvtl conditioru in tht pea! perUxi to a target or accrptabk lfflvtl condition (typically foe-flow conditioru art used). The trawl-timt indcc indicates how much longer a trip will take during a pta! time. For tx4mpk, a travJ ti71ll! indcc of1.3 indicaw tht trip wili talte 30 ptrctnt longer (1.3 timeJ longer). · Pcn:ent of congested tram- This is primarily a syttnn 7111!4SUrt that quantifies the exttnt ofcongestion. A fot-.Jiow spud is used as a congestion "bmch11141'1t" and any trawl on a road stction for a time ptriod that is 1m than the fot-.Jiow spud is tktnmirwi to he congested. 'J.:ht congested trawl is sum71ll!d and thm divided by totiJi trawl esti1114UI.

Travel in this case, is defined as a measure that quantitatively describes the amountof traffic flow. Common examples include VMT or vehicle-hourS:-tcaveled. Delay per penon - Exprmed in pmon-hours peryear, this 71ll!asurt is used to nduce the tofiJi travel tklay va/ut to a figure that is mort rrlatabk to user txptrimct. It also normalizn the impact ofmobility projtcts that handk much higher ekmand

than othtr alurnativts. The measure of delay per persoo, is more appropriate when the traffic flow contains significant transit and HOY components. In regions where the traffic stream is more uniformly composed of passenger cars with a common average occupancy, this measure may be substirutcd by "delay per vehicle. • Reliability Measures (Definitions from Ncudorff ct al., 2003)

Buffer index- This measurrtxpmm the a1111Juntofextra "buffer" nmkd to be on-time 95 pm:tnt ofthe rime (/au I day per month). Trawkrs could multiply thtir average trip time by the buffer irukx, thtn add that buffer ti71ll! to their trip tQ msu~ thty will be on-ti71ll! 95 pm:mt ofa// trip1. An advantage oftxprming the n/iabilil] (or la+k thenoj) in thu way is that a pm:tnt valut is distance- and time-ntutral. Percent "Yariation - Also !mown as the cotjficimt ofvariation, this is the lt1111Junt ofvariabiuty in relation to avtrage travel

conditions. It is calculauJ as the sran4ard deviation Jjvifkd by the mean. Trawkrs cou/J muhiply their avtrage travel tinu by the pm:qzt variation, thm add that product to thtir avtrage trip ti71ll! to gtt the .time ntetkd to be on-time about 85 pm:mt ofthe time (one stan4ard deviation above the 71ll!an). Higher vdiuts indicau kss rtliabiuty. Misery index- This 71ll!a.!Urt atttmpts to quantijj the inttnsily oftklay for only the worst trips. The avtrage travel rau is

subtraaedfrom the upptr 20 pm:tnt oftravel rafts to get tht A1111Junt ofti71ll! btyQnd the avtrage for some 41111Junt ofthe slowtst trips. · When deciding on a system-wide performance evaluation scheme, an agency needs to decide which of these measures it wants to usc to track performance. It then needs to ensure the sensor instrumentation and detector coverage are sufficient to deliver the desired data. NCHRP Synthesis 311 (Shaw, 2003) identifies additional performance measures to supplement the ones discussed above and shown in Exhibit 10-17.

Freeway and Managed lanes Studies • 197

., • (- _ m

,~~wtt;·~~~··~::·· ~~·.:~ ~~t~~~t~~-:~;';.~~l:~w~~~~~~~;-~~¥~;·.~-~ -~ ~ _.A_ o o _ : __ . ·~ : __ - ....:..· ~' Q

o;-

_ •

•

• :

Average delay (total, recurring & incident-based)

Percentage of travel heavily congested

Average door-ro-door travel rime

Person-miles travded

Average speed weighted by person-miles traveled

Persons moved

Bridge condirion

Quality of travel (users' perspectives)

Density {passenger cars per hour per lane)

Quantity of travel (users' perspectives)

Duration of congestion (lane-mile-hours at LOSE or F)

Queuing (frequency and length)

Evacuation clearance rime

Safety

Incident induced delay

Toll revenue

Incident race by severity (e.g., f:.ral, injury) and type (e.g., crash, wearher)

Travel rime predictability

lncidenr response time by type of incidenr

Travel time reliability (% of trips chat arrive in acceptable time)

Incidents

Truck-miles rraveled

Level of Service (LOS)

Trucks moved

Ourcomes (Operational) Performance Measures

Utilization of the system (agency's perspective)

Outputs (agency performance)

V/Cratio

Pavement condition

Vehicle occupancy (persons per vehicle)

Percent of ITS equipment operational

Vehicles moved

Percent of miles operating in desired speed range

VMT

Percent of system heavily congested (LOS E or F) Source: Recommended perfo=ce measures compiled by author from NCHRP Synrhesis 31 1.

4.3 Managed Lane Measures The ev;aluacion of MLs is unique in that it involves an economic data component. The principal mea.sure of interest in an ML analysis is the travel time, and more precisely the travel time difference between ML and GP lanes. For instance, it is that trllvd time difference, compared with the prevailing rraffic demands, that sets the dynamic pricing scheme on a rolled facility. and that affects the driver decision-making process (to pay a toll or co decide to car-pool). Before-after analyses are also common tools used by analysts evaluating converted GP freeway facilities to ML facilities. For instance, an analyst may be interested in enlu.ating an ML facility wing measures such as collisions, speed, travel time, or density (which allows the analyst to determine the associated LOS) to determine the nee effect of converting GP facilities. In this way, it may be possible to determine any poosible benefits associate with converting GP facilities co any ML facility type, and even determine when a facility type should be converted based on one or more measures. It is important that analysts wishing to conduct such before-after studies consider using a comparison group to account for confounding factors that would bias the results of the srudy. For instance, seasonal va.riacions ca.used by weather events may cause speeds or coUisions in the after period to naturally be higher than in the before period. Using a comparison group, and collecting data during the same time periods as the treatment group, would help account for these types of natural occurrences, lessening the bias of unwanted factors. Use of comparison groups for analysis purposes are commonly used and are further d.iscu.ssed in Chapter 17 and Appendix A. If analyst:s are nor &.m.iliar or do not understand how comparison group srudies are conducted, they should consider consulting an expert. More derails on Mu evaluation and analysis are described in Texas Transportation Institute (1998), Neudorff et al. (2003), Shaw (2003), Pen and Sciara (2004), Kuhn ec al. (2005), Jacobson (2006) and orher publications.


: 5.0 REFERENCES ~ American Associ arion of State Highw:~y and Transportation Officials. A Polil) 011 Gtommic D~sigr. ofHighways and Smrtr. ,Washingto n, DC: AASHTO. 2004. Dowling, R. Traffic Analysis Toolbox Volumt \11: Dpnition, lnttrpretatilln, and Calculation ofTraf!ic A1111/ysiJ Tools M~asurn of Efficrivmm. Report FHWA-HOP-08-054. Washington, DC: Federal Highway Administration, 2007. EURAMP •Handbook of Ramp Metering." Euro~an Ramp Mtttring Projm (EURAMP) &porr, September 2007. Federal Highway Adminisuacion. Fruway OptTatillnr Handbook. Report FHWA-OP-04-003. Wasbingcon, DC: FHWA, 2003. Federal Highway Administration. WOrk Zone Operations Best Practicer Guiekbook. Report FHWA-OP-00-010. Washington, pC: FHWA, 2000. Federal Highway Administration. Work Zone Saftty and Mobility R11k. Federal Regisrer 69, No. 17 (September 2004). Hauer, E. Ob$tT'IIarionallkforr-AjitT Studies in Raad S4foty. Amsterdam, Netherlands: Elsevier Publishing.·1997. Jacobson, L., J. Suibiak, L. Nelson and D. Sallman. Ram; Manag~ment and Control Hllndbook. Report FHWA-HOP-06-00 1WashingtOn, DC: Federal Highway Adminisuacion, 2006. Klein, L, M. Mills and D. Gibson. Traffic Detector Handbook. Report No FHWA-HRT-06-108. McLean, VA: Federal HighwaY Adminisuacion, 2006. Kuhn, B. er al. Managtd lAnes Handbook. FHWA!fX-06/0/4160/24. Auscin, TX: Texas Department ofTransporcacion, 2005 · Neudorff, L. G., J. E. Ran and Systtms. Washington, DC: National Cooperative Highway Research Program, 2003. Texas Transportation I.nstirute, Parsons Brinckerhoff Quade & Douglas, Inc. and Pacific Rim Resources, Inc. NCHRP Rcpor1: 414: HOY Systtms ManU4I. Washington, DC: Transportation Research Board, National Research Council, 1998.

'

Transportation Research Board. Highway Capacily ManuaL Washington, DC: TRB, 2000. Transportation Research Board. TCRP Report 1100: Tmnrit Capanty and ~lity ofSmtU-t ManU4L Washington, DC: TRB. 2003.

Freeway and Managed Lanes Studies •

19~

--·

Chapter 11 ... .. ........ .. ... .. .. ... ... ..... .... ...... ................. ...... ......... ...... . .. .. Simulation Studies Bam-]. &brodn; Ph.D. 1.0 INTRODUCTION

201

1.1 Purpose of this Chapter

201

I .2 Limitations of this Chapter

202

I .3 1}tpes of Simulation Models

202

1.4 When to Use Simulation

203

1.5 Definitions

205


206

2. I Sensitivity Analyses

206

2.2 Evaluating Alternatives

208

2.3 Predicting Behavior ..

210

2.4 Emergency Scenario Modeling

212

i.S Safety Analyses

215

2.6 Environmental Studies

217


218

3.1 Model Setup

218

3.2 1}tpes of Measures

221

3.3 Input Calibration

223

3.4 Output Validation

227

3.5 Proc¢ure Summary

•228

4.0 DATA REDUCTION AND ANALYSIS 4.1 Concepts of Stochastic Variability

229 229

4.2 Determining the Required Number of Simulation Runs 229 4.3 Reporting Simulation Results

231

4.4 Documentation

232

4.5 Animation ~nd Visualization

232

5.0 SUMMARY

233

6.0 REFERENCES

234

1.0 INTRODUCTION

1.1 Purpose. of this Chapter 1'hi.5 chapter presents material related to tr.lll.Sportation studies that use traffic simulation modeling software. It has three main objectives.

1. to provide guidance on how to design studies using simulation tools as an alternative tO field data coUeccion; 2. to provide guidance to simulation-tool users on how to march field data to simulation inputs and outputs through calibration and validation; and 3. to provide nonusers of simulation tools a basic undemanding of assumptions that go inro a simulation study and the relationship berween simulated and field-measured data. The focus of thi.l chapter i5 primarily on microscopic simulation, which is a commonly used form of simulation in uaffic engineering and planning practice today. The chapter will reference mesoscopic and macroscopic simulation models as necessary, especially in cases where their applications differ from microscopic models. The differences berween these three cypes of simulation models arc discussed below. The chapter focuses on stochastic uaffic simulation models, which represent complex representations of traffic systems in a computing environment. A stochastic traffic simulation analysis typically involves one or more of the foUowing: • the aim ro evaluate a complex uaffic condition that cannot be readily addressed through deterministic (equation-based) analysis techniques; • the availability of very detailed data to properly calibrate and validate the model; • an initialization period to populate the simulated network before extracting any data; • the use of multiple simulation runs and the statistical analysis ofstochastic outputs; and • input parameters that are preoptimized using an external tool. At the time-of this writing, simulatio~ tools (for example, cycle lengths, splits, offsets) typically do not optimize; other tools or methods are needed to obtain the ~t possible set of input parameters.

1.2 Limitations of this Chapter This chapter is intended as a general reference for the use ofsimulation models for uansporution studies. Therefore, it does not explain in detail how to construct a simulation model and is not sofrware-specific. Other reference documents provide general guidance for the applicacion of microsimulation tools, including the FHWA Traffic Analysis ToolbbX (Dowling, 2004). The forthcoming HCM (2010) also contains extensive guidance on the use of alternative tools for transportation analysis and highlights where simulation-based tools can supplement deterministic H CM methods. Sofrware-specific user manuals provide detailed descriptions ofind.ividual models, and sources in the literature offer operational comparisons of different simulation tools in various applications.

1.3 Types of Simulation Models There are three principal types of simulation models that vary in their level of computational detail and accordingly are suited for different types of applications. Microscopic models represent individual vehicle movements based on vehicle-lcvd behavioral algorithms such as car-foUowing, lane-changing, or gap-acceptance. These models are very detailed and generally offer flexibility in input and ourput options. Applications range from individual intersection and surface streets to integrated freeway and nerwork models with multiple modes of uansporcuion. As a result of the algorithmic detail, microscopic models are also very computationally complex.

Mesoscopic simulation also models individual vehicle movements, but predicts vehicle behavior in the next time-step based on macroscopic uaffic stream models (speed-flow-density relationship). Computations in a mesoscopic model are performed at a more aggregated level and are thus more efficient. These models lack the necessary detail for microscopic behavioral modeling (for example, gap acceptance), but have broad application for nerwork-levd analyses. Mc:Soscopic models are commonly used in relation to dynamic route choice algorithms that predict driver response co short-term or long-term sources of congestion. Macroscopic simulation models aggteg-4te vehicle flow (not ind.ividual vehicles) based on macroscopic traffic Bow principles. Macroscopic models predict traffic conditions for segments rather than individual vehicles. They offer 202 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

a less-detailed analysis, but are more computationally efficient than the former rwo cat~ories. Macroscopic simulation is ·. based on modeling the facility as a sceadysr:ace system and describing its behavior using a long-run average representation of an analysis period measured in minutes or hours. This approach is fundamentally different from microscopic and mesoscopic simulation, which modd individual vehicles and driving events on a second-bysecond (or even subsecond) basis. Exhibit Il- l shows a schematic representation of the three simulation analysis levels. It is also import~t co !iistinguish all three types ofsimulation from deterministic traffic analysis tools. Simulation is a stochastic process char randomly generates rraffic and driver behavior. Simulation models estimate traffic behavior based on distributions of many behavioral parameters. Since simulations are stochastic, multiple mode;! nms with different random number seeds will result in different solutio!?s that need Source: TSS-TC20Sp0rt Simulation Syscem.s S.L to be averaged or analyzed statistitally before reporting. Deterministic tools are based on empirically or theoretically-derived equations that give a unique and constant solution from a common set of input parameters. Examples of deterministic models are those in the Highway Capacity Manual (HCM) (TRB, 2000), and software packages that are based on HCM theory. The stochastic nacure ofsimulation and principles for reporting simulation cesulcs are discussed in more detail in Section 4 of this chapcer.

1.4 When to Use Simulation When co perform a simulation study depends on the analysis goals. Simulation cools ace generally more expensive, require more data, have a fairly steep learning curve and are more time-consuming to set-up and calibrate than de~e.r ministic models. A simulation study is therefore generally restricted co complex geometries and networks, and unique problem stau;mencs char go beyond deterministic (HCM-based) ~ethods. The 2010 HCM formally acknowledges the computationallimir:ations of its methods and provides guidance' regarding when the use of alternative simularioObased tools is appropriate. ln some cases, trarcsporr:ation agencies have adopted guidance for when a simulation srudy is required and how it is co be conducted. The reader is encouraged to explore specific guidance and requirements applicable to the area or agen.Of for which the analysis is being made. The additional cost and effort to perform"a simulation is generally supported if the analysis involves one or more of the following: • uniquegt~ configuratilmJ not addressed by deterministic procedures, such as innovative intersection and interchange designs, or closely spaced intersections with interacting queues; · • assessment of opemtional issues that go beyond the limitations of deterministic analytical tools, including oversaturarion, queue spillback and blocking of rum lanes; • multiple cransportatioo facility typs, such as a freeway intersecting with swface streets; • different intn"sl!ction configuraJions, such as a roundabout, in a signalized arterial corridor; • variow mode oftransportation such as cars, transit, pedestrians and bicycles;

Simulation Studies • 20:3

• analysis of driwr behavior changes in response w incidents or sources of reoccurring congestion; • need for visualizalion of a project and animation ourput for public involvement; and

•'ir.t'iTflliilffin~:J.o.-11.~..;.~.......,.., ,

NGSIM Algorjthm Development Process

desire for measures of effectiveness, (MOEs) not produced by deterministic procedures, such as che proportion of time a queue exceeds available storage. Once the need for a simulation srudy has been identilied, the user has additional Bcxibilicy in selecting from a hose of available rools. There are a variety of simulation pacbges. and the differences between packages are beyond che scope of this chapter. For many applications, there are multiple software options that can be applied equally well. In addition to private software dcvdopers, the U.S. FHWAsupporrs the simulation indwtry through the Nc:xt Generation Microsimulation (NGSIM) effort and the

Tra/fic A.nalysit Toolbox resowces. Through NGSIM, FHWA dcvdopcd data sets and core algoriduns that are open-source and intended to promote simulation practice while also encouraging computational consistency among privately devdopcd software applications. Exhibit 11-2 shows the conceptual NGSIM algorithm development process (FHWA, 2004).

~~ext

Generation S§ t~ulation

So wee: FHWA. NGSIM TllS!t E. I:
,\J:_t:.,·~;~ Crn~t.~,l:

1'/.ln:ttn··. I ),•..,t!:n j1r O['l"r.lt:ur<~/~ ·,1ll"lfllll ion

..........

•looWo
•s._.,.

• CDnYJ«/

Smlll-\

·~

··-

-

•SOV •HOY

•f.-.yWpt

.-.,.

• 11&0 •Tn4

•ln
• HOVI.oM

•·~ 8icydo

In-

• Mtriol

·~.,.,

.....

(2.3.:1<)

··-... .,

·~ew-

o A..oliouyl.oM Lerw

·~

• Moria! "'""'

"=r""' lone •Wed

•SpcE..,.. • APlS •ATIS • DKwcriic:

,.,_,.

•r-nctt..w •lluot.erw

•QX

•ToG ftu& •Uptii&Ol.lne

•• AVC!I> w.tt.r,......

•CVO •TOM

....

Dmnloo • r..Trip

..

.~

• !of'!'i<$iolll

..__, .,..,.,_

·~ a..,.

r=::

---

• Tool Q;>iool Coot • 611oo1{Coot/

• IW
•• ......,_,w.o.

•LOS

• Speod • TdftiTimt •VolwN • Tnt't'C1 Diltlnct

• AVO • •/cRAtio . • Domlty

• YMT/I'WT

• VHI'/l'liT • Dolay

•·~~ •so.,.

·~ • n RdlabiUty

·-

·~

._,c..t • t.lod.eSpllt

Tnl~Ws)

e-c~u..

·-....TNIOid

-

a..-..

•Do•

·~·

..

·~ ~ •· u ... s.ppc.t ·~=-:

.......,_

• Oelaalt vaN.

So)lt'CC: FHWA TftZjfo .ANJyris TIJ(JIJJIJX VtllutM I. Trttjfo ANtJysis Tools Primn, FHWA-HRT-04-038, page 17, figure 2.

1.5 Definitions The following terminology is used in this c:b2ptcr co describe demerits of a. simulation study. A simulated 1Jffem describes the extent of the analysis dolmin in space, level of compkxity a.nd temporal duntion. The nnworlc rcfecs to a collection of stm:ts, highw.ays a.nd inr.erscctions dw defines che physical c:haractcristics included in thc simulation srudy. The network carries simulated automobile, transit, &eight, bicycle and pedestrian traffic, depending on che scope of the analysis. A simulation ti!ol refets co a. piece ofsofiw:ue that is a.wilable (for pwchase) to perform simulation mulies. Each simulation tool has underlying~. which are sets of rules char describe a particular component of driver, pc:dcscrian, or bicyclist beha.vior {for example, car-foUowing, Ian~ or gap-acceptance). A mcJJ generally ref= to a computational approach for ~ traffic How, which includes dmnninistic {equation-based} procedures found in the HCM, as wdl as st«hastic {probabilistic) ~ods found in simulation. In the context of this chapter, a. 1710CklWill be used to describe a simulation srudy lile, including ill dements of the simulated network. traffic inpua, beha.viora.l rules and con.figuntion ofoutput. The concepts of calibration and valida.tion a.re central tO the quality of a simulation srudy . While chese are cliscussed in greater detail in Section 3.0, ba.sic explanations a.re given here. Calibration represents the underuking of specialized field studies to meuure specific traffic How parameters (or cliscributions) that are used as in~tJ in a simulation model to replica.te driver behavior. The pa.runeters needed for calibration, and the method of fidd measurement, are ofun specific to the model being calibrated a.nd can vary by geographic location, facility cype, or rime of day. Once measured by a.n agency a.nd provided to the model, the agency can use the model at vuious locations in its jurisdiction knowing the simulation model should replicate driver behavioc. IXfotdl pa:ram= tmd distributions are often a.dded to a model in case the user cannot obtain calibration data. Modd defaults should only be used as lase resorts,' or if prior analyses have established defaults are acceptable for the simulated purpose. Ana.lysts should not blindly rely on defaults, a.nd should scrutinize ea.ch modeling pa.ramecer, especia.lly if they are new users of a. particular software tool. Validation represents the comparison of estimates obtained from the simulation model (t>Utputs) with similar measures obtained from fidd data for a common location (for example, intersection, street, highwa.y, etc.). The lcvd of agreement obuined from this comparison SetVeS as evidence of how a.ccuntdy the model reflects the observed conditions. If the a.ccuracy is not acceptable, then calibration is needed. If the valida.tion a.ccuncy aftet calibration is not accept· able, the model is repla.ced a.nd an alternative ev;aluarion method is sought.

For more general deli.nicions of other tertns, the rea.der is referred to the Glossary {Chapter 2).

2.0 TYPES OF STUDIES Chapter 1 of this manual generally emphasizes the need to plan and prepare for any cype of srudy, and a simulation· based study is no exception. Simulation models are time-intensive to code and calibrate, and different software tools arc appropriate for differcm applications. It is critical that the analyst is aware of the goals of the simulation study and the limications associated wit:h the selected model. This section discusses different cypes and objectives of a simulation srudy. The different cypes of simulation studies arc presented to give the reader an overview of simulation uses. The discussion highlights specific (field) data needs that arc necessary to set up, calibrate and validate a model with different objectives in mind: Section 3.0 provides a more detailed overview of model coding and data collection procedures in a simulation context. While simulation has virtually an unlimited number of potencial applications, six general types ofsimulation srudies are emphasized here. They represent applications where the simulation model is used as a data collection tool, because future conditions don't o:isr, because field data collection would be time or cost inefficient, or because field experimentation would involve undue risk to rhe traveling public. I. Sensitivity Analyses

2. Evaluating Alccmatives 3. Predicting Behavior

4. Emergency Scenario Modeling 5. Safccy Anal~ 6. Environmental Srudies These six rypes ofsrudics are not mutually exclusive and arc presented as separate approaches for pwpose ofdiscussion and reader understanding. In practice, there can be significant overlap between these studies and, typically, a simulation srudy will address more than one of the objectives listed above. Each rypc of simulation study is presented in rwo parts: srudy overview and analysis steps. The studies are presented in order of increasing complexity. Therefore, analysis steps discussed in a prior type of study will nor be repeated and the reader is instead referred to the earlier section. It is important ro note the analysis steps presented for each of the srudies are considered guidelines. The liscs arc nor necessarily exclusive and ocher steps may be necessary depending on the specific application. The reader is encouraged to consider steps listed in all rypcs of srudies described here and consult with other sources (such as FHWA Tnzffic Analysis Toolbox) for further guidance.

2.1 Sensitivity Analyses 2.1.1 Sttuly Overview Simulation models are often used co answer the "what if' question regarding changing traffic demands and background traffic grOWth. What if rraffic doubles in a future year? What if 30 percent of drivers are diverted due to improved traveler informacion systems? Will this system still operate if traffic forecastS are within a 20 percent margin of error? Simulation ~dies are well-suited co handle these rypcs of questions. A sensitivity analysis can also be used to explore parameter serrings within the simulation cool itsel£ Exhibit ll-4 shows an example sensitivity analysis presented in FHWA Toolhox ~fum~ IV (Holm ct al., 2007). The exhibit shows the e.ffect of different car-following sensitivity settings on vehlcle throughput as a function of entering volume. Ir is evident from this analysis, for e:wnple, that the car-following setting m.arrers most at high entering volumes of 2,000 vehicles per hour per lane or more, as the modeled f.l.cility becomes more congested.


~~\-:p:_~

oU!l-t•

2400.-------------------------------------~

~ 2200r------------------------~~~--_J ~ 2000 r-----~----------~~~~::==0------1 ! 1800r--------------~~~~~======~-----1

1::

'S 1600

t

,------"')i;;!!!~~~::;;;:::~--=-=-===;

1400 +------~

:I

e

1200 12:'..... ..... 1000 1000 ..c:

Car Following Se 1'15itivity M.Jitiplier Values:

1

I

1500

2000

2500

Entering Volume (veh/hr/ln)

1

•

1oo

--<>- -125 ---.-150 -

.... - 175

-74E--2oOJ

Source: Holm et al. FHWA Traffic .Analysis Toolbr»c \.11/unu IV: Guitkli= for Applying CORSIMMicrosimulazion Modeling Software. FHWA-HOP-07-079, 2007. 2.1.2 AmJysis Steps · 1. Define obj«UYe: The objective should be clearly defined at the ou~et. It is hdpful to lay out whar me goals of the srudy are and how the study will be undertaken. The analyst should think about the types of output and performance n=.sures chat will be needed. Ar this stage the malyst also lays out data needs

and, if necessary, plans supplemental field data collection to use for model calibration and validation. Th~ objective should be defined in light of an understanding how MOEs are expected to vary as model inpuc parameters are modified. 2. Set 11p base model; Every simulation srudy should use a bau 11UJtU/ that reflects existing or baseline con eLitions. Later analysis steps then modify·the base model in some fashion. The base model is what the analyst: uses for calibration and validation. The base model generally includes network geometry. traffic volllllle.s (all modes), traffic composition, routing information, behavior rules including speeds, traffic conrrol stra..:-. egies and data collection demenrs (such as detectors) that~ provide the actual simulation results. Modc:::l serup is discussed in more derail in Section 3.1 and gui
The reasonableness check should be performed in light of known or expected capacity constraints a.nd include a visual inspeaion of the model prior to performing multiple iterations for analysis. Severely oversaturated simulation models sometimes do not record accurate performance measures since congestion may prevent a portion of vehicles from entering the network. While these dmied tntry whicks are typically recorded in an output file, they are generally not included in delay and queue length calculations. 6. Perform multiple runs: Simulation is a stochastic process and all results are subject to random variability in vehicle generation, driver behavior and other variables that are drawn from a distribution. It is therefore critical to perform multiple iterations or runs of a simulation model. Results arc then reponed as averages of all runs with a measure of the variability between runs. More discussion on the number of runs and sample size considerations for simulation are presented in Section 4.0. 7. Analyu resulu: At the conclusion of a simulation study, results are compiled and a.nalyzed. Depending on the specific software applications, some analysis may be performed in a post-processing operation using spreadsheet software. Sections 3.0 and 4.0 provide additional detail on the types of results that can be obtained from simulati.on and the means of reporting and displaying results in tables, figures and animations of the simulation model. When conducting a sensitivity analysis, it is sometimes useful to compute the change r11tio as the change in an output measure divided by the change in an input measure of interest. In econometrics, this ratio is also called elasticity. In calculus, it is a diffn-rntial.

2.2 Evaluating Alternatives 2.2.1 Study Overview A common application of simularion (or any tnffic analysis) is the comparison of transportation alternatives. For aample, an analyst may be interested in whether an intersection performs better under signaliz.ed control or as a modem roundabout, or may test di.fferent control strategies on freeways, including variable speed limits or ramp metering. Simulation models are commonly wed to test :md compare diffuen.r interchange configurations and to assess the impact of light rail or bus operations in a corridor. The common thread is that these are applications where there are no applicable decenninistic methods, or are increasingly complex: and thus aceed the limitations of determini.stic methods. To evaluate alternatives with simulation, the analyst needs to identifY a clear scope of the problem and set boundaries of the model. Input data include geomeuy (existing and proposed), traffic demands, speed distributions, trip patterns and inc:erseaion control in the model. Funher, the analyst generally needs to calibrate and validate the model to a known benchmark. This can either be existing operations on the system or a theoretical solution. The analyst then codes alternative ttansponation strategies by modifYing some components of the base model. Exhibit II-5 shows an example microsimularion comparison of alternatives fur the K Stteet bus way in Washington, DC, USA (Kittelson, 2004). When evaluating alternatives in simulation, it is critical the analyst selects a limited number of parameters to evaluate, leaving all other demc.nts in the model unchanged. This is necessary to ensure the effects of a particular parameter change on traffic operations can be isolated. Changes should be made to the same base model, to ensure nonsensitivity parameters are in fact kept constant in the study. Alternatives are evaluated based on some wee-defined MOE as discussc:d in Section 3. Due to the scochastic nature of simulation, each alternative is simulated multiple times and results are averaged for comparison. Seaion 4.0 provides more detail.

An imponant caveat in the evaluation of alternatives is related to traffic conttol strategies that require some form of optimization .(for example, signalized intersections, or ramp metering on freeways}. To ensure different alternatives are compared adcqua~ely, each should be simulated in an opti.mal manner. If the signalization is suboptimal, then the interpretation of the differences between alternatives is confounded by unknown differences in signalization quality, as discussed in NCHRP Repon 457: E1111/uating Intmtctii)1J ImpT1111t7M11t:s: .An Enginming Study Guide (Bonnesoo, 2001). Since many simulation tools do not include optimization routines, analysu often have to perform a separate optimization in a deterministic macroscopic model and then use these optimal parameters in the simulation-based coq~parison of alccmativcs. In a variation of this srudy, the analyst may perform a strrztegy optimizfuion. The objective is to identify the operational strategy that provides the best operation at the subject facility. The analyst applies traffic operational factors (such as speed limit, signal timing, ramp metering) and varies inputs in an iterative manner to find the optimal operating conditions. The analysis may include simple design changes (fur example, add a turn bay, drop a lane, lengthen a ramp).

Person Throughput K St Connecticut Ave to 18th St: WB '.'.=/,'~,·

,"{'c~

I'l _:z:, <·.'·

0

:.};~

i:{..

),

e

F

}~

,-

-;--'

Source: Kindson, W. FHWA Tmffo Analytis Toolhox ~lume V: Tmffo AtuJysis Tools Cue Stwlks: &Mjiu tm4Applications. FHWA-HOP-C~05. Figure 4.3, page 16, 2004.

Rather than comparing competing alternatives for a corridor, the objective is to optimize ics operations with a mix of strategies. Note that simulation tools typically do nor automatically optimize in the way some deterministic (signal timing) tools do. Optimization is achieved by iteratively applying and modifying parameters (that is, tweaking) until the objcctive function is optiQ:llzed. .. 2.2.2 AnAJ.ym Sups 1. Define objective: (For derails, sec step 1 of Section 2.1.2.) 2. Define alternatives: The analyst defines the speci.fic altcni.ativcs to be tested in the simulation srudy. Alternatives may be conceptualized, but typically arc not fully defined. For example, the analysis may compare a modern roundabout to a signalized intersection, but only early design plans arc available for the CWO options. The simulation study can be used to test both alternatives and make a recommendation about which one should proceed to firu.l design.

In this type of study, the analyst commonly has to make assumptions about the specific configurations fur the alternatives (for example, ·diameter of the roundabout; cycle length of the signal). It is critical these assumptions arc docwncnted. In some cases, a sensitivity analysis (as presented in Section 2.1) can be~ to test variations of certain parameters. In other examples, a design analysis or deterministic operational tool may be used to prcscrecn altcmativci prior to implementing them in simulation. 3. Define aoalpis scope: Once the alternatives arc defined, a study scope needs to be defined that matches the objectives for doing a simulation study and the hypothesized effects of any variation in sensitivity parameters. This is important, since ir has implications for the types of measures to be extracted, the physical extents of the model and the number of simulation runs required. 4. Detme tole.nlllce: A study evaluating alr~rnatives considers tradeotfs between different scenapos. including cost, del:a.y and other measures defined in the hypothesis. For a successful simi;Liation study, it is im~ru.nc

to define the tolerance that would be considered a notable impact. For example, a study may investigate freeway operations with and without HOY lanes with the hypothesis: The HOY configuration will reduce travel time for all travelers. The effect size of interest then would be defined as the reduction in travel time that would be considered a noteworthy impact. The specilied effect size has direct implications on sample size calculations co determine the required number of simulation runs, as discussed in S«tion 4.0. 5. Set up base model; (For derails, see step 2 of Section 2.1.2.) 6. Calibrate and validate base model: (For details, see step 3 of Section 2.1.2.) 7. Code alternative models: Once the calibrated base model is in place, the analyst codes the alternatives. While data may not be available to fully calibrate and validate all al~mative models, reasonableness checks will need to be performed on all scenarios by visual inspections of the results and animarions, or by comparing the performance to other sites and sources in the literarure. When coding alternative models, it is important to consistendy define any data collecrion dements. Ideally, the data collection dements used in the base model will .remain unchanged in the scenarios. However, in some cases differences in model geometry may result in slight differences in ddinitions. For aample, intersection approach delay may be measured from the stop bar at a signal, but from the yield line at a roundabout. Further, if either configuration is modeled with a free-How right-tum movement, neither a stop nor a yield line exists for that movement. It i.s of critical importance the user understands how a simulation model defines "performance measures~ and how that definicion may affect alternatives. The 2010 HCM offers a detailed discussion of simulation performance measures in comparison to deterministic methods. 8. Check reasonablenes.s: (For details, see step 5 ofSection 2.1.2.) 9. Perform multiple runs: (For details, see step 6 of Section 2.1.2.) 10. Analyze results: (For details, see step 7 ofSection 2.1.2.)

2.3 Predicting Behavior 2.3.1 Study Overview Another type of simulation study i.s to estimate driver behavior in response co an incident or a source of recurring congestion. An analyst may be interested in how drivers react (short term) to an incident on a freeway, or how they chan&e (long term) their travel habits as a result of a road closure or work zone. Simulation studies to predict driver behavior are also important for the evaluation of ITS, which may employ strategies such as variable ~essage signs, pretrip driver information and en-route guidance.

Driver behavior and resporue to these messages have a big effect on operacioru in areas where an alternate route is available. In some cases, the alternate route may be a coUed section of the same facility, such as a HOT lane. In these applications, simulation may be used to predict driver response to travder informacion and dynamic pricing schemes. Since the focus of chis type of srudy is on the broad-level nerwork impact, it is generally not necessary co conduce amicroscopic simulation study. While many microsimulation tools have dynamic traffic assignment fearures, mesoscopic simulation has more computational efficiency. In modeling driver behavior, it is important to distinguish between a one-time solution and steady-state optimization. In the first case, drivers react to an incident on the network based on predefined parameters. These include access to traveler informacion, responsiveness to the informacion and a measure of driver patience and value of time. The simulation estimates the behavior of all drivers based on stochastic distributions of these parameters. In a steady-state optimization, the simulation accounts for a long-term karning ejfict in individual drivers. lnitially, drivers still respond co predefined parameters. However, they chen further adjust their behavior based on their driving (delay) experience on that initial trip and all subsequent trips. Through repeated simulation, a steady-state solution is obtained that presumably optimizes the overall network performance. For example, chis cype of study was conducted to evaluate strategies for a downtown event in Minneapolis, MN, USA (Kwon, 2005). Exhibit 11-6 shows the mesoscopic network used to eval_uate evacuation from a sellout crowd at a downtown uena. 210 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITION

~·

~

..

.:-~

~~

II

~

Source: Kwon, E. and S. Pitt. Ev2luation of Emergency Evacuation Strategies for Downtown Event Traffic Using a Dynamic Nerwork Model: From Transportation Rmarr:h &corrfjournal ofthe Tramportation Rcearch BotzTd. No. 1922, Figure I, p. 150. Copyright, National Academy of Sciences, Wa.shington, DC, 2005. Reproduced with ~rmission of the Transportation Resea.rth Board.

--

2.3.2 kuzlysis Steps 1. Define objective: (For details, see step 1 of Section 2.1.2.)

2. Focmalate incidents and traffi<: management strategies: As part of the srudy scope, the analyst defines the specific features that drivers react co, ranging from a shore-term incident in one lane on a freeway to a. full-scale and long-term road-closure. Additionally, traffic management suaregies that affecr rou ting beh2.~ ior are implc;mented. These could take the form of variable message signs, or pretrip traveler information alerting driv~rs of congested routes. The analyst then defines the spatial and temporal extentS of the event and identifieS potential diversion routes. These determine the extents of the base model, so it is importan c to select them carefully. When evalu.atin.g the effects of a work zone, the analyst needs to selecr which construction stage of the work zone is of interest. Over the life of a work zone, conscruction activiciCli = Y' include multiple traffic configurations, among them narrow lanes, speed reductions, lane closures and crossovers. Modeling all potential configurations is time-<:onsuming and may not be necessary depending on the analysis question. The focus of the srudy is on the effect of bottlenecks and incidents on driver behavior. The objective is to let the simulation model predict changes in behavior using dynamic craffic assignment (DTA) algorithms and user-defined informacion sources. DTA refers to a class of route choice algorithms, in which the simulation tool dynmnica/Jy assigns traffic (individual drivers) to d.iJ!erem routCli in response to congestion on the simulated network. It is important to note the same congestion effects can also be analyz.ed using the approach discussed in Section 2.2, if the analyst is willing to make assumptions about how drivers may react. Simulation Studies • 21f'

3. Identify coding approach for behavior: The specifics of coding driver behavior differ by software tool, but different strategies arc typically available. Many simulation tools allow the analyst to define 1tatic :md dynamic routes and may have specialized features to modd Mls :md road closures. Static routes arc fixed path assignmentS between an origin :md a destination. Dynamic routes are flexible, and allow a simulated driver to select a route dynamically as a function of uaffic conditions. In a driver behavior analysis, traffic volumes arc typically entered through an origin-destination (0-D) matrix that leaves the actual path choice Acx.ible. The reader should refer to the specific software manual for details. Other sources in the literature give more discussion on driver behavior with respect to routing (Zh:mg, 2008).

4. Set up base model: (For details, sec seep 2 of Section 2.1.2.)

5. Calibrate and vali.UU base modd: (For details, see seep 3 of Section 2.1.2.)

6. Perform one-time solution: As discu.ssc:d above, a one-time solution modds driver ~havior in response to the •ru1es• specified by the user. These may include the percentage of drivers that respond to ITS treat· ments, the penetration rate of on-board navigation systems, or the willingness of drivers to detour from their preferred routes. A one-rime solution may still include multiple simulation runs (see step 9), but all using the same base input for behavioral parameters. In ocher words, a one-time solution is used to obtain average driver behavior in response to the same (shon-term) incident.

7. Perfonn .repeated iterations for steady-state solution: Depending on the application, the' congestion effect in question may cx.ist for an extended period of time. It is therefore expected drivers will alter their behavior over time. Starting with an initial set of behaviors (step 6) drivers will continually test different routes until they find one that optimizes their trips. The overall system (all drivers on all links) will eventually r~ a steady-sate solution where all drivers have settled on a new preferred rou~. AgW, this can be done multiple times to obtain average behavior. The imponant distinction from step 6 is that here drivers rest and learn new behavior over time. The steady-state solution may therefore differ from the initial behavior set identified by the user. The iteration process toward a steady-state system solution is computationally inrcnsc and not all software tools may be capable of performing this step. This step is only necessary to evaluate new and (semi) permanent sources of congestion such as a work zone or road and bridge clo$ures. 8. Check reasonableness: (For details, sec step 5 of Section 2.1.2.)

9. Perfonn multiple runs: (For details, see step 6 of Section 2.1 .2.) 10. Analyze .results: (For details, see step 7 ofSection 2.1.2.)

2.4 Emergency Scenario Modeling 2.4.1 Study~ Simulation srudies can be used to test and evaluate emergency J:llan2gCment strategies. Emergency scenario madding in this context mostly refers to evacuation from a natural d.isaster or terrorist arrack, but may also include general emer.gency management strategies from ocher unusual traffic events (including sponing eventS or parades). For aample, the principles of this section can be applied co post-event conditions at large venues, such as scadiwns co analyze •drain time: In the after~ath of natural disasters, such as hurricanes. :md in times of heightened national security awareness, evacuation strategies have undergone much scrutiny. Simulation can be used co estimate the time it takes co evacuate a particular region i.ri :m emergency or disaster event. This allows local and state governmentS to tnake decisions on when to order an evacuation. Simulation srudies can be used to test alccmativc traffic strategies during evacuation eventS including freeway l:mera:ersal plans and the use of ITS. The results of a simulation study can detcrmi.ne the optimum evacuation strategy and convey the viability of the plan to decision-makers and public scakeholder1. Emibit 11-7 shows the results of an aample emergency scenario modeling study for downtown Houston, TX. USA (Brown, 2009). The exhibit shows both forecasted speeds (shading) :and volumes (link width) resulting from a dyrwnic traffic :assignment modd for the Houston region.

Source: Brown. C., W. White, C. vanS~ and J. Benson. Dcvdopmcnt of a Strategic Hurricane Evacua.tion-Dyna.mic Traffic Assignment Modd for the Howton, Texa,s, Region: From Trruupttrtarion Res~arch Record: journal oftM Transporralion &search Boarrl, No. 2137, Figure 4, p. 52. Copyrighc;National Academy of Sciences, Washington, DC, 2009. Reproduced with peim_ission of the Transportation Research Board.

2.4.2 ANUysis Steps

· I. Define objective: (For details, see Step 1 ofSection 2.1.2.) 2. Define evuuarlon scope: The scope of an evacuation study is determined by the geography of the studied region, the sale of the eme~ey event and the c:xpected augnitudc of evacw,cion traffic. The geographic atent of the evaruation study depends largdy on the locarion and the availability of strategic corridors suitable for evacuation uses. Traffic volumes dwing evacuation scenarios may not readily be available. If the evacuation routes are on managed systems with sensor instrumenta-tion, then historical traffic volumes during foe- ... aicr evacuation events may be consulted, although direct C01Uparison is ~cnging as the (real or perceived) severity of an emergency event and associated evacuation behavior varies. Alt.emarivdy, the traffic demand for evaruation an be obtained in an atcmal analysis step using assumptions about population behavior i.t) ~n Jmuuu/mnes. Similarly, the temporal distribution of evacuation demand is forecasted using ~ ~ t:u111t1 char predia how many people leave the area at what time. Both the augnirude of evacuation tnffic and its distribution over time arc aitical simulation inputs. Exhibit 11-8 shows an enmple evacuation response curve used for an evacuation study in Florida, USA. More resources about hurricane behavioral studies can be foWld at the "Comprehensive Hurricane Data Preparedness Study Web Site~ of the Federal ~cy Man2ganent Agency at http://chps.sam.usace.army.m.il

~im11l<~tinn '\tou1 iP~

• 213

::.;;..:t.~&~'-~~~ ..~-· ~~~~J!~~~!#i-t'"~~~~· --

-· ~L~-- -"-~.

Behaviora l Cumulative Evacuation Curves

,::I -

5

I I I _.. -·1

80 .0

/

. ~

f

TO.O

,'

•

60.0

..

<40 0

I

0..

;

I

' /

J

lJ

,' ,

: ,

.

. /

30 .0

/ 20 .0

.·

••

,'' ,""

,............... ~

10.0

•

#

l

.,,._,. _,.

/

/

+·-/l/l/1 ,/

'

. /

at eo .o

'

/

'

R

~

I

,""'

/

" .,..--"'

"

~/

/

,

I

/

/

,/

1

/

/

,-"

-

,"'

.

LOng "-apon. . Med•um R oaponea

----·

------

~..,

-1

------- --· 0

I

Leg•nd

,,.,..... I_?~

0.0

1

/

I

2

I

,

:s

"'

11

e'

Rapid Reapon7

8

9

I

. t-toura __ Reta_d'Wi__ to_~~cuedon Ord•r at 0

Source: Apalache~ Bay &gicn H~trricaN Ew=atWn StuJr-T«hnical Da/4 Rrport. Homeland Seau-ity Digital Library. www. hsdl.org. Figure 4-1, 2007.

3. Identify evacuation scenari05: The analyst needs to identifY the evacuation scenarios to be modeled. For c:xample, many hurricane evacuation plans in the United States use freeway lane-reversal plans, where borh travel directions are used to evacuate. Research on simulation studies of evacuation behavior determined the critical bottlenecks of an evacuation corridor cypically lie at its starting and end points (Tagliaferri, 2006). The analyst needs to have a dear understanding of the geometric configuration and the type of uaffic conuol at the crossover, and may need to test different alternatives. 4. Set up base model: (For details, see step 2 of Section 2.1.2.) S. Calibrate and validate base model: (For details, see step 3 of Section 2.1.2.) 6. Identify approach for driver behavior: Depending on rhe cype of analysis, driver behavior may be an input in the simulation from external forecast methods, or it may be derived from within the simulation through merhods discussed in Section 2.3. The analyst also needs to consider that drivel'$ will likely behave differently in an evacuation and that driver behavioral algorithms may need to be adjusted. · 7. Code evacuation/emergency scena.ri05: With assumptions about the evacuation configuration and driver behavior in place, rhe analyst codes the actual evacuation model. Given the sensitive nature of this srudy and its implications regarding human life, the analyst should be very comfortable in all model assumptions and perform sensitivity analyses on critical parameters. Analysts should also consider an emergency evacUation may be associated with an increased likelihood for crashes that inuoduce unforeseen sources of delay. By inoorporating some incident likelihood in the simulation study, a more conservative estimate of evacuation time may be obtained. For example, analysts could design a sensitivity analysis (see Section 2. 1) of varying frequency and duration of incidents (broken-down vehicles) on the evacuation routes. Considerations for projected incident-dearing times should consider difficulties for first responders accessing the incident location on a congested network. 8. Check reasonableness: (For details, see step 5 ofSection 2.1.2.) 9. Perform multiple runs: (For details, see step 6 of Section 2.1.2.) I 0. Analyze results: (For details, see step 7 of Section 2.1.2.) 214 • MANUAL Of TRANSPORTAllON ENGINEERING S1\JDIES. 2ND EDITION

\2.5 Safety Analyses. ;2.5.1 Study Ovn-uiew. Conventional safecy studies described in Chapter 17 have limitations because collisions are rare events, and collision data arc sometimes unavailable or incomplete. Also, collision studies arc reactive (performed after the collisions), and Chapter 18 identifies alternative safecy studies which use surrogate measures of transportation safety that can be proactive, but also have limitations. Simulation can be used to perform simil.ar surrogate safecy studies by estimating measures correlated with transpof!ation safecy and collisions. Conceprually, simulation outputs can be analyzed wich respect to near collisions between mulriple vehicles, bicycles, or pedestrians.

One potential surrogate safecy measure is the "time-to-collision" (TTC), which is rhe (theoretical) rime until rWO vehicles (or other simulated entities) would have occupied the same space at the same time without evasive action by one or more of the entities. In simulation practice, the TIC and other surrogate safecy measures are calculated from simulation trajecUJry dAta. This is a detailed record of vehicle position and speed for each simulation time step, which can be post-processed to extract the safecy measures. One example of this post-processing approach is the Surrogate Safety Assessment Methodology (SSAM), which was developed by FHWA (FHWA 2008, Gettl)'lan 2008). Using.the detailed vehicle rra:jecrory data output, SSAM or other post-processing applications develop a vector representation of vehicle movements. A confliCt is registered if the veCtors of two vehicles (nearly) overlap, depending on user-definable safety thresholds. At the time of this writing, the process of surrogate safery analysis from simulation as well as the SSAM application are new and need further validation. However, the approach generally suggests simulation has the ahil.icy to be a proaCtive safecy ·assessment tool. It is presented here as an approach to simulation safety analysis under the assumption that the analyst has properly calibrated simulation inputs and safery performance thresholds. Exhibit 11-9 shows a screenshor from the SSAM user guide showing a conflict analysis at a four·legged signalized intersection. Before setting up a safecy analysis i~ simulation it is vital to recognize that .the validicy of the results is a direct funcrio·n of die quality of the input. While this is true for all simulation studies, it is especially important here since rhe resultS are very difficult to validate. By de.6nition, conJ\iets occur at intecsections and merge poinu. These locations are controlled by software-internal algoriduns, which in some cases are designed to prevent noncomp~ancc (and chus conRicu). For example, while real-world signalized intersections have an elevated risk of high-speed :mgle collisions

Source: FHWA. Surroga~ Sizftty Assmmmt Motkl (SSA.i\1). TechBriefNo. FHWA·HIIT-08-049, 2008. Simulation Studies • 21!$

due ro red-lighr running. a simulated signal typically results in perfect driver compliance. It is theorerially possible to also code some noncompliant drivers, bur then the safety performance of the intersection becomes a function of analysis assumptions and not necessarily of the true safety of the intersection. The safety outputs are limited by the underlying model assumptions, and behavior calibration and validation are essential to assure the validity of the rcsulcs.

2.5.2 Analysis Steps 1. Define objective: (For details, see step I ofSection 2.1.2.) 2. Set up base model: (For details, see step 2 of Section 2.1.2.) 3. Calibrate and validate baae model: (For details, see step 3 of Section 2.1.2.) 4. Identify data collection locatioo.s in model: Trajectory data are complex and therefore very timeconsuming to collect. Even with fast multicore processors, the running speed of larger networks tends to reduce drastically when the model has to record the posicion of each veh.ide at every simulation time step. More veh.ides (in la.rger models) therefore result in long analysis times, as well as la.rge output Iiles. Depending on the application, it can be use!Ul to divide the model into smaller pieces, especially if the safety analysis is only performed at a Limited number of locations. Alternatively, advanced users ofsimulation may be able to configure custom output that only gives veh.ide trajectories or event time stamps (Sch.roeder et al., 2005) at the locations of interest. In any case, it is helpful to identify conflict regions of interest early in the analysis process. S. Define safety measures and thresholds: Safety measures include the TTC discussed above, or other measures defined by SSAM (FHWA, 2008), including:

• rninimwn TIC: minimum post-encroachment; • initial deceleration rate; • maximum deceleration rate; • maximum speed; • maximum speed differential; • classification as lan~ge, rear~d. or path-crossing evenr type; and • vehicle velocity change h.ad the event proceeded to a crash {Source: http:// amc.gov/safety/ pubs/08049/index.htm). The above measures are used to specify user-definable dutsholds ror vehicle conB.ica &om tbc: simulation output. The partia.dar measures used vary by application and the sdeaion should be done according to SSAM documentation. The reader should carefully review lirerarure on d.ilkrcnt measures, thresholds and calibration and validation ofsimulation-based safety analyses, which were largdy unavailable at the time of publication. 6. Ron sample and post-process resulu: With a completed and calibrated base model and a conflict analysis Strategy, it is recommended a sample simulation with trajectory output is performed and the rC$ulCS are post-processed to obtain safety statistics. & indicated, a trajectory-output simulation run can be time.consuming and the analyst should thus make sure the results are reasonable.

7. Check'reasonableness: (For detaili, see step 5 of Section 2.1.2.) 8. Perform multiple nuu: (Foe details, sec step 6 ofSection 2.1.2.) 9. Post-process resulcs for safety measures: The safety post-processing program is applied to each of the simulation runs to obtain an C$cimate of conflicts for all iterations. ~ults are then averaged or otherwise analyzed for reporting.

10. Analyze resulu: (Foe details, see step 7 of Section 2. 1.2.)

?·6 Environmental Studies ~.6.1 Struly Overview Simulation moc:lds have application co environmenal sruclics. R=rch has linked vehicle emissions to cravd patterns and lW shown emissions an be described as a function of vdllcle sped and acceleration patterns for different vdlldc types (Frey, 2002). The input variables these emissions models need arc avaiWJle from microscopic simulation: vdllclc ~· speed, acceleration and displacem.cnt over time. Many simulation tools estimate emissions (C01 , NO and others) as a d.iJ:ecc output. The reader should arcfiilly review d!e software documentation to understand the empirical basis for the emissions model and its limitations. Often, dc.&ult emissions models may be based on limited or outdated vehicle fleet samples, or may have been developed based on fuel and emissions scanda.rds from other countries and arc dicrefore not univcrsalJy applicable. When trajeaory data are avaiWJle from simulation, other emissions models can be applied in a post-proc=ing seep wing spreadsheets. Given increasing focus on emissions nxluaions and the requ.irements lOr quantifying gnxnhousc gas emissions, simulation an be used to evaluate emissions benefits ofdifferent traffic engineuing strategies (lOr cample, modal shift, signal progression, roundabouts). Research has dcmonscratod (Frey. 2002) emissions arc not evenly distribuccd over time or space, but are conccnttaccd during peak periods and areas of rapid 3.f(deration. A:; traffic demand and operations change througho!lc the day. so will-emissions. Exhibit 11-10 shows an cample df an environmental srudy perfonnc4 in simulation.

.

i

r

, ,..,-' ;

~·

_,..---·~ ..

_.,.. _., ,.•

--~

Source: Sartin, B.; S. Mudigonda and I<. Ozbay. Impact of Electronic Toll Collection on Air Pollution Levels. Estimation·· Using Microscopic Simulation Model of Largc.Scak Transportation NetWOrk: From Tnznsporttaion &st4n:h Rt=J: j~~Urnai oftht Tnznsportatio11 &se4n:h &tml, No. 2011, Figure I, p. 70 and Ftgure 2, p. 74. Copyright, National .Academy of Sciences, Washington, DC, 2007. Reproduced with permission of the Transportation Research Board. ~impl~tif'\n <:tt~~·4iac

•

')1 "t

2. 6.2 Analysis Steps 1. Define objective: (For details, sec step 1 of Section 2.1.2.) 2. Set up base model: (for details, see step 2 of Section 2.1.2.) 3. Calibrate and validate base model: (For details, see seep 3 of Section 2.1.2.) 4. Identify environmental meastUeS of interest: Emissions models are available for a number of emission types, including carbon monoxide (CO), carbon dioxide (CO) and nitrogen oxides (NO). The analyse is encouraged co review liceracure on emissions modeling (U.S. ~PA, 2006 & 2009) co make a proper selection and should also consult Chapter 21 of this manual for guidance. 5. Define emission$ model: A variety of emissions modd.s is available and the analyse should make the selection carefully. Many simulation models have built-in emissions models mat are applied for an average vehicle rype. Depending on the level of detail needed for the analysis, d.iffecent en;Ussions models can be ·applied for multiple vehicle rypes, including passenger cars, SWs, trucks and buses. A!tcrnarively, emissions models can be applied in a post-processing effort using vehicle trajectory files. 6. Check reasonableness: (For derails, see step 5 of Section 2.1.2.) 7. Perform multiple runs: (For details, see step 6 of Section 2.1.2.) 8. Analyze raulu: (For details, sec step 7 of Section 2.1.2.) 9. Extrapolate lC$Ulu for system and over time: Reporting of cmissions.daca is different from conventional traffic operational data, since escimates are typically aggregated co the system level and c:nrapolated for an extended period of time.

3.0 DATA COLLECTION PROCEDURES This section describes the procedures for using simulation as a tool for generating (artificial) transportation data and for recording performance measure data for analysis purposes. The specific coding process varies between softwue applications and the discussion in this chapter therefore emphasizes general coding principles that are universal across multiple tools. The section discusses the general coding process and discusses performance measures from simulation studies. Special emphasis is given to the calibration and validation of simulation models, which are essential steps to ensure that the simulated traffic operations match field observations or theory.

3.1 Model Setup 3.1.1 Overttinu ofa SimuLztUm Study This section provides a basic explanation of how to do a simulation srudy. The general srudy procedure follows the different rypes of studies identified in Section 2.0. More general guidance on conducting a simulation srudy can be found in FHWA Traffic Analysis Too/b()X (Dowling, 2004).

In a simulation study the analyst must clearly define the problem scope and collect any necessary field data required for coding the model. In the absence of field data, lheoretical traffic Bow relationships or expert judgment may be used. It is 'of critical importance all assumptions and data sources are properly documented and reflected in the projecr report. Next, an initial base model is developed, representing a working simulation model that has yet to be comp~d to actual data. Following base model development, the essential calibration and validation step is performed, in which simulation algorithms and inputs are calibrated until simulation outputs are validated. Only if the results from this effort demonstrate an acceptable match between the simulation and field data or theory should the simulation srudy proceed. The calibration and validation effort should be repeated and model (input) parameters adjusted until a match is obtained.


With a fully calib rated and validated simulation model, the actual simulation study can proceed to meet any of the objectives discussed in Section 2.0. The model development process is generally consisten t for all types · of simulation models, including chose ac the microscopic, mesoscopic and macroscopic level. T he difference 'between these types of models lies only in the level of derail of the model and calibration/validation parameters.

3.1.2 Components ofA Model Traffic simulation tools can model transportation operations at a great level of derail. While all models make some basic assumptions about traffic characteristics and driver behavior, nearly all parameters can be customized through user input. Only through customizadon of model parameters (calibration) can a model b e configured to replicate traffic operations for a particular geometry, geographic region, or country. The tradeoff to extensive calibration is additional coding time and the poten tial need for more field data co llection to obtai n estimates of calibration parameters. Each simulation model srarrs with geometric input as the analyst defines links, nodes and connections based on existing or proposed future conditions at the analysis location. Geometry is typically based on aerial photography, engineering design drawings, or a combination of sources. Following coding of netWork geometry, it is necessary to define speed dutributWns on the variow links (speed lirrut signs), for different driver or vehicle ryp.es, as well as geometric speed constraints in curves or on grades. Exhibit 11-11 shows an assumed normal speed distribution with mean 45 kilometers/hour (km/h) and standard deviation of 1.5 km/h translated i nto a simulation speed distribution. An essential part of any simulation model is the traffic control strategy at intersections and merge points. Simulation can represent signalized and unsignalized operations. Sig11alized control includes pretimed and actuated signals, ;tnd can fuccher emulate coordinated and advanced traffic control strategies. Unsignalized control includes yield and stop-controlled intersections, as well as merge areas. Next, the analysis codes trajjic dnnand.r in the form of traffic input volumes on different links or an 0 -0 matrix of demand flows. The specific vehicle, pedestrian, or bicycle inputs are typically coded wing a composition of vehicle types; that can include different percentages of entities of varying Jengch, acceleration and d eceleration behavior, or vehicle occupancy (for HOY lanes). Compositions are similarly defined for any nonmotorized modes ·of transpomuion in che simulation, as well as special transit and freight modes. The traffic input demands are then allocated to links on the nerwork in the vehick routing step. Vehicle routes in simulation can be static (same sequence oflinks over time) or dynarruc, as drivers may switch to alceroace routes when facing congestion. Traffic demands, routing and speed distributions all typically vary by vehicle class. At this point the analyst has a working simulation model. Depending on the type of simulation tool, it may further be necessary co explicitly define tlal4 co/kction ekmmts that monitor and record traffic behav-

Q3 Q 27

azs+----------------1._----------------

1: r

Q 21 •

QU

0

I Q04 1400, ' :1 , I, ~

111

Qll

a • • • a

at l

QOS

Q2l.

~

•

•

•

•

•

Q()1

I, I, I ,1,1,1, 4! 1 ,ooo, E

~

s.,...._*

~

;lf~~ir9d,Sp~~ Oh\;iburlon

~

·

~

.. .

E

~

'"··, · ~

,., •

~· I 45 1Gm~ObtrbJion No.:E3J :?1

~

.•

·1:-,.-~....-· j .::··

'C§J·~. .-.;,.--- .

.

;

X~

;.,;

· :· .·.

-

•.

.· -· :. .~·.

';' i. :v;:·.,

~~}<_~·

.,

- :. ::. :~:

~~.: . ~·. :

.

..:..

:"'::

- ~.~~~i ...:

~

1

,., ,

I

;-.

,/'!I

.:lkldo :····· .••.: ..l .. .. .... ,;

')' .)~. 1.-t·-·~ .. :'·<-~. ·,:·~~;.• :··

. ..r-·~·;;K· ::J.-c .s:~""l \ '_ .• :·.

•• 4 '

,(

~....

'~.

'

'

••

Simulation Studies • 219

ior for specific poin!S, segments, nodes, or the network as a whole. These demen!S are necessary to obtain customized output from simulation. The definition of data collection demen!S is conceptually similar to designing a volwne, speed, travd time, or other field study with automatic data collection equipment. In the final step, the simulation model is executed to obtain resui!S. Typical simulation resul!S include visual observation of traffic that can be recorded as animation for distribution to clients and public stakeholders. The analyst .also obtains data from data collection elements in caw form or aggregated in a selected time interval. Depending on the type of model, aggregation and analysis of results in tables and charts may occur auromuically, or may require some manual post-processing in spreadsheet sofrware. Through this discussion, it is evident any simulation model has a lot of Ba.ibiliry for custom-user input, as network geomerry, volumes, routes, speeds and traffic control can be calibrated and compared to field data. In some cases, the analyst can choose to rely on model defaults for increased coding efficiency, bur great care should be taken when doing so. The analyst needs to fully understand the implications of different default parameters and should pay close acrentiori to the validation of model outputs when defaul!S are used. In general, simulation defaults are designed by the sofrware developer to provide a reasonable coding baseline. But since many tools are used internationally and for a variery ofspecial applications, default values ofrentimes will not satisfy local or regional traffic behavior. The discussion below gives more detail regarding when defaul!S are acceptable and when additional calibration is necessary. 3.1.3 Systnn BoutU!my A common challenge in designing a simulation study is to define the system boundary. The system boundary decision affects both the spatial extent of segments and nodes, and the level of detail included in the analysis. Both of these are important considerations; they directly affect the validity of the simulation results as well as the computational efficiency of the model. Larger and more detailed models increase the computatio11al requiremen!S. On the other hand, a system that is too small may not caprure all facets of the analysis.

Asimulation system should be large enough ro contain all queues within its boundaries ar all times. Ifat any time a queue spills beyond the spacial cxrents of the network. the analysis results are no longer coma, since only vehicles that enter the nerwork are included in delay calculations. All queues should also be ront3lned within the remporal.limits of the analysis. In other words, all queuing etfects and overcongested operarions should be rontained within the boundaries ofthe time-space analysis domain. The network also needs tO be large enough 10 assure traffic behaves realislically at points ofinterest. For example, ifoperations at a two-way stop-controlled inrerseaion are afkaed by vehide platoons from an upsa= signal, that signal needs to be included in the analysis (or the platoon effect represented through customized vehicle inputs). The level ofdetail needed for the simulation network is a function of the relative amount of rraffic contained on minor links. Chapter 4 recommended areawide counts can ignore minor links on the network. provided the total flow on all these minot links is less than 5 percent of the overall traffic demand. Following this same guideline, the modeled Simulation network should contaln enough detail tO include 95 percent of demand traffic. The remaining 5 percent of ttaffic from minor generatOrs is not ignored, but allocated to bigger links. This guidance is approximate. In face, it can be very challenging to get link-specific volwne data from simulation, since volume inputS are

Source: Tagliaferri. TFHRC Newsletter on NGSIM. Pages 49-50, figures 3-ll and 3-12; Tagliaferri, A. W. 1-40 l'..ll1u RnmJ Tl'lijftt MA/ytit. FHWA-NC-2005-14. bnps:/lapps. dot.state.nc.us/dot!directory/authenticared/UnitPage.aspx?ide8781. 2006.

::l::lO a MANilA! OF TRANSPORTATION El\l(;INFERING STUDIES 21110 EOIT101\l

typically 0 -D based. In general, the network should include enough detail to include all links that have a un ique and ' significant dfect on network performance. If a minor link has some sort of unique effect on the overall system it needs to be included. Examples of a unique effect include significant queuing from a small 4riveway or a minor link that ' cawes spillover at a signalized intersection. & a rule of thumb, all links that are cawes of ~urrlng congestio n need to be included. in the analysis. Exhibit I L-12 shows a simulation network as represented in two different analysis tools. 3.1.4 SimulAtUm Resolution All simulation cools .w e an underlying time step to up
I

i i

!

I

I I

!

.I

3.1.5 &eJing Period At the start of every simulation run there will be zero vehicles in the modeled network. Consequendy, performance measure estimates in the first few minutes of a simulation arc not going to be representative of the true effect of input demand vo.lumes, until the network has ~ad a chance co fully populate. Each simulation run requires a warmup. or seeding, period during which no performance measures are evaluated. The length of the seeding period should ensure the total number of vehicles ~- the network has reached a stable condi.tion where total entering and exiting flows are about equaL However, this is not alwa~ possible, either because the model is oversaturated and the total n umber of vehicles continues to grow, or because of user-specified varying demand lc:vds throughout the simulation period. In these cases, a rule of thumb is to set the seeding interval to twice the amount of time it takes to traverse the network (or the longest route through the network} under free-flow conditions (Dowling, 2004). For ex:unple, if it takes roughly 5 min. to uavd a modeled 8 km (5 miles) freeway at a speed of 100 km/h (62 mph), the seeding interval should be approxim2tdy 10 min.

3.2 Types of Measures Data collection in simulation can occur at different levels of aggregation. Simulation studies can be performed :at a singl.e-point location or can be aggregated co the segment, node, or even network level. The level of aggregation and analysis detail depends on the objective of the study and whether tl).e focus is on'~ isolated intersection treatment or a network-wide control strategy. While ~egacion to node and sYstem-wide measures can be hdpful in comparing system alternatives, these measures are inherently diflirult to compare to 6dd study results for model validation. The analyst should ensure calibration and validation at the point or segment level before extrapolating conclwions to the node and network levels.

In the following sections, common measures arc discussed for the poim, segroem, node and network levels. The variow input data, calibration parameters and output variables have predse definitions in each simulation tool and these definitions tend to yary from product to product. The HCM attempts to provide general guidance co simulation developers.to encourage consistency across tools. However, in the absence of national standards, the analyst must be aware of the precise definicion of each variable and parameter used in the tool chosen for the analysis. Coo.sequencly, any data that is input to the model mwt be consistent with these definitions. The interpretacion of output variables should be based on a complete understanding of what the measure is (and is not) quantifying. 3.2.1 Poial DllliZ Co/Jeaion Similar to 6dd studies dcsaibcd in Cbaptca 4 through 6 of this manual, simulation studies can be performed at a single-point location in the necwork. The ability to define wee-specific data collection points varies between simulation tools and ic is recognized chat some can only aggregate measures to the segment, node, or necwork levels as disawed below. Point data collection gathers data on traffic volumes, spot speed estimates, or vehicle delays ac_an approach ro an intersection, or at a midpoint location along a segment. Simulation models typically allow the analyst to code a Simulation Studies • 221

data collection point that will record these and other data items at the selected location. The user also specifies a level of aggregation that tells the program the bin size at which data should be stored. Point data from simulation can be especially useful for model validation, since .field data are usually also collected at a spot location. In this case, the ~rcgarion interval needs ro match the field data. For delay studies iris important chat the analyst uses the same definitions, since results will be different for stopped delay, queue delay, or approach control delay. For definitions of these terms, refer to Chapter 6 or the Chapter 2 Glossary. 3.2.2 Segment DatA Collection Some traffic performance measures require that data are collected over a segment of roadway. These measures include travel rimes, delays measured over a distance or route and queue lengths on a freeway or on approaches to an intersection. While the field studies co obtain these measures can be extensive, in simulation the analyst selects the appropriate measure from the menu of available performance measures. As with point clara collection, the analyst needs to carefully select the se.gmenc length and clara aggregation interval to match the study objective and the corresponding field study if applicable.

Simulation segment data can provide travel rimes along study corridors under different traffic, geometric, or traffic control conditions. The analyst can compare uavcl time results to field-measured estimates using GPS technology or other means as discussed in Chapter 9. For freeway simulation, the length of a queue caused by a bortleneck is a common segment measure (sec Chapter 10).

In a freeway application, simulation segment data collection can estimate me traffic SMS, flow and de.nsiry along a freeway segment. Ar. a shore analysis interval (for example, 5 min.), a simulation model can quickly generate many data points to plot a speed-flow-density relationship for the freeway segment. Since segment capacicy ls not an input in simulation, but a result of traffic operations in the model, a visual plot of the speed-flow-density is an important validation parameter. If the resulting relationship does not march field obserntions (or traffic How theory) the analyst can adjust driver behavior parameters in me car-following, lane-changing, or gap acceptance algoricltrns to calibrate the speed-Bow-densicy relationship. Exhibit 11-13 shows an example of a speed Bow curve for a freeway segment from field measurements. and as modeled in simulation. The twO •clouds" of data show comparable free-Bow speeds during uncongested flow periods (less cltan 8,000 vehlb in this case) and a reasonable represenracion of segment capa.cicy (at approximately 11,500 vehlh). Analysis methods beyond visual inspection are statistically rigorous and are discussed elsewhere (for example, Hollander ct al., 2008, or Zhang er al., 2008 I and ll).

Measurement

:..:.....,::~=....-::...:.J!9.9!:.!!SI~-~~..g.V;~..§J~I!~~L-=-.: ;:c::;:;:~:.

:~~~--~----~~~ I .

70 :<:' 60 Q.

.§.

50

..

"i 40

_:.

i- 30 20 10 0+---~--~----r---~--~----r---~

0

2000

4QOO

6000

8000

10000 12000 14000

Simulation - ! .. . . .

r:--:::-::::~ . ~-~.~.,_::-_=:-~

.... ---=...-...:.::..-:~{.......'1.-J.• - -

90--------------------~ 80 70

..-

J:' 60

~50

.....

1

40

• 30 20 10

l . ...... y= ~~~2000

4000

8000

now

[Ve'.'fhJ

Source; jim Dale, P.E., P1V America, Inc.


I.

6000

I

I

I

10000 12000 14000

3.2.3 Node-Aggregated Data

If a simulation scudy is performed to compare incersection alternatives (for example, a signal versus roundabout) ~ node-aggregated clara analysis measure is useful. A node-aggregate measure can give, for example, the total control ·delay for an intersection under different traffic control strategies, or in a volume sensitivity analysis. Node-aggregated data are more difficult to compare co field observations, since a field study for a node performallce is very involved and usually not cosc~ffeccive. It is therefore more common that the opera.cions of individual appr~ches ace evaluated in the model calibration and validation efforr before resulcs are aggregated to the node level. 3.2.4 Network- Wule Measures

When a simulation study is performed for a larger network it is sometimes necessary to gee an overall network-wide performance measure under different system configurations. For example, a simulation srudy may evaluate systemwide impactS of ITS strategies and changes in driver routing behavior, or the overall reduction of vehicle emissio~s for a corridor after installing a coordinated signal system. For the evaluation of these types of large-scale scenarioS. It is helpful to look at net:wOrk-wide aggregate measures. The rwo major problems with nerwork-widemeasures.are chat operational derails are los.t with extensive aggcegati~n, and that any model comparison with field scudy resulcs is virrually impossible. Therefore, while a study may rcqlure the analyst to report network-wide measures, it is always important to look at the 'operational performance of subelemencs in the system. When reporting network-wide resulcs it is therefore common to also report node-level resultS and highlight the worst intersection or spot location in the network.

3.3 Input Calibration . This section presents procedures fur calibl;'ating the inpucs of simularion models co match field daca or tesoUICCS on traffic flow theory. Calibration to field data is generally preferable, since it best repres~nts local driving cultUre. for proposed sires, calibration may be performed to an existing comparison site (preferably in the same region). Calibra-

tion to traffic Bow theory relationships (such as the HCM speed-Bow curves) should only be done if field escimll.tes are impossible to obtain, or would add unreasonable expense and time to the modeling dfon. Calibration ensures t:he sp~ed rules for driver behavior and other computational algorithms match user expectations based on field d:;~.ta or resources in the literature. A national guidance document on calibration procedures 'for simulation lists five core components (Zhang. 2008). I. project scoping and error checking 2.

~lobal

parameter calibration

3. local parameter calibration 4. deparrure/route choice parameter calibration 5. global calibration

In general, global calibration refers to parameters chat apply to the entire modeled system. For example, in some cools the composition of the vehicle Beet or driver behavior (car-following. ere.) are global parametetli. Local calibratiOn parameters are those that are manipulated at the individual intersection, segment, or approach level, and can inclucle speed distributions, signal timing, or gap accepcance pacameters. Calibration of model parameters is an imponant aspect of any simulation scudy, since reliance on default values is likely to ignore regional or international differences in driver behavior. The following paragraphs present the m:J. n simulation inputs that are cypicalJy calibrated in more detail: volumes, speed, gap acceptance, signal control anddriv-e=r behavior.

3.3.1 Traffic Demand Volumes Demand volumes are a basic calibration criterion. Demand volume calibration is a simple matter of matching simul~ cion inputs to.fidd-measured traffic counts. Traffic input tkmana is modeled in simulation based on traffic COWles~! traffic forecasts. However, traffic demand is somecimes affected by congestion and may vary from field-observed O"affi.. c: Simulation Studies • 22~

uolurms if the count location is downstream of a botrleneck. The input demands $hould therefore always be mea$uted upmeam of any congC$cion and similarly coded to enter the simulation network on uncongested links. The calibra· cion of input demand volume$ is fundamentally different &om the validation of $C!Vcd traffic volume$ internal to the network discussed in Section 3.4.1.

3.3.2. sp~~Js The distribution of vehicle speeds on a segment in the simulation modd is another means of calibrating the model. The methods in Chapter 5 can provide fidd C$timatC$ of the speed distribution. The analyst calibrate$ speeds by delining target speed distributions in the simulation. Global calibration changes vehicle input speed distribution, while local calibration changes speed behavior on specific segments in response to sharp turns or steep grades. Some simulation tools are limited in the extent to which they allow speed distributions to be user-defined, while others allow full flexibility as to the type of distribution used.

Speed calibration should only be used to reflect speed patterns caused by (speed limit) si~ng and geometry, never congC$tiOn. In other words, analysts Th(IU/.J only C4librau fm-fow speeds on the segment based on corresponding .field measurements. Speed calibration should never be used to force drivers to slow down in congested operation or to force botdenecks. Ifa real-world segment shows speed reduction due to high demand the simulation model should predict this congC$rlon {and the resulting drop of speed) as an output. If necessary, the analyst can calibrate driver behavior as described below to achieve more or less aggressive behavior (and thus higher or lower capacity). In this case, the field-measured congested speed becomes a validation parameter as discussed in Section 3.4.2. 3.3.3. SigruJiuti Control Modern simulation models offer a variety of options for coding signalized intersecQons, ranging from built-in controller logic, to so&ware-41-the·loop (SIL) and lwdwarc·in·thc·loop (HIL) emulation of aaual signal controll~. In SIL modeling, a separate program handles the signal operations. program runs •paraJid" to the simulation and the two programs exchange information during every time step. In particular, the simulation passes signal detector calls to the SIL program. The SIL then interprets the effect of these calls on signal operations and (if neccs· sary) passes a signal phase change back to the simulation. The updated signal phasing in rurn affects vehicle operations in the simulation, which again affects the detector call in an iterative process.

nus

In HlL, an actual traffic signal contrOller (rmt is, the hardware) replaces the SIL program. H[L gives the analyst confidence the simulated traffic signal truly represents me contrOller logic used in the fidd, but comes with an added expense. It further limits the simulation to run in real-time to match me physical controller.

Befora Calibration

80 70

eo -

!50

140

J:

· a.- CF Panometers • HCM FFS• 60 Curve

10 0 0

1 000

2000 3000 Flow Rllte [pc:JMnJ

4000

After Calibration

eo 70

eo ~eo .§.40

Signal conuol calibration should be done in accordance with the signal design and phasing plans. Additional validation of signal operations (green times, cycle lengths) may be necessary to build confidence the mod· ded conuol !ogic matches a.crual conditions.

\.!

l

~ k>?~" I 0

1000

'

• Modlfted CF • HCM FFS• 60mph

'

2000 3000 Flow R - [pc::IMnJ

'

4000

: Given the complexity of signalized control, many sowces of error may affect the operations including detector settings, field hardware delays, and coding errors.

l

'.3.3.4 Drivn- &havW-r The calibration of driver behavior can be complicated, but is likely to have the biggest impact on the performance of the simulation model of all calibration parameters. Driver behavior in simulation is handled through algorithms for car-foUowing, lane changing, gap acceptance and others. These algorithms often have multiple options for calibration parameters. Unfortunately, the actual algorithms are often proprietary so the analyst is left with a set of calibration par.lpleters thar are presented out of conrext. Software user manullls may provide insight in how these different parameters affect driver behaviors, and developers are usually available to assist the analyst with calibra.rion questions.

Because driver-behavior ~ritluns are the underlying rules of the simulation, they are an important calibration roo!. If an analyst observes unusulll queuing at a freeway merge area, for example, an adjustment in lane-changing behavior may be necesury. Similarly, if a freeway segment appears to carry too much traffic before reaching capacity, the carfoUowing algoridun may need to be adjusted to make drivers more conservative. However, the relationship becween the driver behavior parpnecers and calibration measw;es (for example, capacity) is usually indirect, so some trial and error is needed.

3.3. 4. I Car-Folklwing In their most basic form, car-following models predict a mponu (the acceleration or deceleration rate of the driver) as the function of some stimulus. In the original car-foUowing models, the stimulus was strictly the relative difference of two (foUowing) vehicles' speeds. The stimulus is further multiplied by a smritivity rerm that prescribes more or less aggressive foUowing behavior. This original class of stimulus-mponst car-foUowing models was later expanded to what are known as saftty-dirt4nct mod'els. Safety-distance models further refine the following relationship in .~at the sensitivity of the desire
3:3.4.2 laM Changing · . Lane-dwtging algorithms are related to car-foUowing algorithms, in that a vehicle with a high desired speed may wish to pass a slower-moving preceding vehicle in a voluntary lane change. Lane changes may also be rtquiml if a vehicle has co shift lane to, for example, turn left at an intersection or nit a freeway, in which cases lane changing depends also on a simulation route choice algorithm. In its most basic form, a lane-changing algorithm involves some driver reaction time, and safety paramerers that describe aUowable headway to the preceding vehicle, as weU as the acceptable size of gaps in c.raHic in the ~jacc:nt lane(s). A lane-<:ha.nge maneuver fu.rther includes some acceleration and deceleration behavior as cars speed up or slow down

0

~

i

::I: ~

0

~

Mean Reaction Time vphl • vehicles per hour per lane

1 mi/h. 1.61 bl/h

Source: Dowliog, R. S. FHWA Tniffic An4lysis Toolbox V.,/ume III: Guilkli= for Applying Trtrj}ic Mi=simuiAtion Motkling Softwan. FHWA-HRT~4-040. Figure 18, page 117, 2004.

to make the lane change, or to allow another vehicle to change into their lane (coopmztive behavior). The availability of these paramerers varies by simulation tool and the user guides should be checked for more detailed guidance. The goal of a lane-change algorithm calibration can, for example, focus on operations at a freeway merge area (or other bottleneck), and may~ be compared to field dara or theoretical uaflic flow relationships. Exhibit ll-15 shows a dual-objective calibration exercise to get a model ro predict a mean freeway bottleneck speed of35 mph (±2 mph) (56 krnlh ± 3.2 km/h), while processing 2,200 veh/hour/lane (:tlOO velticles) through the bottleneck (Dowling. 2005). The calibration parameters are mean reaction time and mean headway in a lane changing algorithm. The aample in Exhibit 11-15 illusuaces that, often, multiple calibration goals (in this case, speed and capacity) are competing. For aample, as driver reaction time increases (slower reactions), the observed velticle speed decreases, as does the observed capacity. The solution rtgion represents the realm where both calibration rargets are wit:hi.n an acceptable region. 3.3.4.3 Gapkceptance

·

Gap acceptance algorithms in simulation describe operations at stop- and yield-conuolled intersections, including modern roundabouts, freeway merge sections and the interaction between drivers and pedesuians. Gap acceptance theory in deterministic models uses the parameters csirical gap and follow-up time; Chapter 6 described methods to obtain these parameters &om field data. Gap acceptance behavior in simulation typically employs a minimum gap threshold. The distinction between critical and minimum gap is important, because it makes it difficult to directly calibrate simulation.gap acceptance behavior in simulation. Furthermore, the follow-up time parameter does not exist explicitly in microsimulation, since the likelihood of subsequent vehicles accepting the same gap is an implicit function of the car-following algorithm. A more aggressive car-following model results in the equivalent ofshorter followup time, but the calibration cannot rely on a field estimate of that parameter alone. For mesoscopic and macroscopic simulation the saturation headway can approximate the follow-up time for gap acceptance. Given the d.ifficulties of calibrating gap acceptance behavior &om field-measured da~a, this particular calibration is genctally closely linked to validation of other measures. For aample, a frequently used approach to calibrating the gap acceptance behavior ac the enuy leg of a modem roundabour is to start with an assumed initial ser of gap acceptance parameters. Using this initial ser, the analyst runs the model and validates the roundabout enuy capacity under a range ofvolumes. In an iterative ptocess, the analyse can·then adjusr the gap acceptance behavior to achieve higher or lower enuy capacities. 226 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Diffcrcnc simulation tools allow varying degrees of calibrating gap acceptance algorithms. Some allow full Aexibiliry to modify gap acceptance parameters for different locations and different driver types. Ochers constrain gap acceptance to a global set of parameters, rhat may limir che analyst to differentiating parameters by the type of maneuvec (such as '. right turn, left rurn, or through movement), and not a specific location.

3.4 Output Validation _ Validation typically focuses on model output. Validation parameters are performance measures, obtained from the simulation, that are compared to corresponding field estimates. Their main difference from calibration parameters is they are nor directly coded by the anaJysr, bur rather arc a product of the simulation based on analyst input. In ocher words, they arc nor explicitly defined, but rather implicitly obtained from the interaction of a host o f user-defined and built-in model algorithms. Conceptually, input calibration happens before the analyst hits the "run simulation" butron, and output validation takes place during (visual validation) and after the simulation. In the language of the HCM, validation metrics are usually referred to as service measures, but are also called MOEs. However, not all validation parameters-are MOEs. For example, validation could be performed for a model that predicts actuated phase duration at a signaliz.ed intersection, which could be validated using field-measured phase duration data. 3.4.1 Network-Intert~~tl Volumes

Volumes in simulation can be an input or an output. Section 3.3.1 discussed the calibration of demand volumes as simulation inputs. Volumes internal to the simulated network are a result of many simulation parameters and algorithms and should therefore be treated as outputs. They can only be observed and measured after the analyst elects to run the simulation. Network-internal volumes can be validated against field measurements of prevailing nerwork conditions, which may include congested Row periods (discharge from a signal, or Bows downstream of a freeway bottleneck). The validation of-network-internal volumes is also important in relation to driver behavior srudies (Section 2.3) that apply dynamic traffic assignment. 3.4.2 Operating Speeds

Similar ro volumes, speeds can be a calibration or validation parameter. Section 3.3.2 discussed the calibration of uncongested, free-flow speeds from field data. Opetating speeds are outputs of simulation after the analyst elects w run the.simulation and are a function of many parameters and algorithms. Operating speeds are an important validation pararneter for congested Bow conditions on freeways and arterial streets. The measurement and aggregation of speeds should mirror whatever field srudy was performed to estimate the benchmark speed daca. In this regard, speed validation can be performed using the TMS at a point location (see Chapter 5) or the SMS measured over a segment (see Chapter 9). Similarly, speed validation may be performed across all lanes on a segment, or may focus on a particular element, such as speeds through a channelized turn lane at a signalized intersection. Sample size calculations for spec: d comparisons and rests of significance are consistent with those discussed in Chapter 5. 3.4.3 Point Delays

A common validation parameter is a delay measurement at a point location. The analyst may want to validate the delay at an intersection or at a freeway ramp. It is important to note simulation models almost always define delays as the difference between desired and acrual travel time. In comparing delays to field measurements the analyst needs to be mindful whether the delay includes geometric speed constraints or not. For a.ample, a single vehicle tnvelin.g through a modern roundabout experiences geometric delay because the driver is forced ro slow down to navigate the circle. However, in simulation the geometric delay is typically included in the estiiiillte of desired or free-How uav-el time and the delay is ,therefore zero. This distinction determines whether the field data used for model val.idation is t):l.e rot:al delay (including geometric delay) or the control delay only. Details on performing an intersection or point dcl:;;J.Y srudy are given in Chapter 6. The issue of delay dcfin.jtion is also common when comparing simulation to outp~ cs from deterministic anaJysis models. Some simulation tools may not allow user-defined point data collection elemen a to estimate delays, but may be constrained to measures reported to the link or node level. 3.4.4 Segment Travel Tnnes

Travel time over a segment or through a facility is a good validation measure because it takes into account a 'IUieCY of factors inBuencing the model operations (for eumple, signals, traffic congestion and speeds). Travel rimes are relatively easy to obtain ftom the field using in-vehicle stopwatch measurements or GPS travel time logs as describe d in Chapter 9: GPS data have the added advantage of allowing the analyst to not only validate travel time,.but a.ls .co compare average running speed, delay times and accderarionldeceleration behavior between the model and realiry. Simulation Studies • 22?

3.4.5 Qulue LmgthJ Queue lengths arc important MOEs due are cxcncced from simulation. Many times, analysts turn to simulation models to overcome limitations in decermini.scic analysis procedures. For example, HCM signaliz.ed intersection procedures are limited in their ability co assess the operational impacts of turn-pocket queue overflows onto mainline operations, or of queues fi:om one signalized incerseccion spilling back into another signal upsueam. For this reason, simulation models arc a powerful alternative co conventional HCM analyses. However, care must be taken in the comparison of field-measured and simulation-estimated queue lengths, because the de6nition for the latter varies widely among simulation models. Differences between software packages include di.fferent thresholds for when a vehicle is considered to join or leave a queue, di.fferent spatial extent of the queue (current link only or including downstream linlc) and the calculation of total and average queue (longest queue on any link, or aggregation of all vehicles queued upstream of the congestion section on any linlc). Queuing measures used for model validation may include the di.suibution of average queues (50th percentile), maximum queues, or another percentile of the distribution. In practice, 85th percentile queues are fi-equendy used to describe queuing patterns that are more severe chan the average queue, occur less fi-equendy than the mean, but are still more frequent than the maximum queue. Simulation tools vary in their ability co produce these different queuing measures and the software user manuals should be reviewed for specific guidance.

3.5 Procedure Summary This section discusses data collection procedures for simul.ation studies, including modd set up, and the calibp.tion of

simulation inputs and v.didation ofsimulation outputs in relation to field data. Exhibit 11-16 relates the ~bration and validation parameters to common data sources for each measure. For field data, the cable gives the chapter references for thi.s manual. Other common data sources arc in the literature, including theoretical traffic Bow relationships and lessons learned from prior simulation analyses for similar problems statements and using the same analysis cool. Validation parameters ofsimulation analyses are also often compared to estimates obtained from other analysis tools, including well-established deterministic models. ·

Point Deb.ys

Field Studies per MTS Cbapccr 6 Licuacure Ocher Software

Segment Travel Tunes

Queue Lengths

2.2a •

·1Field Studies per MrS Chapters 6 and 10

MliNI Jb.l nF TRAN<\PnRTATif'\M 't\lr:1t..tt co'"'r.

<"T''"'',.,. "'''"' ~"''"~"'""u'

Ve!Ude Mean Speed 70

60

-~1

-pteeea-002 -pnx:ea-003 -proc:ooss-004 -proce$$-005 lo-008

50

40

sE

30 20

10

<

o~~TMTM~TM~~~~~~~~~~~~TM~~~~

ro'!<...~ro't~ro-Y~ro~~r;;it?~r;;1'~"''"'~ro~~.,..Pro~.,.~ ...r;;
~

Simulation T1me

Source: Dowling. R. S. FHWA Tr4JU Ana/ytis Toolhux ~fume lll: Guitklines for Appfyjng Traffic Mimlsimularion Modeling Software. FHWA-HRT-04-040. F~gUCC 14, page 106, 200·i.

4.0 DATA REDUCTION AND ANALYSIS 4.1 'Concepts of Stochastic Variability Many simulation parameters (such as speeds and hea.dways) are randomly distributed and the ~buces of an individual vehicle are determined based on a stochastic experiment. Each vehicle is randomly assigned a percentile_and will be assigned the parameter set that matches that percentile. For example, an 85th percentile vehicle may bave 85th percentile headway, 85th percentile speed and so forth. Some simulation programs are capable of implementing multiple random number seeds that allow the same vehicle to be associated with different percentiles for different parameter . sets. Because variables describing driver behavior are stochastic, there are no determiniscic (single number) results of a simulation modd and any two runs will result in different estimates of each MOE in question. With a set of input parameters, a deterministic (HCM-based) analysis cool will always result in the same answer. A simulation modd wi~ a sec of inputs will arrive at different answers depending on the rantlfm number sttd used to draw from the parameter distributions. Exhibit 11-17 shows an example of the variability in ~ulacion analyses in the form of 1-min. vehicle speed estimates for a freeway-to-freeway interchange for a coral ofsix simulation runs (called •process· in the figure). While the runs are comparable early in the simulation, the range of results by the end equals 25 percent of the mean system speed (Dowling. 2004). The range of speeds ac the end of simulation time in this example is close to 10 mph (16 km/h), and it depends on the particular application if this degree of uncertainty is acceptable. The concept of stochastic variability is important because it requires the analyst to perform multiple iterations or runs of a simulation modd to arrive at an average result. The main mocjvari.ons behind performing multiple runs are to get a sense of the vari:ibilicy of the estimate across tuns, and to test owlier runs wich unusual results. In che. same way real-world traffic varies from day-to-
4.2 Determining the Required Number of Simulation Runs One of the common questions asked in a simulation study is how many iterations or runs are needed. Analyses a.re frequently don.e.wich five or 10 simulation runs as a rule-of-thumb. However, the accu.a1 number of required runs is described by statistical theocy and depends on the situation. The calculation is conceprually similar to ;unple -size cal-· culations for determining significant differences in variables as discussed in other chapters of this manual. In general, the required sample size is a function of three faccors.

• The effeq size of inceresr: How much of a difference is considered noteworthy? Consider a study that compares different alternatives for traffic contwl on an arterial street. The intuitive measure of e.ffectiveness is the average travel time on the faciliry. The question of the effect size of interest then depends on what difference in travel time is considered a noteworthy difference. For example, a 5-mile (8-k.m) segment posted at 50 mph (80 km/h), has a theoretical free-flow travel time of 6 min. (360 sec), but due to signals and congestion the ex.isting average travel time is approximuely I0 min. For this example, a noteworthy effea size may be a travel time reduction of 20 percent to 8 min., or a 2-min. difference. • The variabilitY berw:cn runs: The stochastic nature of simulation means che performance measures of interest vary &om run to run with differenr random number seeds. The improved average travel time of 8 min. in the example above may therefore vary between 7 and 9 min. across different runs. The measure of variability is cbe standard deviation of the travel time estimate. In principle, if the travel time is .F.Urly constant across runs, a lower sample siz.e is needed. However, if random effects in the simulation (signals, parking) result in large differences across runs, more iterations are needed. • The significance !eye!: As with all statistical testS, the significance level is specified by the user depending on how much confidence he/she wanes to have in cbe estimate. For example, the analyst may want to be 95 percent sure that the travel time difference is at least 2 min. Given cbe above factors, it is impossible to give a general requirement for the number of simulation runs required. The effect size of interest will vary depending on the goal of cbe analysis. The variability between runs ean under some circumstances be estimated, but more often requires the analyst to perform some test runs. And finally, the significance level may also vary depending on the problem at band. The rdationsbip between the factors above is given mathematically by equation I 1-1 .

X > tXSp

fi

Equation 11-1

where

X -= effect size of interest, defined as the absolute estimated difference in means of a variable between cwo alternatives n ., the number of model runs per alternative tested t .. t-statistic for a confidence levd of [(1-alpha)/2] and (rn-2) degrees of freedom (froin Exhibit C-12 in Appendix C)

s

1

the pooled standard deviation of the variable csti.mace from multipk modd runs of the two sa:narios to be compared

"

The values oft for confidence levels are determined &om cables of the e.-distribution rbar are·included in most statistical tc:nbooks (for example, Washington et al., 2003) and repeated in Appendix C (Exhibit C12). For an estimate with 95 percent confidence, and a cwo-sidc:d significance: test, the column for alpha- 0.025 should be used. The standard deviation, s , is the pooled standard deviation of all runs for both scenarios in the comparison and is calculated from Equation'll-2, assuming an equal number of iterations in each of cwo scenarios x andy.

Sp

Jsi ;s~

=

Equation 11-2

where

s,

s

the pooled standard deviation of the variable estimate from multiple model runs of the two scenarios to be compared

.

s = cbe sW!dard deviation of the variable estimate from scenario x


. s1 = the standard deviation of the variable estimate from scenario y ·The required number of simulation runs can then be determined from Equation 11-3.

n>2x~ x'

Equation ll -3

where all variables ase as previously defined. Two cb.allenges with sample size calculations arc the pooled standard deviation is not known prior co performing chc simulacion study and the value of 1 in Equation 11-3 is a function of the sample size, n, itself. It is therefore necessary to perform some initial sample runs to estimate the pooled standard deviation, using Equation 1 1-2, and chen co iterativdy solve for n as t varies. For exasnple, assume an analyst wants co test the impacts of an increase in traffic volumes on the mean l'Unning speed on an arterial street. Th~ analyst is interested in showing at lease a 1.0 mph (1.6 kmlh) difference in speeds as a noceworthy impact. The analyst' further ~rformed five sample runs for each of the two scenarios and calculated apooi:d standard deviation of speeds of l .5 mph (2.4 km/h) using Equation 11-2. If the analyse wanes to show a difference 1n mean speeds of 1.0 mph (1.6 km/h) with 95 percent confidence at an estimated standard deviation of 1.5 mph (2.4 km/h), the initial esti.mace of the required number of simulation runs is: n

> 2 X (2.57l)'(t.S)' {LO)'

29.7 runs per scenario

TI1e value of t>-2.571 was obtained fi;()m Exhibit C-12 assuming a sample size of five runs per scenario. Since the escimace of n changed. to 30 runs per scenario (after rounding up), the new estimate oft is 2.042 for a 95 percent confideflce lcvd. Recomputing n with the revised tyic!ds :l.rt estimate ofl8.8 (or 19 runs), which in cum changes the estimate oft co 2.093. In the third iteration of Equation 11-3, the estimate of runs is now 19.7 (or 20 runs), at which time the solution converges (t no longer changes significandy enough to impact the result of n). Consequendy, the analyst needs to perform an addirionall5 runs for each scenario to add to the original five runs each. After perfonning all40 simulation rufl.S (20 per scenario) the analyst should recompute rhe pooled standard deviation to ensure th~t the estimate didn't change·

Ifthe analyst strictly foUows the sample size procedure outlined above, the consequence may be a large number of ruf'lS are required to fully satisfy the criteria. Multicore computers and batch run processing can reduce the time requir
4.3 Reporting Simulation Results Results of simulation models should always be reported in terms of averages and a measure of variability. As di.saJ.ss~d above, the outputs from simulasion analyses are stochastic, and multiyle simulation runs ase performed to gee an cst:j mace of the vasiahility:of the estimate. The analyst should report the standard deviation or the srandard error (mn!ard deviation divided by the square root of the sample size) so that the reader is fully aware of the confidence levels of tb.-= estjmace. As discussed above, the standard deviation is also used for sample size calculations to determine the nec=d number of replicacioos. Exhibit) 1-18 shows results of multiple simulation scenarios. For each (hypothetical) scenario, both the mean and standard errors of aavd time ace shown, calculated from multiple iterations or runs. This form of display allo"WS!he user co visually compare the performance of cwo scenarios, while acknowledging the variability in the estimate. The ahibit shows that both scenarios 3 and 4 result in a low cravel time along the reseed corridor. However, the error bar& show the variability across runs is less for scenario 4. So, while scenario 3 resulrs in the lowest overall mean, sceruri~ 4 may be prdi:rred since it also enhances travd time reliability. ·· Simulation Studies • 231

Travel Time Comparison &ror Bars at One Standard Errcr from 10 Iterations

•hi&•

! ! I ~

1

35 30

25 20 15 10 5 0

1

Scenario1

Scenario2

Scenario3

Scenario4

ScenarioS

4.4 Documentation Bcause of the inherent variability of stochastic simulation results and due to !,he ability to "tweak• simulation rc· suits duough calibration efforts, prop~f dcxumentation of$imularion analyses is csscncial. It is fair to say simulation outputs are highly sensitive to the sdected inputs and calibration parameters in behavioral algorithms, and almost any conceivable result can be obWr\ed by adjusting the parameters, even if the a.djustmcnts are not sensible! Proper documentation therefore ensures the integrity and validity .o f the submitted analysis. The ducc key components to proper modd documentation are: 1. a detailed listing of input variables (such as geomeety, volumes, signal timing) in the form of screen shots, input tables, or other format; 2. a record of behavioral algorithms and any modifications done for purpose of calibration (for example, car· following, lane changing, gap acceptance); and 3. a write-up of modd validation results, demonstrating the modded base scenario matches field observations or theoretical performance metrics. With these items in place, an cxpeticnced user and reviewer of simulation work can attest to the validity of the analysis. This requires that reviewers at the funding agency arc familiar with simulation practice and have experience with the particular modd in question. In the absence of a qualified reviewer, a peer-Teview process with a third-party con·

tractor can be used to evaluate the study.

4.5 Animation and Visualization One of the biggest selling points of simulation tools is the ability to generate 3· or 4-dimcnsional anim2tion files of the moddcd tnffic operations. In effect, simulation tools can be used to create video clips of what the operations on a facility will be like under different tested scenarios. This feature is hdpful to the analyst to identify bottlenecks and congestions points in the network, and to error-check the work. It can further be a valuable asset in public meetings and Stakeholder workshops for projects, as it clearly conveys whar can be complex engineering concepts to the layper· son. Through the usc of 3D dements (buildings, landmarks and trees) the animation can be eustomi:z.ed to a specific region and help the audience puc the simulation into a familiar context. However, while animation files arc a nice foturc, they should nor be wed exdwively, ~d ~e no substitute for a quantitative reporting of results. It can be difficult from a visual "snap shot" of operations to truly appceciatc the 232 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

:performance of traffic Aow and to compare different scenarios. Therefore, visual animation should only be used as a ~lupplementary form of reporting simulation output. i

If animation is used, it is common to record a predefined "Hy-duough" course duough the network to highlight key points. Some guiddines for generating such video dips arc: • Start with a network overview to allow the audience to become familiar with the scope of the model. • Slowly room lnro any sub~ections that are to be shown in more detail. A lay audience is less used to watching animation and may not be able to follow too-Ease transitions.

• Rest on any desired detail long enough to show operations. Consider whatever narrative may be desired by the presenter. • Assure an overall dip length that is :tppropriate to show the intended results, without being overly long. For public podium presentations, dips longer than approxim:ttdy 1 min. are discouraged as,too long to hold . the attention of the audience. The recommend:ttion for a dip length of approximately 1 min. is given in light of temporal constraints for many presentations. Clearly, longer dips ue feasible if time permits. Especially in a walk-in type public presencuion of a project, longer dips arc feasible md in lice desirable to give the audience more time to take in the information presented. Longer dips also reduce the :tbilicy to mask or hide problems that may occur at other times in a simulation. The general recommend:ttion is to tailor dip length to show what needs to be shown to fully communicate the craflic operational issues under consideration. Exhibit 11-19 shows screenshots from a sample simulation By-thrOugh for .an urban roundabout and screet redevelopment project. Sdected 3D bull~ were created to hdp the audience recognize fUniliar features. The arrows were added to trace a university bus duough the network.

S.OSUMMARY Thil chapter presented an overview of studies performed using cra.ffic simulation tools and was intended as guidance for users of simulation tools, as well ~ those charged with reviewing results of simulation studies. There is much additional detail on simulation studies, and espe. dally the underlying simulation algorithms, that go bey~nd the scope of this chapter and the reader is encou.raged to refer to other material in the licerarwe, such as FHWA Traf fie Analysis TotJibox (http://ops.fhwa. dot.gov/ trafficanalysistools/index. hem), for funher discussion.

(";_ ... .. . : .... C" . ..... : ....

. . . . . ...

6.0 REFERENCES Alsnih, R. "Review of Procedures Associated with Devising Emergency Evacuation Plans.• Tmnrportation Rmarch Record: journal ofthe Transportation RescardJ Buard 1865 (2004): 89-97.

Bartin, B., et 2!. •Estimation of the llhpacc of Electronic Toll Colleccion on Air Poilu cion Levels Using Microscopic Simulation Modd of a Large-Scale Transportation Nerwork." Tmnsportlltibn &March &cord: journal oftM Transportati4n &uarrh &ard 2011 (2007): 68-77. Bonneson, J. ct al. NCHRP Report 457: Enginming Study Guitk for Evaluating lmmecti4n lmproiJCTimtJ. Washington, DC: Transporcation Research Board, 200 I. Brown, C. ec a!. "Devdopmenc of a Stttcegic Hurricane Evacuation Dynamic: Traffic Assignment Model for the Houston Region." Transportation R.escacch Board 88th Annual Meeting. Washington. DC: Tcansportation Research Board, 2009. Courage, K. NCHRP Project 03-85: Guwna on the Uu ofAlumativt Traffic Ana(rsis Tooi.J in Highway Capacity Amtiym. Washington, DC: Transportation Research Board, 2008. Dowling, R. S. Tmffic Analysis Toolbox Volume Ill: Guitklines for Applying Traffic MicroJimulation Motk/ing Softwarr. Washington, DC: Federal Highway Adminisuation, 2004. Federal Highway Administration. NGS!M Task E.l: Corr Algorithm AJrwmmt. Publication No. #FHWA-HOP-06-0009. Washington, DC: U.S. Deparunc:nt ofTransporcation, FHWA. 2004. Federal Highway Administration. SurrogauSafttyAJsmmmt Motkl (SSAM). Tc:chBriefNo. FHWA-HRT-08-049. Washington , ' DC: FHWA, 2008. Frey, H. C., N. M. Rouphail, A. Unal and J. D. Colyar. Emimon Redudions Through &na Traffic Managanmr: An Empirical Evaluati4n Bastd Upon On-Road Mt4Sumnmts. FHWY/NC/2002-00 1. Raleigh, NC: Department of Civil Engineering, North Carolina Scare University for North Carolina Department ofTransporcation, 2002. Gerunan, D. P. Surrogau Sttfay Assmrot Motkl and Vaiid4sion: Final &port. Washington, DC: Federal Highway Administration, 2008. Hidas, P. and P. Wagner. "Review of Daca Collection Methods for Microscopic Traffic Simulation.• IOtb World Con~ on Transport R.esearch. 4'on, France: World Conference on Transport Research Sociery, 2004. Hollander, Y. and R.. Liu. "The Principles of Calibrating Traffic Microsimulation Models." TranJportlltion: Plannng, Policy, Rntarch, Pmdict 35, No.3. springerlink.com/contem/40387v6350097010/. Holm, Peter, Danid Tomich, Jaimie Sloboden and Cheryl Lowrance. Traffic AsuJysis Toolbox Volume IV: Guitklin6 for Applying CORSJM Microsimultrtion MotklingSoftwarr. Federal Highway Adm.ini.stration. Publication No. FHWA-HOP-07-079. Washington, DC, 2007. Jcannone, K. C. Trttffic Analyr# "[qq/box Volume JJ; Deciwm Suppi!Tt MahotkJ!JJgy for Sekcting Traffic Anlllysis Too/;. Washington, DC: Federal Highway Administration, 2004.

1Gm, T. E. "Operational and Safety Performance of a Nontraditional Intersection Design: The Supersueet. • Transportation Research Board 86th Annual Meeting. Washington, DC: Transportation Research Board, 2007. Kictdson, W. Traffic Analysis Toolbox Volume V: Traffic Analyris Tooi.J Cast Snuiks: &ntfitt andApplicarioru. Washington, DC: Federal H ighw:i.y Administration, 2004. Kwon, E. et al. "Evaluation of Emetgency Evacuation Strategies for Downtown Event Traffic U$lng a Dynamic Network ModeL • Transport41Wn Rnearch Record: founus/ ofth• Trtmrport4tion Rntarrh &ard 1922 (2005): 149-155. McCanhy,J. Traffic Ana/:pis Toolbox. Washington, DC: Federal Highway Administration, 2005. Olscrom, J. and A. Tapani. Omparison ofCar-Fol/Qwing Motkls. Publication No: VTI meddelande 960A. ISSN 0347-6049. Linkoepig. Sweden: Swedish National Road and Transport Research lnstiruce, 2004.


I.

Schroeder, B. and N. Roup hail. ''A Framework for Evaluating Pedesuian-Vehicle lmera,cdons ac Unsignalized Crossing Facilities in a Microscopic Modeling Environment." Proceedings ofthe 86th Annual Meeting oftbe Transportation Research Board. : Washington, DC: Transponacion Research Board, 2007. Schroeder, B., N. Rouphail and R. Hughes. "Towards Roundabout Accessibility- Exploring the Operational Im pact of Pedescrian Signalization Options at Modern Roundabours.• ASCE journal ofTransportation Engineering. 2008. Tagliaferri, A. W 1-40 Lane &venal Traffic Analysis. Report# FHWNNC/2005-14. Raleigh, NC: North Caroli na Deparcmem ofTransporcadon, 2006. Theodoulou, G. and B. Wolshon. "Aicernadve Mechods to Increase the Effectiveness of FreewayConuaflow Evac uation." Tramportation Rmarch Record· journal ofthe Tramportation Research Board 1865 (2004): 4&-65. Transpon:adon Research Board. Highway Capacity Manual, HCM2000. Washington, DC: TRB. 2000. Transportation Research Board. "Traffic Signal Systems and Regional Systems Management 2006." journal ofthe Tramportation Rerearch Board No 1978. Washington, DC: TRB, 2006. U.S. Environmental Protection Agency. Greenhouse Gas Emissions from the U.S. Transportation S«tor, 1990-2003 . Washington, DC: U.S. EPA Office ofTransportation and Air Quality, 2006. U.S. Environmental Protection Agency. Prifmedl&commmtkdMotkls. Washington, DC: Technology Transfer Nerwork Support Center for Regulacory Aunospheric Modeling. U.S. EPA Web site: www.epa.gov/scraroOO!Idispersion_prefrec.hcm. Washingron, S.P., M.G. Karlaftis and F.l.. Mannering. Statistical and &onometric Methods for Transportation Data Analyris. Boca Racoo, FL: Chapman & Hall, CRC Press LLC, 2003. Zhang, M., J. Ma and H. Dong. Droeloping Calibr41ion Tools for Microscopic Traffic Simulation Final Report Part 1: Overview Methods and Guidelines on Project Scoping and Data Co/kerion. Berkdey, CA: Partners for Advanced Transit and Highways (PATH) Working P<~.p«; O!Jifomi<~. Dep<~.nmem ofTramponation, 2008. Zhang, M. M. Droe/oping Calibration Tools for Microscopic Traffic Simulation Final Report Part II: Calibration Framework a Calibr41ion ofLocaUG/obal Driving Behavior and Departure/Route Choic~ Model Paramtm. Berkeley. CA: Farmers for Advanced Transit and Highways (PATH); University of California; California Department ofTransporcation, 2008. Zhang, M. M. Droeloping Calibration Tools for Microscopic Traffic Simulation Final Report Part Ill: Global Callibration-0/D/ Estimation, Traffic Signal Enhancement and a Cas~ Study. Berkeley, CA: Partners for Advanced Traosit and Highways (PATH); University of California; California Departmenr ofTransporcacion, 2008.

Simulation Studies • 235

Chapter 12

·......... ........ .... ' .... .................................................. ....... i

Pedestrian and Bicycle Studies Origirud by: H. Doug/its Robemt~n, Ph.D., P.E. FAiteJ by: D4nul]. FnuJJey, P.E. Bastilm]. Schroeder, Ph.D. 1.0 INTRODUCTION

237


238

2.1 Volume Studies

238

2.2 Pedestrian Walking Speed Studies

246

2.3 Gap Studies

246

3.0 PEDESTRIAN BEHAVIOR STUDIES

250

3.1 Introduction

2SO


251


253


257

4.0 REFERENCES

259


2S9


261

4.3 Other Resources

261

1.0 INTRODUCTION here are several types of pedestrian studies designed to capture some aspect of pedestrian and bicycle beha.vio.r or pe.rform.ancc. Engineers use the resulcs of these studies to determine if traffic signals are warranted, devdop exposure cb.ta for c:alcu.l.acing nonmotorized crash rates, locate and design sidewalks, crosswalks and trails, design and implement ~ty improvements, and analyz.e roadway cro6Sings to derermine appropriate controls and control operations for ex.ample, school crossing protection or signal timing. In recent years, measures describing userperceived quality of service of the pedestrian and bicycle modes have.received increasing attention. Methodologies are now available IX> preditc a wer-perceived pedestrian and bicycle LOS at crossings and along trails and sidewalks. Srudy methods apply to bicycles and pedestrians in many situations, particu.l.arly on shared-use paths. However, bicycles are typically required co operate in the roadway a,nd follow the same laws as vehicles; this affects the ability to conduct some of the studies included in this chapter.

T

The behavior or performance of the pedestrian and cyclist is generally evaluated by one or more of the following: • volume • walking or travd speed • gaps in traffic

• conflicts wirh vehicles • understanding of and compliance with traffic control devices • exhibited behaviors (running, stopping, retrcating,loolcing) • user perception • accessibility to pedestrians with physical or vision impairmentS In general, pedestrians and cyclists are differenc from vehicular craffic in that they are more vulnerable and require special consideration when evaluating their travel conditions and barriers to travel. The pedestrian and bicycle modes should further be considered as imponant for accessing other modes of transponation, including transit or vehicle traffic (to and from parlcing facilities). Virtually every trip starts and ends wirh some wallcing activity. This chapter focuses on the common methods for capturing these measures. Discussed are issues of srudy design, sampling, equipment, personnel, field procedures and applications. Examples of field-data collection and summary forms arc shown throughout the chapter. Appendix E provides additional sample forms suitable for copying.

2-0 TYPES OF STUDIES

2.1 Volume Studies Pedestrian and bicycle volumes are obtained by recording the number of pedestrians or bicycles passing "il point, entering an intersection, or using a particular f.tcility such as a crosswalk, sidewalk, or bikeway. Counts arc usually samples of actual volumes, although agencies may conduct continuous counts for certain situations or circumstances. Sampling periods usually range from 15 min. to several hours. The length of the =piing period is a function of the type ofcounc being taken and the eventual uses of the pedestrian or bicycle volume data. Agencies usually count these modes in good weather, unless the purpose of the study involves certain environm~tal conditions. The two basic methods of measuring pedestrian and bicycle volumes are manual observation and automatic recording. ln the past, nonmotorizcd travel was often undercounted because it is less frequent in many places, and generally more difficult to track using automated methods than vehicle travel. Traditional traffic clara collection equipment based on radar or magnetic inductance technology discussed in Chapter 4 is not applicable to nonmotorizcd modes. Today, a number of new technologies can be used to better measure the travel activity of people rather than vehicles, including pedometers, accelerometers, GPS transponders, location-tracking mobile telephones and laser counters suitable for measuring traffic on paths and trails. Therefore, while manual observations were historically the only option for counting oorunotorizcd travel, automated technology is becoming more readily available and cost-effective. 21.1 Mtmual 0bJenJ4tion 2.1. 1. 1 Purpou and AppliC4tion Most rypes of pedestrian and bicyclist counts arc taken manually through direct Ob$crvation. Several types of counts require classifications and are more easily and accurately obtained with trained observers. Examples include counts by age group, sex, physical handicap and type of behavior. Other studies focus on special behavior (signal compliance, jaywalking) that is hard to capture through automated technologies. Other reasons for conducting tlWlual counts arc time and resources. A number of srudies that use pedestrian and bicycle volumes often require less than 10 hours of data at any given location. Thus the effort and expense to set up and take down automated equipment, plus the time required to manually reduce the data, are usually not justified. Count expansion techniques, such as the one described larer in this chapter, offer a way to obtain reliable estimates from manual shott counts for less cost than continuous sampling. A common application for pedestrian studies is the Manual on Uniform Traffic Control D~im (MUTCD) (FHWA, 2003) pedestrian warrant as discussed in Chapter 7. 2. 1.1.2 Equipmmt The simplest means of conducting manual counts is to record each observed pedestrian or bicycle with a tick mark on a prepared field form. An example of a field sheet for a crosswalk pedestrian count is shown in Exhibit 12-1 and similar forms can be developed for other applications. The form allows for any desired classification. A watch or stopwatch is required to cue the observer to the desired count intervals. Observers tally their raw counts and summarize or key them into a computer upon return to the office. This method is low cost and is easily adaptable to different geometries 238 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

.., ~l~~

·''ofi:j{Q~.

-~ ·

~·· -------------------------------------------T~

~

w~---------------------

~--------------------

Da AdWU

0\i'*-1

tN

~

I

-

II

I I .

I I U

II

l

I

--

l

~

.. . . . I -"'""""'

I

U

-

IJ

and count types. However, its application is less common today with the availability of electronic count boards afld laptop computers. Battery-operated, handheld, electronic count boards are currendy the most common device to aid in c:he collcctiotl of traffic count da~a, which can include pedestrians and bicyclists. Electronic count boards arc compact, lightweigl:1~ handheld computers with different buttons allocated to different movements at ~intersection and the abiliry to claSsify groups separately (such as pedestrians, bicyclists, etc.). They~ much simpler in design and visual display thaft a laptop computer, and feature rugged casing and long battery life. Electronic count boards con!ain a.n internal dock that separates the da~a by a specified interval, so no field forms are needed. Many electronic count boards are capable of supporting crosswalk, classibcation, gap, gap acceptance a.nd pedestrian behavior studies. For agencies requiriog more than occasional manual pedestrian and traffic counts, the electronic count board or handheld computer is a cost-effective, labor-saving tool. A battery-efficient lapcop computer can be substinned for a handheld count board in many applications. Often, lapcop computers are ~dy available co the analyst, making this alternative more aruactive than the purchase of nev<-' hardware. Commonly available spreadsheet software can be used to record time stamps of different types of events uSing a macro routine. Appendix E contains an example and discussion of how to code a time-stamp maao in Miaosof1: Excel a.nd V!sual Basic. The benefit of using a macroenabled spreadsheet to collect volumes (or ocher temporal M:r:L -c: data) is ic can be customized co the specibc needs of the user. Also, ma.ny commercially available count boards are resoicted to output aggregated d:ua. Alternatively, this approach allows the analyst co obtain individual time-st2.Dlpec:l events. The downside is some software coding and post-processing analysis are required. Consequencly, this approacP may be more interesting in research settings and other special applications. Manual counts_ can be performed in real-time in the field or in a post-processing operation from video observationS: . in the office. The use of videos may be practical if video is already available, or if it is easy co ob!ain, For example, modem traffic ma.nagement centers often have live video feeds from permanent field cameras (at signalized intcaecPedestrian and Bicycle Studies • 139

lions or other locations) to a central office, where they can be recorded. Video-based counts can also minimiu staff requirements, if the same analyst can replay the video to count different movements. A well-chosen camera angle, ideally from some overhead vantage poinc, is critical to ensure usefulness of the video. Video image-processing sofrwue can in some cases automatically provide volume data even for the harder to detect pedesuian and bicycle movements discussed later in the chapter. Alternatively, observers can record their counts with a handheld count board, with tick marks on a tally sheet, or directly into a computer. The abilicy to conduct other studies with video, including the intersection and driveway studies discussed in Chapter 6, can lead to very efficient uses of the labor of field crews. A more detailed discussion of the advantages and disadvantages of various counting equipment, as well as exhibits featuring commonly usd devices, is in Chapter 4. 2.1.1.3 Penonn~l &quirtti Trained observers are required co perform accurate manual pedesuian counting. They muse be relieved periodically to avoid fatigue and degraded performance. Breaks of 10-15 min. should be scheduled at least every 2 hours.lf the data collection period is more than 8 hours, breaks of30 to 45 min. should be allowed every 4 hours.

The siu of the data collection team depends on the length of the counting period, the type of count being performed, the number of crosswalks or bike lanes being observed and the volume level of pedestrians and bicycles. One observer can easily handle a four-way signalized intersection with single approach lanes and low volumes as long as special classifications and/or directional counts are not required. As any or all of the foregoing parameters increase, the complexity of the counting task increases and additional observers will be needed. The c:xact number needed can be determined by conducting pilot studies at the locations of interest. Duties may be divided among observers in different ways. At a signalized intersection, one may record the nonh and west crossmlh while a second watchC$ the south and east crosswalks. In chat way, only one crosswalk is active for each observer at any given time. Another way co divide duties is for one observer to record certain classes of pedestrians, while the other counts coral pedestrian volumes. At complex sites, individual crosswalks or classifications may be as· signed to individual observers. Also at complex sites, one ob~rver may have the sole job of relieving the other observers on a rotating schedule basis. ·- !

2.1.1.4 ~arruion

An accurate and reliable pedestrian count begins in the office. A locally developed checklist is a valuable aid, even to experienced teams, to ensure all preparations for the field study have been completed before the team arrives at the site to be counted. Exhibit 1-1 is an overall checklist analyses can modify or add to for local conditions. Preparations should start with a review of the purpose and type of count to be performed, the count period and time intervals required, any information known about the site {for example, geometric layout, vehicle or pedestrian volume levels by time of day, signal timing. etc.). This information will help determine the type of equipment to be used, the field procedures to foUow and the number of observers required. If the purpose of the study requires ideal weather conditions, analysts must prepare criteria for canceling the count or procedures for dealing with inclement weather. The selection of equipment will dictate the type of data forms noeded, if any. Header information should be filled in co the extent possible in the office and the forms arranged in the order they will be used by each observer in·the field. The preparation checklist should include equipment items such as pencils, batteries, stopwatches and blank videotapes. Returning to the office to reuieve forgotten items may delay the Start of the srudy or cause it to be poscponed. An inadequate number of forms to complete the study could also invalidate !he study, resulting in wasted resources. An office review of the study procedures and a check of the proper operation of all equipment completes the prcpara· cion stage. Bicycle counts shoul
II ~!

Multi-Use Path Data Collection Sheet Dote

Weather_ _ _ __ _ _ _ _ _ _ _ _ _ __

StlrtTimt

EndTim•·---------------~

~ti·~----------------------------------------------- Mode u.. Grovp Oi..alon Ale Gonder Htlmtt l.oation

Wlleeled-Oevic:o

Pedestrian

lime

~ r

2

Li

.____

...

f

!~ 1 l

" 0

~

J

ili Li li l' 1 h1 ! I ~ I! I ~ li

-'--

~

£ ;) I ~

~

.., ~

j

I

' · ,__

Commems ____ _ _ _ __ __ __ _ _ _ _ _ __ _ _ _ __ _ _ ___

~ -----------------2.1.1.5 Obmwr L«ation Observers must be positioned where they can most dearly view the pedestrians or bicycles they are responsible for counting. Observers should be located well away &om the edge of the travd way, as a personal safety precaution, to avoid obsuucting pedestrian movements and to avoid discracti.ng dcivers. A position above the levd of the street and dear of obstructions usually affords the best vantage point. If several observers are coimcing at the same site, they must maintain visual concac;t with one another and be able to communicare so as to coordinate their activities. Protection from the elements is an important consideration for the observer. Proper clothing to suit prevailing weather conditions is paramount. Safety vests should be worn if the observer is near traffic at any time. Observers may count from inside vehicles as long as their view is unobsnucted. Outside, observers may use chairs to prevent fatigue and umbrellas for protection &om the sun, as long as these devices are not distracting to travelers. A sign indicating a traffic count is under way usually satisfies driver curiosity.

2.1.1.6 Da1a &cording . If manual field observation forms are used, it is critical for the study team to keep the data organized and correctly labeled for a. successful pedestrian or bicycle volume study. The counts may produce a large number of data- forms. · Each must be clearly labeled with information such as the count location, the observer's name, the time of srudy and

the conditions under which the counts were made. The form itself should clearly indicate the movements, classifications and time intervals. Location descriptors are critical for later data reduction, including a north arrow, sneer names and landmarks. The observer must concentrate his or her attenrion on accuratdy recording each counr in the proptr place on the form or with the proper button. Special care muse be raken with dearonic count boards or handheld computers to ensure chey arc properly oriented co the geographic and geometric layout of the intersection. lime intervals must be accurately maintained and coordinated when rwo or more observers arc working together. When mechanical count boards are used, the observers must have time ro record the accumulated counts and reset the counters at the end of each interval. Two procedures may be used to accomplish this: the short-break and alternating-count procedures and are described in detail in Traffo Enginemng by Rocss, Prassas and McShane (2004).

2.1.2Auto71Ultic Counts 2. 1.2.1 Purpor~ tJnd Application . There arc some applications of pcdesuian and bicycle volume dara that do nor require complex classi.ficarions, or on the other extreme are so complex they must be recorded for slow- or still-motion analysis. The simple counts may be needed for extmfkdpmods oftim~ (such as days, weeks, or even months) at busy intersections or in remote locations on paths and trails. The use of observers for such purposes would be cost prohibitive. When complex behavior classifications are required, the actions {for example, head movements) may be too quick for observers to see and record. Automatic video recording provides a means of gathering these pedestrian or bicycle data at a reasonable expenditure of time and resources. 2.1.2.2 Equipmmt . There are several types and models of automatic volume data collection equipment. This equipment generally 'includes rwo basic components: sensors to detect the presence of pedestrians or bicycles, and a data recorder. Sensors may employ active or passive infrared light transmission and detection, Piez.o 61m, time-lapse video, in-pavement loop detectocs and pneumatic tubes (Schneider, eta!., 2005). These technologies can reduce labor costs compared to manual counting methods. Classification of user cypes can be difficult with automated techniques, but can provide extended counting periods. N.ew technologies can further be used to record more detailed travel activicy of pedestrians and bicyclists, including pedometers, acceleromete.rs, GPS transponders, location-tracking mobile telephones and laser counters suitable for measuring traffic on paths and trails. These technologies allow the analyst to collect much more than just spot volume data, and other studies are discussed later in this chapter. 2. 1.2.3 Pmonnel &quired The only personnel required for making automatic counts are chose needed to install, calibrate and recov~ the equipment. Crew sizes of one to two are usually sufficient to deploy and recover most counting equipment. For complex video studies, personnel are needed to process video recordings in the office.

2. 1.2.4 PrqJaratWn fu mentioned previously, 6.eld work $hould never be undertaken without proper preparation in the office. A locally prepared checklist is an invaluable aid even for the most routine task. The purpose of the count will drive the type of equipment to be used and the deployment procedures. AU equipment should be checked to see it is functioning properly and appropriately positioned. An ample supply of accessory items (such as nails, clamps, tapes, adhesive, chains, locks and batteries) and aU necessary tools should be provided. Field personnel should be prepared to provide business cards of the principal investigator of the project and an authorization letter from the sponsoring agency with contact information of the appropriare official. Analysts also cannot forget weatherproofing for cameras. If night photos or video is desired, a camera applicable to nighttime settings should be used. Interference of the equipment with pedestrians and bicyclists should be minimized. 2.1.2.5 Selecting the Count Location The street or highway on which the count will be made and the generallocation-midblock or incerseccion-where the counters or cameras will be placed is decided in the office and is a function of the type of study being performed. The exact locations of the cameras, count recorders and sensors are usually determined in the field. In the case of cameras, the most important factor is the field of view. The location of the camera should take into account adverse weather and reduced visibilicy from shadows. For studies that investigate compliance at signalized intersections, a supple!Jlental camera may ~e needed to capture signal changes. Multiple cameras can be synduonized in the office 242 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

.using a video multiplexer device. Additional informacion on count location selection, installation and .rerrieval of 'counting devices and count periods can be found in Chapter 4. Chapter 7 contains informacion about pedestrian rolume warrants for traffic signals. .

2.1.2.6 Data &duction and Analysis Following collection, raw data must be placed in a form suirable for analysis. This reduction of the data usually consists of converting cally marks to numbers, summarizing !he data by calculating subtotals and totals and arran ging the data in a format for performing analyses. Eleccronic count boards or handheld computers preclude the data reduction steps and automatically produce the data in a summary format. The analysis may range from a simple extraction of dcscrip· tive information to a sophisticated Statistical treatment of the data, depending on the type of study bei ng conducted. 2.1.2.7 Sampk Counts and Count Expansions All counts are samples. Even permanent-count stations represent a sample of specific locations among many locatio!lS in a given area. Count periods are also samples of the overall long-term Bow. Tune and resources do n or permir the continuous counting of every pedestrian or bicycle facility. Consequently, sample counts are taken over shorter cirne periods at specific locations.These counts are then adjusted and/or expanded to produce estimates of the expected cr~ffic Bow at char or similar locations. Shore counts, known as coverage counts, may be expa.nded by use of a control statio~ (see Chapter 4). lf several sample counts are needed in a relatively small area, analysts select one location representative of the area locations to be sasrt· pled. It is important !he control station service the same type of facility and variations of uaflic being sampled at rhe other locations. The control station is counted continuously during the entire sampling period using the same collrlt interval (for example, 15 min.) as at the sampled locations. The counts taken at a sampling location are called control counts. Both the coverage and control counts are taken at midblock to avoid the complexity of turning movemeotS· Each link or segment to be sampled should be counted at least once during the sampling period. The counts may be made manually or with automatic counters. The mnrrol-counc data establishes the volume variation pattern for the entire sampling period. The pattern is quanti· lied by calculating, for the control-count data, rhe proportion of the total sampling period volume occurring during each count interval. Assuming this pattern applies co all of the sampled locations in the study area, the full sampli.llg perio,d volume for a coverage-count location is obtained by
2.1.2.8 ExpamWn ofShort Counts An expansiqn model technique developed for the Federal Highwar, Administration (FHWA) uses sho re coums i!'l conjunction with empirically derived models for a variety of rime petiods and counting intervals to expand the shor-t: counts into pedestrian and bicycle volume estimates for the time period of interest. Ranges for these estimates, based on the standard error about the mean, may also be calculated. The accuracy of the technique is sufficient for screeniCJ.~ purposes or to confircn volume levels based on qualitative evaluations. The technique is briefly summarized below. A user manual containing details and examples is available from FHWA (Mingo, 1988). The seven-step procedure is summarized below. The first step is to select the time period for which pedesuian oC bicycle volumes will be estimated. To estimate hourly volumes for us~ in signal warrant analysis, 1-hour time perioclS are selected. For uses involving average daily volumes, choose 2-, 3-, or 4-hour periods that represent variations ~ pedestrian How throughout the day. The procedure can also be applied ro bicycle counts. Seep 1.

Select the time period for which volumes will be estimated.

Seep 2.

Select the count interval.

Seep 3.

Develop the data collection plan.

Seep 4. Step 5.

_ .Collect the data. Select the expansion model coefficient and exponent. Pedestrian and Bicycle Studies • 243

Step 6.

Compute estimated volumes.

Step 7.

Determine estimated volume ranges.

The second step is to select the count interval. The choices are 5, 10, 15, or 30 min. of counting in the middle of each time period selected in Step 1. The uade-off in the choice of count interval is between economy and accuracy. The accuracy of the estimate incrases with the length of the count interval, whlch in turn increases the cost. If, for example, a likely outcome is being verified, a shorter count interval may be satisfactory. In Step 3, the analyst chooses the order of counting and the specific time periods for each location. In Step 4, data are collected. One observer is usually sufficient depending on the volume level. At an intersection, the observer simply conducts short counts of each crosswalk or bike lane in turn according to the schedule. Check the user manual for specifics on dealing with signalized locations. Step 5 is tci select the expansion modd coefficient and exponent. From Exhibit 12-3, choose the proper values, based on time period and c:ount interval, of a and b for the expansion model: volume - a X counf'

Equation 12-1

where:

vo/umt : estimate of pedestrian volume for the 1-, 2-, 3-, or 4-hour period of interest . Count

= number of pedestrians counted during the count interval a

a

= derived parameters (from Exhibit 12-3)

b

a

derived parameters (from Exhibit 12-3)

Seep 6 is co compute estimated volumes. For example, if 25 pedestrians were counted during a 10-min. interval in the middle of the time period from 7:00 and 9:00 a.m., the estimated volume for the 2-hour period would be: volume • 20.9 x 25o.au =296 pedestrians Since the numbers calculaced from the expansion models are estimates, the range of values within which the acruaJ volumes are likely co fall may be established with Seep 7, determine estimated volumeTanges. Exhibits 12-4 through 12..7 contain range factors (in percent) by pedestrian volume level and counc interval for each of the four dme periods, respectively. For the example above, the factor would be (from Exhibit 12-5) :t 32 percent. Thus the actual volume for the 2-hour time period would likely lie between 201 and 391 pedestrians. Thls range may seem large; however, there are many siruations where dais levd of accuracy is sufficient. Simply !mowing whether a crossing location has a low, moderate, or high pedestrian volume may be adequate to select the proper pedestrian conuol or accommodation.

t

·f. i

\;

Source: Min&Q er al. Mt4Niring Pt~n Yofwnes; A UHrl Uznu4 Yolui7U' IlL

FHWA-IP-88-0~0.

FHWA. Table 5, l988a.

Sowce: Mingo ct a!. Mus11ring P~rlenrUm W.lwna: A Usn-'s Manll41, W.illmr 1/J. FHW.A-IP-88-030. FHWA. Table I, 1988a.

Source: Mingo et al. M~llJuring Pdestrian W.luma: A User's Manll41, W.lumr 1/J. FHWA-IP-88-030. FHWA. Table 2. 1988a.

Sowcc: Mingo et a!. M~IIJIIring P~destriim W.illma: A Usn's Manll41, W.lsmu 1/J. FHW.A-IP-8~30. FHWA. Table 3, l 988a.

Source: Mingo et al. M~llSUring Ptdotrian W.illmn: A Usn-'s ManiUil. W.lwnt 1/J. FHWA-IP-88-030. FHWA. Table 4, 1988a.

Pedestrian and Bicycle Studies • 245

2.2 Pedestrian Walking Speed Studies Walking speed is a parameter used in a number of pedestrian srudies. Examples include gap acceptance, school crossing and signal riming srudies. Walking speeds are affccred by a number of facrors, including: • Volume of pedesrrians • Age of pedestrians • Sex of pedesrrians • Level of physical fitness of pedesrria.ns • Densiry of oncoming pedesrrians • Steepness of grade • Width of crossing • Width of path • Distance of oncoming vehicle • Speed of oncoming vehicle • Weather condition

Walking speeds typically range &om 2.8 ro 5.7 feer persecond (fps) for fully abled pedestrians and from 2.0 to 3.7 fps for disabled pedestrians (Dewar 2007). Recent rese:uch regarding pedesrri.an walking speeds has found that slower wa!J
'I

·Ii

Also, based on the same research, guidance is added that the rota! of the walk phase and pedestrian clearance time should be long enough to allow a pedestrian to walk &om the pedestrian detector ro the opposite edge of the tnvded way at a speed of3.0 fPs. This change will ensure that slower pedesuians can be accommodated at longer crosswalks ifthey starr crossing at the beginning of the walk phase. If this calculation finds that sufficient crossing time is not available, additional time should be added to the walk interval.

i

:

The srudy should be performed at the locacion ofinreresr under the conditions of interest. One or more o~servers may be used based on how much the conditions vary over time and the number of classes of data desired. The observers should be positioned where they have a clear field of view and do not distract passing pedestrians. Observers mark a measured distance along the path tnveled by the pedestrians and then simply time individual pedestrians through the speed trap. A sample of 100 observations is generally adequate. Analyz.e the data by first calculating each individual average walking speed by dividing the uap distance by the observed time, then classifying the observed speeds, and 6nally plotting the cumulative percentage of observations by class. This will produce a cumulative speed curve from which values of various speed percentiles may be derived (see Chapter 5). The 15th percentile speed is a generally accepted value to use in timing signals for pedestrians (KeU, 1991). Bicycle travel speed studies are performed similarly, by measuring the time required to travel a predetermined distance.

2.3 Gap Studies 2.3.1 Purpose andApplication Gap studies refer to the determination of the number of available gaps, in traffic passing a point, that are of adequate length to permit pedestrians to cross. In this coneat a gap is defined as the time that elapses from when the rear of a vehicle passes a point on a roadway until the from of the next arriving vehicle (from either direction) passes the same point. ~s are nor~y expressed in units of seconds.

246 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES. 2ND EDITION

,Gap studies consist of measuring the predominant pedestrian group size, d etermi ning the length o f a minimum ~dequare gap. measuring the gap sizes in the uaffic strea m and derermining the quantity of adequate gaps. The principal application of the study results is in analyzing roadway crossings by pedestrians to d etermine appropri· :1.«: traffic controls and safety improvements. Applications for bicycle traffic are rare, but may be adop ted from the foUowing discussion as necessary. The results of gap studies are used in traffic signal warrant analyses and school crossing studies. In addition to the techniques described bdow, the procedures for determining gap acceptance characreristics for drivers of vehicles entering or crossing roadways described in Chapter 6 may be ad apted tope· desrrian gap and bicycle applications.

2.3.2 Unsignalized Intersecticn Crossi11g According to HCM procedures, the foUowing analysis can be utilized to determine the single pedestrian critical gaP• group critical gap and dday. This methodology applies to a pedestrian crossing againsr free-flowing traffic or an approach that is not concroUed by a stop sign (TRB, 2000) .

2.3.2. 1 Dmrmining the Critical Gap The critical gap is defined as the time (in seconds) below which a pedesuian on average will 0ot attempt to begin crossing the sueet (TRB, 2000). Pedestrians are assumed to accept gaps when the gap is longer than the critical gap and reject gaps that are shorter than the critical gap. The critical gap for a single pedestrian can be calculated as:

L tC =s-+1I

Equation 12,2

p

where t, = critical gap for a single pedestrian (sec)

S, ." average pedesuian walking speed (ftfsec) L

=· crosswalk length (ft.)

t, • pedestrian sran-up time and end clearance time (sec} Commonly assumed values for some of these variables include:

sp ..

3.5 ft/sec

t, : 3sec

From the data in Exhibit 12-8:

t = 48ft + 3sec = 16.7 sec. ::: 17 sec. • 3.5fl /sec

Pedestrian and Bicycle Studies • Z4?

~

,I :F.ii1!1mnv~:

-

I

•

•

.. ..

~.

·r.~

:•

•

~

East Blvd at Dilworth July 23, 2009 48 feet

Location Date Crossing Distance

Gap Size (sec}

I

..... .

'·

.!~·.

Time Walking Speed Critical Gap

Tally Total 10 11 12 13 Discard Gaps less Than Critical Gap 14 15 16 17 1111 18 J.lii')IHII 19 Jilt' I 20 Jilt 1111 21 1111 22 lJIIt 23 24 II

8:15AM to 9:00AM 3.5 feet/sec 16.7 sec

I

4 12 6 9 4 5 0 2

0

25 26 27 28 29 30

. ...... .,..,. . . ' - ~~~~~~

lUI

4

I Ill

1 3 0 1 51

I Total Adequate Gaps

'

If p~atooning of pedestrians is observed in the fidd, the spacial disoibucion of pedestrians should be computed using Equation 12-3, to determine group critical gap (TRB, 2000}. Othetwi.sc, Equation 12-4 ca.n be used to estimate the platoon size for usc in Equation 12-3. Group critical gap is determined using Equation 12-5. If no platooning is observed, spatial distribution of pedestrians is assumed to be one. · NP=INT[

·0(::-l)]+l

8

Equation 12-3

where:

N,

• spatial distribution of pedestrians (p}

N.

" total number of pedestrians in the crossing platoon (p)

WE

=

effective crosswalk width (ft.}

8.0 = dcf.wlt dear effective width used by a single pedestrian to avoid interfetence when passing other pedestrians

J!c= - - --

Equation 12-4

'ol(here: \

N,

= sitt of a typical pedestrian crossing platoon (p)

'· vP

= pederuian Bow rate (p/s)

v

• vehicular How rate (veh/s)

r.

=

single pedestrian critical gap (s)

tc &t.+ 2(Np-l)

Equation 12-5..

where: tc

= group critical gap (s)

t,

• critical gap.for a sin'gle pedestrian (s)

N,

= spatial distribution of pedestrians (p)

2.3.2.2. Dan-mining tht Dtl4y.

,

The measure ofdfecriveness for a pedestrian crossing is delay. The average delay per pedestrian is a function ofcritical gap and the vehicular Bow rate as shown in Equation 12-6. Exhibit 12-9 relates the calculated average delay per pedestrian to an appropriate level ofservice (LOS) and the likdihood of risk-taking behavior following the HCM (fRB, 2000).

d, =.!.(e"• -vt0 -1) y

Equation 12-6

where:

a,

= average pedestrian delay (s)

v

~

t0

= group critical gap from Equation 12-5 (s).

vehicular flow rare (vehls)

Note: a • LikelihOod of acceptance of short gaps Source: Highway Gzpacity Manua/2000. Copyright, National Academy of Sciences, Washington, DC. Exhibit 18-13, page 18-15. Reproduced with permission of cbe Transportation Research Boa.rd.

2.3.3 Muuuring Gap SU:u The next pan of the field study is to measure the time l~ngths of the gaps in traffic. This may be done simply with a sropvn.tch and the field sheet shown in Exhibit 12-8. Only the gaps thar exceed the minimum adequate gap are of interest; therefore, it is not necessary to record every gap. The observer can develop a •feel• for gaps that are close ra or exceed the minimum adequate gap by observing the distance between and speed of vehicles while measuriiig the gap time. With some experience, the observer will be able ro eaprure the majority of adequate gaps.

Measured gaps arc rounded to the nearest second. A rick mark is placed in the rally column corresponding co rhe measured gap size char equals or exceeds rhe minimum adequate gap. The tally marks arc then totaled for each gap size. The sum of these corals is the number of gaps of sufficient length to accommodate rhesafe crossing of8) percent of the pedesrrian groups using the crossing at a day and time and under the conditions similar co chose of the srudy. The example shown in Exhibit 12-8 indicates chat a total of 51 adequate gaps were recorded. Gaps may also be measured using electronic count boards or laptop computers in place of the stopwatch and tally sheer. The observation procedure is essentially the same as described above. [merna! clocks in the computer record the times. Observers push the appropriate burrons co record gaps in the traffic. The primary advantage with this technique is chat computer software reduces the data, rhus saving time. This approach can also be adapted co record accepted and rejected gaps and from those data field-estimate the critical gap using methods described in Chapter 6. To evaluate the study results, analysts compare the number of gaps equal co or exceeding the critical gap co che number of minutes the gap measurement study is conducted. The appropriate criteria are chen applied co che result. The length of che srudy depends on the type of application for which che gap srudy results are being used. For example, in the MUTCD warrant 4 for traffic signals requires that in addition to the scared minimum pedestrian volumes, there shall be fewer chan 60 gaps per hour in the traffic stream of adequate length for pedestrians to cross during the same period when the pedestrian volume criterion is satisfied. Another MUTCD criterion (warrant 5) states a traffic signal may be warranted when the number of adequate gaps in the traffic stream when school children are crossing is less than the number of minutes in the same period (FHWA. 2003). If the analyse applied chis criterion co the data shown in Exhibit 12-8, the signal would not be warranted since the number of adequate gaps (51) exceeded the number of minutes in the srudy {45).

3.0 PEDESTRIAN BEHAVIOR STUDIES 3.1 Introduction In addition co the aforementioned general pedestrian and bicycle studies, two ocher studies are increasingly common for both research and practical applications: behavioral srudies and user perception srudies. Studies on pedestrian and bicycle behavior caprure characteristics of nonmororized road users that do not fall within the classic volume, speed and gap study categories. Behavioral srudies provide an undersranding of the needs of pedestrians and bicyclists, and identify the human factors' relationships that arc critical to mobility and safety. The studies may be grouped into three general categories: pedestrian/veh.icle confiiccs, understanding of and compliance witli traffic control devices (TCDs), and exhibited behavior srudies. They are frequendy prompted by research and devdopment of safety countermeasures and design considerations for pedestrian and bicycle accoffi!11odacions for road crossin~, for sid~al.ks and on-Street bike-lanes and for off-road path and trail facilities. [n all cases, driver behavior is a critical dement, since many confficcs and other behavioral patterns are a direct resul t of the interaction with motorized traffic. Some aspeccs of driver, pedestrian and bicyclist behavior in terms of compliance with TCDs are presented in Chapter 8. In addition to behavioral srudies, recent trends in the evaluacion of nonmocorized transportation modes, as well as auto and transit modes, have emphasized user-perception-based quality of service (QOS) measures. In a paradigm shift from traditional delay-based performance assessment, these new measures quantify the road users' experience based on factors of safety, convenience and comfort. These quality of service measures prompt the need for two additional types of studies chat address: 1. how the performance measures are derived from survey methods; and 2. how agencies can apply the developed models co their jurisdictions.

This chapter gives an overview of both and refers the reader to ocher sources for details on the devdopmcnt and application of very elaborate user-perception based models. NCHRP Report 616 (Dowiing ec a!., 2008) gives a comprehensive overview of literarure on user-perception measures and model development.


3.2 Types of Studies .3.21 Cotiflicts ,The scudy of conflicts is motivated by the desire to idenrif)r a surrogate measure for crashes that would allow pot~n tially hazardous situations ro be deale with before crashes occur. Chapter 18 of chis manual offers derailed discuSSIOn on co~Hiccs and other surrogate safety measures.

A pedesrrian/bicyde/vehicle con.Bict occurs when one of the interacting agents has to cake some action, such ~ a change in direction, speed, or bodl, in order co avoid a collision. In the definition of a conRict it is key char a colliSIOn was imminent in the absence of such action. Assertive pedestrians and bicyclists may sometimes exhibit behavior char may appear "risky~ co an observer, but that is still wit!Un their conuol. Enmples.include delibcratdy "forced• yi~d events (in which case the pedestrian could have stepped back if the driver had not reacted) or imcntional accepting of a shore gap in traffic by rurtning. The most commonly applied conflict scudy is on pedestrian/vehicle conflicts. Bicycle/vehicle conRicr st udies are rare. Researchers have mer with difficulty in establishing a causal rdacionship between pedestrian/vehicle conHicts and acrual crashes. While some .evidence of such a relationship has been uncovered, the complexity and rdativdy rare occurrence of pedestrian-vehicle and bicycle-vehicle crashes at a given location has to date prevented a clear conclusion. Despite the difficulty in predicting crashes, pedestrian/vehicle conBicrs remain a useful measure of relative differences among pedestrian safety alternatives. A number of studies have used conflicts as a measure of effectiveness for idencifri ng pedestrian safety problems, evaluating TCDs, and comparing pedestrian accommodation designs {Campbell er a!., Zegeer et al., 2005). With devdopment of the Transportation Research Board (TRB) Highway Safety Manual, much forus !w been on developing pedestrian safety prediction modds (TRB, 2008), the details of which are beyond the scope of this chapter. Exhibit 12-1 0 shows common pedestrian/vehicle crash and conflict types (Harkey and Zegeer, 2004):

zoo4.

Source: Harkey and Zegeer. PEDSAFE: Ptdmritm Safoy Guitk and Cormln7MilS1m! Stkmon Syrrnn. FHWA-SA.C4-003. Chapter 3, 2004.

Pedestrians crossing a roadway may encoWiter through vehicles, right-ruming vehicles and left-turning vehicles. Th~ bas.i.::::; conflicrs ocrur when the projected paths of the pedestrian and the vehicle intetsect and either the pedestrian or the \lhi:l- ~ muse change ~on and/oc speed co avoid a collision. Variations and rdinernents to these basic conBicrs have been~ -. In addition, the severity of the conflict, as determined by the strength of the decderation or accderation, th~.speed diHe~ rW and how closely spaced the involved parties are, has been used e£fecrivdy in further de.6ning conflia: MOEs. Pedestrian and Bicycle Studies • 2S~

A particularly common and risky conBict is known as the multipk threat confl.ict, where a yielding vehicle in the nar lane blocks the pedestrians' view of a vehicle in the far lane and vice versa. The situations are dangerous, because the pedestrian alrady made the decision to cross and may not be paying attention anymore. The situation is especially risky at a multilane two-directional crossing, where the pedestrian may already be screening traffic in the opposite direction. This case is very common at crosswallcs at multilane roundabouts.

3.2.2 CtmrpliAnce mul V'wl4tiom Pedestrians, bicycles and vehicles may not comply with TCDs and the rules of the road. Vehicles may run a red signal or srop sign. Pedesttians may anticipate the walk or green signal or start to cross during a clearance or prohibited signal indication. Pedestrians also commonly cross informally outside the crosswalk and thereby violate jaywalking laws. Bicycliscs are considered vehicles in most States and many countties and therefore have to comply with the applicable motor vehicle codes. In practice, bicyclists are often observed skipping queues at a red signal, running a red light, or using the sidewalk insread of the street. In short, compliance and, in turn, violations to TCDs are important measures to describe behavior of nonmotorized road users. One simple way to determine pedestrians' and bicyclists' understanding ofTCDs is to ask them. Engineers have often used surveys to identify pedestrian problems and evaluate alternative control devices. Appendix B provides guidelines for survey design. Another way to measure pedestrian understanding is to observe pedestrian and bicycle compliance with TCDs. While some may understand a device and choose to ignore it, compliance is often an indicator of the degree of understanding by the pedestrian or bicyclist, particularly when coupled with other measures such as surveys. Compliance is usually measured by observing and recording violations, such as violaring a red or don't-walk signal indication, crossing illegally, or riding the bike on a pcdestrian-<>nly sidewalk. Research studies have used compliance to good i:lfect as. a mea.Sutt of effectiveness (Schroeder eta!., 2009, Harkey and Zegeec, 2004). In the conteXt of compliance, it is important to explore driver compliance and behavior in relation to rionmotorizcd road users. Common examples are drivm yielding compl~ce at marked crosswal.ks and behavior of vehicles passing bicyclists on a narrow roadway. Compliance is often tied to before-and-after evaluation of countermeasures or treatments. Several national resources (Harltey and Zcgecr, 2004, Fitzpatrick et al., 2005, Hunter et al., 2006) summarize research on treatment effects and various behavioral artributes at signalized and unsignalized crosswalks, as well as bicycle facilities.

3.2.3 Tr~~Wl P11th Studies Another type of study is a pedestrian walk path or bicycle aavd path study. These studies trace movements by pedestriaiu or bicyclists on a map/figure in a downtown area, along a commuter route, across a Street, through a terminal, through a park network, or through a public plaza area. Data collection can be perfonned manually by literally tracing the travel path on a map or on transparencies, which may be done during a public meeting or stakeholder workshop. With advances in modern GPS technologies or decttonic ttansponders, travel paths may further be recorded directly using a sample of volunteer pedestrians and bicyclists. The overlapping pedestrian or bicycle path tracings then highlight the pcefened travel paths and can assist the agency to target improvements for nonmotoriz.ed transportation users. These studies can be helpful to explore concerns of network (sidewalk, path, trail, bikeway, etc.) connectivity. 3.2.4 Network Ctnmmivity Strulies A special type of study for the pedestrian and bicycle modes deals with issues of network connectivity of sidewalks, bicycle lane, paths, trails and crossing poincs (runnels or bridges) of major obstacles for these modes. In times of increasing recreational and commuting trips by bike or foot, these become critical considerations for many agencies. This group ofstudies involves concerns for safety, which is di.scussed in detail in Chapter 18 under the heading of (pedestrian) road safoy 11uJjts. Connectivity studies also employ principles of ustr-pn-criwd quality ofst'1'1Me as is discussed in section 3.2.6. They may also use the w11/Jtability or biJttability chtdrlists presented in section 3.3.5. Network connectivity studies can further involve comprehmsiw trrlvel surveys that arc specifically targeted toward n_onrnotorized users of the transportation netwOrk. Often, traditional aavd surveys bias against these more vulnerable road users by overlooking or undercou.nting short trips, noncommute trips, nonmotoriz.ed links of mulrimodal trips and ttavel by children or other travelers who do not have ready access to motorized ttanSportation. A related concept is the principle of univm~~l design, which can be defined as •facility designs that accommodate the widest range of potential users, including people with mobility and visual impairmentS (disabilities) and other special needs.~ (VrPI, 2009). Methods are available to quantify walking and bicycling accessibility {for c:nmple, Litman, 2008, or lacon.o et . . ..

•

t. ' . . .

,,I A'

nr TO AM('OnOTATif"'\... 1 C:Mf.1MC:C01Mf; <;:TI tnu:<: ?Nn FniTinN

J \

i

I

I l

: al., 2008), but a detailed d.iscwsion goes beyond the scope of this chapter. Special attention should be given to sueet \crossing barriers (Wdlar, 1998, Litman, 2009), as well as to end-of-trip facilities, such as (secure) bike parlci ng and :.even showers and clothes-changing areas for commuting traffic. 3.2.5 Other &hibiutl Beh11,U,r In addition to coolliccs and compliana with TCDs, other pedestrian and bicyclist behaviors luve proven reliable to varying degrees in identifying problems and evaluating safety countermeasures. Examples of these behaviors for pedestrians include failure to look left and right before and while crossing, hesitating in the roadway, running, jaywalking, use of the signal pushbutton and n:turnlng to the curb after starting to cross. Examples of these beluviors for bicyclists · are the use of hand signals, bicycle helmets and nighttime lights, or bicycle speed, positioning, use of the signal pushbutton and gencnl driving style on the roadway.

These types of behaviors represent undesirable or unique actions that can reflect some degree of threat to the pedestrian or bicyclist. Acro~odations and/or TCDs that reduce these behaviors are generally regarded to be safer. Use of these measwu is documented in a nwnber of research studies (Harkey and Zcgeer, 2004, Fiapatric.k et al., 2005, Hunter cc al., 2006, R.o!legercs ec al., 2007). 3.2.6 Usn-Percei11d Qruzlity ofService Me~ User-perceivod QqS measures represent a o~ paradigm in the transportation field. These QOS measures shift away ftom traditionally used performana measures (such as delay and uavel time) and towards measures describing the aperiena of the user-the customer. These meas11res are popular with nonmotorized road users because they directly target their travel experiences. They are also gaining acccptana in the engineering community and are anticipated to actually rcplaa some methodologies in the 2010 f<:lease of the HCM to describe the levels of servia for nonmotoriied road users. User-peraption-based QOS methodologies are available for the pedestrian (Dowling ct a!., 2008, Peuitsch ec al., 2008) and bicycle modes (landis et al., 2003, Pcuitsch et al., 2007, Dowling et al., 2008), :md have recently been calibratctl to u.s:"national.dacascts in an NCHRP project (Dowling Ct al., 2008) for pedestrian and bicycle experieoa on urban meets. Another 'FHWA research project has developed corresponding models for off-street shared-use paths and cra.ils (Hummer et al., 2006).

The QOS methodologies are commonly developed from ratings .o f video clips showing different travel experiences. The racers are accual pedestrians and bicyclists who assign a letter score to each clip based on their perception of the portrayed quality of service. The research team then correlates participant ratings with. variables shown in the clip, inclu
3.3 Data Collection Procedures The experimental ddign for behavioral and user-peraprion studies is much the same as for other studies. The issues revolve around sampLing, site selection, scheduling data collection and developing the analysis plan. Appendix A contains further guidance on experimental design. Engineers must carefully supervise data collection to ensure the data collection plan and schedule are followed closely. Observers must be relieved frequently to avoid fatigue and subsequent errors in judgment. Observers must carefully note conditions at the da~:a collection sites to prevent an atypical event or situation ftom confounding the srudy. It is import211t changes in pedesuian and driver beluviors be explained by exactly whar caused them. Selection of MO Es depends on the purpose and objectives of the study, the situation and conditions at th.e ~i.J;es to be studied and the resources (time and money) available. It is not necessary to include all types of behavionl measures. Pedestrian and Bicycie Studies • 153

The following discussion highlights special considerations for behavioral and user-perception studies for the pedesuian and bicycle modes.

3.3.1 Characteristics ofBehavioral Measures For a behavior ro be useful to the experimenter, ir must possess certain characreristics. The following checklist is essential to ensure behavior can be measured accurately and reliably by multiple, differently trained observers. • The behavior must be definable in terms of objective, observable evencs so coding is reliable. • The behavior must occur with sufficient frequency to permit an efficient data collection schedule. • The behavior must have an association with pedescrian/bicycle safety or Bow, either theoretical, empirical, or assumed. • The behaviors must be sensitive, which implies the ability of the measures to discriminate reliably between certain variables ofincerest (to discern a difference where one exiscs). • The behaviors used in the srudy must be meaningful and believable to the users of the study resulcs. If the above conditions are nor met, the analyst runs the risk of wasting valuable data resources by collecting insufficient, inaccurate, or otherwise unusable data.

3.3.2 Preparingfor Data Colkction Data for behavioral studies are collected through manual observation or from video observations. Manual observation is the most common method used because of the added expense of reducing vid~o data. If the behaviors chosen are difficult to observe, video may be the only feasible method. Most behavioral data is collected by t:allying the frequency of a certain event. The craining of observers is perhaps the most critical aspect of performing behavioral studies. This is true for both manual and video data collection. The behaviors are coded by observing the actions of the pedestrians and vehicles and recording the MOEs of interest. It is critical each observer code the same behavior (MOE) the same way. This is referred to as inurrater rc/Jabiiity. Agencies can check imerrater reliability by having two or more data collectors observe the same events, independently code what they see and compare their resulcs. Differences are resolved by a third observer, who ove.csees data collector training. The data collectocs must practice until they reach an agreement level of 95 percent or higher on every MOE. This training is best done using a video of a pedestrian crossing similar to that to be studied.

3.3.3 Suppliu and Technologies Since behavioral data are most commonly in the form of counts of particular events, simple tally sheets are an adequate means of data collection. Engineers usually must custom design data collection forms for each specilic study, since rhe MOEs and site geomecries may vary considerably among studies. The rally sheecs should concain sufficient space to record tick marks for each observed event within an analysis period or can even include checkboxes for particular events. Typically, a stopwatch is needed to keep track of analysis time periods. . In the design of data collection sheets, it is helpful to have most event-types predefined for observations. For example, a conflict study at a signaliz.cd intersection could have a graphical representation, of the conRict types shown in Exhibit 12-10, that allows the analyst to quickly identify differem event types. This approach is likely more accurate and reliable than one char relies on custom narrative for each evenL Depending on the type of studies, handheld count boards or laptop computers can be used to tally evencs. These have the advantage of automatically keeping track of time, and facilitate post-processing of data in rhe office. Using the directional arrows of a standard count board (see Chapter 4), the observer can easily crack a number of different events, aggregated to user-defined time intervals. In that case observers should mark each button for the MOE it represents si~ce it is likely to differ from the predefined function. Ttme-scamped macros on laptop computers can be customized for more complicated data collection protocols. For example, one analyst may code pedestrian behavior while another keeps track of signal phase changes. Using a common time scamp (or even the same computer), these data can be used to relate pedestrian behavior to signal phases.


. The aforementioned example illumates a common problem of behavioral studies. Depending on the p roblem sClltC:· \ ments, it may be the case that one observer cannot reliably record all necessary factors, either bccal!se of a high co~r1 l ·_ tive load or the observational field of view. In addition to the use of multiple observers, it is possible to use mult'ple video cameras co capture different angles of the same event. For example, Exhibit 12-11 shows a screenshm from a picrure-in-piccurc view of a pedestrian signal compliance study (Schroeder ct al., 2009). One camera captures che entire crosswalk and pedestrian behavior, while a second camera is zoomed into the signal phase. The picrure-in-piccure (PIP) display can be created in the office using a video multiplexer device that is common to security camera systems.

Source: Schroeder et al., 2009. 3.3.4 Ob1er~~~~tio1UZI Stwlies vernu Controlled Experiments The discussion above focused on observational studies of pedestrian and bicycle behavior, where an observer sialplY watches behavior of existing road users without intervening. While: this is accep4ble for most observational srudi~5 • some research applications require a grearcr degree of control. For ~pic, an observational srudy investigatingdri...rer yielding behavior can make sratements about behavioral dilfecences between different sites (for example, Fir:zpw-ick et al., 2006) or for before-and-after treatment evaluation (for example, Harley and Zegeer, 2004). However, th~~e observational datasets have too much "noise: which are uncontrolled factors that affect the outcome of the ap~-'" 1 ment, or in this case the willingness of drivers to yield. These uncontrOlled factors make it very difficult to isolate tJ-:lC effect of the treatment in question. Controlled experiments can be helpful to isolate a specific effect, such as alii.J.d-y of whether drivers are more likely ro yield to pedestrians in the crosswalk versus waiting at the curb (Gerusc:hat a.P-d Hassan, 2005).

The experimental design and sample size considerations of controlled experiments follow principles discUS!~ ;..n Appendices A and C. The guiding principle is to control for only one or two factors at a time, and control for ctb- e::r souxces of variability. 3.3.5 D~~t~~for Usn--Perception MnhotlologUs The input data needed for the QOS methodologies describe aspects of comfort, convenience and safety of nollJil..c:::>toriz.ed road users on a facility. Typical variables include traffic flow factors (vehicle volumes, speeds, heavy vehicks :::> • roadway design characteristics (crossing width, number of lanes, presence of bike lanes), characteristics of TCc::::-' 5 (signal timing) and aesthetic appeal (presence of trees, pavement condition). Exhibit 12- 12 shows t!te list of iap'-!-t variables in the auto, tranSit, pedestrian and bicycle QOS models from NCHRP Report 616 (left column) . Tbcpo..Jf'Pedestrian and Bicycle Studies •

zs; !!f!!!S

ect developed user-perception-based LOS

models for bicycle, pedesuian, transit and auto modes. While only the firn two are directly rdated to this chapter, variables for transit and auto modes are included for comparative purposes.. The "X)OC markings indicate interactions between modd input variables and factors on the facility like design, traffic control, or volumes.

-.

-.-.- ...-.-. -

_-- ·-- -

--.-. - .-. - -.- .- - .-.- - .-

.-

.

-_" -. .. .. .. - --"- ,_ ' '~-· ~~~l<~~~~~

_,__

1_ ........ _

Foolllly.,....

,ocllly Coolrol

TrtMit~

.....

'-LO.lOU<

-~·

_.....,._.......

--

"""-

~~

lOU<

XXX

"""

lOCI XXX XXX

...........

"""

"""

XXX

"""

XXX

-~

"""

_.,_...., -c..-

XlCX

lOCI XXX

XXX XXX

-LOS__

~,...,....~

XXX

0..--

XXX

-Oorollr

_.,

XXX

-~

.,.,... fi
XXX XXX XXX

0>.-,..,.,.~

XlDt

Prllencl oiT,_

XXX

-__ --eo--o---

XXX

lOCI

XXX XXX

_,.,,_.......

OloiMco1li-U..

XXX

XlDt

DX

I

XXX

NuN- On Red

XXX

XXX

DX

XXX XXX

lOOI

XXX

"""

XXX

""" XXX

XXX XXX

XXX XXX

IX1

XXX

IX1

xu

lOOI

=

XXX

XXX

XXX

lOCI

lOUt

XXX

xu

"""

"""

XXX

xu

o-.Doloy

......-~

XXX

"""

-lMo'll S9*Ctdo fAoGI\

lQQI XXX

)00[

XXX

0..-U..

DX

.....a.-n..o

""" XXX

XXX

.... LOellrodll•t

--llua-

......... (orlleloyt LoftlVnU..

XXX

-LOS llodoil 2

_

IT-1.011-lOS I -l)oo

..... ..-...,_

XXX

lluolpood

""" XXX

JCa

JCa

"""

JOCX

xu XXX JCa

""" """

Source: Dowling et al. Interaction of Modal LOS Modd Inputs. NCHRP Report 616, MultitriiHW Ln!e/ ofSmtict for Urb~tn Strttts. Project 3-70, page 93, exhibit 99, 2008.

The datt clements listed in the first column of Exhibit 12-12 are addressed by datt collection methodologies described el.scwbere in this manual. Chapter 4 provides methodologies for volume nudies, Chapter 5 describes speed nudies, a.nd delay-based measwes are discussed in Chapters 6 and 9. The reader should also refer to Chapter 15 for guidance on roadway inventories and the collection of asset management data.

A3 an alternative to the formal QOS methodologies discussed above, agencies commonly develop more qualittcive survey- or interview-based methods ro assess the user-perception on, for c:xample, a panicular multiuse path or a downtown area. Exhibit 12·13 shows a walkability checldist {FHWA, 2009) that can be used as a basis for designing custom survey forms (www.walkinginfo.org/library/details.cfm?id"l2). Researchers b.ave conducted large-scale applications of on-street surveys and daca collection with many volunteers who walked or bicycled a particular course with intersectioru, sidewalks and crossings to rate their experience (for c:xample, Peuitsch eta!., 2005, Landis ct a!., 2003). A comparable bikeability checklist is also available (www.bicyclinginfo.org/libruy/details.cfm?id•3).

• • - -

• • · · - - - - • ••• • • ..,., ,..,,,,,..,.,..,,,.,.. '"TI lnt.-t"

"""'lr\ Cr'\ITII""'._I

r

,.

Take a w~lk and u s~ this checklist to rate your neighborhood's wa!kablllty.

. ..,....,... . ........

.

Rating Scale:

Location of w a l k - - - - - - - --

_,,,

1. Did you have room to walk? 0 Y•

0 Soo'w ptd;llcat.J; D N.-..lb Of' poW la.I'Ud and cop~d 0 Sldrwalb we~• brcbo OC' cracbd SW.W.Ib ....... b&ocbd •'kh pokof,

c 0 0

.w.bbe-Jt.. . .pit. . . etc.

"*"'·

No ~,....,orr thouldltn "lOo M~Kil cnJik

~~~-~~~~~~~~~

Locod... ot.,-1....~

Jladllcl(cla<;if.,.•> ' 0

12J4S"

0

Son. p.obt«n..: [] lloa.4 ..-u roo wW4I Cl T~ . . . . m..de u _ . -.o kJoscw 4jd

ar-

_.,...••acatPiiiDI>tDaoou

.-co

0YC'S:

ONo

C.O.tat.~lborwt.•youc:ould

ONo

ov..

=:t~~=~na:.::: dt~

CNo

ov..

ONo

=~·:::.:c:~~

... .anctt.c...,.by•;.en'P

l..oatioaa oC~c

_._ . --,..-

3. Did drivers behave well? DVio-s

D

Some,__,

Dftwoft..

0 - - · . , . _ , . wlcho• _...,.

0 DW aot: yW4 CIO ..,opW c:n.lna the .......,. 0 '1\ined iato p.tOplc QOA\ag cb• tAIWd'

oo......--

... ~~~ k ._F_.~"

o~

4..-. ~ cndk Meta! 0 Somed:l.iag . . - t.·~ o£pll0bkau:

a.~,

(ebr:J. oa.) 1 2 . . . 5~

k4

Crou wkh che UaJK! Location. ot'probt.l:ll'll:·- - -

Rac:&.CJ (F&.d. ont)

1 2. .J
0 Y• •

0 Som,e t.anp'-nf tlt.\np: Q NHd-.l lllON pa.M. ............ OC' tnon CScuydop OScuyCNo<_.up..d 0 DlftY.Iot• oltinu orcr-h 0 Dlrqo a1r ••• tD a.ron.obla. cduv.a

·

c - . , . - __

ce;:: ~".,

--~···"''

"''"' •••e:

ov...

blod
~~

• .-.d

.

..

- -..-

4. Was It easy to follow saf*'Y rules? Could you and your child ...

....a.:~ 0 Pad»4 can bJocb4 . - vMw ~trdc 0 N.-..t ... -

O-n..

· -·

s. Was your walk pleasant?

2. Was It easy to cross streets? 0.........

_,..

..... _ _ t.ocoo-.ofpooblcaoc

OSo.._~..,_

Radac•fl:·l «k-) l1J "I '

How does your neighborhood stack up? Add up your ratings and dedde.

'·--

···--:t. _ . _

~·--

ToiOI _ _

2'-".JO

a.- . .....,._._ .,._,.. Cddnae.l ,.,_ bn'e

2 a...2.5 CeJebale • led& '\f:Mu 16-20

1t- 1 S 5-tO

••\Jbborhood II .,.,.-,. po«l

~ k.c

k liMed& ..-k. l t.,...&.oc•oi"""'''k.,....._......_ bcclt..rdua ct.c. l('tadilalt«b~

Now that you've identified the problems, go to the next page to find out how to fbc them . .· ' ' ! ' \ .

Source: FHWA Wallcabilicy Checklist. FHWA. Pedestrians and Bicycle Information Center. www.walkinginfo.org. Accessed August 2009.

3.4 Data Reduq:ion and Analysis Analysis of the d.ua dloukl follow a pmconccivc:d pbn. Most behavioral smdies employ a "before/after with conttol" type: of design or a simple comparison across different sites. The assistance of a statistician in designing an analysis plan is generally wise. Appendix C contains guidance on data analysis. 3.4.1 Qnwplll4tUmAJ Post-Prouuing For most behavioral and u.sc:r-pc:rcc:ption data, a significant amount of post-processing of data is necc:ssary to make it accessible to the audience. Many bc:h.avioral studies include obsecvc:d frequencies of different c:vc:nt categories that can be: displayed in histograms and bar charts. In some cases, multiple cime-synchroniud data sees need to be combined before analysis, which can be: performed F.Urly easily with modern spreadsheet software. For example, a video-basc4 user compliance study at a signalized intersection may combine rwo data sees of signal phase changes and peaestrian behavior into one continuously time-stamped file for analysis (Schroc:d.er et al., 2009). Pedestrian and Bicycle Studies • 257

3.4.2 Principk.t ofStatistical lnfernu:e Most pedestrian behavior srudies deal with rhe comparison of two conditions, either in a before-and-afrer study, or in the comparison of nvo or more sites. Tn addition, mosr observational data is in rhe form of proportions, such as the fraction of bicyclists running a ree light, the percentage of drivers yielding to pedestrians, or portion of pedestrians ~rossing our.side of rhe marked crosswalk. The analysis of proportional data requires special tests of significance described in Appendix A. It also raises the question ofsample size. As with all significance tests, the statistical comparison of rwo observations requires the mean and standard deviation of the estimate. This means a behavioral study needs to perform multiple observations at each site, at different times of day and/or over multiple days, to obtain a representative sample. As discussed in Appendix A, the sample size requirements are guided by the (expected) difference in means, the variabiliry in the observed data and the desired confidence level. User-perception srudies rypically result in some way of quantifying the performance or qualiry of service of a point, route, or nerwork. Most of the methodologies presented in the literature do not give an estimate of variabiliry in the predicted performance with a few exceptions where models can be used to predict a range of LOS scores (for example, Dowling et.al., 2008). In practice, the results are rypically aggregated to a single LOS score for each point or segment. The comparison berween alternatives then emerges more or less qualitatively from the summation of all individual LOS scores. Commonly, the results are conveyed to the audience using mapping technology as discussed below.

3.4.3 Yuual.izing Behavior and User Perc6ption Dat4 Behavioral data lend che.rrudves tO display in tables or bar charts that show compari.son of the observed frequencies or proportions of different event rypes in different categories. General da.ta di.splay principles discussed in Chapter 3 apply. For studies evaluating multiple dimensions of behavior, more complex graphs can be useful. Exhibit 12-14 shows a way to show pedestrian crossing volumes at 10 signalized crosswalks (numbers) and in seven rnidblock sections (letters) over a period of 6 hours (7:30 a.m. untill :30 p.m.), created in a standard spreadsheet program (Schloeder et al., 2009); thi.s was di.scussed in Chapter 3. The volume of pedestrians (in pedestrians per hour) is shown in different shades of gray (or color) along the studied srreet and over several time periods. The graph is a contour plot of pedestrian volumes over time and space that gives a vi.sual understanding of pedestrian travd patterns.

KlllsboroUJIISII'III, Ralelch.NC·Southbounc!Podutrloi\Croulnas C.. . . ·~ D . . · - • . _ CIIIO-UII •t.»J.e ·~JM at-..~ . . . . .

~l

I

I

~

!

~

~I

.

II lj ,

I ~ ~,;:~· ' A

f~ --~

-

. ,, l i-~' I~L~~· , _ '. 41 ••,! 'I'. Ir 1~~• Jl~ •

'•11 1

_

'il' " --...,~-~ .. 111 •I , 1 ~I!

"":.:'ll·':·: ~~l

II'V ' \

S

J

I

1

•

S

C

t

t

'

I

t

t

r----t ~ 'r-''r-':r--TT---T'.. .. "' J

J

l

J

J 1

.3

..

J

Source: Schroeder er al, 2009.

User-perception studies are commonly combined with spatial GIS analysis to provide levels-of-service maps of the studied regions. Different colors can illustrate regions of particular good or bad LOS scores. The same approach can be taken fur other QOS measures. Exhibit 12-15 shows a map of "bicycle ease of use• for downtown Austin, TX, USA. Different shading represents Low, moderate and high ease of use, as weU as barriers to bicycle connectiviry.

258 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2NO EDITION

\

~TmFAcunes

.

.

~-- --.=""'L_ ...,_,~ __:._~'_

....:... .. ...·,· ~----

Soiuce: City of Austin, Public Works Deparuncm, Bicycle Program.

4.0 REFERENCES 4.1 Literature References Campbell, B. J., C. V. Zegccr, H. H. Huang and M. J. Cynecki. Report FHWA-RD-03-042: A RrvitW ofPedestrian Saftty !WMreh in tht Uniud St4Rs and Abrolld. Washingron, DC: Federal Highway A.dminstruion, 2004.

Dewa(, R. E. and P. Olson. Pedemians and BU:yclist~, Human Facton in Traffic Saf~. Tucson, AZ: Lawyers&: Judges Publishin~ Company, Inc., 2007: pp. 568, 571. · Dowling. R. F. NCHRP Report 616: Muhimlldai Ltutl ofSmtict Analysis for Urban Sutm. Washingwn, DC: Natioru.l Cooperative Highway Research Program, Federal Highway Administration, 2008. ~ckral

Highway Administration. Manual on Uniform Traffic Control Dnlicn. Washingwo, DC: FHWA, 2003.

Federal Highway Administration. Manual on Uniform Traffic Control Devicl!. Washingron, DC: FHWA, 2009. Federal Highway Administration. "Walkabiliry CheckliSt.• Pedestrians and Bicycle Informacion Cenrer. www. walkinginfo.org/. Accessed August, 2009. Fitzpatrick. K. et a!. TCRP Report 112/NCHRP Repon 562: Improving Ptdertrian Safiry at UMgnalizul lntmections. Washington, DC: Transportation Research Board, 2006.

Gerwchar, D. R. and S.E. Hassan. "Ddver Behavior in Yidding to Sighted and Blind Pedestrians at Roundabouts." journal of Vuuai Impairment and Blint.Wss 99, No. 5: 28~302. . Harkey D. L., and C. V. ZegeeL PEDSAFE: Ptdertrian Safoty Guidt and Counttnn4a.:s-t= Stkction Syrttm. Report FHWA,SA-04-003. Washingron, DC: ~decal Highway Administration, 2004. Pedestrian and Bkyde Studies •

25~

Humme~, J. E., er al. Evahuztion ofS4fity.

Design. tznd Opmttion ofShartd-Ust Ptuhs • FiMI R.tpon. Report PHWAHRT-05-137. Mclean, VA: Federal Highway Adminisuarion, 2006.

Hunter, W. W., L Thomas and J.C. Stutts. BIKESAFE: Bicyck Coun~1'7MIZSUrr &kctU!n Systnn. Report FHWA-SA-05-006. Washington, DC: Fedenl Highway Adminim:uion, 2006. Iacono, M., K Kriuk and A. EI-Geneidy. kess to Destiruuions: How Close is Close EMugb? .&timllting.A=rate Distana DeCAy Funailms for Multipk Modn tznd DijformJ Purposts. Report 2008-1I. Minneapolis, MN: University of Minnesota Center for Truuporration Studies, 2008. www.ets.umn.edu/Publication.s/ResearchRqx>rts/pdfdownload.pl?id=916.

Kdl, J. H. Tnmsportlllion Plmming Harulbook. Chapter 2: Transportation Planning Studies. Englewood Cliffs, NJ: Prentice Hall, 1991. landis, B. W. et al. "Inteneetion l..eve! ofSetvice for the BiqdeThrough Mo-anent." TI'IINJ'Drtlllion &starch Rwml:founlill of tht Transportlltion &wuch Bocrl1828 {2003): 101- 106. Litman, T. EwdU4ling .Acctstibilityfor Transport4tion Plllnning. Victoria, British Columbia: Victoria Transport Policy ln.sticuce, 2008. www.vcpi.org/aa:ess.pdf. Litman, T. ~ Ejf«ts, Tnmspvrtlllion Cost tzndjknrfo.ANJysis. Victoria, British Columbia: Vicroria Transport Policy Institute, 2009. www.vtpi.org/rc:a. Mingo, R., H. D. Roberaon and S. E. DavU. Mt11111ring htiettria-tr VOlumes: .A Usui M~»~IUI!, VOL IlL Wasjllngton, DC: Federal Highway Administration, 1988.

FHWA/IP-88~30.

Petritseh, T. A. ec a!. "Level-of-Service Model for Pedestrians at Signalized Intersections.• T11Wpol14tion Restllrrh P.=rrl.·]ownud o/tht T'rtmSportlllion Resum:h Botmi 1939 {2005): 5~2. · Petritseh, T. A., et al. •Bicycle l..eve! of Service for Arterials.• Tnmsportlllion Restllrrh &corrl: ]()1jrn41 vftht Tnmsportl#ivn &surch Boart/2031 (2007): 34-42. Petritxh, T. A., ec al. "Pedestrian Level-of-Service for Arterials." Transportlllion &starch Recortl: joUrn4/ oftht Transportation &stJtrrh Botmi2073 (2008): 58-68. Pline, J. L Trtifjit Enginming Htuulhook. Chapter 2: Traffic Studies. Englewood Cliffs, NJ: Prentice Hall, 1992.

Rodegcn:s. L ec aL NCHRP Report 572: ~in tht lJnitl,J States. Washington, DC: Transportation Resean:h Board, 2007. Roess, R. et al. Tmffic Engintning. 3rd ed. Upper Saddle River, NJ: Pearson Pre.tltice Hall, 2004. Schneider, R., R. Patton, J. Toole and C. Raborn. Ptdatrian tznd Bicyck Data Coll«tion in United St4tn Communidn. Washington, DC: ~enl Highway Administration, 2005. Schroeder, B. J., N. M. Rouphail and B. A. I..ehan. Observational Study of Pedestrian Behavior Along a Signaliud Urban Corridor. Transportation Research Boaro 88th Annual Meeting. Washington, DC: Transportation Research Board, 2009. Transportation Research Board. Highrv4y CAptzcity Mmrwal. Washington, DC: TRB. 2000. Transportation Research Board. NCHRP Web-Only Document 129: Phast III Pttlestri4n S4fity Prrdiaion Mttho®loo. Report FHWA-HRT-04-100. Washington, DC: National Cooperative Highway Research Project. McLean, VA: Transportation Research BoarQ. fedenl Highway Administration, 2008. VtaOria Transport Policy lnstilll!e. UnivmJ Dmgn-T~ ~that~ .AlJ Usm.lndwlint Ptttpk wiJh Disabi/itks, anJ Other Spma/Nd. VICCOria. British Columbia: Vicroria Truuporc Policy Institute. www.vcpi.org/tdm/aim69.han, Accessed January 5, 2010.

Wdlar, B. W.Jbnt S«urity IIUiex. FiNd Rrpon, Ottawa. Ontario: University of Ottawa Geography Oqlarunent, 1998. hccp:l/aix1.uottawa.cal~wdlarb/Dcsign%20Swdy%20Publications.br:rn.

Zegeer, C. Y. et a!. S4foty Effms ofMarJud vmus UnmarkM Crosswalh at UncontrDlled LoctUionJ: Final &port anJ &commmdtd Guuirlina. McL:an, VA: Federal Highway Administtatioo, 2005. www.tfhrc.gov/sali:cy/pubs/04100/inda.htm.

.,4.2 Online Resources )"Bikeability Checklist" and other resources. Bicyclinginfo.org. www.bicyclinginfo.org/libraryldeails.cfm?id=3. Retrieved 'January 5. 2010. BLOS/PLOS Before & After Calculator Form. (October 15, 2003). League ofUiinois Bicyclists: www.bikdib.org/roads/blosl blosform.htm. Retrieved January 5, 2010. Victoria Transport Policy Institute. Univmal Design Resources. Victoria, British Columbia. www.vrpi.org/tdrn!tdm69.htm. Reuieved Januaty 5, 2010. "Walkability Checklist" and other resources. Walkinginfo.org. www.walkioginfo.org/libraty/details.cfm?idRl2. Retrieved January 5, 2010.

4.3 Other Resources Dharmaraju, R., D. A. Noyce,J. D. Lehman. An .f1valuatWn ofTechnologies for AutDmated Daectio,_. arui Clastification of htkstrians and Bicyclim. Washington, DC: Federal Highway Administration, 2002. Dowling. R. NCHRP Web Document No. 128: Mu!Ji11Wtial Lrve/ of&rviu Analysis for Urban Stret1J: Uso- Guitk. Washington, DC: National Cooperative Highway Research Program, Federal Highway Administration, 2008.

Flannery, A., D. McLeod, N.J. Pedersen. ·cuswmer-Based Measures ofl.evd of Service." ITE]ournal76, No.5 (2006): 17-21. Patten, R. S., R J. Schneider, J. L Toole, J. E. Hummer and N. M. Roupbail. Shared-Use Plllh Ltw/ of&rvi&t Calcuiatot-A lhn-s Guitk. Report FHWA-HRT,.05-~38. Washington, DC: Federal Highway Adminsuation, 2006. Vandehey, M. NCHRP 03-92: Production oflhe Year 2010 Highway Capacity Manual. Washington, DC: Transportation Research Board, 2008.

Chapter 13 •••••••••••

0

.

....

...

..

.

.

. . . . .

...

. . . . . .

.

.

. . . . . . . . . . . . . . . . . .

••••••

..

..

. . . . . .

.

. . .

..

.

.

...

.

.

Public Transportation Studies Original By: Humme-r, Ph.D., P.E.

]os~ph E.

EJit~dBy:

Bastian J Schroeder, Ph.D. 1.0

INTRODUGION

264

t. t Purpose and Limitations of this Chapter t .2 Transit Quality of Service

264 265

1.3 Interaction with Other Modes

265


265

2.1 Problem Identification

265

2.2 Transit Performance Measures

265

2.3 Transit Field Data

268

.2.4 Use of Existing Data

271

2.5 Choice of Study Method

271

3.0 DATA COLLEGION PROCEDURES

272

3. t Manual Data Collection

273

3.2 Automatic Data Collection

281

3.3 On-Board Transit Surveys

283

3.4 Office Studies

285

4.0 STATISTICAL ANALYSIS

285

4. t Sampling

285

4.2 Sample Sizes

286

4.3 Selecting a Sample

288

4.4 Estimating Mean Values and Proportions

288

5.0 REFERENCES

288

5. t Literature References

288


289

5.3 Other Resources

289

Public Transponation Studies • 26:J:

1.0 INTRODUCTION

P

ublic transportation studies provide operat.ors with the information they need to make inteUigent choices about services. Studies provide the numbers and trends whlch may indicate changes in operations are needed. Studies indicate whether patrons have responded to changes. Studies also provide data for comparisons berween agencies, to measure the quality ofservice and to provide the basis for funding decisions. It is' important that public transportation studies be conducted properly. A poorly designed or executed study may be worse than no study because the numbers may provide operators with c:xrra confidence in a mistaken course of action. In addition, studies may consume large quantities ofscarce funds. A mff of over 30 full-rime data collection personnel (referred to as "checkers" by public transportation professionals) is not unusual in a large agency. Automatic data collection equipment can be expensive, but agencies increasingly make the investment to obtain more complete data on transit system performances.

1.1 Purpose and Limitations of this Chapter In thls chapter public transportation studies are introduced to novice public transportation professionals, checkers and students. Experienced professionals can use this chapter as a quick reference. Agencies or companies spending large sums to conduct important studies should refer to TRB Transit CP.pacity anti QJtality ofService Manual (TCQSM) (TRB, 2003) for a detailed description of analysis methodologies for transit operations and associated data needs. Other reference material is available through the Federal Transit Administration (FTA) (Carter, 2002) and other literature sources (TRB, 2005, Kittelson, 2003, Lede, 1998). Larger agencies with a mix of transit modes, large transit management centers and specialized transit services such as bus rapid transit and dynamic scheduling are referred elsewhere for a more comprehensive discussion (Schwenk, 2002).

Thls chapter applies primarily to srudies of fixed-route bus transportation. Many of the techniques discussed also apply for other fixed-route modes such as streetcars. Demand-responsive (for example, parauansit) servit;es have different purposes and vasdy different ways ofoperating than do fixed-route services, so their study techniques are generally different. Those interested in studies of paratransit operations are referred to resources provided by the Easter Seals Project ACITON (Accessibk Community Tranportation in our Nation) and specifically its bus stop checklist (Project Action, 2008). The studies discussed in this chapter are primarily those whlch are useful for operators studying their own services. Studies are also conducted to meet the needs of outside funding sources, investors, the media and the public. In the United States, agencies that receive federal funding conduct studies to provide National Transit Database (NTD) data to the FTA. The types of data agencies are required to report include (mdprogram.gov): • Operational Characteristics: vehicle revenue hours and miles, wilin.ked passenger trips and passenger miles, etc. • Service Characteristics: service reliability and safety, etc. • Capital ~enues and Assets- Sources and uses of capital, Beet size and age and fixed guideways, etc. • Fmancial Operating Statistics: revenues, federal, state and local funding, costs, etc. Transit operators conduct srudies under vecy rigid sets of rules that are nor repeated in this chapter. Thls chapter further focuses on the conduct of the study rather than the establishment of data collection programs or the manipulation of study results. Fmally, the chapter is limited to studies requiring actual field data collection and does not include any discussion of the coUeccion and analysis of routine management informacion. Counting revenue, counting transfers, conducting an inventory of equipment and parts, collecting personnel data, compiling cost informacion and other continuous management informacion efforts are outside the scope of this chapter.

.,.:A a

lAAMt l.t\1

1"'1~

TDAM<:tltii:!Th.Tit"\td C.._lf:tt..II:I:DIMt": <:TIIt"'\IC<:

"U,Ir\ t:r.I"TI/"'1.._1

, 1.2 Transit Quality of Service )Many analysis methodologies found in the Highway Capacity Manual (HCM) (TRB, 2000) for the evaluation of performance of traffic systems for the auto mode have traditionally focused on measures derived from traflic flow theory concepts, including the wdl-escablished speed-Bow-density relationship for freeway operations. Recently, a variety of measures have been introduced that incorporate measures that describe the user perception of the quality ofservice of. a particular transportation system. For the transit mode, the user-perceived performance is even more important, since customers directly pay for the service and greacly value the availability ofservice, schedule adherence and reliability. These travder-perception measures are a central focus of the TCQSM (TRB, 2003) and the reader is referred to that document for an extensive discussion of transit quality of service (QOS) concepts and associated performance measures. For the purpose of this chapter, it is assumed the transit agency has previously established a performance evaluation framework and a policy approach for providing safe and reliable transit service to its users. The customer-focused analysis paradigm is reBected in the studies described in thls chapt:er. It presents data collection approaches needed for the TCQSM QOS frunework, forusing on r.ransic availability, comfort and convcnienet.

1.3 Interaction with Other Modes Transit trips inherently rely on other modes for riders to access transit facility on either trip end. Most predominantly, transit facilities are accessed by pedestrians and bicyclists, but, often, park-and-ride facilities also need to be considered. This chapter does not repeat details of these other studies, since pedestrian and bicycle studies are discussed in Chapt:er 12 and parking studies are presented in Chapter 16. The data collection procedures for these studies are similar to the on-screet studies 'Oi.scussed in those chapters, bu.t the focus is often on provisions within (or near to) the transit station. Example data items ot iiueresc are the processing rate or queue lengths for rumsriles, crowding on escalators, stairs and fare controls, or parking availability for vehicles and bicycles near the scation. Additional details on station design, including time-space analyses for platforms and pedestrian ac= routes to transit, are discussed in theTCQSM (TRB, 2003). Additional guidance is also given in Chapter 12.

2.0 TYPES OF STUDIES 2.1 Problem Identification The type of public transportation srudy to be conducted depends on the problem facing the operator. Operators could have problems planning future services, scheduling. evaluatinll and concrolling current services, or reporting to funding sources and investors. The first task when conducting a public transportation srudy is, as for other studies, to identify the problem to be solved. An agency may be interested in monitoring the general performance of its tranSit system, or may need to address a specific need for one route, one segment of a route, or even a single transit stop. The metrics needed to answer the question of transit performance ultimately guide the design of the transit study.

2.2 Transit Performance Measures ]usc as with automobile traffic, there is a range of perfoCIIWice meastU'CS that ean describe and quantify the ope.n tions of transit. In faa, this manual devotes many chapters to varying aspects of data collection. and uansporcation performance measures, many of which apply to transit. However, one of the important difference between transit and (private) automobile traffic is that transit is a serviet riders pay for directly. A transit system is therefore under more direct pressure to provide high-quality service to its traveling CUStomers.

The customer-service-based approach to transit performance measur~ is reBected in the TCQSM user-peretption measures. That manual emphasizes the use of measur~ that are useful to the operating agency, but also reBect the travel (or waiting) experience of the transit user. The TCQSM defines transit LOS for three levels of analysis: the transit stop, the route segment and the overall sysrcm (definitions &om TRB, 2003).

Public Transportation Studies • 265

• Transit Stops: measures addressing transit availability and comforc and convenience at a single location. Performance measure values in chis catego ry will rend co vary from one location co another, since these measures depend on passenger volumes, scheduling, routing, and stop and station design.

• Roucc Segment:;/Corridors: measures chat address availability and comfort and convenience along a portion of a transit route, a roadway, or a set of parallel transportation facilitie.o> serving common origins and de.o;tinarions. The.o>e measure values will tend to have less variation over the length of a route segment, regardless of conditions at an individual stop. • Systems: measures of availability and comfort and convenience for more than one transit route operating within a specified area (that is, a district, city, or metropolitan area). System measures can also address doorto-door travel. In an effort to incorporare user-perception into the evaluation of transit systems, the TCQSM defines smnct mttzJures for each level and for each of two QOS categories: availability, and comfort & convenience. Service measures are chose mecrics char are used to define transit LOS. The TCQSM also provides ocher performance measures that are useful ro the agency, but chat are not directly used to derive the A-F letter score for uansir operations. The TCQSM recognizes both service categories are important: Passengers expect transit services chat have broad temporal and spatial coverage (availability), and services that are comfortable, convenient and safe (comfort & convenience). Exhibit 13-1 shows the service measures used in the T CQSM to define LOS for the thiee analysis levels, as well as other performance measure:; that can be useful to transit agencie.o>. Exhibit 13-2 gives definitions for the six service measures as defined in the TCQSM. Note that information in Exhibits 13-1 and 13-2 applies to fixed-route bus service and that the TCQSM defines many additional measures for ocher forms of public transportation,

Availability

Other Measures

Reliability: on-time Service Measu.re

Reliability: headway adherence Missed Trips

Comfort and Comoe.o..icnce Other Measures

• Measwes adopted from Trrtn!it Capacity and Quality ofStroict Manual, TRB, 2003. • Complicated and requires spatial GIS analysis.


~~~. • '· · ~ • -.

-

.

p~~~-···· ~~-··~~~~~ ~• ...: 1 • o. . • ·-· ·- ' ..· -"·"'·'. '' ' ~jt§o ~ . •. f.l..- :1o.....· t ' J. ~ ~ . . ~ ..... " ' ·~..-~

-~~...........

" ........~;··-~'·';i.;i!'; ,~.:.::-4:~! ".!{:tr~f~\~~ . . . . . . ~.Jo.a:. .. "1- ..

,.p; ..-,•.•• , '

o

'

Service Me:a.sure

Definition•

Service frequency

The number of transit units (veh icles or trains) on a g~n route or line, moving in the same direction, that pass a given poi"t within a specified interval of time, wually I hour; see also head way.

Hows of service

I. The number of hours during the day betwee n the start and end of service on a transit route, also known as the service span. 2. For calculating transit level of service, the number of hours during a day when service is provided ar least hourly on a uansit route.

Area, coverage

In mruit operations, the geograpbical area that a transit system is considered ro serve, normally based o n acceprable walking distances (e.g., one-quarter mile, 0.4 km) from loading points. For suburban rail transit that depends on automobile access (park-and-ride or kiss-and-ride), coverag<: may extend several miles (kilometers). Coverage is usually computed for uansit·supportive areas.

Passenger load

The number of passengers on a cransit unit (vehicle or uain) at a specified point.

~liability

How often tranSit service is provided as promised; aHtcu waiting time, consistency of passenger arrivals from day to day, toc:a.l trip time and loading levels. The service measure of route-levd comfort and convenience in the TCQSM qualit)" of service framework.

Travel time

The door-to-door difference between automobile atld uansi c uavel times, including walking, waiting and transfer times as applicable. A quality of service measure representing how much longer (or in some cases, shorter) a uip will cake by tranSit.

\

d.ifferen~

-----

-----

-

-----··-

-

Note: The equations wed 10 estimate some of the service measures in Exhibit 13· 1 are estimated from a range of input vari· ables that go beyond the scope of this chapter. The reader is referred co the TCQSM for dewled discussion of the methodology. The focus of the following discussion is on performing field studies that can be wed to evaluate fixed-route bus services, including srudies to validate the service measures predicted by the TCQSM.

-

-

• Source: Kittelson & AssociateS Inc. et al. TCRP R.tporr 100: Transit Capacity and Quality ofSnvice Manual, 2nd Edition. Put 8 "Gio.Ssuy", p. 8-i, 2003.

Public Transportation Studies • 260- 7

2.3 Transit Field Data The transit performance measures described in Section 2.2. can be estimated from fidd data, using a variety of data collection procedures. The list shows in parentheses the sections where these are discussed in more detail. 5. Driver Study (3.1.1)

6. Point Check (3.1.2) 7. Ride Check (3.1.3) 8. Trail Car (3.1.4)

9. Survey (3.3)

10. Automatic Vehicle Location (AVL) (3.2) 11. Office (3.4) 12. Observation The first five studies represent actual field measurements of transit operations. All can be performed ~¥ing manual data collection procedures, and most can be automated to some extent. AVL is only an automated study that uses on-board GPS technology to directly report a number of transit performance measures. The category •office~ includes a range of analysis tasks that are performed off-route using various software applications and database analysis tools. Transit planning applications rdy exclusively on office studies using pre-existing data for other transit routes and sys~cms, or inore commonly, are performed using software fo{ scll«iuling. ~vd forc:.;asting and ~mpping. The final type of transit study combines various studies that rely heavily on (qualitative) obstr~~ations of a particular aspect of transit operations. No formal study procedures are described for observation studies, since the scope, purpose and methodology vary greacly depending on the application. Examples include general observations about boarding: and alighting practices, or challenges to special rider populations. · The different srudies are described in detail in Section 3 of dU.s chapter. Exhibit 13-3 relates the performance measures to the applicable 6dd studies.

"tea

a

A.AAt..llll\l 1"'\C TOA ... ICOAOTATlr\ft.l

I:Mt:.l.._li:::~Ofto.lt":

C'TIInlf:C'

"'U.In

~r"\ITII"'to"- 1

r~rn.n~o ..

·-. · ~

.

::,;,:_~.

••

·;o'i~::ii~~~:·

.

'

..

~.

-"'t<:i -~'1!..">!£1

<

-?r-"i""'

.'j

~-

Study Method Dmu Study

Data Item

Point

Ride

Ch«k

Ch«k

Trail Car

Survey

AVL

Office

Observation

TCQ.SM Senice Measures

Service frequency Hours of service

,; ,;

.J

..J

.J

..J

..J

,;

..J

..J

..J

..J

.J .J

A=., coverage

.J

Passenger load Rdiability

,;

Travel time difference•

i

,; ..J

Othec Measures

Customer Satisfaction

..J

Safety and Security Performance

.J

,;

Missed Trips Freq. of Mechanical Brcakdown.s

Travel Speed, Travel Tune and Delay

.J

..J

..J

..J

..J

"

..J

Wait Assessment

..J ..J

.J

Transit Travel lime

.J .J

. ..f

..J

..J

..J

Passenger miles

..J

.J

Passeogec origin and destination pattern '

Pedestrian Crossing Difliculty

..J

' .J .J

Transit-Auto Travel Tune Ratio" Boarding and Alighting (by fare category)

..J

.J ..J

Pedestrian Alx.ess

..J

Bicycle A=

.J

Parle. &

Ride~

..J

As:a:u for PersoiU with

Disabilities

'

Passcoger Loading Ease Transit Stop Amenities

..J

" ..J

• Requires mea.rurement of passenger-car traffic!


These performance measures represent a sample of the most commonly used transit attributes. TCRP &port 88 (Kitte~on, 20Q3) idenrilies additional transit service attributes that are sometimes used by agencies as shown in Exhibit 13-4.

Source: TCR.P Report 47: A HAndbook for Measuring CustQmer Stuisfaction and Qflality, 1999.

It is a challenge for an agency to select a subset of these attributes to use for evaluation of their transit service. Oearly, the ones required by the NTD for reporting are critical. Measures used in rhe TCQSM to define transit LOS are useful and may be required by region. But other measures may be of particular interest to a transit agency. Fmally, measures describing the accessibility of transit systems to persons with disabilities are useful to ensure the service is in compliance with the Americans with Disabilities Act (ADA) legislation. One approach for selecting additional transit service attributes is to survey the users of rhe system about their perceived preferences. TCRP &port 47 (TRB, 1999) presenrs a methodology to assign an impaa scOt? to each attribute based on resulrs of a customer sat:isfaction survey. The survey principally asks respondents to rate transit performance for a range ofattribute$ and also asks whether a person experienced a problem with each attribute during the trip. The approach rhen calculates a gap sco", which is defined as the difference between the average attribute rating for those who experienced a problem and for those who did nor. While the impact score approach can be helpful, it also requires signilicant (survey) data collection prior to collecting other items. In a more general guidance, rhe selection of a data item for study is a function of the srudy objective and in som~ cases may be deterrp.ined by a specific complaint or issue with the service. Ifseveral data items are relevant to 270 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDI110N

! the

problem, it may be possible co collect them simultaneously. Analysts must take the rime co understand the prob\lem thoroughly, however, or they may overlook a relevant data item. For example, analysts evaluating adve rtisements :for public transporration services will usually collect hoardings, and may be interested in hoardings by fare catego ry. ~ measures of behavior. Since advertising usually has the objective of changing attitudes and awareness as wdl as behavior, analysts should also collect data on passenger attitudes and awareness in chis case. Data items selected for srudy must be closely related to the problem or to the objectives of the program being evaluated. Whichever performance measures-are selected by an agency, it is critical that recording of these measures is done accurately and consistently to be able to reliably track performance over time. Iris useful for agencies to develop their own data collection standards and best practices affecting specific data elements or certain aspects of their tra nsit service-

2.4 Use of Existing Data After identifYing the data items needed, the next task in conducting a public transportation study is to decide ..,.,.!Jjch data items are to be -collecteQ in the field. Before starting an extensive field data collection prograin, the t ransit agertCJ should investigate if orher data sources exist that already contain some of the needed information. Examples include prior data collection efforts or existing travel demand modeling data from the regional planning agency. Transit field data collection should be used to supplement existing data sources. For agencies desiring to monitor transit perfo=ce over time, many measures need to be collected at regular inter-

vals. Data required by NTD need to be updated annually, but agencies may wish to collect these data more fr~uendY· To reduce the need for ·ad hoc" data collection e.lforts, some agencies develop continuous data collection prograr11s. These programs often consist of a baseline phase and a monitoring phase. The baseline phase provides a comprehensive one-time •snapshot" of system ·operations. Data items collected simultaneously during the baseline phase provide all the data necessary for long-term planning and other infrequent major efforts, provide conversion factors for auxiliary items and provide the beginning points for trend analyses. During the monitoring phase, key data it~ms are updated as needed, using conversion factors whenever possible. During the monitoring phase, operators should alsO make periodic checks co reveal problems that require attention.

2.5 Choice of Study Method For some data items, there is no choice of study method. For other items, there are several choices. The prime dete.Crninants when a choice is presented are availability of resources, measurement errors and cost. Availability of resources means that questions such as the following must be a.ddressed. ·

• Driwr studies: Do driver work rules permit data collection? : • Point and riM checks: Are sufficient numbers of trained checkers available? If not, can checkers be hired and. trained? • Studies with autcmatic data colkction tquipmmt: Is the equipment available? If not, can the funds be procuced and allocated to purchase and maintain the equipment? • Surwys: Is the expertise to write, administer and analyze a sticvey available?

• Trail Car and ObstrVaMn: Are sufficient numbers of observers and uail vehicles available? Can data items be combined to ma1te most efficient use of resources? • Office: Is the software and expertise available in-house? Are funds available to outsource parts of the analysis? If the resources are not available to conduct the study using a particular method, analysts must explore alternativ-e methods, change the data item to be collected, or abandon the scudy.

Public Transportation Studies • 271#

Measurement errors occur with every study method. For driver studies, increasing passenger loads and heavy uaflic are major factors in causing error. For point checks, many standees and vehicles with tinted windows lead to measurement error. For ride checks, simultaneous boarding and alighting, two or more doors in use, a multitude of fare categories and crowded vehicles cause error. Automatic data collection eqwpmem can fail systematically (due to large crowds, passengers with large packages blocking sensors, etc.) or can fail fiom mechanical or electrical problems. Finally, surveys are extremely sensitive in the choice ofsample, format, wording ofquestions and many other ways, as described in Appendix B. When considering a study method, analysts must consider any problem particular to that method and the data item needed. For a boarding count, which is the most common type of public ttansporracion srudy, automatic data collection eqwpmenr will generally provide the most accurate data. The ride check method will be close to automatic data collection equipment in accuracy, and point check data will be less accurate (UMTA, 1985b. Deibel and Zumwalt, 1984). Cost is more important than measurement error in choosing a study method. This is because all the methods discussed provide reasonably accurate data and because increased sample sizes can reduce the effects of many measurement errors. The choice of a srudy method is easy when costs are considered, since driver srudies are generally the least expensive srudy method and point checks are the next le;m expensive. Asuggesred prioritization of study methods is: 1. automatic data collection systems, if available

2. driver studies, for items that cannot be collected automatically 3. point checks, for items that cannot be collected automatically or by drivers 4. ride checks only for items that cannot be collected automatically, by drivers or by point checks Observational ~dies and analysis in the office typically apply for different types of studies and are therefore not included in above list. A survey is a relatively expensive and risky last resort when none of the other common methods can deliver the needed data items. There arc a few circumstances when ride checks are less expensive than point checks, such as when many simultaneous point checks are needed along a route or when hcadways are long.

3.0 DATA COLLECTION PROCEDURES Following the identification of data items to be collected, the analyst chooses a study method. Most studies arc performed by checkers in the vehicle (ride checks), by checkers stationed along the route (point checks), by drivers, or with automatic data collection eqwpmem. Analysts usc surveys during some studies. Some data items require the use of computer software for travel forecasts or mapping applications. Exhibit 13-3 relates performance measures to different types ofstudies. This section describes the data collection procedures of different manual and automated means of collecting tranSit data.. In the question of whether to use manual or automated data coUecrion, a recent synthesis of practices found many small uansit agencies scill rely on manual data collection (Boyle, 2008). Reasons cited include the high cost associated with equipment procurement and concerns about rdiabiliry. The report also found many agencies apply a combination of manual and automated means of data coUecrion in their transit performance measurement protocol. The most common forrn of automated data collection are automatic passenger counting (APC) systems and AVL in the form of on-board GPS technology. It is further common to rely on the fare box to keep track of the number of paying customers and hoardings. However, the fare box is less useful in areas where tickets are purchased at the station and are not automarically checked upon entering the uansit vehicle. Exhibit 13-5 shows the distribution of APC, manual and rue box methods among agencies surveyed in TCRP SynrheJis 77(Boyle, 2008).

Sow= Created by author wing data based on Iioyle. 2008.

The following discussion is divided into ~anual methods (driver study, point checks, ride checks and observational srudies), a.utomatic methods (wing APC and AVL technologies), surveys and office srudies for planning applications.

3.1 Manual Data Collection Manual data collection methods require transit agencies to have sufficient staff available, or to hire, temporarily; trained observers and checkers, to perform the tasks at hand. For agencies with many.rouces and vehicles, the staffing requirement can be extensive for a system-wide study. Depending on the study objectives it may be preferable to contract out an annual daca. collection prognm, supplemented by periodic spot checks carried out by agency mff. 3.1.1 Dmer$tutliu

Having drivers count boarding passengers or bouding passengers by fue category is a very efficient and :~.ccurate study method. The method is efficient because no labor coscs for checkers ue incurred. Driver counts ue a.ccurate becawe they ace kept simple, becawe drivers are monitoring boarding passengers anyway and because drivers ace Wn.il.iac with the vuiow fare categories being recorded. Driver studies ue impossible on unmanned vehicles and when p:.I.SSCngers can board the vehicle tluough more than one door. In these cases, transit agencies must we other study methods. The driver cao keep the count on a mechanical counter or on a counting system integrated into an dectronic fare box. Usually the count boa,rd presents no more than six keys to the driver·so some fare categories may have to be recorded together. The driver oould write the total count or the count per fare category on a summary form, like that given in Exhibit 13-6, at the end of each trip, or the driver could enter the code for •end of trip" into an electronic count board.. Appendix E provides blank data. collection forms suitable for copying. Statistics produced from driver srudies include the mean number of passengers per ttip and the proportions of passengers by fare category. The analysis of statistics produced in public transportation studies is di.scwsed later in the chapter.

P1 1hlir Tr~n~~;nnrt~tinn C\t11rliPc; • 27:1

~~~~~:~~1}~~~!:···:~~~~~~-:-;l~~---h:~;;~tJ~~~~~~t4:~~~JY!f~··~~ Boarding Count field Sbeet Block Number

Route Date

Day

Weather O bserver

Route Segment From

Boarding Passengers To

Foil Fare

Reduced Fatt

Transfer

Full + Transfer

Reduced + Transfer All Passes

3.1.2 Point Checlu Chrocers conduct point checks while standing along a route, usually at a stop, watching boarding, alighting and on-vehicle activity. Point checks are efficient if the point has moderate to high levels of public transportation activity because the checker will be able to collect informacion from multiple vehicles on different routes. Point checks are also efficient if both directions of a meec can be monitored by one checker. Point checks are reasonably accurate, although not as accurate as ride checks or automatic data collection equipment. Point check accuracy will deteriorate during busy boarding and alighting times or ar stops with multiple transit vehicles arriving at the same time. In these cases, multiple observers may be necessary, at which point ride checks may become a viable alternative. Operators should not consider point checks when the check locations are subject to bad weather or if crime is a concern.

Point checks primarily provide load co unt:; and schedule adherence. Exhibit 13-7 provides a convenient data form for these primary items. Checkers will need an accurate warch ro record schedule adherence. Secondary purposes of · point check studies include: • boarding and alighting counts at the particulat point; • deceleration, stop and acceleration rimes at the particular point; and • data on passenger behavior, such as arrival times, wait times, directions of uavd before or aftec uansit trip and modes of travel before or after transit trip at the particular point. Operators can alter the data form in Exhibit 13-7 to include columns for these secondary elements if they are of interAppendix E contains blank data collection forms suitable for copying.

est.


.

·-

.

:;;:

.·;

--~,.,.,

.

•... .,. .. .

·-~

.. ~~~~W~'"f~-:~~~~!~~'*-~~i;~~~-.~~ Point Check Field Sheet

Route(s) Day

Bus Stop Number Date

0 Departing Load

Weather Observer

.

0 Arriving Load Block Vehicle Route Number Direaion Number Capacity

Arriving Tune

Scheduled

Actual

Passens ers

-

-During most point checks, the checker estjmates the load while standing ouuide the vehicle. An efficient technique is to count the passengers if the load is light, count empty seats if the load is moderate and count standees if the Jo:im check forms. However, in some cases pc:rsonal data assistanu or some types of•smart phones" may be custom.iud for data collection and allow for time savin~ 5 in data processing. Optical scanning may be useful for other surveys as well. Standard point check data provide several useful plors and statistics. Exhibits 13-8 and 13-9 show cwo of the mor.e useful plots with load data and with schedule adherence and load data, respectively. The vertical axis in Exhibit 13-B is the loadfoetor in percentage of seating capacity, rather chan the acrual number of passengers, co remove chc dfect:5 of different sizes of vehicles. In Exhibit 13-9, both passenger load and schedule adherence are shown on the saJP e graph. Later in this chapter the analysis of relevant sta.tistics, such as 'rhe mean load and the proportion of vehicles o~ schedule, is d.iscussed.


I

I:

~:'-1 II ~

····-··

~

Pc&ylDad

t; : ::::::::

~

~

~ OJ{.

1

2

3

4

5

6

7

8

w

9

ll

~

VeHde Nu'ri;)er

A.M. P.ak

I I I

I I I I .I I

Pa. . .no•r toad-

I I I

~

I I I

I

I

I I

I I I I I

II

:' Il . II

I I

II

I I I I I I

. I I

I

I

I

I I I I I I I I

I I

I

I I

I I

I·I III . II

l

I I

I I I

I

7:00

7:10

7:ZO

1

7:30

7:.c>

I

7il0

8:.10

Tl""' a

I·i

i :00

ll:20

8;30 0

lndlcet.cf by eolld lfn• on tv

e:.c>

I I I I I I I I

I I I I

II II I I I I

1 I I

e:ao

eili<

I

I

8ctledu1ed amval

• Actuar am..,.l

Source Urban Mass Transpo=tion Association, 1985.

3.1.3 RUle Checlu A chcck.cr conducts a ride check py riding in the vehicle of interest a.s it coven its route. Ride checks provide more acrurate data than point checks and arc comparable in acruracy to automatic data collection equipment under most typical loading conditions for many data items. Ride checks are usually much more expensive than driver studies and point checks because a checker is gathering data on only one trip at a time. In addition, ride chcckecs lose considerable time finding or waiting for the assigned vehicle and traveling between assignments. For long articulated buses with multiple doors, two ride checkers may be needed to ensure accuracy.

0 :

\.,.

Ride chcc.ks are Ba.ible and provide seven! useful dau. items. Ride checks most ofttn provide boarding and alighting volume~ the vehicle arrival time at each stop and at other key points. Exhibit 13-10 shows a fonn suitable for recording 771: '" Ml1NIIo11

()~ TRo1N<;Pf"IIITATif1~1 '~lf:l~"""•""

-..u-. <"OTt"•'

Ride Check Field Sheet 'BiockNumber - -- - - -- - -- - --

Route Number - - - - - - -- - - - - - - : -______________________________

~~~er

Day

Date - - - - - - - - - - --

Direction ofTrip _ _ _ __ _....,_ _ _ __ _ __ _

Observer - - - - - - - - -- - - - -- - -

Scheduled SwtTune - - - - - - - - - -- -

Location

On

Off

Load

Tune Check

these basic items. A checker must wear an accurate watch to record arrival times. Progr.uns are available that allow data to be recorded on a handheld or portable computer and later tranSferred to a personal computer for analysis. Ride checks sometimes provide more detailed data on vehicle movements than simply arrival time at stops. Ride checkers can also record deceleration times, dwell times at stops, acceleration times and dwations and causes of delays. Exhibit 13-11 shows a form for these speed and dday items. If they are busy, checkers df) not need to compute column 7 for dday time in the field. Colwnn 2 for the time at conuol points allows schedule adherence and various speed measwc:s to be calculated. A ride checker using a form such as the one shown in Exhibit 13-11 can carely collect data on passenger boarding and alighting at the same time.

Ot.hli..-. Tr~rt.r,.."l""t ~t i "n <.tt tl"'i or a

.,..,.,

~-~mw:::··. ·~:;..~· ~~-.~·: -·_·o:. ~-·

•

""!

~

"~::~· ~

D::l6i'?l

·

o.

'""''~' -•. · ·;~;7·~ ., ••• . ',: ' ~·''·J .... '' ::..u.;: :..:;;<':~·~ ~- ~i;ltk~;:>.;;;~·:,4.~'c.; s • .- . ' • •. ... • • __ ,.

"

-_

':1-~~-

. ••. .

)'~~

~

Public Transportation Vehicle Delay Field Sheet Day

Date

Route

Direcrion ofTrip

Weather

Vehicle Type

Observer

Method Trip Number

Trip Start Time (1)

(2)

Location

Tune at Control Point

(3)

Trip End 1ime (4)

(5)

(7)

(6)

Tune Time Jlowerthan faster than walking wal.ldng speed StopTwe speed Delay Cause

(5)- (3) Delay time (sec)

I

Total DdayTtme, SecondJ Symbols for delay cause: P • passenger loading, S ~ craflic signal, SS • srop sign, PK • parked cars, DP =double parked, PED • pedestrians, RT • right rums, LT = ldi: rurns, T . general congestion, Kf • inrentionally killed time, 0 ~other (explain) R.emarlu:

Ride checks can also provide many of the data items that driver studies and point checks provide. Checkers can count hoardings by face category during a ride check. Load counts and boarding counts per trip can be made direccly during a ride check or can easily be derived from a ride checker's boarding- and alighting-by-stop data. Ride checkers usually occupy the seat direccly behind the driver. Boarding and alighting from the front door, rhe causes of delay and the rype of face paid are all best observed from that sear. Boarding and/or alighting through a rear door, with m:my standees, may require a change in position or may mean the checker stands to conduce the counts. In rare instances, more than one ride checker may be needed on a vehicle. Checkers must be familiar wirh the causes of delay or the possible fare categories, depending on the data items being collecred. Ride check data are analyzed in sevetal ways. Data are mosr commonly displayed in a simple bar chan showing boarding and alighting by stop. The bar graph can further be modified to vary the width of the bars by the distance between scops. Ic i.s also common with boarding- and alighting-by-stop data to calculate the passenger miles per trip. The p=ger load lxtween twO stops is multiplied by the distance between the scops; the number ofpassenger miles per trip is then the sum of these quantities for the entire route. Operators examine tranSit vehicle time and delay data by producing a space and time plot as in Exhibit 13-12 or a bar graph with magnitudes ofdelay by type as in Exhibit 13-13. Useful statistics from ride check data include the mean number of passengen boarding at stop x, mean p=nger miles per trip, mean delay ar sigJul y, mean ovu.ill ttavd speed and the proportion of delay due to ruming vehicles.ln a Iacer section the analysis of typical statistics from public transportation studies will be discussed. With new visualization techniques, analysts have been developing new and innovative means of displaying transit data. Exhibit 13-14 shows an example where transit load, boarding and alighting and searing capacity are all displayed on a spatial plot of the route in Google Earth.


I

SfCOHDS PER STOP

.

~.

.•.

..

Sowcc: Created by Graemc Brown for Tran.slink.

3.1.3.1 Passmgn- Origin an4. Desti711ttUJn Operators can use ride check chta to estimate passenger origin and destination paccerns by stop on a. roU[e (cha.t is, not "crue· door-to-door origins a.nd destinwons) at Jea.n four different ways, including: • A checker stands a.t the rear of the vehicle a.nd notes on a seating cha.n the stop where each passenger bouds a.nd alighu (Simon and Furth, 1985). • A checker distributes a cud to each passenger as he or she boa.rds. The card is coded with the originating stop. The checker colleca the cuds as passengers a.ligbt a.nd notes the destina.tion stop on the c:ard. (Simon a.nd Furth, 1985). A ride checker (or automatic da.ta collection equipment) provides boa.rding a.nd alighting da.ta by stop. Then a relatively simple a.lgorithm provided by Simon a.nd Furth (1985) estimates the origin a.nd d~a.tion pa.rtem. The a.na.lyst uses point check loa.d data a.nd a. minimum of ride check boarding and a.lighcing by stop da.ta. in the a.lgorithm from the method immedia.tdy above (sec Furth, 1989).

Each of the four ride check methods of determining origin a.nd dcscina.tion patterns by stop on a. route bas adva.nta.ges a.nd cl.isa.dva.n~es. The first twO involve more direct measurement, so they a.rc generally more a.ccurate but more apensive tha.n methods three a.nd four. Method one is appropriate only on trips with no sca.ndees and limited seating turnover. Method two bas a. high response rate but may ca.use initial confusion among passengers. The checker must be able to c:xplain quickly. possibly in more tha.n one la.ngua.ge, the purpose of the card a.nd wba.t the passenger is to do with it (Simon and Funh, 1985). Instructions can a.lso be printed directly on the card, but it shouldn't be assumed all riders can qUickly read a.nd understa.nd printed (English) material. If a.na.lysu can tolerate somewha.t less accuracy 280 • MANUAl OF TRANSPORTAnON ENGINEERING STUDIES. 2ND EDffiON

and have the needed ride check or point check data on hand, methods three and four arc the best ways to estimate o1gin and destination patterns. Operators also usc surveys to estimate origin and destination patterns, particularly when they need •uue" doo r-to· door origins and destinations. On-board surveys arc discussed in Section 3.3, and general survey methods are described in greater detail in Appendix B. 3.1.4 Trail Car A Checker in an auwm~bile trailing the transit vehicle of interest can also collect the speed and delay items on the form shown abcve in Exhibit 13-11 for the ride check. The trail car method is advantageous be<;ause the checker is (perhaps) out of sight of the transit vehicle driver, so no bias is introduced by changed driver habits. However, the trail car method is more expensive than the ride check method because it always requires an auwmobile and under heavy. traffic conditions may require a second checker. The trail car method also docs not allow a full view of the causes of ddays. Because the disadvantages of the trail car method rudy outweigh the advantages, operators use the ride check method more often for speed and dday data items. With the advances in AVL equipment discussed in the next section, the mil car med_lod ~ rudy liSM in practice.

3.2 Automatic Data Collection. Automatic data collection for transit srudies involves rwo main technologies: automatic passenger counters (APC) and AVL. The first, APC, automatically counu passenger boarding and alighting; the second, AVL, uses on-board GPS technology to record position and speed data on transit vehicles. With both technologies, it is important to check for accuracy and perform frequent validation of the obtained count or travel time data. While advances in both of these data collection technologies have.gready improved accuracy and rdiability, there arc still sources of error including equipment malfunction. · While the cost of i.nsta1lacion is significant, oftentimes APC and AVL technologies arc installed as part of a broader ITS deployment in a city, which reduces the relative cost of APC technology. Usc of both technologies requires good communication among municipal departmentS, because the technological infrastructure spans across jurisdictions. TCRP Synth~is 77 (Boyle, 2008) provides a more detailed overview of advantages and disadvantages of these technologies, as well as guidance for good practices and lessons learned with installation. 3.2.1 AuUima#c Pasmrger Counters . APC equipment includes data acquisition, recording, transfer and reporting componenu. In most installations, a series of photoelectric beams near the door or pressure-sensitive pads in the stairs provides boarding and alighting counu by stop. Computer algoritluns decipher the activity recorded by the sensors and decide, for aarnplc:, whether a sequence of signals represents a boarding or alighting passenger. The equipment records tra'ld distances (through the vehicle odometer) and times to index the countS and ro provide schedji.(c adherence and similar data items.

Agencies may usc APC technology to obtain both ridership and travd time data along routes. The main advantage of APC systems is that data are available in a permanent form allowing the transit agency to track performance on a continuous basis. The primary benefits of APCs cited in a recent survey of transit agencies (Boyle, 2008) were data disaggrcgated at the srop. segment and trip lcvds: better quality ridership data; availability of running time data to adjust schedules; and a better basis for decision-making. The main problems with APC systems included issues with reporting software. data. processing and analysis, data validation and h;udware problems. Approximately one·quarter of all respondents reported either no problems or only minor start·up issues (Boyle, 2008). 3.2.2~J Ko.bi& lACAIWft

With increasing availability and cost-dfectiven~ of GPS technology, AVL insall.ations arc being used more commonly by transit agencies. Through on-board GPS installations, an agency can automatically gather a lot of data including uavd times, schedule adherence and ddays. With a continuous data stream the agency can readily implement a performance monitoring system and can identify common sources of recurring congestion (for example, a particular intersection or a particular time interval). An agency pn usc these data to improve upon the service provided and enhance the qansit experience for customers.

AVL dara can also d irecdy benefi t customers through t:he use of real-rime transit data, disuibured to rhe ridership through displays at transit srarions, Web sites and even integrated in applications for cellular telephones. The technology investment therefore serves the combined need of collecting data and providing a service to customers. Exhlbir 13-15 shows an example of an online route visualization system for a major university. Exhibit 13-16 shows a snapshot of data (five trips) ob r.Uned from one of the transit routes indicating what segments of rhe route are most correlated wirh travel rime vuiabiliry.

Source: C20 10 TransLoc Inc.

SpacefTlme Trajectories - Loop

O r----~-----~----.-~--------.-~------~.-~-------,------~ --~r-----------------------------------------------------~----~R~oundabout 0 .5

Gym

C heck 2

~

::E 1. 5 Che ck 3

2

~

2.5 0

2000

4000

T ime (sec)


eooo

8000

3.2.3 System Acquisition and DatA Processing To accurardy process auromared data, additional equipment infrasrrucrure is required at scops and the agency needs tO ·develop a back-end daca processing mechanism through software and database managemem. One of che additional infrasuuceure ueatmencs are signposts, which are radio transmitters mounted on the roadside along a rout e that provide the on-board processing unit with location information as the vehicle passes. Signposts eliminare the need for difficult decisions in data processing about which boarding and alighting data belong to which stop. Without signposts these decisions often cannot be made, which results in discarded data. Signposts have other applications besid es daca collection that may help justify their coscs as well. Sensors to record when the vehicle door is open and keyboards char allow extensive manual inputs co the data record are other features of some automatic data collection systems .

Automatic data collection equipment is expensive, but ir can be inscalled on only a representative fraction of the vehicles in a fleet. Deibel and Zumwalr (1984) concluded a useful, valid, reliable database can be main tained if :tboUl 10 percent of the vehicle fleer is equipped for automatic data collection. Scheduling an equipped vehicle to a trip on which data are to be collected is challenging due to a variety of problems, including vehicle servicing a nd incompatible vehicle sizes. For example, operators cannot schedule a data coUeceion vehicle with a capacity of 40 passengers on a route char nee9-l 60-passenger vehicles. However, equipping I 0 percent of the fleet providcl an adequate mugin for most of these problems." If the objectives of technology installation include public visualization of transir data. as discussed above, mosr ifnot all of che vehlcles need to be equipped with AVL technology. Automatic data collection equipment produces much more raw data than manual study merhods. The data processing and report generating capabilities of the system must be sufficient to handle this load. With lase microcomputers and other advances in computer technology, though, this is more an administrative than a technical challenge. The reports themselves are no different from ch.e types of reports produced with rrianual methods and described elsewhere iil this chapter. Data collected automatically aie usually more accurare than data collected manually, but bias can be a problem. ALlto(llacic systems tend to undercounr for several reasons, including mechanical or eleccrical failures, environmen~ factors and passengers blocking sensors. If bias is suspecred, the operators can compare the automatic count to acarefu.]., trusted manual count. If the operators .find a bias, they can repair the system or correct for the bias machernaricallY during data processing. Beciuse the costs ofAPC inscallations vary so widely, it is impossible to State exaccly which public transportation ageocies would fi.nd the equipment cosr-dfeccive and which would not. However, experience and cost analyses indicate

• for larger agencies (over 400 vehicles) the equipment will usually be i:ost-effecrive • for midsized agencies (100 to 400 vehicles) the equipment may be cosc-effeccive depending on data needs • for small agencies (under 100 vehlcles) the equipment will wuaUy not be cosc..effeccive When the availability of funds, the data accuracy needed, the data rurnaround time and other faceors are considered with the cost-effectiveness, deciding whether to invest in aucomaric data collection equipment becomes very compl~·

3.3 On-Board Transit Surveys Surveys provide infor!Jlation on passenger demographic traics, door-to-door origins and destinations, travel d~ires • attitudes, future plans._and ocher characreristics. Operators cannot coUece these data items with the ocher study methods presented in this chapter. Surveys are relatively expensive to develop and conduce and are probably less accurar~ than the ocher srudy methods discussed in this chapter. Therefore, if operators can coUecc a data icern with a srud::V method ocher chan a survey, they should do so. Analyses conduce many surveys for public transportation agencies b::V telephone or by visiting respondents' homes. Appendix B discusses these types of surveys. This section highligh!S th~ features that make on-board surveys distince from ocher surveys. The sampling and analysis rechniques for on-boarcl surveys do noc differ much from other study methods discussed in this chapter. Those techniques are presented lare.C in the chapter. In recent years; online uansit surveys or even surveys done via cellular phone text messaging are bec9ming increas~ ·

ingly popular. Online surveys rely on riders completing the survey after completing the .trip, once they have access too Public Transportation Studies • 283

a computer (although many may be able to access the Web while riding via laptops and smart phones). The resulting response rates may therefore be lower than traditional in-vehicle surveys, buc the cost for administering the survey and printing material is much reduced. Furthermore, much of the survey analysis is automated &om online database tools. A survey done via text messaging uses readily available cdl phone technology to · vote" on a specific transit survey question, much like the viewers of variow tdcvision shows and polls. Analysts gencrally we two types of •traditional" on-board surveys: hand-in and mail-in. For both survey types, survey workers distribute the forms with a quick word of explanation to passengers upon boarding. Hand-in surveys are collected &om passengers as they alight. Hand-in surveys arc short, consisting of only a few questions, because there is limited time for passengers to complete them. Passengers answer hand-in surveys c:xclusivdy on the vehicle, so only information passengers remember readily can be asked for. Hand-in surveys are printed on a card that can be written on and are distributed with pencils. Passengers may complete mail-in surveys· on the vehicle or elsewhere. Thus they do not need; to be as short as hand-in surveys. They can pose more difficult questions, since passengers have time to ponder questions and seek assistance for answers. Mail-in surveys arc printed on ordinary paper and do not have to be dist.ributed with pencils. The forms can be stamped and addressed on the reverse side, or a stamped and addressed envelope can be provided with each form . The choice of survey type depends primarily on the number of questions the operator must ask. Shorter surveys arc generally berter, but on-board surveys can be longer than postcard size if the questions provide needed data items. Response rare is another factor that in.fiuences the choice betwttn hand-in and mail-in surveys. Hand-in surveys have higher response rates, which could approach 90 percent usable returns, while mail-in surveys generally achieve 30-60 percent usable responses. However, a high response rate does not necessarily mean an unbiased survey {Brog and Meyburg, 1981, Doxscy, 1983). Fmally, if passengers are on board for a very short t(ip or if there is a signi6caht number ofstandees, scv~ bi;~ses will occur with hand-in surveys. An intriguing option that combines the best features of hand-in and mail-in surveys is the two-part suivey (Stopher, 1985). A two-pact survey consists of a short hand-in form and a longer mail-in form that are handed out together. The hand-in form asks for very basic informacion on passeriger characteristics and should have a high return rate. The mail-in form asks the detailed questions that arc truly of interest to the survey sponsor as well as the same questions askai in the hand-in form. The amlysr can compare the characteristics of passengers who rerumed both parts to the characteristics of the passengers who only returned the hand-in pact, and will thereby gain insight into biases in the sample that answered the important questions. Bamford et al. (1984) described a useful innovation for hand-in surveys. Instead of distributing pencils with a hand-in survey, they created a survey form on which passengers could rub off responses with a fingernail or coin, similar to the instant lottery tickets sold in some states. They reported a high response rate and that the eight-question survey took only 30 to 60 sec to complete. The main disadvantage they noted was the need to prccode all possible answers, so a survey with detailed questions could not use the format. Stopher (1985) offers several exceUent suggestions for conducting on-board surveys. First, he recommends a count be maintained of the number of survey forms distributed for each trip. This count is essential for measuring the response rate. Second, Stopher argues drivers should never be asked to distribute survey forms. Distribution of forms by drivers may bias questions about the driving. may not allow the control of forms noted above to be maintained and may not provide passengers with the assistance they need to complete the survey. Next, he states that separate forms should be available for each predominant language in the survey area. A quick greeting by the checker distributing survey forms should be enough to indicare which language a particular passenger can respond in. Stopher also advises boxes should be placed at each exit door for the return of forms, with large signs posted nearby to draw actention to them. The chccket distributing survey forms may be busy with boarding passengers and may not be available to coUect every completed form. Finally, Stopher urges every fare-paying passenger be given a survey form, since there are many problems with sampling a subset of passengers on a vehicle.

·i


3.4 Office Studies , Some of the data that are needed in TCQSM melhodologies or dut may be of interest to agencies o.n be ob rained in . the offiee without extensive field d:~.ta collection. While not the focus of this manual, these office studies arc recogniz-ed here, because they ue pertinent to many transit applications. Common offiee studies include planning-level analysis of transit schedules, route mapping and travel forecastS for tr:~.nsit planning purposes. These to9ls employ computertravel demand models to perform a four-step forecasting of travel demand, including trip generation, trip distribution, traffic assignment and mod:c split. The latter component is central to planning for tr:~.nsit systems and may be a vital component to procuring federal funding for transit improvements. More information on planning s tudies is given in Chapter 20 of this manual.

4.0 STATISTICAL ANALYSIS In this section sampling. sample siz.es and me:~.ns and proportions are discussed. These issues apply to all the study methods presented above. More detail is given in Appendix A.

4 .1 Sampling A sample is necessary for almost every public tl:lnsportation study because the cost of a complete census is pro hibitive. Public transportation agencies face increasing financial pressure, especially for expenditures perceived as nonessential, such as data collection, and cannot tolerate waste. Forrun.ately. the ability to generaliz.c: about an entire population from a sample allows useful data to be collected and analyz,ed for a reasonable cost. · ..

After determining !:he study methOd to be used, the analyst conducting a public transportation study must determine the sampling unit. The sunpling unit is the entity tha.t represents one data point. For most publ.ic rransportacion scudies, the analyst chooses me trip-one vehicle in revenue service traveling from the start to the end of a route-as the sampling unit. For example, in determining whether hoardings have risen due to the introduction of a new fare pass, driver scudies are to be conducted. Each driver participating in the study reports the number of passengers on each trip. From these data the analyst can calculate mean boudings per trip. In tum, multiplying this mean by the total number of trips will produee the desired quantity for comparison. Once a sampling unit is chosen, the analyst devises a sunpling plan. Histotically, analysts have used simple random sunples in most public tr:~.nsportation studies. With a simple random sunplc, each unit in the entire population of units has an equal chance ofbeing selected for measurement. The sample size and analysis formulas for simple random samples are straighcforwud. However, the plan sometimes causes inefficient use of resources. A simple random sample of trips measured by ride checks, for example, may mean that check~ will be scrunbling aU over the route necwo.rk to be on their assigned trips. A simple random sample of passcngeis measured using an on-board survey may mean distributing survey forms to only a few passengers on each vehicle. For more efficiency, analysts o.n conduct many studies using stratified or multistage sampling plans. Strarifod plans require the analyst to divide the population of units into several identi6.:~-ble groups and then sample randomly from within those groups. For example, analysts could distinguish radial routes from crosstown routes and could sample a ips from those groups. Another common stratification is by time of day: One could draw separately from A.M. peak, P.M. peak, between-.(leak, evening and overnight "owl" services. MultistJJg~ puns are more common and usually more efficient than stratified plans. Most multistage plans require the random selection of.

• routes from the entire sysccm of routes; • dire<:tions of travel from the selected routes; and • time periods from the selected routeldirection combinations. With typical multistage p!:l.ns, point checkers can stay in plaee throughout a peak period, for example, or survey distributors can give a form co every passenger boarding a vehicle. Stratified and multistage pl:~.ns intrOduce more order g,,hli,- Tr"\nrnr.rt,tif'r"'

..

,.c

into the data collection process, assure a more accurate estimate of the statistic ofinrerest and can result in a reduced sample siz.e. The sample siz.e and analysis formulas for suatified and multistage sampling plans ate more complex than for simple random samples. Refer tO Cochran (1977) and other sampling texu for details. Analysts contemplating stratified or mulrisrage sampling plans should seek advice from a professional statistician.

4.2 Sample Sizes Once analysu have planned the sample, they can calculate the necessary sample size. The sample size necessary generally depends on the desired confidence levd of the e;timate (t), the coefficient of variarion in the sample (v) and the tolerance on the estimate (d), which represenu the allowable range around the estimate (fot example j: 15 percent). For estimating mean values wirh a simple random sample, the necessary sample siz.e (n), can be estimated using: n=

(t~)

2

Equation 13-1

where t

= a constant that depends on the confidence levd needed in the e;timate

v

=

d

= tolerance on the e;timate

coefficient of variation (the standard deviation divided by the mean)

Analysu can conservativdy assume values fort as 1.8 with 90 percent confidence and 2.2 with 95 percent confidence, based on Urban Mass Transponation Association (UMTA) (1985). A tolerance, d, of 0.15 with a 90 percent confidence level means that 90 percent of samples of the size n will have mean values that fall within 15 percent of the estimated mean. To apply Equation 13-1 and to estimate a sample size, advanced knowledge of the coefficient of variation is necessary. Exhibit 13-17 therefore prov]des codlicients of vamtion (v) for common data items if estimates are nor available locally. Tolerances recommended by UMTA (1985) range from 10-30 percent for most common data items.


Load

Boardings, passenger-miles

RWlning time

"Owl (late night) default values are the same for weekdays and weekends. Source: Urban Mass Transportation Association, 1985b.

Equation t3-t above is use& for data that are in the form of a simpje random sample with a sample mean andsom.~ variation about that mean. However, many data items collected in public transportation studies are proportions, such as the proportion of passengers paying special student fares and the proportion of trips more than 5 min. behind schedule. The sample size, n, for proportion data using a simple random sample, can be estimate using n-

t 1 p(l- p)

dZ

Equation 13-:Z

where t

= a constant related to the confidence level

p

= prior estimate of the proportion of interest

d

• tolerance on the estimate

r

Values of in use with Equation 13-2 are approximately 3.0 for 90 percent confidence and 4.0 for 95 percent confi..dence. If no prior estimate of pis available, analysts can use a conservative value ofp =0.5. In public uansporncion&srudies, a cominon value of tolerance for estimating proportions is 0.1. for values of p near zero or 1, analystS oeecf-· asymmetric tolerance ranges (see UMTA, 1985). Public Transportation Studies • 287

4.3 Selecting a Sample After analysts determine needed sample size they select the units to sample. They can use random number tables and computer programs that generate random numbers. Stratified and multistage sampling plans will lighten the burden of this task compared with simple random sampling plans. Operators schedule checkers to study che selected units while considering their work rules, the need for breaks and travel times berween study locations. Foul weather and special events may force analysts to drop cenain units from the sample. If analysts want an estimate of annual ridership, for example, they should not have to alter che sample plan due to foul weather and special events. However, ifanalysts need an estimate ofschedule adherence for scheduling purposes, they should delete trips from the sample that arc affected by foul weather or special events.

4.4 Estimating Mean Values and Proportions In earlier sections of this chapter graphs were presented that were useful for presenting the results of studies. Along with a graph of key results, many public transportation studies also require an esrirnate of a mean value and knowledge of the tolerance associated with the estimate. The best estimate of the mean value for a data item in a population is the mean value from a sample. The tolerance associated with the estimated mean value is found to be: v d=t-

""

Equation 13-3

where the variables are as defined for Equation 13-1. One can compute the coefficient of variation, v, for.Equatlon 13-3 from the sample (that is, sample sta.n
Analym can estimate che tolerance on a proportion from a sample using: d = tjp(l:p)

Equation 13-4

where the variables are as defined for Equation 13-2. The sample provides the value ofp, and the value oft is as described for Equation 13-3. Presenting estimated means or proportions and the tolerances on those estimates graphically is usually of great assisrance to decision makers viewing the study results.

5.0 REFERENCES 5.1 Literature References Bamford, C. G., R. J. Carrick and R. MacDonald. "Public Transport Surveys: A New Effective Technique of Data Collection." Trtrjfic Enginuring and ConiTOINo. 6 Ounc, 1984): 318 ct seq. Boyle, D. Passmgtr Counting SJ11nN-.A. S]nthais ofPnzaUt. TCRP Synthesis 77. Washington, DC: Transportation Rc:search Board, 2008. Brag. W. and A. H. Meyburg. "Coruidcration ofNoruespoo.sc EffectS in Large-Scale Mobility Surveys" Tranrportlllion Rnearrh &cord: journal ofthe Trrmsport:aJion RntarGh Botm/807 (1981).

Career, R. T. ~ .A.uvnmar and Quaiity ConiTOI GuiJelma. Washingroo, DC: Feckral Transit Adminisuation, 1987. Cochran, W. G. Sampling Ttclmiques, 3ni ed. New York: Wiley, 1977. Deibel.• L. E. and B. Zumwalt Modular App~h 111 On-BoarrJ AutonltttU Dllla Coli«tion ~· National Cooperative Transit Rcscarcli and ~opmcnt Program Report 9. Washingron. DC: Traruportation Research Board, 1984.

Doxscy, L Rnpondmt Trip Frrqumcy Bias in On-Board Surveys. Transport4tion Researrh &cord: Journal ofthe Tramportuion : Rnearrh &art/944 (1983). ·Furth, P. G. Updating RUU Chedu with Mulrip/4 Point Chec!ts. Transportation Rnearch &cord: journal ofthe Transportation Rnearch Board 1209 (1989). Kittelson and Associucs, Inc. A Guitkboolt for Dtvtloping a TT111Uit Pirfonnanu Me~ZSUmnmt Systnn. TCRF &port 88. Washington, DC: Transp0rt4tion.Rnearrh Board, 2003.

l.ede, N., L Yu, C. A. Lewis and K. Godui. A Manwzlfor Evaluating Pmonaliud Public Trzznsit SyJtnns. Dallas, TX: Texas Southern Univcnity, Federal Transit Admini.stntion, 1998. Schwenk, J. Ewzillation Guitk/ina for Bus RApid Transit Dnnonstration Program. Washington, DC: Federal Tr.uuit Administralion, 2002. Simon, J.. and P. G. Furth. "Generating a Bus Route 0 -D marrix from On-<:~ff Data..• ASCE Journal ofTrzznsportalion Engineering 111, No. 6 (November 1985). Sropher, P. R. "The Design and .Execution of On-boanl Bus Surveys: Some Case Studies. • New Surwy Methods in Trrvuport. · · Umcht, The Necherlan<:U: VNU Science Press, 1985. Tr.uuporation Research Board. TCRP Report 47: A HIZ1I.dlxJolt for Measuring Cuctamer Slltisfacrion and Servia QJuzlity. Washington, DC: TRB, 1999. Tran.sporation Rcsearch Board. Hith=J Czpacity Manwzl. Washington, DC: TRB, 2000. Transportation Research Board. TCRP Report 1100: TI'Ztiiiit Capacily and QJudity ofServia Manwzl. Wasbi.ngcon, DC: TRB; 2003. . Tn.n,sporation Research Board. TCRP Synthesis 63: On-Board and lntnrq>t Trzznsit Surwy Techniques. Washington., DC: TRB, 2005. Urban Mass Transportation Amninistration.. "Review ofTraruic Data Collection. Tedtniques: Final Report." DOT-I-85-26. Washington, DC: Urban Mass Traruportation Adminisuation, 1985. Urban Mass Transportation Administration. "Transir Data Collecti.on Design Manual, Final R.eporr: DOT-l-85-38. Wasbi.ngcon, DC: Urban Mass Transportation Adminisrration, l985b.

5.2 Online Resources Projecr Action. "Toolkit for the Assessmenr of Bus Stop Ac=sibility and $akry: Wasbi.ngcon, DC: Easter Seal.s Project Action. from: http://projecraction.easrrrseal.s.com/site!PageServe:~pagenameaESPA_BusStopToolk.it, R.erricved December 4, 2008.

5.3 Other Resources Box. P. C. and J. C, Oppenlander. Manual ofTrrzffic Engineering Stutlies, 4th ed. Wash.ingcon, DC: Instirute ofTransporation Engineen, 1976. Camus, R. L "Estimation ofTran.sit R.elial>iliry lcvd-<:~f-Service Based on Automacic Vehicle Location Data.." Trtznsport4tion Rnearrh &cord:J~unuzl ofrlu Trzznsportation Rnearch Board 1927 (2002). Carter, R. T. QJudil] .A.snmiMe and QJudil] Control Guitklines. Washin.gcoo, DC: Fedetal Transit Adminisuation, 2002.

Furth, P. M. "Designing Automated Vehicle Location Systems for Atchived Data Analysis." Trtzruportation Rntarrh &cord: journal ofuu Trzznsportation Rnearrh &ani 1887 (2002)._


Golani, H. "Use of Archived Bus Location, Dispatch, and Ridership Dau for Transit Analysis." Tramportation &uarch &cord: journal oftlu Tr1Z11lportation Board 1992 (2002). Kimpel, "f. Data Vi111ahzation aJ a Toolfor Improved Decision-Malcing within Transi& Agencies. Seattle, WA: University of Washington, 2007. Kimpel, T. S. "Automatic Passenger Counter Evaluation: Implications for Nacional Transit Dacabase Reporting." Transportation &uarr:h &cord· journal ofthe Tranrportation &search Board 1835 (2003). Perk, V. and N. Kamp. Handboolt ofAutomated Data Collection Mtthods for the National Transit Database. Tampa, FL: National Center for Tr:msit Research at the University of South Florida, 2003. Robenson, H. D. Manual ofTranrportation Enginming Srudm. Washington, DC: lnstirute ofTransporracion Engineers, 1994. Urban Mass Transportation Administration. &vemu Based Sampling Proudum for Obtaining Fixed Route Bus Operating Dat4 &quirtd uiukr the Section 15 Reporting Synnn, Circular UMTA-C-2710.4. Washington, DC: Urban Mass Transporracion Adminisuuion, 1985.


Chapter 14 ........................... .............................................. . ..........

Goods Movement Stud ies Original by: Marsha D. Anderson Edited by: Daniel]. Findley, P.E. 1.0

INTRODUCTION


291 291

2.1 State of the Industry/Existing Data Sources

293

2.2 Route Studies

298

2.3 Loading and Unloading Studies

299

2.4 Vehicle Weight Studies

302

2.5 Hazardous Materials Studies

304

3.0 REFERENCES . 3.1 Literature References

305 305


307

3.3 Other Resources

307

1.0 INTRODUCTION oods movement, the distribution ofraw materials and finished produCtS, is most frequently handled by cruck, train and :Urplane. Some conunod.ities move by ship or pipeline, but these modes tend to be less visible to the ~e cran.spornuion planner. Incorporating &eight planning into general transportation planning activities is bco>rniPg increasingly important as the regional and 10<:21 road necwod
G

Many views are represented with respect to freight movement: those ofshlpp~ and carriCI'S, public agencies such as ckpa.-ecments of cranspornuion at alllevds, plann~ and c:ngino:rs, public officials, citizens and the general motoring public. '[b.e role each plays when ~y study is being conducted will inspire the study design and therefore the results. The question of whether one is looking for short-term, praccical solutions to ccisting problems or projections for long-rerm planningpu..1"· poses will also influence the types of srudies conducted. lnvestigacioos of.&eight movement may lead to changes in o~ra.-ring practices, wning or devdopment regulations, design scand.ards, fo:s, tariffs, or caxes. It may also influence the timing o f public or private projects and the dollars expended. In this chapter we address the process of coileaing d:ua. the impaas o f route characrcristics,loadinglunloading studies, vehicle weight studies and hazardous ma.ce.rials movement.

2.0 TYPES. OF STUDIES The goals and objectives of a study will dictate the types of data and quantity needed for drawing reas'onable concl~ sions. The impaas of freights movement on a community can be extensive and ace a likely subject area to srudy fo I Goods Movement Studies • 29 ......-,

mitigation options or policy implementation. An examination as part ofNCHRP Report 320 described the following commwtity issues from the various freight modes (Suauss-Weider, 2003). Trucking community issues include: • congestion generated on local roads, highways and at customer facilities; • large tractot-uailers making deliveries ro customer facilities-inslifficient loading dock space, leading to double parking and meet congestion; • movement of heavier trucks on roadways adversely affecting automobile speeds; • damage caused to pavement, especially from heavier trucks and more frequent truck movements on local roads; • hazardous materials spills and accidents caused by truck movements; • collisions involving trucks; • diesel emissions (impact on ait quality) derived from truck operations; truck hours of operation affecting peak period traffic flows; • noise and vibrations generated by trucks; • potentially negative impaccs on property values from truck activity; • lack of available truck parking and rest stops resulting in trucks parking on shoulders and along roads, causing potential safety concerns; • light and noise pollution generated by nighttime operations at loading docks and truck terminals; • potential new development on cxlsting truck terminal properties; • inadequate truck access to matitime and air cargo terminals affecting the competitiveness of these facilities; and • on adequate road geometries, turning radii, grades and turning lanes to accommodate trucks.

Rail ~mmunity i$ues includ.e: • facility shut or rail line abandoned, resulting in the atea being deprived of service and economic development opporrunity; • inadequate capacity to accommodate the rail freight needs of the area; • facility location impodes economic development goals; • hazardous materials spills and accidents resulting from rail freight operacion.s; • other land uses encroaching onto rail rights of M-y; • noise and vil:>ration.s resulting from train operation.s; • diesel emissions resulting from idling locomotives;

• lack of a buffer zone arOWld the rail yards; • wtdesirable odors from the rail yards; • light and noise pollution generated by nighttime operacion.s; • impact on property values along rail rights of M-Y from increased train activity and/or lack of maimenancc of right of way; l .t A alii A I

1'\.- Tn A ltrnr\nTITI,..• •

.,_. , , . , . , , . . , . , , • •

.,_, · - · - •

• inadequate truck access to rail yards; delays at at-grade crossings and resolving congestion and safety issues; • trespassing on rights of way/accidents; • confficts with commuter/passenger rail service on rights of way; and • train cars stored on rights of way near residences. Maritime cargo community issues include: • The -environment-
• Access co waterfront-access to the waterfront for nonfreight uses is increasingly desirable, particula rly in wban areas. This creates competition for land in these areas. Air cargo community issues inClude: • Hours of operation and noise-Air cargo operations rend to occur during nighttime hours. A:; a result, noise issues are more pronounced. • Truck traffic on access roads--Similar to maritime cargo facilities, the volume of truck traffic typically increases as cargo activity grows at an airport. • Theft and security-Criminal and terrorist activity can occur at air cargo operations.

2.1 State of the Industry/Existing Data Sources Documenting. the facilities, equipment and activities related to freight movement occurs at many levels with numerous existing data sources. At a national level, sources for volume and value of freight movement in the United States include the Federal Highway Administration (FHWA), U.S. Bureau of the Census, U.S. Bureau of Transp ortation Statistics, Sucface Transportation Board, the Army Corps of Engineers and the Energy Information Administration as shown in Exhibit 14-1. Internationally, many countries release similar documents quantifying freight movement data.

·~ ~*~. ~:~ ~· ~ ~ ·:·1_"':

· tr~~::~i~iilt?F-i.'.r~:t~idt~'!·~~,,~~J'.¥-~~:\~~t"

DataSet

Description

Volume/Value of Freight Movements

1

Freight Analysis Framework (FAF)

Commodity movements among states and metropolitan areas by value, weight and mode for 1997,2002, and forecastS for 2010 through 2035. www.ops.fhwa.dot.gov/freightlfreighc_analysis/faf

Commodity Flow Survey (CFS)

Origin/destination of commodities shipped in U.S. by mode, value and weight. Updated every 5 years. www.census.gov/econ/www/cfsnew.hrml

Transborder Surface Freight Dara

North American merchandise trade data by commodity and mode. Geographic detail for U.S. exports to and importS from Canada and Mexico. www.brs.gov/transborder

U.S.-Canada and U.S.-Mcxico Border Crossings

Number of trucks, uuck containers, train and rail concaine11 crossing imo the U.S. through land ports on U.S.-Canadian and U.S.-Mexican borders. www.bts.gov/programs/internarional/border_crossing_cntry_data

Carload Rail Waybill Sample

Origin/destination of commodities shipped by rail, weight, number of catS involved, length of haul. Data based on the proprietary Carload Waybill Sample of Class I rai110ads. www.stb.dot.gov/stb/industry/econ_waybillhrml

I

1

Domestic Watetbome Cornme.rce

TonJUge and uips by commodity for major poru/waruways and origin and destination data on warerborne cargo movements by waterways and harbors. www.ndc.iwr.usace.army.mill/wcsc/wcsc.hun#2008%20Waterborne%20 Commerce%20of%20tbe%20Unired%20Staces%20%28WCUS%29

FQreign Waterborne Commerce

Value and weight of cargo by type of service for U.S. waterborne imports and expom. www.iwr.usace.army.mill nddusforeign

Oil Pipeline

Oil movements by multistate regions. www.eia.doe.gov

Air Traffic Statistics

Air tnflic, connage and revenue ton-miles data for large air carriers and by airport. www.bcs.gov/prograrn.slairline_information

ln&astructw"C National Transportation Adas Database (NTAD)

Geosparial attributes of infrascrucrure for all modes and facilities. www.bts.gov/prograrns/geographic_information_services

Highway Performance and Monitoring System (HPMS)

Extent, condition, performance, use and operating characteristics of U.S. highways. www.thwa.dot.gov/pol.icy/obpilhpmslindex.cfm

National Highway Planning NetwOrk (NHPN)

Miles of current and planned roadways. www.lhwa.dot.gov/planninglnhpn

Railroad-Highway Grade Crossings Grade crossing location and safety data. Other rail network characteristics in NTAD. hrrp://safetydata.fra.dot.gov/Officeof5afety/ U.S. Pores and Waterway Facilities Database

Physical characteristics of coastal, Great Lakes and inland U.S. portS, terminals and locks. www.iwr.USi.ce.army.mil/ndc

Fteight Vehicles FAF Highway Capacity Database

Truck Bows at highway segment level for 1998 and forecasts for 2010 and 2020. www.ops.fhwa.dor.gov/freightl&eighr_analysis/faf

VTRIS-W

Number of trucks weighed and vehicle weight information by type of vehicle and highway functional class. www.fhwa.dot.gov/ohim/ohimvtis.cfm

Highway Statistics

State truck registrations, moror vehicle and motor carrier ux receiptS, and disposition of rax receipts. www.lhwa.dor.gov/policy/ohim/hs02/mv.htm

Commercial Motor Vehicle Safety Data

Commercial motor vehicle crashes, fatalities and injwies. www.fmcsa.dot.gov/faets-researchlfacts-6gureslanalysis-Statistics/cmvfacrs.hun

Waterborne Transportation Lines

Inventory of U.S. vesseJj moving waterborne commerce. www.iwr.wace.army.mill nddveslchar/veslch.ar.hun


I

·• · >[~~~ . _-. · -' ~iiL~cY~_tt""'•~ _ · ·•· ''•w~·· · ,_.. • ~r&'i.mt4t • It • 4

._. •

•

•

....... ~ \~t.;;~~~:~r~~~~-.:~'~'1~~~;~, .. ,.. ~ ~ .!~~':)~~,~~~, ,,.,/ .~.~:~'. '"':0..<.;•!'~'-- .. ~?~ , ;.;:~~~~::~·-

' Economy U.S. Economic Census

Economic
Regional Economic Accounts

Gross state product, personal income, population and employment at sruc: levd. www.bea.govlbealregionalldaca.htm

U.S. Census County Business Patterns

Economic activity at the county level by indwuy. www.census.gov/econ/cbp/iodc:x.htm!

Wages, Earnings, and Benefits

Data categoriz.ed by geographic area, occupation and indusay. www.bls.govlblslwagcs.hun

Productivity

Output per hour of labor; international comparisons included. www.bls.gov/blslproduccivity.hun

Compreheosi.-e Tabulations

Freight Facts and Figures

Volume and value of freight flows, network characteristics, economic, safety and energy usc data, and' environmental effects. www.ops.fhwa.dot.gov/frcight

National Transportation Statistics

Overview of the acent, condition and performance of U.S. uansporntion system. www.bcs.gov/publications/national_transporcation_stacistics

North American Transportation Statistics

Information on. transportation and related activities in Canada, the Unitc:d States and Mexico and berwccn the three counuics. www.brs.gov/publications/north amcrican_transportation_atlas_data__

-

--

Source: Adapted fcom FHWA, Freight Planning, www.fhwa.dot.gov/frcightplanning.

-

Numerous additional W.ca sees, from W.ca service bureaus or agencies specializing in the topic of inccrcsr, in addition to those listed in Exhibit 14-1, may be helpful for goods movement sturues. Affiliate state trucking association organizations also publish deWied data on activities in their regions, such as the largest manufacturing fuciliries, number s of vehicles by type and taxes paid, agriculrural p roduction and magnitude of freight movement by mode. Exhibit 14-2 presentS an example of the information presented on truck Aeet mixes from American Trucking Association {1987) . The Arizona F~t NetWOrk Analysis Decision Support System (Rul.wan, et al., 1988) was devdoped as a result of inCXJrporaring the &eight movement and highway carrier attitude survey findings intO the planning process, and is represented by Emibit 14-3. The major steps in the process include mOdal decisions, forecasting and simulations and management and strategic planning capabilities. Several effon:s have resulted in the development of databases relating truck trip generation to land use (Fischer and Han, 2001). TheauthoNhavecreatedasynthesis of available truck trip generation data so~m:es, data characteristics, ex:unples ofstudies and the current state of the practice. Data coUea:i9n was done in many cities in the United States, with the rcsulrs reported in proportion to acre2Se, employees, or 10,000 square feet (sq./fi:.). of development. An ex:unple of truck srop generation is sum.marized in Exhibit 14-4. Other trip generation data is available in the ITE Trip Gmntl~Um Handhook (ITE, 2008). Special activjcy centers have aucking ncc:ds that may be atypical, and standard trip generation rues or weight c:x.pcctacions may not apply. A study done by the Texas Transportation Institute (Middleton, 1990) used man-

,;11,.; .

WI

- ~~ ~cs·- swr

Flatbe d 45% Dump 11%

Tank 9%

Van 7%

Uvectock 6% Between 1982 end 1987, the number .of large Wyoming r~tstered trucks engaged Iii commerc:lal activities decreased by 4,000; this Is • 15% reduction trom t he 27,000 ve hic les registered In 1982.

Source: American Trucking Association and Wyoming Trucking .Association, Trudtingin Wyoming, 1987. Goods Movement Studies • 197

( J) MO":clo.l~CTOAY

(II FMI""T JOOYOCI
rJ.-Tc'm\'J.ON-=.1

'

N
~COUNT

OATA

Jr-.I...'W'~"::.

Source: Radwan, A. Essam, el. a!. A !Rcision Support Systnn for Frright T1111Up0rt in Ari.uma. Prepared for the Arlwna Department ofTransportation. Transjx>rution Research Record #1179, 1988. .

Source: Ad.aptt
: ual counts and poruble weigh-in-motion (WlM) scales co assess the chan.cteristics of truck mffic for timber, produce, ~ grain, beef cattle, limestone and sand/gravel. Interviews were used as a followup to quantify the typical conditions. ! 'Exhibit 14-5 presents the specified permissible weights and vehicle dimensions for the countries sucveyed (Nagl, 2007). The range of gross vehicle maximum weights ranged from 36 metric tons in Japan co 62.5 metric cons in Canada.

The Commociity Flow Survey, conducted on a 5-year cycle as part of the economic census by U.S. Cernus Bureau, provides data on the discribution of products by class and shipper group. A sample of manu&.ccuring establishmentS and a sample of their shipping documents are used to develop the database. The data elements in the survey include weight and value shipped by North American Industry Oassificarion System (NAlCS), dace, destination and means of transportation. The Annual Survey of Manufacturers, along with Cu17T1Jt Jntiustrial &ports and other publ icacions by the U.S. Census Bureau, provide measures of total sales, market share, quantity and cost of materials consumed, production hours and transportation modes and location information. Over 100,000 establishmentswere sampled during the 2007 Commo4ity Flow Survey (RITA, 2007). Reports generated include a series by industry, mode of uansporcation and distance, weight and commodity shipped.

Table 3.1: Truck weight and dimensions for selected coWltries 18

(The values for vehicles with extended dimensions are in parenthesis)

Australia

Brazil"' Canada

China Germany" India Japan Russia Sweden Thailand .UK USA:u

Permissible slnglt axle wdi'ht · . {tonn,_;f' 9.0 10.0 9.1 10.0 10.0 10.2 10.0 10.0 10.0 9.1 10.0 9.1

Permissible groSJ weight tractor+ uml-trnl/er

Mn:dmum "tllfde dimn1sfons fmetersT

letlglh

Height

ll'idth

{tonnes]

45.5 (125.2) 45.0 (74.0) 62.5 40.0 44.0 44.0 36.0 .44.0 44.0 (60.0) 37.4 44.0 36.3 (S9.45)

19.0 (53.5) 22.40 (30.00) 23.0 {38.1) 18.0 18.75 18.0 18.0 20.0 18.75 (25.25) 10.0 : 18.75 19.8 (35.20)

4.3 4.40 4.12 4.2 4.0 4.2 3.8 4.0 4.0 4.0 4.0 4.1

2.~

2.6 2.6 2 .~

2.6 2.7 2.5 2.5~

2.S5 2.~

2.S~

2.6

11

Sources: Asia: ESCAP (2002), p . 7f; Brazil: RESOLU<;A.O No 68, DE 23; Canada: Schulman (2003), p. 13f; Sweden, Gmnany: ECMI" (05-12..01): UK: "The Road Vehicles (Construction and Use) ResuJations 1986" and "The Road Vehicles (Authorised Weight) Regulations 1998". Europ~ regulations can be found in EC Directive 96153/EC; USA: Code of Federal Regulations Title 23 Part 6S8 (Re'
2.2 Route Studies R.'l.il and truck movement of freight are often affecred by similar facrors, and one area where rhis is rrue is in the routing of vehicles. For rail shipments, the ownership of certai n segments of the rail nerwork can influence rouring. For rruck shipmenrs, the designated truck nerworks, or resrcicted segments of a road nerwork, exert some influence. Trucks can comprise one-third of the traffic stream in ordinary situations on access controlled roadways or as much as half in severe cases. On arterials, the percencages are cypicaUy lower. The utilization of GPS receivers and GIS software can be useful for a route study. Laws and regulations regarding the maximum weighr and length specifications issued ar rhe national level, and adopted by many states, have resulted in some use of double and criple trailers. Concern for the infrastructure, roadway capacity, truck productivity, route conti nuity and safety of the motoring public led to the designarion of a Narional Truck Nerwork. This nerwork includes most limited-access facilities and a majority of major arterials and collector streets. The access rules for local roadways vary considerably amongjurisdicrions and can have a large impact on the size of the available truck nerwork. In some states the impact of route designations is simulated in com purer modeling software to determine the impact on the overall road nerwock. Such simulation srudies may be used to depict the rate at which nonessential truck traffic is discouraged from using streets in congesred areas, particularly in peak hours. With growing concern for air quality, such analyses may rise in importance and help answer such questions as: Will the increase in i:ruck vehicle miles traveled (VMT) create more or less pollution than the added congestion expecred without specified truck routes? When defining a srudy, it is necessary to understand the fleer composition of candidate vehicles. The Comprthmsiv~ Truck Siu and Wright Study (U.S. DOT, 2000) includes gross vehicle weight, amount and configuration of axles, maximum axle weight, pay load, length, width ofload, number of trailers, cab cype and overall truck type. TQpology, geom=ics and uuck atuibutes in an area must be assessed simultaneously when drawing conclusions about operating characteristics with respect to freight movement.

In Vancouver, British Columbia, Canada, truck route rules are applied only to vehicles with more than rwo axles or greater than 32,000 lb. gross vehicle weight (Swan, 1979), A test was run to compare the existing route nerwork to a highly concentrated scheme of four north-south and four east-west arterials. The evaluation looked at the additional cost to the trucking industry in terms of driving time increases, vehicle miles traveled and noise impact. The data collection dforc was an 0-D study, from which travel patterns were derived. For coding purposes the study area was divided up into 70 zones. The classifications used for the rest of the coding detailed the body types, gross vehicle weight, fuel cype, use of vehicle, land use classifications and commodity classifications. The Standard International Trade Classification of the United Nations provides the youpinS$ for motor transport traffic in Canada. A computer model was used to rest the original alrernatives and subsequent variations, using aU-or-nothing assignments. Route srudies are sensitive to roadway capacities and in particular the ability of trucks to navigate roadways and intersections. To determine the suitability of a particular roadway or intersection to handle trucks of different sizes, field observations may be made. A srudy conducted for FWHA (Hummer, er al., 1988) involved more than l ,100 trucks. The characteristics inventoried included number of legs in each app[oach, average lane widths by approa~, width of lane from which turn was made, average curb radius, signalization at intersection and phasing of signal, where appropriate. The data were then stratified based on truck cype (semi 40-ft., 45-ft., 48-ft. and double 28-ft.). Turn times were measured as well as encroachments and conflicts caused by truck maneuvers. Wisconsin applied the Caltrans turning template software, together with overhead photography and ground-level photo-triangulation, to determine the adequacy of various ramps and road segments to handle erucic traffic. Computer simulations were run for each srudy site and compared to field observations. Truclc-rdared factors examined included angle of turn, radius of rurn, tractor wheelbase, trailer lengrh, trailer width and axle width. Standard roadway geometries were examined. Modeling similar to that done for road nerworks can be done for railroads, with ownership being added as one of the critical variables for each link in the system. The impact of tail line abandonments and changes in ownership can be evaluated by modeling freight Bows and alternative path choices. The intermodal implications and highway issues are tied directly to the railroad component of the commodity movement.

298 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2NO EOrTION

; 2.3 Loading and Unloading Studies : T he movement of trucks through urban areas is essenrial for the pickup and delivery services provided, bur this san:e , movement is a component of the urban congestion. Further, much of the downtown freight loading and unloading rs handled curbside, adding yet another dimension to the importance of adequate facilities planning. At the local level, data collection is usually undertaken in response co the identification of goods movement problems, which can be grouped as presented in Exhibit 14-6, which cices operational, economic and external impact areas for downcowu. terminal and network applicatio~. These impacts can be influenced by local loading/unloading policies. Models for determining optimal on-street loading and unloading space distribution are available (Aiura and Taniguchi, 2005). To service businesses, on- and off-sueec loading spaces must be designated. How many spaces, what type of f;!ci~ ity and where they can be located must be determined. Ic may be necessary to impose rime restrictions. T he specificMtOn of those time ranges should be based on the local objectives, supported by an appropriate data coUcctio n effort. Truck uip generation rates, the result pf one type of study, are used to derive the number and type of spaces. As dlsc¢sed previously, these rates will vary by land use and category of esrablishmem. Pickup and delivery activities include at least the following steps: • finding the appropriate parcel if it is for a delivery; • locating the precise address for the interaction; • receiving the item if it is a pickup; • identifying the responsible person;

• walking between the truc15, a_nd the location; and

• cqmpkting the docwnentation.

Systems/Network Traffic counts with detailtd vehicle classification ______ __ ___

________ -- - .--

Operational

Speed and delay A. System segments B. Rail at-grade crossin~

C. Moveable

StreetS

Survey of shippers, receivers and

operators

I Survey of tennlnal operatOr$

Economic C. Water

D. Air

I

usmg volUme aaca

External Impaccs

. . Morutonng

A. Air B. Noise

I

swrounwng area

M . . oruronng

I

Monitoring A. Air_ B. NoiSe C W: . arer

Source: Gcndell, DavidS., et. al., Urban Goods Movnnmt Comidmttiom ;,. UrbiVI Transporratum Plannmg for the U.S. Deparunenc ofTransporcacion, Engineering Foundation Conference, 1974. Goods Movement Studies • 2~ ~

If the ddivery location is not located conveniently, the dwell time can be excessive and there can be a measurable impact on the adjacent road network. The typ~ of vehicles used for urban deliveries include everything from private passenger cars, taXis, pickup trucks and so on, through tractor-trailer combinations. According to a cenrral business district (CBD) study in Dallas, TX, USA (Christiansen, 1979), the majority of delivery vehicles are single-unit ttllcks (40 percent), with vans accounting for an additional 27 percent and passenger cars providing another 18 percent. Tractor-trailers were only 3 percent of the delivery vehicle population. Sample data collection forms are included as Exhibits 14-7 through 14-9 for individual trucks, single. loading docks, or curbside pickup and delivery operations. Srudies of single-truck operations are conducted ro determine such f.u:rors as trip-length characteristics and dwell times by land use type. The investigation of an alsting loading dock allows the derration of the number of trucks per day, peak period and peak-period percentage md dwell rimes, all for a given land use. Curbside studies also allow the determination of space requirements by land use.

Truck Daily log Sheet Date~--------------

Gty/ke:'---------------------- -----

Beginning Odometer Reading (Start of First Trip)_ _ __ _ __ __ _ __ __ __ _ __ __ _ Ending Odometer Reading (End of Last Trip)_ _ _ __ _ __ __ _ __ _ _ __ __ _ __

Stop Location

I

Arrival Time

Depanwe

I

Time

Notes

\

L

~

•

"AllNIIlt.l

ni=TAAN<;Pf'IRTI\Tif"'l~l t:Mr:IMCrCU~Ir: <''TJU'\Ir"l'"

"'"'n ,.,...,..,...'"',..

I

I

. · ~lli1iTit.B1·_.: •

..

0 I

'

Lo.di.ng Dock Operations Log Sheet Date

"··· ~

Location

Size of Buildin~

sq. ft.

und Use

Perccm of Occupied Space

.

Company Name

Tune In

Tune Out

Vehicle 'JYpe

Commodity Deliftmi

Type

Q.uanticy

Commod.ity Pic:lad-Up Q.uaaticy

Type

Not~

I

~

Commencs

Observer

Str=L----------------------= Company Name

I

A.rriftl TUDe

I

Depamue Tune

u

m

Da~----------------------------

~

Pickup and DdMry Operadoo.s Sheet· ~

i

i

__________________________________

Cicy/Aro

N o ta

-'',..r.····

.,.

~-

1~:3'1iTic!Il..'II~\Olll

... . .

•"..tl('

.

:..

.

""-·• D

--- ~~:" ~ ~~---~· :~>· ~·-·-.: -~-:~$i7~-- -~-:~_;~~~#.m~~:-~,

Percentage of Space Sizes LandU" Office ~tail

and Personal Services

~tail ( > 60,000 sq. ft.)

CommerciaUindusrrial HotcUMotd Food and Beverage Secvices

;

55 ft.

35ft.

20-25 ft.

-

40%

60%

60%

40%

25%

25%

50%

40%

60%

1 space

-

75% ·I space

25%

40%

60%

Source: Walters, Carol A., 1980. CBD DallaJ: A Cau Study in DnJekJpmmt ofUrban Goods Movnnnu Rrgula:Wns. Prepared for the City of Dallas, Texas, Office ofTranspornuion Programs, Dallas, Texas.

Past srudies indicate most pickup and delivery trucking activities occur between 6 a.m. and 6 p.m. (Christiansen, 1979). Activities tend to peak between 10 a.m. and 2 p.m. (Habib, 1980), with dwdl times ranging from 11.5 to almost 20 min. Ocher studies have supported these observations of a peak period between I0 a.m. and 2 p.m. with an additional afternoon peak around 5 p.m. (DDOT, 2004), with food and beverage ddivery services experiencing an average parl
2.4 Veh icle Weight Stud ies Data on truck weights are collected for many purposes, including pavement design, revenue estimates, motor carrier enforcement, highway cost allocation and other planning and engineering activities. The vehicle weights are reported by motor freight companies as pact of their reporting requirements; the recipient varies by state but is frequently the motor vehicle department through the registration process. Roadside weight checking is conducted with either permanent or portable scales, usually as pact of an enforcement program. Following the changes in trucking regulations in the 1980s and the resulting changes to fleer mixes, truck size and weight studies are often conducted to evaluate the impact on types of trucks being used, pavement and geometric requirements and changes in industry efficiencies. In WISCOnsin, the Department ofTransporration used statistical criteria to locate their weigh stations (Gardner, 1983). FHWA suggestS considerations for site selection to include avtnge daily traffic (AD1) volumes, percent trucks, percent trucks by type, p~t trucks by commodity, seasonal areas, int~tate versus inrrasrare trips, land use characteristics as related to trip origins and destinations and sire suitability, and nearby alternative routes (Wmfrey, 1976). 2. 4.1 Statie Mellnll"mUtlts

The use of static scales is required for the certification of truck weightS in the case of a cited violation. The accuracy of !hese scales, when calibrated, makes the data defensible, if necessary, in a coun of law. Typical Pc:rmanenr stations are often costly to construct, equip and operate. From a transportation network perspective, having to bring the truck ro a complete stop often causes backup and safety problems, as a weigh station may be prooessing many thousands of trucks in a single day. The data collected at permanent weigh stations is often biased, due to general knowledge of their location and operation. Oversized and/or overweight vehicles are suspected of circumventing the stations. As a result, weight enforcement programs often focus on known detour routes around weigh stations. 302 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2N O EOrrlON

....... ·•

-y~~-~,.;;~-~1\:~-~...>(

"'"'•'"'-'

-

.. -'

. ..

,_.,._

'. .

'

Technology

Quam Piewdecrric Sensor

Performance

Can Meet

·-

.. -. ...

-.,o~l

. --

Polymeric and Ceramic Piezoelectric

-· ..~~,... · ~,

~~~·~ri~~'0~;· ,,,_.~ . - ;,:~·

~.: ·.. ~~1• '• ~~'~'

Bending Plates

·.',.J,' - .

~.~~_r _o_ ;·~~~~i~:~: ::!~:rl ... .~~;f~ r:t~

Loadulls

BridgeWIM

Not At:ceptable- Can Meet Temperature Enforcement Sensitive Requirements

Can Meet

More Researcll Needed to Verify Accuracy

Snull Road Cuts I Day to Complete

Small Road Cuis I Day to Complete

Significant Road Cut.with Proper Drainage Required. Multiple Days to Complete

Maintenance

Must maintain surface smoorhness and seal properly . to achieve satisfaCtOry performance

Musrmaint:ain surface smoothness and seal properly to achieve satisfactory performance

Inadequate Corrosion ofload cell if nor scaled drunage can cause bending correctly p~ tO come out of rhe roadway. Required 6-month checks and annual inroad in.specrion

Minimal

Safety Issues

One Day for System Insu.Uation and During Periods ofin-R.oad Maintenance

One Day for System Installation and During Periods ofln-Road Maintenance

Significant safety issue if bending plate comes our of the roadway. Multiple Day System In.scallation and During Periods ofln-ROad Mainre~ance

Multiple Days for System Installation and During Periods ofln-ROad Maintenance

None

Cost of System Low-Medium Including Installation Cosr

Low Cost

Medium-!:Ugh Cost

High Cost

Low Cost

-

Marure I Proven Technology

Yes

Yes

Yes

Not in United S12res

-

Senso~

Enforcement ~uiremencs

Enforcement ~uirements

wilh 2 rows of sensors. Bcuer accuracy can be achieved wirh 3 rows rhrough averaging out of vehicle dynamics In.sta!lacion

Yes

Significant Road Cut wich

Proper Drunage Required. Multiple Days to Complete

-

Non -Intrusive ln.strumenr:a.tioo·

-

Source: Connecticut Academy of Science and Engineering.

2.4.2 Weigb-in-Moi'Um WlM is usd in many areas to detesmine if a truclc is traveling within a reas9nable range of the legal limits as adopted ~ a local government. The dara collected may include all or some of the following: gross vehicle weight, axle weight anC:f. tandem axle weight. In the United Srates, local and federal departments of tranSportation or other responsible agmcie~ monitors rhe data collectt
2.342

.7

57

KEY: - Repr=nts ari estimate equal to zero or less than 1 unit of measure. S = Estimate does not meet publicacion standa.tds because ofhigh wnpling variability or poor response quality. ~Truck"

as a single mode includes shipments tharwae made by ooly private truck, only for-hire uuck, or a combination of prmu: and for-hire uuck. 2 Escimates for Diodine e:zclude shipments of crude

1

Sou.rc.c: U.S. J)qlartment ofTransponacion, Raearch and Innovative Technology Adm.i.n.istratioo, Bureau ofTran.sportacioo Statistics and U.S. Census Bureau. 2002 Commodity Flow Surwy. H.u.!mt Daca. Table la. December 2004. .

.

304 • MANUAL OF TRANSPORTATION Hlr.IMHO,.II': e-n"''"'

~""

,..,....,,.,.,

(Shipments should be routed to optimize r.raruit time and reliability, minimize accident probability and minimize the
w

A risk assessment is the nc:xt step in the process. Data from Materials Transportation Board and local crash records arc valuable f
3.0 REFERENCES 3.1 Literature References Aiura, N. and E. Taniguchi. •pbnP.ing On-street Loading-unloading Spaces Considering the Behavior of Pick-up Delivery Vehicles." ]DIITIIIIl Dftht E.astnn AriA Sti~IJfor TtrtnspDTt41itm Snu/Us 6 (2005): 2,963-2,974. American Trucking Association, Wyoming Truclcing Association. Trudang in lr)ooming. Burlington, MA: American Trucking Ass~tion and Wyoming Trucking Association, 1987.

Bureau ofTransportation Statistics. 2002 EctmtJmic Cnuus- TnznsportllliM: H4Z4rdous Maurials. Washington, DC: Bureau of Transportation Statistics. U.S. Department ofTransponation, 2004. Christiansen, D. Urban Tnmsportatitm Planningfor Goods and Smtiar. Washington, DC: Fed=! Highway Administration, 1979.

Christiansen, D. DttJJ.u CBD C.txis 11Ni Smtices Dimibu:Um ~jm. Dallas, TX: Gcy of Dallas Office ofTransportation Programs, 1980.

DC (District.of Columbia) Department ofTransportation. Di.ttrict DfColumbia Mbt11r Ctrrin- Mmwgrmmt and Thro# Ass~t . St...ty. Washington, DC: Volpe National Transportation Syscans wccr: U.s; DepartDWtt ofTranspon:uion, 2004. Federal Highway Admittisaation. Frright A1ralyrit ~ft. Washington, DC: Federal Highway Administration. Freight Management and Operations, November 19, 2008. http://ops.fhwa.dor.gov/freightlfreight_analysis/faf/index.hun. Reuieved January 5, 2009. Federal Highway Administration. Fmtln ~t and OpmuiDm. Washington, DC: Federal Highway Administration. http://ops.lhwa.dot.gov/freightlindcx.cfm. Retrieved January 7, 2009. Federal Motor Carrier Wecy Administration TIN NiUilm41 H~lr4n~Dws Maurials Rouu &gistry. Washington, DC: Federal Motor Carrier Safety Administration, June 10,1009. www.fmcsa.dot.gov/safery-security/l=mat/oational-hnmat-rout~aspL FISCher, M. ]. ar~d M. Han. NCHRP Synthesis 29S: T~!t Trip GmmzliJJn. Wa,dllngron, DC: Transporration Raearc:h Board, 2001.

Gardner, W. D. "Truck Weight Study Sampling Plan in WISCOruin." Trrzmporr.tllilm kuTch &ami: foliTIIIIl Dftht T~" ~tmh IJolfnJ 920 (1983): 12-18. Gcndell, D. S., ct al. Urb1t11 Gotxis Mtn~m~mt C4mi4n#Wm ill Urban Trtl11Sport41Wn Pl4nning Studin, FHWA 32-01-23 RFP 397. Washington, DC: U.S. Department ofTransportation, Engineering Foundation Conference:, 1974.

Habib, P. "Transportation System Management Opcions for Downcown Curbside Pickup and Delivery of Freighr. "Tramporrarion &search Record: Joumal ofth~ TTl111!porration <arrh Board 758 (1980): 63-69. Hummer, J., C. Zegccr, and E Hanscom "Effects ofTurns by Larger Trucks at Urban lncersecrions." Transponarion &srarch Rmrd:joumal oftht Tmns~rttttion &uarrh Board 1195 (1988); 64-74. lnstiwre ofTransportation Engineers. Trip Gmtration, 8th ed. Washington, DC: ITE, 2008. lnstiture ofTransponation Engineers Technical Commlrree 5D-6. Trucking ~ighing-in Momm, Informational Repon. Washington, DC: ITE. 1986: p. 7. McMillen, R., M. Anderson, and C. Cerbone. Traf!U and Transportation Analyris: H=r®us Wastt Disposal Facility. How ron, TX: Nacional Conference on Hazardous Wastes and Environmencal Emergencies, 1984. Middleron, D. Rnults ofSptcial-Ust Truclt Dlllll CollNrion, FHWArfX-420-3F. Washington, DC: U.S. Pcpanment of Transportacion, Federal Highway Admlnimarion, 1990. Nagl, P. Langer Combination Vth~ks (LCV) for Ana and tf1t Padfo &gion: ~=Economic lmplicalions. Bangkok, Thailand: United Nations Economic and Social Commission for Asia and the Pacific, 2007. Pines, D. and C. Fang. A Study ofWtigh Swion Ttchnolot,Us and Practim. Hanford, CT: Con.necticut Academy of Science and Engineering. 2008. Radwan, A. E., J. Cochran and M. Farris. A Decision Support Sysum for Freight Tramport in Ari.7Jlna 1179 (1988): 2}-30. Radwan, A. E., M. Rahman and S. Kalevcla. Freight Flow and Attitudinal Survey for Arizona. Transportation Reuarr:h R«~rd: journal oftht Transportation Restarch Board 1179 (1988): 16-22. Research and Innovative Technology Administration. Commodity F/Qw Survey. Washington, DC: Research and Innovative Technology Adminisuation, Bu=u of Transportation Sratisrics, U.S. Census Bureau, U.S. Depanment of Commerce, 2007. Stammer, R., C. Wright and J. Donaldson. "Conducting Truck Routing Studies from a New Perspective.• Transportalion Res~arch Record:,]oumal ofthe Transportation Restarrh Board 1038 (1985): 59-63. Scrauss-Weider, A. NCHRP Repon 320: Jn~grating Frtight Facilities and Operations with Community Goah. Washington. DC: Transportation Research Board., 2003. Swan Wooster Engintering, Co. Led. Evaluation ofUrban Trucking &aionalization in Vizncouvtr. Phases/ and 2, W>L 5. Montreal, Quebec: Urban Transportation Research Branch of Canadian Sumce Transportation Administtacion: pp. 2-15. Transportation Consulting, MDA,lnc., d/b/a Street Smans, Duluth, GA. 1992. Tnnsponation Research Boatd. Mtasuring Pmonal Tra~l and Goods Movtmm~ Special Report 11277. Washington, DC: Transponation R=aJch Boud, 2003. U.S. Census Bureau. Econam~ Cnuus. Washington, DC: U.S. Depattmenc of Commerce. Bureau of the Census, 2007. U.S. Department ofTranspomtion. Comprthmsivt Truck Sizt and weight Study. Washington, DC: United Stares Depattment of Transportation, 2000. Walrers, C. A. CBD Dallas: A Cast Study in Dtvtlopmmt ofUrban Goods Movtmmt Regulations. Dallas, TX: City of Dallas Office ofTransportation Programs, 1980.

Winfrey, R., P. D. Howdl, and P. M. KenL Truck Traffic Volume and Wnght Dara for 1971 and Thtir Evaluation. Washington, DC: Federal Highway Administration, 1976.


\3.2 Onli ne Resources (Available as of January 5, 2010) ·Border Crossing Data ~- rranstats.bts. gov/BorderCrossing.aspx Carload Waybill Sample www.stb.dor.gcw/srblindusrry/econ_waybill.hrml Censu~ervice Annual Survey www.census.gov/econ/www/servmenu.hunl

Oass 1 and Rfgional Rail Traffic Data www.aar.orgf~/media/AAR/Indusuy%20Info/Statisrics.ashx

Commodity Flow Survey-RlTA www.bts.gov/progranulcommodity_Row_survey Economic Census w\vw.ccnsus.gov/indcx.hunl Freight Analysu Framework www.ops.fhwa.dor.gov/frcightlfreight_analysis/fof/index.htm Freight impactS on air quality W'f'W. fhwa.dot.gov/ cnvironmcntl&eightaqfindex.hun

North American Transbordcr Freight Dar:a www.brs.gov/programs/inrernational/rransborder/sea.rch.hunl RITA. Data and Statistics www.brs.gov/programs/frcight_aansporration/ Urban Goods and Intercity Freight Movement www.wsdoLwa.gew/R=arch/~ports/300/373.l.htm

Waterborne Commerce Statistics www.iwr.usacc.arrny.mil/nddwcsc/wcsc.hun Waterborne Trade Statistics www.rnarad.dot.gov/library_lmding_page/daca_and...sratistics/Data_and_Scatistics.han

3.3 Other Resources Bolger, F. and H. Bruck. Urban Goods Movnnmt Projms and D414 Scurr:er. Washington, DC: U.S. Department of Transportation, Office of Systems Analysis md Information, 1973. Carey, D., H. Mahm:wani, and G. Toft. "Air Freight Usage Patterns ofTechnology-Based Industries.• Tran.rportation <arriJ &corri:]oumalofrhe Trrz~n Rnrarr:b Boarri]oumal oftht T~n Rtsurch &tmi 1179 (1988): 33-39. Christiansen, D. Urban TranportarUm Planningfor Goods and Smlices. Washington, DC: Federal Highway Administration, 1979. Chrinianscn, D. Da/Jas CBD Goods and Sn-Wer DimiburUm Projta. Dallas TX: City of Dallas, Office ofTransportat:ion Programs. Federal Highway Administration. Chicago Area Transportation Srudy, 1991: Operation Grtm/Jgh~ Freight MovemtnJS and Urb47¢< Cmgtstion in th~ Chicago Arra. Washington, DC= FHWA, 1991: pp. 9-10. Hanscom, F. Trtljjic Optrrztions, ~£ II: The Effict ofTrw:lt Siu and Weight on .kcidmtExpnim« and Traffic Optrrztions. FHWA/ RD-80-136. Washington, DC: U.S. Depasunent ofTransporr:ation, Federation Highway Administration, 1981. ' Goods Movement Studies • 307

Saito, M. and T. G. Jin. E~~t~fuating rh~ At:curacy Lewl ofTrudt Traffic Data on Stat~ HighwllJS. Provo, liT: Brigham Young Univcrsiry, 2009. Teal, R. Estim.aJing tiN P,(J Economic Costs ofTruclt lnaamts on Urban FruwllJS. Falls Church, VA; AAA Foundation for Traffic ~ery,l988.

TEE Consulting Services, Inc. FraT1UWOr!t for Urban Goods Movnnml lnfomliltion in Catimia. Vol. 9. Urban Goods Movement Rcpon Series. Montreal, Quebec: Urban Transportation Research Branch, Transpon Canada, 1979. U.S. Department ofTransporution. Urban Goods Movnnmt Input to NawMI Trrmtporr41wn Ptm.. FiMf &porr. Washington, DC: U.S. Department ofTransportation, Office of the Secretary, 1976.

-

'- • AUI I A I f"U:'TOf't .. t ('M/"'\DTATI/"'1111 r""tl"" l .. lrrnu.tr I"TIIf"'\11'"1"

'"'t •tn ,.,..,,...... ,,.H,t

Chapte r 15

......................... ........................................................... '

Inventories OriginAl by: H. DougfM Robertson, Ph.D., P.E.

EJiuJby: Dtmulj. FUu/Jey, P.E. 1.0 INTRODUCTION

309

1. 1 Word of caution

310

7.2 Purpose of the Inventory

310

1.3 Choice of Data Elements

311

2.0 STRUCTURE OF THE INVENTORY

3.0

311

2.2 Inventory Location Reference Systems

316

2.3 Inventory Qassification

316

2.4 Access to the Inventory

317

2.5 Storage

317

2.6 Retrieval

317

ESTABLISHING AN INVENTORY

317

3. 7 Step 1: Determine the Purpose of the Inventory

317

3.2 St,ep 2: Select the Data Elements to Be Colle~ed

318

3.3 Step 3: Select a Data Collection Technique

318

3.4 Step 4: Prepare a Data Collection Plan 3.5 Step 5: Collect the Data

4.0

311

2.1 Means of Recording and Displaying Inventories

318 i 320

3.6 Step 6: Construct the Database and Data Displays

320

MAINTAINING AN INVENTORY

320

4.1 Software

320

4.2 FrequencyofUpdates

5.0 REFERENCES

;

321 321

1.0 INTRODUCTION n inventory is a catalog. lisring. accounting, record, or display of &ctua1 information that describes aisting con
A

lnH.c.nttvioc

a

~

• regulations, laws and ordinances • roadway and roadside • intersection • traffic control devices (TCDs) (such as signs, signals and markings) • parking • lighting • transit routes •

traffic generators

• wiiing (fiber optic, copper, overhead detection)

A general procedure for conducting street and highway inventories is presented in this chapter. Discussions pertaining to performing ocher types of inventories can be found in the chapters devoted co those subject areas. Additional information on inventories is available in the Trrmsport41ion PlAnning Handbook (ITE, 2009), the Traffic Enginming Handbook (ITE, 2009) and NCHRP Report 437 Colkction and Pmrot4tion of Roadway lnvrotory Dat4 (Karimi, Hummer and Khattak, 2000).

1.1 Word of Caution Inventories can be useful and productive tools for the traffic engineer and planner. However, inventories may also prove to be cosdy, unreliable, difficult to maintain and cumbersome co access. A successful inventory has a cleai!y seated purpose, provides useful and needed information, is easy to access and extract data &om and can be kept current at a reasonable level of effort and cost. The significant effort and cost of an inventory includes the planning and execution of the study, as well as a wdl-designed database ro store, update and query the data. If these criteria an: not met, it is likdy the money and effon expended to produce the initial inventory has been wasted.

1.2 ~urpose of the Inventory Before an agency conduces an inventory, it must addicss several important questions, including; • How will the inventory be used? • What specific information and/ or data will serve the purpose of the inventory?

• Can those data be obtained more effectively by means other than an inventory? • Does the informacion ilieady exist in another form? • How large is the srudy uca and what will be the extent of the effon to complete the invenrory? • Who will collect, enter and analyz.e the dua, and what is their level of expertise? Agencies must identify the intended uses of inventory information so they can collect relevant dara in a suitable form. For example, if the inventory is to be used co schedule maintenance, a sec of data elements should be identified that relate to condition and service life. On the other hand, if the inventory is to be used co crack maintenance costs, elements related to time, labor, equipment and materials will be required. Some typical uses of inventory data ue: • illustrate street classifications • locate TCDs • specify condition o~ service life of devices 310 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

• schedule maintenance • manage costs • depict application oflaws and ordinances • assist in evaluation of traffic operations • provide baseline conditions for use in other studies • locate traffic generators • compare measures of effectiveness with before and after conditions • track deterioration and service life of transportation assetS Of course, the inventory may sen•e more than one purpose. Agency or governmem accounting procedures might require the storage of cer~ inventory characceriscics for the calculation of infiascrucrure costs. Agencies must avoid collecting data dements simply because they are there and might be useful in the future. If after careful srudy :an agency cannot identify a use for a data element, the element is nor needed. Many inventory efforts fail because dJ.ey attempt to collect too much information.

1.3 Choice of Data Elements The choice of data elements is crucial to che success of che inventory. Daca elements are often chosen on che basis £h:at

having those data in che inventory Will preclude a visir to the field co obtain che information at a later time. Agencies m!J-St weigh th.i.s savings against how frequendy' and quickly they will need the data dement and whecher a field trip would be 'necessa.ry anyway to collect or verify other data. Occasional field trips may be more cost-effective than mainwruo.g certain data elements in an inventory. Inventory data collection and maintenance are costly. Although computers a$e valuable and necessary tools for coostructing inventories, they are a4o associated with the possibility that coo muchcfat:a are collected and the pracrical utility of the inventory is compromised. Extensive amounts of electronic data can be difficult £O format, update and access. The fundamental rule is: Ketp it simpk and to th( point. Elements that are crirital a!J.d are used frequently should be inventoried. Ocher data dements should only be collected "as needed."

2.0 STRUCfURE OF THE INVENTORY Inventories are usually unique to the agency or jurisdiction chat cond11crs them. While two inventories may have coif).mon purposes and data dements, the StrUcture of the inventories will almost always be different because of differences iP the size and layout of the street system, the equipment and resources available to conduct the inventory and local policieS •

2.1 Means of Recording and Displaying Inventories Agencies can record and display inventories in a number of ways. Hard-copy forms, carcls, maps, graphs, cables, digicaJ. photographs and vide? logs are typical means of recording inventory data. Any of these means m.ay suit the pwpose5 of a given agency, particularly the smaller ones. Today, many agencies use computer files and databases instead ofhardcopy forms and cards. From these files, analysts can easily son the data and produce listings, rabies, graphs, computergenerated maps and interaa:ive Internet-based maps. Exhibits 15-1 through 15-5 illuscrate various ways of prescncin~ inventory data.

Inventories • 311

~" ~

.•• •• .••• .

.

.• -:-:~t'!::....:~!!!'::~-:...~-.- .

.• •=

~

'''

I'

'' I I

' I I I

t

' I I

~:t!tt I I I I I I

' I I I I I I

I

!

LEGEND:

======= .

30 MPH

• • • • • • • • 35 MPH - - - 40MPH - - - · 45 MPH Sour= Bor. and Oppcnlandcr, 1976.

314 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDmON

"" 0

~

~

~

;

" . ~ ~

~ .....

·1\' · ~' q4 1

•i

i

"" "" '"!" .... >-I 9

. <>

r .•

rJ

(:

0

0

0

0

"'

"'

~

"'

•§ ~ ~

-+

i

e I a

i

-

.· • :

:

tS

~~ @l[·,gl~, ~w 0~

•

-

~ ·

· ri ~

-

. .

----

0

0

0 (X)

0

2.2 Inventory Location Reference Systems Almost all uaruportation inventories use some geographical location system. Next to sc:lc:cting the datll dements, the system chosen to locate the data dements is the most important ingredient in an inventory. Location systems range from simple references using known points (such as the speed limit ·on Main St=t from Oak to Pine is 35 mph (56 km/h) to more complex direction and distance schemes. An ccarnple ofa more complex scheme is when each traffic warning sign may be located by direction and distance from a reference point, such as a milepost or intersection {for example, a •siGNAL AHFAD" sign is located on the: south side of Maple Street, 375 ft. west of the interseaion with River Road). In manual synems, data are filed alphabetically by major Street and then direaionally by link (section) and node (intcrsc:aion) along tbe major street. Locations by strett name work relatively wdJ in a small hard-o:>py system, but agencies often convert the: names to numeric or alphanumeric codes when entering them into a compurer database: system. A popular computer tool is a geographical information system (GIS), which provides easy-to-access databases for locating, visualizing and storing inventory data dements. The rypes of location reference systems include the following georeferencing technologies (Karimi, Hummer and Khattak, 2000): • satellite-based global positioning system (GPS) receivers (handheld and vehicle mounted) • distllnce measuring instrument • inertial navigation system • rangcfinders

•

ground~

• remote sensing

2.3 Inventory Classification Numeric or alphanumeric codes allow the computer to search for and/or sort data clements quickly. The codes relate to the street network using a link-node system (McShane and Roess, 1990). Exhibit 15-6 illustrates part of such a system. In the sample link-node coding system shown in Exhibit 15-6, each street has a unique two-digit number. Thus each intersection is identified by combining the east-west (EW) street code with the north-south (NS) street code (always in that order). For example, node (intcrsc:ction) A is coded 3467. Node B (a five-leg intersection) is coded 3369, using only the codes of the two major streets. A separate data dement called numkr of 11ppro"hts idencifi~ the intersection as having five legs. Such an element also identifies three-leg intersections, such as node 3468. Linh arc identified by the codes for the nodes at each end of the link For c:xample, link Cis 33663466. If it is desirable to identify the link by direction, the order of the node codes would indicate the' direction. For a ample, link C northbound would be 34663366, and link C southbound would be 33663466. A link may span several intersections (that is, the span of t...S IIWJftBft ............ sn;grs three EW links along street 35 would " '·" IMAOC»CAL a"D.Bn'l be CX?
2.4 Access to the Inventory !

I

. The importance of accxss to inventory data has already been mentioned. The means of accessing data and information . must be a primary consideration when establishing any inventory. The method ofstorage and access must be tailored to the frequency and type of usc to be made of the data. The cost ofsofrwate licenses can affect the total price of che inventory and should be considered carefully when selecting a sofrwate package and when considering how many access poincs to the invenrory need to be provided to users. Sofrware training expenses and the technica,l expertise required for analysis should also ~ considered.

I

il

I:

2.5 Storage Inventories ace stored either manually or in a computer. Manual systems may be in the form of maps, card files, file folders, boUlld listings, ~cro6cb.e, photologs, or videos. These systems are kept in 6ling cabinets or on boo kshelves. They work reasonably well with smaller inventories, but larger systems become unwieldy and consume space. Computer inventories may be stored on hard drives, computer storage media, or on a server. Computer systems consume less space than manual systems do, so they are better- suited to large inventory systems, including picrures and vi4eo. However, computer systems must be appropriately archived on a regular basis to reduce the potential loss of valuable data in the event the operating system crashes, 6les become corrupted, or users unintentionally delete data. The use ofshared systems across departments can provide a reduction in storage costs.

2.6 Retrieval Nearly all inventory systems are stored by location (that is, intersection, street segment, or GPS coordinates). Therefore, one musc know the location of incerest to access inventory data on that location. In some manual systems, all inventory data for a given location may be kept in the same 6le. Other manual systems store data first by type of inventory and second by location. Computer data systems may be organized in a similar fashion. These systems are capable of searching and sorting data by data elements other than location. For example, if there is a need to examine the installation dates of all stop signs in the system for maintenance purposes, each intersection in the manual :file would have to be checked to see if stop signs were present. In the computer system, the computer seeks out stop sign locations. Additionally, the computer could sort and retrieve only those intersections with stop sign installation dates prior to a given cutoff date, thus saving both time and effort. The comp11ter can also generate repons or listings containing only the information desired and therefore avoids the need for manual extraction of data. This capability applies to periodic replacement ofdevices, spare part inventory control, stock replacement and ordering, dispatch of work crews, maintenance of history files and possible liabair:y defense. In addition, work orders may be generated automatically.• Agencies typically design work order formats to enhance inventory updating upon completion of the work. ·

3.0 ESTABLISHING AN INVENTORY The following is a suggested step-by-step procedure for setting up an initi.al inventory. Not all of the steps apply to every type of inventory and thus some steps may be omitted. Careful thought and planning before starting data collection should lead to a useful and cost-effective invenrory. ·

3.1 Step 1: Determine the Purpose of the Inventory The purpose(s) of the inventory will determine the applicable steps to be followed and should result in an inventory that best meets the agency's needs at the least expenditure of resources. Review the beginning of the chapter for typkal uses of inventory data.

3.2 Step 2: Select the Data Elements to Be Collected Remember the cautions mentioned earlier about collecting the data elementS that will be used and omitting those that are just "nice to have. • The following list illustrates the type: of data clementS that may be collected for a street or highway segment. Typical data elements for other types of inventories are described elsew~ere in this manual (such as inventories related to collisions, goods movement and roadway lighting). • Unique link identification by name or number • Classification of street or highway • Number oflanes by direction • Width of lanes • Parking conditions • Speed limit • Pavement markings by type of material and condition • Traffic signals • Signs by type:, location and condition • Bus stop locations • Bicycle routes • Sidewalks • Transit roures • Lighting by type and location • Driveway entrances by location • Drop inlet location and condition • Adjacent land use

• Length of the segment • Date of last information update

3.3 Step 3: Select a Data Collection Technique Data collection techniques refer to the methods used to record inventory data elements. Observers may record the dements dU:ectly and in the fidd, or they may record video of the dements in the field and then record the data in the office. Once a technique is sdeaed, agencies can identify and obtain the equipment needed to record the inventory data dements: The equipment niay range from a map, aerial photography, data forms, pencils and voice recorder to a specially equipped vehicle with cameras, video recorders, distanco-measuring devices, laptop computerS and even bicycles with handheld GPS receivers. The choice of technique is a function of available budget, time, and personnel. NCHRP Synthesis ofHif!!way ~ 157 presents a detailed discussion of data collection techniques for sign inventories (Cunard, 1990). The cost-effectiveness of alternative sign inventory procedures is discussed in a report by FHWA (Datta er al., 1985).

3.4 Step 4: Prepare a Data Collection Plan The data collection plan should specify how the agency will collect the inventory data. The plan should contain in detail the equipment needed, the time during wruch data will be gathered, exactly how each element is to be measured 318 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDffiON

and recorded, how ro handle unusual ciccumsranccs and che rules ro follow rhar wiU ensure consisrency in measure:· · menrs and dara recording. If agencies collect dara manually, rhey may design forms or program laptop compu;us so ·.observers can record data in the field directly. These actions preclude the daca reduction ~teps required in the office if ·video is used. Exhibit 15-7 shows a sample of a sign inventory data collection form. ·,· "'';

Sign Inventory Sheet Date

City

Street

between Location

l;-

~ ~

~~

Size

and Condition

.

~

. ?: ·;;

~

~I I

~ .=f!ef! ;;;"8~ -£-c t! ~c: "EijV.:~~ .. ~l;~~;. ~~ ~c o!1., o."D 8~E c...,.,:::."'"'u"ii2"E f:~i!l:!l'§~of.c"il ~ ~ ~-~~ ~... ~ 8 g v ~~~Vi§::i!~ X~~ a:.. .§. z::>E z::>E

-

0

41

a:

\1

4'

u

-

::r

-----------------

Post >

2:~

~

Sign Type

Direction,_ __ __ __ _

z

z

l;

~

....

l;

"0

~

· "0

~

~I

Notes

""c: 1: 1 "'~ 0

z

..

Q. IJ

a:

Inventories • 31 9

3.5 Step 5: Collect the Data Agencies should schedule clara collection during good weather to enhance accuracy and completeness. Ir is possible, howC{cr, to condua inventories, particularly manual studies, during less than ideal weather conditions when other more weather-sensitive data collection is impossible. The data collection team should we. a cheddist such as Exhibit 1-1 \:o ensure it has all of the necessary equipment and materials prior to leaving the office. The team should note any unwual circumstances that arise during data collection and discuss them with the supervisor by cell phone or upon returning from the field. Although it is possible to conduct field inventories with one person, the preferable practice is to work in teams of rwo or more. Such teams allow for quality control as data are recorded and provide fur greater safety in that one person can drive while another observes and records. Specifications and definitions should be written for each element being collected. These instructions will create consistency among variow data collectors and among time periods. Managers should conduct a quality control review on a sample of the data. This quality control review is particularly weful at the beginning of data collection to conect any mistaka that might be discove.red on future data collection effortS.

3.6 Step 6: Construct the Database and Data Displays The means of recording data in the field dicrate the specific tasks needed to place the data in a usable format. If agen· cies record inventory data elements manually. they mwt transfer the data from the collection funris to maps or tables or computer files. If the data are recorded directly in a computer or other automated device in the field, computer software can handle the data reduction and formarting tasks with relative case. The sofrware should contain systematic error-check routines to improve the quality of the data. Agencies need to check their computers and softy
4.0 MAINTAINING AN INVENTORY Keeping an inventory current is critical to its wefulness. ~ an inventory becomes outdated, its utility deteriorates rapidly. The success of an inventory is directly proportional to an agency's commitment to maintaining it.

4.1 Software Numerow computer software packages are available to aid in both establishing and maintaining traffic and highway inventories. Tcaffic maintenance software focuses on pavement markings, signs, signals and street lighting. Highway maintenance software encompasses a road inventory system of highway and bridge segments. Typical data elements in highway maintenance software include pavement, culvert, guardrail, road and sign elements. The strategy evaluation capabilities QfRoadSoft GIS (Michigan Technological University, 2009) can provide the ability to anal}'le the impact of policy and fun~g decisions on the roadway network. This analysis can provide planning and reporting details for maintenance spending over the lifetime of roadway assets. ~cral factors need co be considered based on the specific needs of the agency. The number of users who need access to the sofrware will affect the cost of software licenses; this should be considered in the total price. The synchroniza-

tion of the software files must be considered if multiple users plan to si.multaneowly edit the files. Software data must be appropriately archived on a ~ basis.

4.2 Frequency of Updates i The frequency with which agencies must update inventories depends on how prone the data dements are to change. ·..The issue is complicated by dements with differing service lives, damage inflicred by collisions and incidents and replacement necessitated by changes exre.rnal to the facility itsd£ Such changes may consist of land use growth and development. safety and/or operational improvements and shifts in rcavel patterns and demands. There arc two basic approaches to inventory updating. Perhaps the mosr prudent and cost-effective method is continuous updating. The simple concept is that once an agency establishes an invemory, it is promptly updated each time there is a change, whether that change is an addition, deletion, modification, or replacemenr. This approach sounds easy and very logical bur requires disciplined anencion to detail ro implemenr properly. Computer software has made this approach nuly feasible. Not only can agencies enter updares quickly and easily, bur the software enables fast and convenient examinations of the database to check the accucacy of its contents and ensure changes are being recorded promptly. Continuous updating is most appropriate for inventories that are used frequently and thus require a greater degree of accucacy. The second basic method is periodic updating. As the n~e indicates, the agency conducts a partial or complete fic:ld inventory at specified intervals and records changes that have occurred since the lasr update. The interval length depends on the level of accuracy required and the frequency of changes experienced by the inventory in question . Periodic updating is most appropriate for inventories that have a low usage level, experience rdatively few changes and require a lower degree of accuracy.

5.0 REFERENCES Cunard, R.. A. NCHRP Synthesis·of Highway Practice 157: M.Untmanc~ MaMgmsmt DfStrut and Hifbway Signs. Washington, D~: Trw.sportar:ion R=an:h Board, Nar:ionil R.c::search Council, 1990: pp. 28-43. Dana, T. K, K H. Tsuehiyarna, and K S. Opida. CDn-Effictiw lnvmtJJry Promium for Highw4J lnvm~~~ry D
Edwards, J. D. Transportation Planning HandbDolt. Chapter 11. Englewood Cliffs, NJ: Prentice Hall, 1991. Institute ofTransportation Engineers. Traffic Engin~mng Han4JJDD!t, 6th ed. Washington, DC: ITE, 2009. Institute ofTranspottation Engineers. TranspDrtation Planning Handhoolt, 3rd ed. Washington, DC: ITE, 2009. Karimi, H. A., J. E. Hummet and A. J. Khattak. NCHRP R.:pon 437: Colkction and [>rQmtlltion Df&adw4J lnvmtDry Data. Washington, DC: Transportation Research Board, 2000. McShane, W. R.. and R.. P. R.ocss. Trrtjfo EngiMmng. Englewood Qiffs, NJ: Prentice Hall, 1990: pp. 75-79. Michigan Tec:hnologic:al University. RoaJSqfi GIS. Houghton, MI: Michigan Tec:hnologic:al University, Technology Devclopmenc Group, 2009. Pline, J. L Traffo Enginming HitndbDDit. Englewood Cliffs, NJ: Prentice Hall, 1992: pp. 59~. Sabria, F., B. Bonnet and E. Sullivan. Evaluanon ofTraffic Conrrol Dnnu lnvmtDry Progrwns for Microcomputn'S, UCB-ITSTD-85-2. BeYkdey. Ck University of California, Institute ofTransportation Studies, 1985.

lnuDntnriDc:. •

'1.,1

Cha p ter 16

...... . ........... ...... ....... ... ... .... .... ... ....... ............. ........ ......... . ......... .

Parking Studies OriginAl by: L EOis Ki11g; D. Eng., P.E. Edited by:

Daniel Findky, P.E. 1.0 INTRODUCTION

324


324

2.1 Parking Inventory

325

2.2 Study Locations

325

2.3 Personnel and Equipment

325

2.4 Method

326

2.5 Parking Usage 5tu~ies

329

2.6 Accumulation and Generation Studies

330


332

3.1 License Plate Checks

332

3.2 Parking Interviews

336

33 Postcard/Flyer Interviews

337

3.4 Peflona/ Interviews

338

3.5 Parking Space Counting

339

4.0 DATA REDUCTION & ANALYSIS 4.1 Tabulation

339 339

5.0 SUMMARY

343

6.0 REFERENCES

345


345


345

6.3 Other Resources

345

Packing Studies • 32::::;:::3

1.0 INTRODUCTION

P

arking is one of the essential elements in urban transportation. With few exceptions, automobiles and trucks must be parked at least temporarily at each end of a vehicular trip. Even in areas served by public transit, rhe automobile is a common means of transportation. This chapter focuses on vehicular parking, but many of these concepts and methods can be applied to specialized applications, such as motorcycle or bicycle parking. There are rwo general types of parking: 1. owner-supplied facilities ar homes, apartments, retail centers, industrial buildings, institutions and offices, including public street curbs where non-metered parking is allowed; 2. commercial parking. including private lots or garages where fees are collected; and charge parking. such as meters at curbs or in public off-street lots or garages. Parking s~dies arc typically conducted to check physical needs of the existing parking supply and location or to establish parking policy and parking regulations, such as zoning codes or specific developments. The majoriry of parking studies are performed to determine the need to expand existing parking or adjust management techniques by comparing parking demands to parking space supply. Parking studies can involve central business districts (CBDs), industrial developments, office parks, densely populated apartment/condominium areas, hospitals, universities, sports arenas, cultural facilities and special events. The provision of parking spaces has important economic and environmental effects. According to a study in Tippecanoe Counry, IN, USA, parking spaces consume 6.57 percent of the total urban footprint in the counry and use approximately 20 percent more area than the buildings they service. With appr.oximately 1.7 parking spaces per . registered vehicle in the counry or 83,000 potentially unused spaces, the study found an environmental iinpact of

$22.5 million for the additional spaces (Davis c:r al., 2010). The presence of on-street parking reduces the capaciry and flow of roadways, while increasing the potential for collisions (AASHTO, 2004). However, parallel and angle op-street parking can be useful in specific situations with context-sensitive designs by encouraging pedestrian activity, increasing packing supply and traffic calming (ITE, 2009). On-street parking is generally more favorable for low-speed and low-to-moderate-volume roadways. Advanced parking management systems such as pretrip planning information, quantiry available (by lot, floor, aisle, or space), reservation systems and navigarion systems can ease parking supply issues (FHWA, 2007). Examples of these techniques ace present in many countries. The public dissemination of information about parking space a.vailabiliry can increase the utilization, efficiency and consumer satisfaction of a packing system. The occurrence of illegal parking increases significandy when the packing lor reaches 80 percent of capaciry and is generally perceived by users to be full when 90 percent of capaciry is reached (ITE, 2004). · Parking studies look at supply and demand. On the supply side, inventories ace conducted to detail the current parking conditions. Demand is commonly measured in terms of usage and turnover. However, accumulation and usage studies measure where drivers actually park, while real demand is the location where drivers would prefer to pack- Usage examines number of cars while turnover looks at the speci.£c vehicle using a space. Additional data may be collc~ed ro determine packing generation which is important in calculating future needs.

2.0 TYPE.S OF STUDIES Common types of parking studies ace discussed in the following sections which include: • packing inventory;

.j

• parking usage; and • accumulation and generation.

l

:!· ·:· 1.1

:l·';,

\ ·:

\;

~24

•

MAIIILIAI OF TRAfii<.PORTAT10fll FIII(;IIIIFFRIIIJ(; <.TIIf\IFC:: ?1\lf\ Ff\ITIOIII

2.1 PARKING INVENTORY ' !The parking invenrory assembles information about the location, number and other perrinent character istics of ·.existing parking spaces at the curb and in off-meet areas, including alleys and spaces behind buildings. Every available legal public and private parking space should be identified. The usual informacion needed is:

• number of parking spaces; • time limits and hours ofoperation; • type of ownership, such as public, private, or restricted to employees or customers of a particular building; • rates (if any) and method of fee collection; • type of regulation at curb spaces, such as loading zone, passenger zone, handicapped zone, taxi :z.one, or bus zone; • type of facility, such as lot or garage; • probable degree of permanency-many informal, poorly maintained p~lcing facilities are temporary and may be expected to be replaced with new construction in the foreseeable future; and • building occupancy. In urban areas with populations over 50,000, basic parking or land use data lll2f be available in files of the local transportation agency. The parlcing inventory is highly useful to the traffic engineer in day-to-day activities and is &equendy used by zoning commissions, building departmentS and orher ciry agencies. The inventory is an essential prerequisite to any parking study and sliould be updated periodically, such as every 3 years. The layout, design and co·ns.truction of parking spaces and areas are compla: and specialized. Guidance is available from various sources (ITE, 2004, ITE, 2009, FHWA. 2003, AASHTO, 2004).

2.2 Study Locations If the inventory is for a CBD, the study must include the primary retail core, the secondary retail and office ring around the core, and the fringe area, where employees may be expected to park For a neighborhood business district, parking may be expected up tO 500 feet (ft.) (150 meters [m]) beyond the limitS of the cOmmercial zoning. A fidd inspection should be made to determine the actual extent of this parking. and the study boundaries adjusted accordingly. Similarly, fidd checks are needed to determine the study limics.at a special generator, such as an industrial build~ ing, auditorium, univecsiry, hospital, stadium and so on, which may extend for 1,000 fr. (300 m) or more.

If the srudy involves a congested area, it is desirable to sdcct natural boundaries, such as rivers, railroads, or major routes along which changes in land use oc.aJI. For a major route, it is generally necess:uy to include a distance of 300 to 500 fi:. (90 to 150 m) along each cross street since these areas are potential locations for curb parkers unable to park on the major route.

2.3 Personnel and Equipment When conducting fidd Studies, interaetion with the public through surveys or other parking srudies might lead to inquiries about the srudy. Fidd pecsonnd should be prepared to provide bwiness cards of the principal investigator of the project and an authorization lener from the sponsoring agency with contact information of rhe appropriate official. Study me!D.ods can be conducted with dipboards with pen and paper, handheld computecs, or global positioning syst-em (GPS)-enabled data collectors. For a study of a small area, current aerial photographs ar scales of about I inch (in.) • 50 ft. (1:600) are most usefUl (Syrakis and Piatt. 1969). A I in. = 100 fi:. scale (1:1200) may be used for larger areas, although considerable derail is lost if rhe area· includes tall buildings. Land use maps may be available which &cilitatc the inventory. The fidd reams consist of one to two persons, who are wually on foot in the case of a business district, or in a vehicle for studies of small Parking Studies • 325

areas, and along major traffic routes. A measuring wheel is useful, although most measurements of curb and open parking facilities can be scaled from aerial photographs. Incremental photographs from cameras mounted at a high vantage poinr can also serve as valwble s:ources of data. The use of online mapping rools can be useful for the srudy co indicate the orientation, quantity and overall capacity of parking spaces. These cools can also provide a rough depiction of land uses.

Block and Cub Face Numbering System

cp 0 I

l) 2~®2 .(D 2.4 Method Before beginning the invenrory. a master coding system is established. Each block is given an identification number and excra-long blocks may be subdivided and given cwo numbers. If there is a prior tr.tnsportation or origin-destination (0-D) srudy for che area, which includes a separate number for each block of the parking study area, it may be convenient ro use the same numbers.

. i'

Once a block number has been selected, numbers 1 through 4 are used for identification of curb faces in four-sided blocks, as shown in Exhibit 16-1. However, the maximum number of sides in the most oddly shaped block must be provided for. For aample, if a six-sided block is che worse condition, number 6 is reserved for the sixth curb face. Numbers 7 and upward are then left for identification of individual parking facilities within each block. Each block is numbered separately and should include informal areas off alleys at the rear of buildings, as well as che more obvious parking facilities.

G) 3

3 I

4

®

3 I

.I

24(j)

~

3

I

I

...

CD :D 402 4·® 3 I

2~

®

3

3

I

I

2~. 2 4

I.EOENOo < D·BLOCt< NUMOEA 3 .CUA8 F ACf!. NUM8EA

Source: Box and Oppenlander, 1976. ~;t

11,1.11" PARK~

FACIUTY NUMBERING

1

~t 41

ST.

5

@j~l ~

8

91

1::;

12

At the beginning of the inventory, the exact number of off-street parking facilities within each block is usually unknown; thus numbers must be assigned by inventory personnel wh.o should be provided with the master map 10 giving the block and curb face numbers. These numST. bers may readily be transferred to a map with a scale of ..-. approxima.cely I in. =200 ft. (1:2400) for ease of use. (!)<1.00
lf individual curb parking stalls do not have pavement markings, it is necessary to determine the approximare length of each curb parking secrion. This length is the distance from near edge to near edge of crosswalk in a block that contains no driveways, fire hydrants, or other parking restrictions. However, che usual block contains several areas of

326 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDIT10N

restricted parking and measurements are needed within each separate section. If parking restrictions are readily identi~ fied on a scale aerial photograph, much of the field measuring can be eliminated.

·Estimates can be made of the available number of parking spaces in a given unmarked discance by utilizing the following figures: • parallel packing

23 ft. (7 m) per vehicle

• angle parking

Ii

• 90• parking

ft. (4 m) per vehicle

9.5 ft. (3 m) per vehicle

These dimensions for unmarked stalls reflecc a lower operating capacity than if markings were in place. Counts of individual stalls or linear measurements may be used to determine the number of parking spaces available in a lor or garage. If the facility is a private commercial cype with attendant parking. determination of capacity is more &fficulr because ve~ides may be stored in the aisles. In general, the operating capacity of an attendant-parking (v-aJer) facility is the peak-condition loading, when all space is filled except for what is needed to allow for off-street man~u vers and temporary parking of vehicles (which must be moved to allow access to parked vehicles). The number of parking spaces in each block of the survey area and the number of available parking spaces in similar facilities are tabulated and summarized for the entire study area. Typical categories for the overall summary are 1. curb and alley parking a.

metered

b.

nonmetered

c.

special zones

d.

restricted types (motorcycle, handicapped, 15-min:, etc.)

2. parking lou a.

public

b.

private

3. parking garages a.

public

b. private The block-by-block summaries are tabulated on an inventory form such as that shown in Exhibit 16-3 . Under c.he "special facility cype" columns, individual types of facilities, such as those under a given time-limir restriction or a special regulation such as •loading zone: may be identified separately. In a parking report, maps are frequently used to illustrare invcnrorr. data, as shown in Exhibit 16-4.

Parking Srudies •

12?

?;:~·~wv'."'~'

· ··t-

f

t

I•

•• -~~

.. •

0

-.

·: '>ric~:,

• .. 0 •

·

...;-":'i'-'~;~w~

fi~~"'-~w~~~~~'3<~~;: , .••. ~r:.;: . . : ~ !-'" ~~....,;.: .... ~

..,

~·r~..

" . • ~~- . .

Parking lnftDtory Summary Sheet Weather

Date

Area/Ciry Special Fadliry Type

Block

Facility

Comments Observer

//

Sou=: Box and Oppenlander 1976. 128 • MANUAL OF TRANSPORTATION FNCiiNFFRIN\, <;TIInlf'\ 7Nn miTII"\N

Off-Suut Parking Private

Public

Total Stalls

qrivate lots or garages are those restricted to the use of employees, residents, or tenants of a building; r:U:icab d[ truck storage area, or other facilities not open to the general public. Lots and garages open to some public

parking, as weU as weekly or monthly rentals, are classed as public facilities. Where space is rented to jnd ividuab, it is classified as public space because the space may be rented in the future by other members of the genen.l public. If this study is for public parking only, employee spaces are not included since they are not usable b y the general public. Other restricted spaces include those for named employees, visitors, compact cars, motorcycles, or carpools. When appropriate, the pa.rking inventory may also gather data on assessed valuations of vacant lots or old buildings that might be replaced by parking facilities. Certain elements of traffic control, including one-way s treetS and alleys and restrictions on turning movements at intersections, should also be included. Such controls affect routing of•windshield• surveys, as weU as access to potential locations for development of new parlcing facilities. Inventories of smaU areas and along major traffic routes foUow the general format of the CBD inventories except that aerial photographs and/or st.r ip maps are used to a greater extent.

2.5 Parking Usage Studies There ue two general types of usage srutlies: 1. accumulation and generation . 2. license plate checlcs These studies gcnenUy involve 6~d che.clcs that can be made without publicity or public knowledge of the work, while more speci~ studies may require direct contact with the parkers. Image-sensing and license-plate detection syste.qu can simplify data coUection and sorting. Multispacc metecs or parking kiosks for a set of parking spaces can also proYide valuable information for studies. Typical usage stutlies ue summarized in ExhibitS 16-5 to 16-7.

Source: Litm:~.n, T. Parlting MaMgnnnu: Stmugin. E~n and Plmrning. Victoria Trall.'lport Policy Institute, 2008.

Vehicle-Hours of

Block I

Parking

Accumulation, or parking occupancy checks, plus rurnover and duration studies, are useful in determining what (if any) curb-parking improvemenrs can be made to increase parking capaciry. The duration analysis indicates where long-time parkers are using space inefficiently, and a relative measure of curb-use efficiency ~s provided by comparing hourly parking turnover rates with desirable rates to indicate efficient usage. Parking practices or regulations thar contribute to uneconomical and inequitable use of street space are revealed in this analysis. For example: 1. The srudy may show enforcement of existing time limits is needed to stop overtime parking.

I

2: It may indicate existing time limits are too long or coo short. Consistent 30-min/parking in a 15-min. z.one mighr indicate a need to revise the regulation. Conversely, a preponderance of 30-min. parking in a twohour z.one would indicate the desirability of reducing the limit to 60 min. or less if an increased rumover is desired. Permit parking during restricted hours (7 a.m. to 3 p.m., for example) could displace parking to adjacent areas. 3. H azardous or illegal parking may be revealed. The faccual evidence of such practices may prove helpful in bringing about needed enforcement or voluntary public compliance with regulations. Occupancy checks are useful in determining needed improvements for cruck pickup and delivery service. Consistent double parking by commercial vehicles may indicate the need for additional loading z.ones or for berter enforcement of regulations if available loading z.one space is adequate. Police action may also be required if passenger vehicles are observed to be blocking the existing zones. Where it is proposed to prohibit parking. occupancy checks are helpful in determining the parking demand at the hours of the day during which the prohibition would be in effect. When considering changes from angle to p:uallel parking, installation of parking meters, or installation of curb passenger and freight loading z.ones, a study of the curb parking activity data is useful. These studies are also performed to determine parking efficiency at older industrial plants, and parking lots or garages in business areas. To represent typical conditions, a study should consider any underlying schedules, such as nearby school class schedules, holidays, or special event schedules, to either capture or avoid those conditions as desired by the study needs.

2.6 Accumulation and Generation Studies Accumulation and generation studies are made at relatively frequent intervals on different days of the week to determine hourly variations and peak parking demand. In a CBD, the studies are generally made on an hourly or 2-hour basis between 6:00 a.m. and 8:00 p.m. In a business district, the retail shopping hours must be identified, and if stores are open only on cenain evenings, the studies should include days of both early and late dosing. In smaller communities, retail areas may produce peak activities on Saturday mornings or afternoons. Oftentimes, parking studies may·aJso address church-related activities that may generate significant Sunday morning (or other period) activity in otherwise low-parking-demand areas. When peak weekday-parking-demand hours have been determined, repeated checks should be taken at the peak for each weekday.

If the accur;nulation studies are being made at a particular generator, such as an office building, studies should begin prior to the morning opening of offices and immediately after closing hours. At industrial plants operating on more than one shift, parking checks are needed at time of shift overlap. It is typical for office and administrative personnel to have different hours &om licrory workers, ·and separate checks may be needed to determine these demands for parking. 330 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Along major traffic routes, checks are made during the morning peak hours from 6:30 to 9:30 a.m. , during sever:tl : midday times, and across the evening rush period from 3:30 to 6:00p.m. Tf residential properties a b ut the route , · it is essemial checks be made during the lace-night period since for residemial parking, peak accumulation occurs · berween 1:00 to 5:00 a.m. Demand is relatively constant during this period and a single overnight check should be adequate. The traffic engineer is generally most interested in the peak-parking-demand times as rela ted co currenr supply. ln the case of a major traffic route study, where the incenc may be to impose only morning or evening rushhour resrrictions, checks should be made during these time periods. It is often desirable to prohibit all parking along a major route since this is the most effective way co provide for the total transportation of persons and goods, and co maximize public safety. Therefore, parking checks along maj~r routes usually include hours ocher chan just che rush periods. In all parking checks, it is important tO avoid condttions of tempocuy parking restrictions, such as foe street sweeping. screet repairs, or snow plowing. Parking checks in retail areas are usually not taken during periods of abnormal demand such as the day following Thanksgiving or national holidays. I n SOf!le cases, seasonal parking demand, such as the weeks before Chriscrnas, may be the focus of a parking study. The effect of major sports activities should be considered in the area of such generators, and special checks ace 9ften made to determine peak parking needs. ·· If the accumulation studies arc made co dete.r mine the peak parking demands of specific generator types in cider to develop wning code specifications, peaking characteristic information muse first be secured from cooperadve administrators or other reliable sources. The heaviest hours, days and months muse be determined, and factors s1.1ch as varying visitor hours at h ospitals and nursing homes or shift overlap periods must be ascertained. In the generation-type study, it is essential to know the occupancy or facility usage at the time of the checks. In hospitals and nursing homes, this includes the number of occupied beds as a percentage of total available b~ds, and in the case of subdivisions or apartment buildings, it involves the number of occupied dwellings. In offices, industrial plants and retail ceiuers, rhe .amount of occupied leasable square footage of Boor area as a percemage of che total leasable floor area is used. A s~cond precaution in the performance of generation-type srudies is the need for identification of parked vehicles as related to specific establishments. When patron parking is cominglcd with ocher parking, it may be nccessa.r)' to conduct interview-type studies or make specific observations of patron activities to secure reliable data. T h e parking needs of specific generators often vary among distriets within an urban area, and the availability of pub lie transit is a major factor. An office building in a CBD served by transit will have a lower parking demand than chat of a similar building located in an outlying area. Parking-generation figures calculated for one district should pot be applied arbitrarily to a dissimilar area. Representative trip and parking generation races for vario us land uses and building types are readily available (ITE, 2003, ITE, 2004). However, these average rates should be used wi. t:~ care when applied co a particular situation. The interaction and coordination of parking for multiple land uses .•s a consideration for mixed-use developments (ULI, 2005). Mixed-use developments provide the opponunity £or complementary land uses to share parking facilities due to variatio;ns in hourly, daily, or seasonal vehicle accumulation and walking trips within the developme11t to different land uses. Parking-accumulation studies can be performed using maps that show the oudine of the different parking facilities. In larger areas, a number of sectional maps may be required. The occupancy studies work well using a vehicl e with driver and observer. Colored pencils are used to note the number of parked cars found in each facility, with. a different color used for each time period. In larger facilities it may be easier to note the empty spaces rachCI th~ll the occupied spaces. Since pare of the accumulation checks are made during periods of traffic congestion, it is important to plan c.b-e study route carefuUy. This includes preliminary checks of driving time required to cover the different ar~ a~d establishment of the most efficient travel pattern. In more complex CBD accumulation studies, where the 101..1 ce may be involved and congested with cf'l!flic, it is desirable tO have the survey crews make several practice flllll. J: .D some of the more congested areas, it may be necessary to conduct the accumulation checks on foot. When it _i.s impractical to count accumula_rion within garages by driving through them, the required data are gathered fro~ garage operators or by walking through the facilities.

In some facilities, an accumulation analysis may be made by continuous counting of in-and-out vehicular movc:::=.ments. This requires knowledge of the number of vehicles within the facility at the beginning of the cciu.iies, plus ::a check of the number remaining at the end of the count, in order to verify accuracy of the counting. ln-and-outdac- ~ Parking Studies • 33 ......,

for facilities wim automated gates or attendant-controlled gates may be recorded automatically and made available by the facility operator. This type of study may also include notation of license plate numbers as described in me nc:xt section. Accumulation studies of residential parking are unable to determine whether vehicles are parked inside private garages. Checks are usually limited to vehicles observed in driveways and ac curbs, which makes it impom.nr to alert data users to the incompleteness of me check. In larger areas, personnel limitations may make it impractical to perform all accumulation studies during a single day. As a general rule, it is desirable to conduct additional portions of the checks on the same day of succeeding weeks, provided weather conditions and ocher in8uencing factors are the same. Whenever possible, me complete area should be covered in me initial occupancy checks, even though this may mean a less frequent interval for each time check, to establish a contrOl base. Supplemental studies can then be made at intermediate hours. The accumulation data from studies within a specific area are normally plotted on an hourly basis to show the changing demand. Exhibit 16-8 shows a typical accumulation curve. It is also customary co indicate the number of available parking spaces, including any hourly variations that may occur due co differing operating schedules and regulations. Accumulations may also be shown in tabular form for individl121 blocks and types of &cili ties.

800

100%

700

90% 80%

1600 &!

I!!

70%

500

60%

.... "400 ..

SO%

0

ll

E ::s

z

40%

300

30%

200

~ ~ u

a. c

":. II

1:!

20% 100

10%

0

0% 6:00

8:00

10:00

12:00

16:00

14:00

18:00

20:00

22:00

Time of Day

While accumulation srudies provide information on total numbers of vehicles by location, they yidd no information on the length of time each vehicle is parked or on driver destinations. Information on length of time parked and turnover, or number of times each parking space is used during the day, is determined from a license plate check.

i' :

f:

3.0 DATA COLLEcriON PROCEDURES 3.1 License Plate Checks License plate checks can be used for detailed observation of curb parking. The primary purpose is to determine rumover, which is defined as the average number of cars parked per day during the srudy period, in each space of a given block face. The equation for turnover, T, is:

(: • ;

J, I

• - - - • •• • • • • • • - • . . . , • • ,..,,_,, ,,..._..., ,., ,.

.,.~, ,... ,,,..

'"""'"' r-niTII"'\'1

l;l}:li_liOJk(llj • .. ',

..

..

•

I

B •

•

. . !:\7~:-

~·

-~

' ....~~~:tt~~~~,(~"·:;rt~::

~ .. ~.~!tii;Jljil~<

License Plate Check Field Data Sheet Date

Weather

Cicy

Side of Sattt

between

Sueet

. Spa-ce and Regula.do.o

and

Tune Cin:uit Begin& Notes

Codes: ###: last three digits oflicc:nsc plate number for first obscrnrion of a vehicle { Check mark for repeat vehicle from a prior c.Ucuit - : Dash mark for empcy spaces Comments

Observer

1

= _N_umb-:-:-er_ . -:o..:..f_D....:If;"f::-er_ent77'"_Car-::--s-P_ar_k_td_ Number of Parking Spaces

Equation 16-1

Other reasons for license plate checks include accuring data on length of stay, acru.mu..btion, illegal packing and enforcement which requires special notation as to whether a ticket is found on an illegally parked car. The study hours arc sdect~ to suit the purpose of the study and the liccnsc plate check is most often performed by a person on foot. Because the study is rdarivdy expensive, a sampling technique is normally used. Several block faces are sdeaed which typically represent the different time limits of parking found in the study area. Thus two or three block faces having 1-hour parking, twO or three others with 2-bour 'parking and so on, are checked. The pcrsonnd requirement is dependent on the frequency of checking. which in cum depends on the parking rumover. Since cumover is unknown at the beginning of the study, it must be estimated from known or assumed characteristics of the situation. For example, if a given curb &cc is regulated by 1-hour parking. checks arc needed at 20- or 30-min. intervals. If a curb face has 2-hour parking. checks would nonnally not be needed more often than every 30 min. but no greater than 1 hour. Very short parking. such as a 15-mia.li.mic, requires checks at 5-min. intetvals. In the fringe aceas, where unrestricted all-day parking occurs, hourly checks arc generally sufficient. Data from thc.acru.mulation checks on numbers of cars packed per block arc useful in setting up routes and determining pcrsonnd requirements for license plate checks. Voice recorders have been used succcssfUI.ly in license plite check~ ing, although there arc ~ere limitations and large-lClie opportunities for error in their use. The handhdd computer is Parking Studies • 333

also weful in conducting license place surveys. Data recorded in the field can be downloaded into a patking database after retu rning to the office. A field form for use at typical curbs is shown in Exhibit 16-9. The essential components include the identification of che facility by both code number and srreer names. The column headed "space and regulation" muse have provi· sion for illegal parking as well as legal spaces. This column can also be wed as a curb inventory form. As indicated on the form, a ponion of the license number, usually the lase three digits, of each parked vehicle is enteted in the appropriate column. When the same vehicle is found at subsequent checks, the number is not repeated but rather a check.mark is wed. This simplifies the calculation of turnover, since the number of different cars parked in the block is being determined. Other notations typically used in license plate checks include the indication of expired meters and color notes to idemify vehicles rhat have no license plate. Trucks should be noted. Double-parked vehicles ate identified by puccing a diagonal line across the particular box in the column. The license plate of the curb-parked vehicle is written in the upper part of the box, and the license number of the double-patked vehicle is wrirten in the lower pan of the box. It is sometimes desinble to note full license plate numbers so owners of vehicles may be identified. This information is wed to check the extent to which store owners or employees are wing curb space to patk their private vehicles. It may also be desinble to designate commercial and out-of-State vehicles. The license plate numbers of trucks may be underlined iri areas where the state plate does not give a separate identifying symbol such as a «T.n·Spccialized notes may also be made on size of truck and type of deliveries being made. Knowledge of frequency and location of truck loading and unloading activities are needed in the planning of curb loading z.ones as well as for development of truck terminals, and the degree of truek activity during variow hours is a faeroe when considering restrictions on such activity to increase street capacity (see Chapter 14 for more discussion on goods movement StUdies). '

,I.. I

'!

One person can conduce license plate checks for approximately 60 parking spaces each 15 min. This perm!ts coverage of two to four block faces on each trip after making the initial round, which generally requires additional rime. The degree of park,ing turnover is a major factor inHuencing the acrual time required and consideration mwc also be given co sections the checker has to traverse where no parking is allowed. Other administrative dements include need for rest and food breaks for the checkers, personal security in some areas and need for checkers co act in an unobtrwive fashion. While ic is difficult co conceal the fact that a parking check is being made, the survey results may be biased if•mecer feeders" ate aware of the field study. When questions ate asked, the checker should answer them courreowly and give assurance the study has no relation to law enforcement activicy. Where widespread license plate checks are being made, the wual method is to start routes at a common point to facilitate supervision. The field checker walks down one side of the street and back on the opposite side, covering all curb spaces along the route. In a rectangular street layout, routes may be started ac one intcrscccion and extend out around the block with the supervisor Stacioncd at the starting point. Fidd checkers often work a 5-bour shift, but this length is dependent on the tot:allengch of the parking survey period.

r

The license place check is most often used at the curb; however, it can also be used in lots or garages. ln th~e ~~. fidd observers go through each lot or garage at regular intervals, recording license numbers. Care mUSt be-taken co follow the same route within the facilicy co simplify comparison of license numbers during the summarizing process. If the off-street check is made as a part of the curb check, the intervals between trips arc the same as in the curb check. If the off-scrcct study is made separately, intervals can be increased by 20 to 30 min. or longer since parking durations in off-street locations are commonly longer chan at the curb. Any underlying schedules, such as nearby school class schedules, church service hours, or movie theater schedules, should be taken into consideracion.

·

Where off-street facilities arc too large to check every vehicle in the time allotted, or arc ope.ratcd by people who will not cooperate in allowing the observer to make the necessary observations, the •in-and-out" study is used. This study method requires observers at all entrances and exits of the facility and gives the maximum accuracy of turnover and duration data since every vehicle entering and exiting is observed and the license number recorded. In cases where the volumes of entering and/or leaving vehicles arc unwually high, several observers or video recording may be required. The cime limits of this StUdy should coincide with those of the curb license place check. Ax suitable intervals, the time of day is noted following the lase recorded license place number, on both the "in" and "out" sides of the form. When crafiic volumes are low, the ~c of acrual entry or exit may be recorded for each vehicle. During high-activicy periods,

i


.

lfll~ hmrn_o

••

.

"' •• • ·te-

I!''''•

,.:-~

., -~"~TI;;!"• ~ - -~1r?I~~~\~':; .· .~. ./~ :~:. .-:<'"rA-~~~~-t-.. ,:t·"';_.~~' ~-·:. · - -ittf-Jt:.--r:;

•

Packing Duration Summary Sheet Date

Weather

Scm lime

End1ime

Location

ci~ificatioo TYPe (Pa.rlcing zone, block, or other cla.ssification-Indiaue bdow) Pa.rkiog Duration Range (Indicate Bdow)

Passenger Vehicles

Truckt

Passenger Vehicles

Truckt

Passenger Vehicles

Truckt

Passenger Vehicles

Trucks

Number

Pera:D.t

Number

Percent

Nu.mbcr

Percent

Number

Percdlt

Total Vehicles Total Vehicle-Hours

Average DIIIlltion Percent Overtime Comments

Observer

time should be noted at 5- co 10-min. intervals. When possible, the number and license plates of the vehicles in. a

facility at the beginning and at the end of the study should be checked. Office surqmary and analysis of the license plate field check can provide information on parking accumulatio ri·. The peak acc·u mulation for each block face is determined directly by a co unt of the parked vehicles. Exhibit 1 610 shows a typical dwation summary form. The column headings may classify data by blocks; by type of parkiPg zone (such as !-how curb parking, 2-hour, etc.); or by curb, lots, blocks; and grand total for the area. Park.iog durations may be estimated from the round-trip times of the parking activity checks. For example, if trips a..re made at 15-min. frequencies, a vehicle found on only one trip is assumed to be packed for 15 min. Ifitis seen on two successive trips, the vehicle is assigned w a 30-min. duration group, and so on. When the vehicle is seen on the first or last trip of the observer, its parking dwation is unknown. All durations should be consid(red :as starting within the time limits of the srudy. Total vehicle hours arc calculated by dividing the total number of vehicles seen on each trip by the number of chcclo made each hour. Thus if 10 vehicles were observed three times on 30-min. check intervals, the nwnber of vehicle hours is (3 x I 0/2) or 15 vehicle hours dunng the 1.5 hours of the study. The average duration is found by divi~g total vehicle hours by the total number of different vehicles observed (15/10 = 1.5 hours). This average is slight:J:Y higher than the true mean value, because all packed vehicles are not observed. The percentage of overtime parkers J.S found by adding the figwcs foe all durations in excess of the legal parking time limit. Exhibit 16-11 shows a. typical rucnovec summary form tha.t may be used to swnmarize the turnover for eac::ll block; for each classification of parking zone according to time limit; for lots, garages and cwb space; and for different areas. A summary is made for individual blocks; for classification of time zones; for each lot or garage; Parking Studies • 33 .!5

Suut Puking Turnover Summary Sheet Date _ _ _ _ _ _ _ _ __ S~tTune

vv~mer

_______________________________________________

_ _ _ _ _ _ _ _ _ _ _ _ _ _ __

End lime _ _ _ _ _ _ _ _ _ _ _ _ __ __

Location Total Facility

Block

Number of Stalls

Tum.nu

Vehicles

Hourly Turnove.r Rate (vehicles/staWhour)

Conunents __________________________________________________________________ Obset'Rr

for all curbs combined, all lots combined and all garages combined; and finally, for a grand total. The left-hand column of Exhibit 16-11 is used to indicate classilicacion. The next column contains the number oflegal vehicle parking stalls that are being used all or part of the time on each normal weekday. The number of vehicles observed making use of the parking stalls is entered in the third column. Any vehicle making use of a space whether or not it entered or left during the period of the study is included in the turnover. count. The fourth column is used to record the hourly turnover rate per stall. For instance, if 50 stalls had an observed total of 100 different vehicles in 8 hours, the rate would be 0.25 vehicles per stall per hour [100/(50 x 8)]. The three remaining blank columns in Exhibit 16-11 may be used for various purposes. For example, it may be desired to show probable future number of spaces, allowing for new off-sueet facilities and prohibition of curb parking. and resulting possible turnover. Or it may be decided to show what turnover would be if the turnover rate were increaSed through enforcement of existing time limits. Column headings might be, in this case, "estimated turnover rate assuming compliance with time limits." "total turnover" and "increase in turnover compared to pr~ent." Daily or study period turnover is usually shown. Continuing the example, 100 cars parked in 50 stalls would yidd a daily turnover of 2.0 for the 8-hour period.

3.2 Parking Interviews To determine uip origin, destination, pwpose and walking distance, it is necessary to makt some type of dinxx conr:acr: with the parkers. This inteJView can be conduaed by a teturn-type questionnaire, usually a postcard, or by personal interview (FHWA, 1970). Such studies are used primarily in determining parking demand by location. Interview data are intended to reveal both the pattern and number of driver destinations reached on foot after parking, and to measure the demand for space under the assumption all drivers would like to park near their ultimate destination. Data on driver habits are coUected, such as walking distances as rdated to trip purpose and parking fees charged. This information is essential in determining acceptable locations for new parking facilities. Duration of parking by various trip purposes indicates practical time restrictions which may be imposed on various classes of parkers. If revenueproducing facilities are to be devdC?ped, this information is needed to estimate income. Consideration of integrating public transit and parking facilities requires an evaluation of bow many persons would use joint facilities, where they should be located, and what transit service would be necessary.

~'2~

.a

l.AA t..IIIAI 1"'\1: TDAf\IC:::DI"\DTATJf'\U Cll.ll':lli.ICC:DU,Jr.: C T IIr'\I&:C

.,ll..lr"'\ C:t'\ITII"'\Al

.i3.3 Postcard/Flyer Interviews

)The postcard interview is used to determine charaaeristics only at the peak parking hour and the handout o r place'{nenr of cards u.nder windshield wipers shou.ld be done in a short time span. The time required for card ~istribucion sbou.ld be determined experimentally before the stu.dy begins. Newspaper publicicy is essential for any cypc of inter· view stu.dy, especially when postcards arc being used, and every effort shou.ld be made to obtain maximwn coopcra· tion from the public. The personnel rcqu.ircments for the postcard survey arc much lower than those for the p ersonal interview; however, the low rate Q.f returns and the unknown degree to which dara are biased are nevtive aspects of card usage. Exhibit 16-12 shows a cypical parking-survey postcard.

The Department ofTransportation is conducting an analysis of the parking conditions in this area. -we are asking each driver to complete the following questionnaire. Your answers will be kept anonymous and da~ will be summarized ~o that .the anal~is and n:pott will in no way dw.wc da~ on ind.ividuals. We thank you in advance for 6.lling · this brief survey. Your time is valuable, and the answers vou l!ivc will hdp us imorovc parkine: in the area. l. What is the primary 'purpose of your trip? (check only one)

OWork

Q Tourism/R.caeacion

Q Personal Business (Medical,

Social, eu:.:)

2. What was the specific: location you visited

QSchool

QShopping

QOther

~ecW.tcly after parking your car here?

Street address or business name: 3. How long were you parkt:d

!~cion?

---------------===

hou.rs

---------------------------------- minutes 4. Where did your trip stut (where did you come from just prioi to parking ~ere)? Nearest street intctsoetion:

5. Please share any additional comments you luve about parking

in this area:

Typically. 30-50 percent of the cards distriburelar~ returned. To expand the sample, it is necessary to have a record of the nwnber of cards placed, by locarion and rime. For example, if five cards were returned from a given b lock h.~ indicating these parkers shopped at a particular store, and the rerur;n sample from this block &ce was 50 percent, it cou.ld be estimated there were a toeal of 10 parkers at the peak rime ,..ho actually went to the particu.lar store. Because the sample of rctu.rn.s wou.ld be expcaed to vary from facility to facility, a separate calcularion of =pie size and ex· pansion &ctor shou.ld be made for each parking fa.cil..ity. One method of maximizing the return is to take special care in selecting qucsriotts so as not to ask for informacion some drive~ wou.ld be rductant to reveal. The best illustrarion concerns the home address, which should never be asked since .many persons do not want to give out this information and it is of no direct value in the parlci.ng survey. The needed locarion is some general part of the city or the name of.the cicy for parkers from subwbs or other nearby communities. Appcnaix B shou.ld be cottsu.ltcd for additional suggcstiotts. To maximize the return of the postcard interyicws, pastage shou.ld be included on the postcard. To avoid paying for postcards that are not recucned, business reply mail postage can be used for the nudy. Another method of contacting drivers is to record license plate numbers of parked vehicles, trace mailing addresses on the basis of these numbers and mail return postcards or quesrionnaires to the owners. However, a rime delay of several days is caused by lookup of nwnbers, mailing and ~ccipt of card by the vehicle owner. This resu.lts in fewer returtts and less a.a;:ucacy.

P;u~inn

'\tuniP< • ::1::17

":,!'-

.

·~t.~;p~fii,

N-

·~:"-<• ~~~.:., "'"'"'~~~ ... ~~~ · '.p~:,.,py~c:t~~~~~~~~~~ --..:.· . :~~;!;~ - Jt! :- : .. , . 'it ~· '.:1!

Parking Interview Sheet

Wather

Dace SwtTime

End Tune

Location 1ime

Start (Parked)

Trip Purpose

End (Left Stall)

(Work, Scltool, Shopping, Personal, Other)

Trip D estluation

Office Use OnJy Dur:r.tlon

s~t

Walking

address or

Bwiness Name

Hours

Minutes

Distance

Commenu O~r

Flyers can 21so be used co direct drivers to a Web-based survey. A Web-based survey can reduce data collection costS. However, the survey might suffer from sdection bias from drivers. An incentive for participation can increase the quantity of completed surveys.

3.4 Personal Interviews

~.

Personal interviews of parkers can be conducted at the curb or in a parking facility. They can also be conducted at the entrance or exit of a particular generator, such as department store, office building, or hospiral. The exact questioru; vary on whether the interview is conducted at the origin or the destination of the parker, at the time of parking. or at the time of unpacking. Exhibit 16-13 shows a typic21 form that is used when the interview is being conducted at either time of parking or unpacking. Questions include uip purpose and destination of the parker. Interviews at an apartment garage would add "reside here" as a trip purpose. The parlter may be asked to give an estimate of length of time parked. The form shown as Exhibit 16-13 docs not include space for trip origin; however, this can be added.

=

h :·;.

Person21 interviews be used on a day-long basis, during sever21 hours of the day, or only during the peak hour. The work is not only C:xpensive bur also personally demanding, a.nd careful selection a.nd training of interviewers are essenti21. Persons with previous c:xperience in census taking are especially valuable for this activity. Ea.ch interviewer should wear a distincr armband, ribbon, reflective vest, or other identifying insignia ro minimize confusion, apprehension, or antipathy on the pact of paclters being approached for questioning. The words "Traffic Survey" or other suitable information should appear in large letters on the insignia.

' ··

1' ...

li

!;i; .: .•

338 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND ED!nON

: The interviewer should be on location 15 min. in advance of the starting time. One person can handle aboor 15 ·: spaces, provided rurnover is not too rapid. Thus one or two persons are usually needed to handle one block face. In '. the case of a large off-street F.!cility, it may be necessary to sample parkers by interviewing every second or third pe.r· 'son. The study normally does not include vehicles stopping momentarily to discharge passengers, taxicabs parkin!? tn reserved curb zones, buses and emergency vehicles. Truck drivers may be interviewed, depending on th e informatiOn that is desi.red. Interviews are directed only to the driver of each vehicle. · Vehicles parked when the interviewer first arrives are recorded on the field form with an "inn time 5 min. prior tO beginning the official survey. The drivers are interviewed if they return to their vehicles during the survey period, and it is usually then possible to determine the approximate actual time of parking. The original entry is then deleted and replaced with the correct time. Normal procedure calls for the interviewer to approach a vehicle quickly as it puUs into a parking space and record the "time parkedn to the nearest minute. The interviewer then asks the driver the selected questions and noces ch~ answers on the 6dd form. When the driver rerurns and begins to unpack. the interviewer records the "time leaving on the form if the survey is being made along a block face or :a a point where the interviewer can see the unpulcin g activity. For such cases the 'interview form should also include a column for license plate numbers. For curb checks• the interviewer records the time leaving only on the same line as the time parked if the license number can be found quickly on the form. If not, only the license number and the time leaving column are filled out on the next empty line, and the cross-matching is done later in the office. If the same driver parks several times in the same block during the study, he or she must be interviewed separately each time. If the driver refuses to cooperate or the interview is missed for other reasons, the interviewer writes "refused" or •missed" in the •purpose" column. Sometimes, trip destination and purpose can be determined by observing wbc:r e the drivers walk and which building they enter.

3.5 Parking Space Counting Advanced parking management systems require accurate and timdy informacion on parking space quantities. T'VV"0 types of counters are used, entry/exit counters and space occupancy detectors (FHWA,.2007). Enrry/ cxic counre.rs provide broad levd information for lot, floor, or aisle parking space availability. Space occupancy detectors present motorists with specific parking space availability. Induction loops, video detection, ultrasonic sensing and otbC:r technologies can be used to count or detect the number of vehicles that have encered.or exited an area or whether a parking space is occupied. The primary objective of parking space counting systems is to provide information direccly to motorists who desire to park. Common uses of advanced parking management systems are in large garages such ::as those at airporcs, where the owner wants to minimize customers driving around looking for a space. The information can also be 'stored and analyzed to determine parking demand patterns, occupanCy levels, turnover rates, the need for additional parking F.!cilities and the effects of pricing or management strategies.

4.0 DATA REDUCTION & ANALYSIS 4.1 Tabulation Calculation of parking duration and walking distance is performed in the office. Walking distance is noc •as the bi..C.d Hies•, but rather is determined along the pedestrian sidewalk system available to the nearest entry or ex.it from tb-e building destination. In tabulating specific generator destinations such as retail stores and office buildings, it will oftc:.Jt'l be desirable to assign code numbers to these building. This may be done on a block-by-block basis by using a numb&:" I series beginning with 100 to distinguish clearly between parking f.aciliries and pedestrian generators. In most case: 5 all parking destinations within a given block will be consolidated for the purpose of determining parking-space ho..._,: surpluses and deficiencies. Exhibit 16-14 shows a parking occupancy summary.

ParKing Studies • l3~

Source: Reproduced with pernilision, NC Sta~ University.

Other ~abula.r information, such as ttip purpose related to length of time pa.rked and distance walked related to both trip purpose and facility where parked, will also be needed. Depending on the applio.tion of the study, such as the dcvdoprncnt of a &asibility report for the construction of new revenue-producing pa.rking &.cilitics, it will be neccssasy to determine the number of parlrus by both trip purpose and parlcing duntion for each block. A va.ricty ofspecial forms may be wed, as shown in Exhibit 16-15.

•

..

.

•

!> .. "

.. . ""

..

.~~·~~ ~~~~~-..,"::'~;~~~~

Puking by Facility Type and Trip Pu.rposc Trip Purpo.se ·work

Type of Facility

..

Non-Metered Zones

Tourisaal RKreatloo

School

Shopping

Pcnonal BIISioCSS

Otbu

Unknown

Metered (30-Minutc Zones) Metered ( 1-Hour Zo.nes) Etc.

Parlring by Facility Type and D i.stuce Walked Distance Walked (Feet)

Type of Facil.ity

< 100

1,300 1,700 lOOto SOOto 900to to to 499 899 1,299 1,699 2,000 >2,000

AYC.-.,e

Unknown

Total

Di.staace

Non-Metered Zones

M=rcd ..

(30-Minutc Zones) Metered (l-Hour Zones) Etc.

Puking by Facility Type ...;d ~ Duration Di.staoa Walked (Fcct)

Durarioo

< 100

lOOto .SOO to 900to 499 899 1,299

1,300 to 1,699

1,700 to 2,000

Avaase > 2,000

Uob.cnm

Tocal

D istaace

0 to 14 Minutes 15 to 29 Miitutt:S

..

'

30 ro 60 Minutes Etc.

Puking by Trip Parpo.e and Walld.o.g Di.staacc Trip Parpote

Walldoa

Distance (Feet) I

Tocariam/

Work

Recreation

Shoppins

School

Pcnooal Bu.sincsa

Other

Uobon

~..a:s than 1oo 100 to499 500 (D 899

Ecc. Cmtinud on nca fllfl! -

Source: Box and Oppenlander, 1976. Parking Studies • 341

. .. . ,,

s:-r.. .. • -;:,

•.:...a . . . •

'
..

~

Parking by Trip Purpose and Pa.r!Oog Dur.ation

Trip Purpose Duration

Work

Tourism/ Recreation

School

Shopping

Personal Business

Other

Unknown

Total

Unknown

Total

0 to 14 Minutes 15 to 29 Minutes 30 [0 59 Minures .Etc.

Parking by Trip Purpose and Datlnaton Block Trip Purpose Dcstmatioo

Block .

Work

Tourism/ Recreation

School

Shopping

Penonal Bwiacss

Other

Pukiag by Duration and Desti.aaton Block

Parking Du.ration (Hou.rs) 0.25 o.s Desti.aation to to 1 to 2to 3to 4to 5 to 6to 7to 8to Block < 0.25 0.49 0.99 1-9? 2.99 3.99 4.99 5.99 6.99 7.99 8.99 >9 UnkoOWD Total Average

General SW11.10A.C'Y of Parking Uaagc and Capacity Block Number

Vehicle Stalls

Maximum Expected Total Turnover

Number of Driver Ttunover Observed

Desti.aations

Continu.!d on nextpage Source: Box and Oppenlander, 1976.


Block Number

Numbers of Vehicles

Space-Hours

Parked

Used

Driver Destination

I

Space-Hours for this Driver Destination

Source: Box and Oppenlander, 1976.

5.0 SUMMARY The analysis of parlcing studies should lead co the evaluation of current parking techniques and the potencial mitigation of parking supply issues with various management techniques. A wide variety of parking management techniques are available and must be carefuUy selected for each unique situation or parking issue. A lise of strategies and their respective typical reduction in required amount of parking, and whether traffic volumes are reduced as a result, is shown in Exhibit 16-16.

Parking Studies • 34:;::3

Shared Parking

Parking spaces serve multiple users and destinations.

10-30%

Parking Regulations

Regulations favor higher-value uses such 2S service vehicles, deliveries, customers, quick errands and people with speci;d needs.

10-30%

Adjust parking standards to more accurately rdlea demand in a particular situation.

10-30%

./ Walking and Cycling Improvements

~15%

Increase Capacity of Existing Facilities

~15%

Mobility Management

10-30%

./

./ Parking Pricing

10-30%

Improve Pricing Methods

Varies

Financiallncemives

10-30%

{/nbundle Parking

10-30%

Parking Tax Reform

~15%

Improve Information and Mark.eting

Provide convenient and accurate informacion on parking availability and price, using maps, signs, brochures and the Internet.

5-15%

Improve Enforcement

Ensure that regulation enforcement is efficient, considerate and fair.

Varies

Transport Management Associatio'n

Establish member-controlled organiz.acions that provide aansport and parking management services in a

Varies

Overflow Parking Plans Address Spillover Problems Parking Facllity Design and Operation

./ ./

./

./ ./

./

./ Varies

management, enforcement and pricing to

Varies

lmprove parking &cility design and operations ro hdp solve problems and support parking

Varies

So!J!CC: Litman, T. Parking MIJliiJgmtmt: S~a.

Evaluation and Planning. Victoria Transport Policy Institute, 2008.

.

6.0 REFERENCES ~-1 Literature References American Association of Stare Highway and Transportation Officials. A Policy on Gtommic Design ofHighw4J$ and Street:s. Washington, DC: AASHTO, 2004.

Davis, A.Y., B. C. Pijanowski, K. Robinson and B. Engel. "The Environmental and Economic Costs of Sprawling Parlcing Lots in the Uniced Scares.• LanJ Use Policy fournJl/27, No. 2 (2010): 255-261. Federal Highway Administration. Advanmi Parking Management Syrtems: A Cr11u-Cutting Study-Talting the Stms Out if Parking. Washington, DC: FHWA, 2007. Federal Highway Administration. Manual on Uniform Traffic Contr11/ Devicesfor Strtet:s and Highways. Washington, DC: FHWA.2009. Federal Highway Administration. Origin-DestinatWn Suroep. ~ashingron, DC: FHWA, 1970. Insticure ofTran.sportation Engineers. Parking Gmtrtztion, 4th ed. Washington, DC: ITE, 2010. Instirute ofTransportation Engineers. Traffo Engineering Handboolt, 6th ed. Washington, DC: ITE, 2009. Instirute ofTran.sportation Engineers. Trip Genenuion, 8th Edition. Washington, DC: ITE, 2008. Litman, T. Parlting Management: Srraugies, Ewtlwztion and Planning. Victoria, BC: Viccoria Transport Policy Institute, 2008.

Syrakis, T. A. and J. R. Plan. "Aer~. Photographic Parking Srudy Techniques.• Highway Research &cord 267 (1969): 15-28. Urban Land Insrirute. Shartd Parlting. Washingoon, DC: Urban Land lnstiruce, 2005.

6.2 Online Resources (Available as of January 5, 2010) Victoria Transport Policy Instirute. Parlting Solutions · A Comp~hensive Menu ofSolutions UJ Parking Problems. Victoria Transport Policy Insricute. www.vtpi.org/rdm.ltdm72.btm. Retrieved from TOM Encyclopedia.

6.3 Other Resources Edwards J. D. Pa_rlting lf4ndlxJolt for Small Communities. Washington, DC: ~stirure ofTraffic Engineers, 1991. Shoup, D. C. The High CtJst ofFru Parlting. Chicago, 11.: Planners Press, American Planning Association, 2005. U.S. Environmental Protection Agency. Parking Spaces/Community Places - Finding the Balo.nce thrt1ugh Sm4rt Growth Solutions. Washington, DC: U.S. EPA, 2006• . Vaca, E. and R. J. I
Chapter 17 ~

•

e

e

•

e

•• e

e

••• ~ •

e

•

•

e ... •

.

.. e

e

•

•

e

•

e

0

•• •••

••

f

•• •• e

•••

0

e

•• •

e

.

..... .

f

••

0

•

4

.....

e ... ...

C

•

I

e

•

e

•

-

Traffic Collision Studies Original By: Joseph E. Hummer, Ph.D., P.E. Edited By:


ltiTRODUCTION

'

2.0 TRAfFIC SAFETY STUDIES 3.0

347 347


348

3. 1 Collision Reports

348

3.2 Sources of Data

352

3.3 Collision Data Reduction

354

3.4 Merging Collision an.d Other Data

355

3.5 Recognizing and Dealing witH Problems in Collision Data

356

4.0 ANALYSIS OF COLLISION DATA

4.1 Number of Collisions and Trends

357 358

4.2 Identifying Hazardous Locations - "'Network Screening• 359

5.0

4.3 Selecting Countermeasures

369

4.4 Countermeasure Evaluation

379

REFERENCES

379

1.0 INTRODUCTION raffic collisions exact a terrible toll in the United Scates and ocher countries. In 2008, 5.8 million motor vehicle traffic crashes were reporred in the United States alone (NHTSA, 2008). The collection and analysis of data o n traffic collisions is fundamental to the design of programs to reduce chat toll. AJWyscs use collision data ro bel p understand why collisions occur, co hdp identify collision-prone locacions, to aid in deciding which safety progran"ls or countermeasures should be implemented and to assist evalua.cions of countermeasure effectiveness. In addiciot'l• certain programs of the U.S. government require that parcicipacing agencies have a program of collision data analysis.

T

This chapter discussd techniques for conducting studies with traffic collision data. Fl.rst, the collision report forms are described along with sources of collision daca available co the analyst. Further discussion is provided on how to reduc~ and merge data sees to prepare for analysis, along with typical problems and limitations that come up when prcparin.g data. Next, studies are discussed with particular emphasis on uends, hazardous site identification and the possibl~ countermeasures implemented at those sites. Finally, specific types of analyses ue briefiy described.

2.0 TRAFFIC SAFETY STUDIES Many collision ~ cedmiques have imp~ dramatically in r=t years, due to the widespread use of computers and research into improved study methods. New techniques chat appear practical for many jurisdictions and traffic engi.Traffic Collision Studies • 34::;'

n=s in the United States are d.iscusst:d in this cbapter. Some analysis techniques are nOt mentioned in this chapltt because tbcy require cxpecti.sc, dfon and/or equipment nor readily available to most jurisdictions and engineers. The premise of the chapter is engineers who wish to analyze uaffic collision data that have been collected by others, principally the police. Thus, the chapter does not cover techniques for investigating individual crashes. However, engineers working with coUision data should become familiar with crash investigation and reconsuuction to provide greater understanding of the data. Analysis of trends in crash types, time of day and weathe.r conditions should be distinguished from analysis of detailed crash causes, which is often a secondary exercise for practitioners who may be involved in an expert witness role. Cer· tain generic causal factors, such as driver inexperience (younger drivers), slow reaction times (older drivers), drinking or drug impaired driving and distracted driving such as ceU phones, tcxting, global positioning system (GPS) unit use, etc., should be reviewed if compiled. Standard crash reports, at least at present, do not require all necessary data. Most of the core causal factors must be debated and resolved on a larger policy level. Traffic engineers' predominant role is to design the most forgiving transportation system they can within budgetary constraints. The engineers' mission is mobiUty for all and their mandate is safety. However, when a significant causal factor is present in a large proportion of crashes (for example, distracted driving due to electropic devices or an increase in elderly crash rates), engineers have an obligation to nuke recommendations to polieymakers th:u could reduce their occurrence. Several new traffic collision analysis tools have been prepared in recent years. A primary safety resource in coming years wiU likely be the American Association of State Highway and Transportation Officials (AASHTO's) Highway Stlftty Manual (HSM, 2010). The analysis methodologies presented in the HSM are robust and will likely become srandard for safety analysts in the future. In addition to the HSM, the Federal Highway Administration (FHWA) distributes a useful software tool for safety practitioners called "Safety Analyst," which is based on the m~ules pre· sented in the HSM (FHWA, 2009). Analysts should become familiar with these new tools, as weU as the availabl~ tools presented in this chapter, when preparing to conduct safety studies.

3.0 DATA COLLECTION PROCEDURES 3.1 Collision Reports Collision data used by traffic engineers arc recorded primarily by the police on report forms or laptop computers soon after a crash. One police report form is filled out per collision. Most states have a standard collision form used by all police forces within the state. Exhibit 17·1 shows the collision report form from the state of North Carolina. The forJII requests informacion on the drivers and passengers, the vehicles, the roadway and the conditions at the time of the collision. Most forrns require a sketch of the crash scene showing vehicle paths and Qbjeets struck and a narn.tive describing the evenr.

·lMJS REPOftfllfO~ tHE U8l0Fl!IEDMSION OFIIO'IORYI!taclU. llltDATAIS COLUclmfOR .

r

!Jt:~~~~=-::mw~~OETERIIINAT10N$0f

__,_

(1=T:-

__......_

·-"-·-----------------o:=r:__......._ .J Contin~d on .nexJ pag~

So=; Nonh CuoliM Department of Motor Vehicles, www.ncdot.orgfdmv/forou/truhccportSidownload/DMV349_ lnstruction_Manual.pdf.

·

s'

A

•

..

'ifiil'll ~·" ~ ,>.;.~,,,_.. ,. "· , ,

:'I)U-,j. , ..... ...

.

o· · -·~·

• t; 011 Tli ~ .•, , . __ r 'l'- , _ .·

, _.._ n ~~111AL ""'

,

••

~-

.. _, •

• · ·: .· .

·--Uo)

'ltiUCLEINFO.

.,..,

A ·;

,.,.-...,- -·· ~·- ·=,.~-.5 "< · . ..,.,.__ ~~ _..._,.. •I• jo :;;':'(i; ;:r.\.-~~11!''··>,,,...;.. ,;:,..,j~i~::;.,_.;,-;;,j ...... ~~~-···'·•'··-\ ·~---c \ ~--"'-~r..-.r. 13 ,_;t-~:.) -~

__. _ . .

•...,

:"'''·!~ZOflfRELATEli '.'

ROAOWAYINFQ,

w..i;.i:iioi';r;,:;-~--" ·>~-- ~ . .. . ,

-·,;

,. .... a..-

"-~~:{:-:): "•"' ·

u r.. ~eatn--C.'l.)

lt,..,O.k.lb n Pbal~ Tp

·....

u ....-tl
04 llillloT, _....,_fP.l

n-~

,:i~~·~i ·.~~- ~.;_;

lltmllf9. _·R-UlWdforNIW

•~v.at:t~ U.

nAa:~essc-d

~~~i~:t·._,;

,. r!d:c.."'r'"

~=-¥~: ~~~ ~: ~:.,;~

jcRASH SEQUENCE

-.vc.ta~ tUt:Mti~~

IUokiA'II~ ~. . . . . f I

--~,, • ~-..-·

•• . .•. • ; o .r.P~'f!t!. 4 , -~'! .• ~.

J

a truH~·StcDidEwt'll

..
I

to l!sioolielo.iolwlf"'""Spud

"~or--arqaa

•PG~tC.fhtl'Y•'ct.clt.b:l3

0 0 0

os-e.;.r-alwro

UOW.!--A-~[\ft

•Sctalb· ,._....v.ab '

=-~~~- :~~~- :~:~.·.·

~;'j,;ifi:f ~~;·_) ~···

n li*CtalfO,.

"..,...,,.,.,_'(>._

,,_.,..._, -~ o r...o "'~"-:! :,,:!'C"".=~:_,;:!:'

u~DIIIctl

·:::~"~~~d~A·~:?r~~~.§.B.:~:~·~·:·~i\~~;..,: 1; +. • , ' . • •. , ,

." .-.....:'"': !, o-l',""'!""J •. ~J~·. :. _;~.-~:

...: ;

o

o

•,

I

UrDODD ~-.o -r..., • • r w

u.u

......................

• MIRATM. ,....,...._._.....,_

-

u-

o

-o - r"*"-

I

•

I

I

0

o'

,

o'

---

I

'<

.

I

•

"~·~• -• II ~."'- ~ '·

..::_. 0

I

- - ---- - - - - - ----- --- -"' <--•

I

~--.,.. - --- --- --- --"''"

l

DODO

0... - - --- ---- ---- - - - ~MIIml:lfiW.,...,.I*G '""

-

=====c;; ; w.;;;;=-====== _.._ 11Wnt-.mcll(tl

'".. -------------·-~ Qu

...:.

·~;:r~o:;~~~~v1~~~W5<:: ;=:~~ :;::; . :_ ___

::::.:::~: •

·

·..;;;~~- r$_. .. -t;~.,t:; ::.:-;~-?1 .;~:,

n~ .. l.a"Ja

0 0 0

.·r· .. .... ""-: -! . ·:·

.~

_,_

t

r

't>il*•*"

Oiiol-

Source: North Cuolina Department of Motor Vehicles, www.ncdoc.orgldmv/forrns/cnshreports!down1oad/DMV349_ lnsuuction_.Manual.pdf. · 350 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION.

~~~?k~~tfuMt
' -. • • • ~

Collision Type

Non-Collision

Collision of MotOr Vehicle Wilh ....

CoUision ofTwo or More MotOr Vehicles

S~cific Collision

Code Number I

Ran off Road Righc

2

Ran off Road Left

3

Ran off Road Sttaight Ahead

4

J:u:kknif~

5

Overturn/Rollover

6

Orner Non-Collision

7

Pedestrian

8

Ped~cydist

9

Railway Train/Engine

10

Animal

11

Movable Object

12

Fixed Ojea

13

Parked Motor Vehicle

14 15

Rear End, Turn

16

Left Turn, Same Roadway

17

Left Turn, Different Roadway

18

Right Turn, Same Roadway

19

Rear End, Slow, or Stop

20

Right Turn, Different Roadway Head On

21

Sideswipe, Same Dicection

22

Angle Collision

23

Backing Up

24

Other Collision with Vehicle

1

__

-

-

-

-

-

_J

----

-

-

-

~

Crash type is one of the most important fields on the report form f~r traffic engineers. Various jurisdictions code collision type in various ways. Some states co
Traffic Collision Studies • 35~

1. F (htaliry) or K (killed): The person died within 30 days of the collision as a direa result of injuries received during the collision. · 2. A:. The person experienced serious, incapacitating, non&tal injuries during the collision. Broken bones, massive losses of blood, or more serious injuries are rated A. 3. B: The person experienced a visible but not serious or incapacitating injury during the collision. 4. C: The person complained of pain or momentary loss of consciousness due to an injury during the collision, but no visible sign of injury was evident to the investigator. 5. 0: No injury, which includes "PDO-Property Damage Onll collisions. These can be significandy underreported since they are often handled between the drivers of the vehicles (or by the driver striking the obstacle) without the assistance of the police. Researchers have developed other scales of injury with finer gradations than the FABCO/KABCO scale; however, this scale has gained mainstream acceptance on standard collision report forms. An c:nrople of another injury scale is rhe Muimum Abbreviated Injury Scale, discussed by Miller et al. (1985), which rates injuries from "1" • minor injury through "5" =critical injury and "6" • unueatable, immediately htal injury. Another interesting and useful code on most collision report forms is the vehicle identification number (VlN). The YIN is a 17-digit number unique to each vehicle that is assigned to most automobaes and other motor vehicles. Coded within the YIN is information such as vehicle weight, wheelbase and engine size. The VIN provides information about the vehicles involved in collisions that is not otherwise recorded on the report forms. Computer software is available to decode the YIN for further a.nalysis. . '. In some jurisdiaions, other data are available in addition to, or instead of, a police collision report, such .as individual collision repons of drivers and/or witnesses. A driver or witness form is usually shorter than a police collision report form. For some collisions, police or other investigators collea supplemental information. Photographs, sketches, calculations, notes and other information may be on file, especially for more severe collisions. Police officers usually complete collision repons during the shifr they investigated the collision. Once an officer su'bmits a report, a supervisor or other personnel may check it and rerum an incomplete or obviously inaccurate report to the officer for correction. The proportion of police collision reports completed on a computer and submitted elearonically grows each year. Although every attempt is made to correa ertors in the collision reporting, some can and do still occur. Analysts should review these reportS critically and with caution as they have been prepared using obseiY.lcions and statements of wimesses who are prone to error and often subjective. Although not considered part of the collision report, another useful tool becoming more popular for crash analysis in many countries, and particularly accident reconsuuaion and investigation, is the "black box.· This hardwue device is on most new model vehicles and stores data on events just prior to the collision event (SWb, 2008). In the event of a crash, the boxes will give clues as to why the collision happened. However, many other countries have not moved further into this arena because of fears the box could infringe on drivers' civil liberties.

3.2 Sources of Data AU police agencies keep collision reports as records for use by courtS in legal proceedings, by engineers in traffic safety improvement projec;ts and countermeasure analysis, and for other safety-related activities. Collision records are stored decuonically to allow easy access to the inforcnation by safety engineers, police, research groups and various transportation entities. Computers can store and quickly process collision data from a small city for many years, and seven! software programs are available for typical applications. Typically, the original copy of the collision report is sent to a cenual agency for processing and storage. Stare agencies routindy check for missing data or ertoneous keystrokes. The costs of ente.ring collision data into a database can be significant. Previous srudies have indicated agencies computerize becween 45 and 250 data items per collision, with a mean of 110. States are increasingly using the narrative description of the collision in the report form which allows key word searches such as driver di.straaions (cell phone usage, billboards, etc.), sight distance issues, or rdated issues such as sun glare (FHWA. 2004). In addition to the

already time-consuming collision rep<~rting process, many entities require separate filing of collision rep<~rts for insu ranee companies, adding additional constraints. 'Most agencies can quickly provide analysts with copies of, or electronic access to, requested collision data via software designed to poll various parameters noted by the analyst. Stares will erase data items identifying individuals to prot~r privacy when providing data to outside agencies, a private company, or an individual. Collision report forms are cypically available 4 to 6 months following a collision, but can be available earlier in many jurisdictions depending on the method of data entry. There are various ways in which collision data in a collision report form relate location information. In most jurisdictions, police and motorist report forms ask foe the name of the highway where the crash occurred and the distance and direction of that point from some known reference p<~int, such as an intersection or milepost. The fo rm may also request the address of the collision location. The informacion on the form is usually translated into the standard location sysum of the state at the time the data ace entered into the computer. Instead of a highway name, the agency may enter a code number identifying the highway. Instead of the distance from the intersection nearest the collision, the agency may r~rd ~e distance from the beginning of the highway or the distance from an established reference point such as a major interSection. Many police agencies are now geo-locacing crashes wi.th global p
• Mapping: Most agencies have geographical information system (GIS) programs in place that have various layers of data that could be useful co the safety analyse. ln the GIS, basic data on roadway geometry may be linked co roadway inventory d.:ua on speed limits, signagc, signals, barriers, etc., as described in Chapter I 5. • Traffic Volum~: Tht rnajority of agencies collect traffic dara using various methods such as permanent auwmacic traffic counters, pneumatic tubes, or manual turning-movement counts as described in Chapter 4. Traffic volumes are used in collision studies co capture uexposure," or the chance that a particular entity is involved in a crash.

3.3 Collision Data Reduction Once a database has been chosen, the analyst muse separate the dara needed for the particular study from the remainder of the file. Reducing data is a much easier process now with the availability of the personal computer, database software programs installed on most computers and collision reducing softw:ue used by rransporracion agencies for compiling collision data based on many types of pafa.!Ileters. The rwo most common parameters used for collision data reduction are timeframe and location of the collision (such as the intersection). However, ocher types of data are often uciliud such as weather, time of day, collision type, etc. 3.3.1 Choosing a Tunefrtmu

One of the lim steps in this reduction process is to choose an analysis time frame and discard oudying data. In some studies the choice of a time frame is self-evident (for instance, the study is to see whether the collision rate in 2010 cliffered from the rare in 2009). When analysts must choose a rime frame, 3 years is the most common choice. rtu"ee years represents a compromise becwtW the desire for larger samples and the desire for cirneframes within which conclitions were unlikdy co have changed a great deal. Also, a 3-year cimcfrarne eliminates many problems with cliscarded daca. Timeframes of up to 5 years are common, but for periods longer than 5 years, analysts must use special care co ensure changes in background conclitions (traffic volume, land use, traffic control, geometry, ere.) can be tolerated within the scope of the study. T uneframes ofle:ss than 2 years may be necessary, but a smaller sample siu could constrain the study. It is good practice to check that collision data for parycular locations have not been biased by construction or ocher major but temporary traffic events during the cirnefraine selected. Analyses can consult highway agency records for construction and/or project lectings, but such records are often incomplete or inaccurate and cartnot be trusted alone. Supervisory and/or experienced area highway personnel can estimate whether major construction projects or other aaffic events have taken place, or at least give some hint of construction which can be further investigated. However, those estimates should not inspire great confidence without corresponding documentation. A scan of the collision data co see whether any reports mention construction or work zones is often reassuring. 3.3.2 Analyzing Specific Locations

Many sruclies focus on one location or a limited set of locations on the highway nerwork. One of the analyse's important casks in these srudies is to reduce the database to crashes chat occurred at locations of interest. Analysts usually summarize collliions into those that occurred at spors and those that occurred in roadway sections. Spots are short segments ofh.igbways that help identify problem "point• locations such as intersections, curves and short bridges. The highway cross-section and ocher fearures at a spot should be noticeably different from surrounding spots. Sections are longer, relativdy homogeneous segments of highway convenient for studying cross-sections, pavement surfaces and other longitudinal fearures. Roadway sections will rypicalJy correspond to tangent sections of roadways. Spot lengths of 0.2 to 0.3 miles (.32 to .48 kilomece~ [km]) and section lengths of 1 co 2 miles (1.6 ro 3.2 km) are recommended for most agencies (TRB, 2010). Spots and sections that Boat are usually more desirable than chose that begin at fi..wl increments. For example, suppose that many collisions occurred near a highway fearure at the 0.6-mile (.97-km) mark. Ic is likely the police would code some of these collisions as occurring at the 0.5-mile (.8-km) mark and some as occurring at the 0.7-mile (1.13 km-) mark. If the agency used a fixed spot length of 0.3 mile (.48 km), a spot would end at the 0.6-mile (.97-km) mark, so SO!Jle of these collisions would fall in one spot and some would fall in the next. By contrast, a Boating spot (sometimes referred to the ·sliding window") definition might have 0.3-milllong (.48-km-long) spots beginning every 0.1 (.16 km) mile, overlapping each other. Thus there would be a spot in uding all collisions coded as occurring at the 0.5-, 0.6- and 0.7-mile (.8-, .97- and 1.13-km) marks, and the agen would obtain a more accurate estimate of the uue number of collisions at the feature location (TRB, 2010). 354 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDffiON

Am.lyz.ing crashes ac incerseaions requires a precise defini tion of the limits of rhe intersection. Some agencies include coLlisions within an intersection if they were coded as occurring on che ap proaches, up ro a fixed distance usually In · the range of 150- 500 feet (ft.) (46--152 (meters [m)) from the incerseccion depending on such factors as the :unoum of congestion, queue spillback, the length of rurn bays, or the length of the approach legs in relation to nearby iote~ secrions. A more rime-consuming buc technically defensible definition, especially in areas with chronic congestion. IS to examine each collision on an approach co determine whether the intersection was an influencing factor. With the latter definicion, the intersection size varies and basically extends co the back of the queue on the appro aches. Many agencies now provide a geo-coded reference point for every crash location if the exact location is known· .In addition, in many agencies, Incerner and reporc-based informacion is provided on various regional crash rates for stgnalized and unsignalized imerscccions and roadway links by functional class. If data is provided in this format io your district, it may be easily uti.lized; however, caution should be used to make sure rhe method for ranking is knowll up &one. For instance, if a floating method is noc used along links, it might not be advisable to use chis type of data for choices such as countermeasure deployment.

3.3..3 &during CollisWn Data &ued on Other Va~bles The collision types or vehicle incenc patterns analyzed during a particular srudy should always be those w hich are most closely related to the subjeCt under srudy. The subjects of some studies obviously rdat~ directly ro certain collision types. For example, a srudy of crashes occurring on an approach with a new left-rurn signal treatment should aasnine left-turn collisions or collisions involving left-turning vehicles, among ocher coLlision types. For some srudies it is fl?C obvious which collision types should be examined, so previous srudies and engineering judgment must play a role 10 the sdection. Analysts may wish to reduce their databases based on several variables. They may have an interest in certain times of day, monchs of the year, weather, types of vehicles, driver conditions (such as only drivers coded as "had nor been drinking") and so on. While reducing ~e database, analyses should be cautious nor to delete data that may beco.rne us_efUI.later. It is generally less expensive to sort a larger database several times than to trim down to a smal.l dacab~se and Aiscover needed informadop has been lose.

3.4 Merging Collision and Other Data Some agencies maintain files on roadway features, traffic volumes, intersection characteristics, construction and maintenance histories, traffic control devices (fCDs), or other traits. The ability, to merge these files with coLlision d:oJ.Ca files is essential for many analyses. Indeed, the ability to merge large files quickly is one of the main advanages of clearonic collision data files. An example of a common type of merge is to combine collision and traffic Row data co produce collision rates (collisions per I 00 million vehicle miles for highway seaions and networks, or collisioos pc;.r million entering vehicles for spocs). When combining collision dat:~ with other forms of data, it is very imporunt to remember to save backup files before overwriting or adding other data. To execute a merge, analyses need a data dement in common between the files to be merged. In manual merga, cbis common dement will mosc often be the name(s) of the highway(s). In electronic merges, analyses mosc often usc c:Pe codes and mileages identifying the spots or segments. Ir is best if these codes are common to many different files. I t is also desirable that distances are measured in a uniform way (from the same points with the same units) in differ-= ~t data files. Some departments of transportation have built these common codes and measurements into all their inve~ tory and coLlision dat;a files. In the merging process, analyses must change the format of one of the files. Collision files are usually arranged j_n.. a format of one record (line of daca) per collisiqn. Traffic. highway characteristic and ocher data files, however, areus-&Jally arranged in a format of one record per segment or fearure. Typically, it is the collision data format that must t::::>e changed. Analyses might creace a file summarizing collisions that occurred in particular segments before mergill ~e data with a file providing average annual daily traffic (AADT) by segment. Collapsing coLlision data in this waylos .::s some informacion. When summarizing collision data for a merge, analyses should nor purge too much of the filc.IC .1S less expensive ro have unused information in a file th,.n to discover needed information has been lose during thew:-Jj!§e and muse be reconstrucced from the original file.

Traffic Collision Studies • J~ 5

3.5 Recognizing and Dealing with Problems in Collision Data Many confounding factors occur when a colfuion We$ place; me collision itself is a random event, some collisions are not reported (especially PDO). Errors in the reporting process occur frequently wim limited time for me office to input information into a database. There are a large number of inputs included wim me coUi.sion report, which the police officer must choose from and typically there is little: available information a~ut me collision causal chain. The following subsections outline some of me problems, picfalls and limitations that could affect collision studies.

entire

3.5.1 RAntlom NIUure ofCcllisio'IIS. When analyz.ing collision data, the greatest limitation to recogniz.e is the number of collisions occurring at a location or tO a particular driver is not the same thing as che "safety" of che location or driver. Colfuions are rare and somewhat random evencs-high-collision locations or drivers may be "unlucky" locales or people racher man imsafe ones. At best, colfuion data provide only an estimate of che safety of me location or driver (that is, che number of coUi.sions that would occur in an infinite time span) ifconditions are the same in chc: future as chey were in chc: past. However, because of various ocher problems and limitations on the use of collision data, even this estimate is difficult to obtain.

3.5.2 Unreported Cc/Jisio'/1.1 Motorists do not report all traffic collisioru ro the police. The major reason that collisions are not reported is they wete not serious enough. Most agencies have thresholds of property damage below which police decline to investigate a collision. For many agencies, this threshold currently is around US$1,000-$1,500. Some agencies do ~or prepare reports on any collisions mar do nor involve an injury. Other agencies have established dual thresholds-a higher threshold for police reports and a lower one for reports prepared by motorists. Crashes are unreported for reasons besides a reporting threshold. Sometimes motorists do not report collisions for fear of higher insurance rates. Police probably will not file: a report on a coUi.sion that bccurred on private property. They may also be too busy to investigate collisions even when informed. This is more: common in larger cities. There are two key points for analysts regarding unreported colfuions. FICSt, analystS relying on reported collisions will underestimate the total number of collisions at a location or set of locations. Past srudies estimate {Hauer and Hakke.rt, 1988) chat: • the number of fatalities is generally known within ±5 percent of the true number; • the number of injuries requiring hospitaljza.rion is underreported by about 20 percent; • only about half of all injuries occurring in coUi.sions are reported; and • motorists report fewer dun half of all PDO collisions. Analysts should include an awareness of this undere$timacion in reporu by writing in terms of "reponed collisions· rather dun merely "coUi.sions." The second key point regmiing unrepo.rtod coUi.sions is that me extent of underreporting' varies by coUi.sion rype. driver age, location, time and many othc.r variables. Analysts making comparisons across such variables must assunne the underrepo.rting is constant and negligible within the scope of the particular srudy, or they mwt correct f~r underreporting. Analysts sometimes correct for me most obvious reasons for underreporting, such as changing reporting thresholds, but rudy for less obvious reasons. As often as practical, agencies should determine che propo.rtion of collisions reported for impo.rtanc variables and avoid naively assuming uniform reporting.

3.5.3 Emnu1ous Dat~t Collision reports often contain errors, and the number of errors grows with every step of processing. In fact, researchers have found an average of between 1 and 2.2 discrepancies per coUi.sion between the collision repon and me coded file {Hauer and Hakke.rt, 1988). The errors arc not evenly distributed throughout the ccpo.rt, so it is very difficult to generalize about erroneous data. Often, the hard copy of the repon is wed co confirm input variables used in the analystS' 6.nding1. lUnge and logic checks will reveal some errors in colfuion files. Analysts conduct range checks wich programs that compare the values of a variable to acceptable limits for chat variable. The program will Bag a particular report if its value is outside the limits. Logic checks compare the values of two or more variable$ within a collision record, looking for intcrrul consisrcncy. For example, lcft-rurn collision types arc impossible on freeways aO!pt where: vehicles can make U-turns through median openings. When range and logic checks turn up erroneous data, analysts could ,.,,

...

• A 1\ Uti At t"\t: TO 1\~I('Qf"\QTf'\TifUI C:t-ll":lltoiC:C:Dif\lt": CTIIt"'llt:C

")~If'\ S:niTlt"'U\1

delete the offending collision report or delete the offending value. If the error occurs in an important variable or if a ·limited sample of collisions is available, the analyst may need to consult a hard copy of the collision report and read · the narrative or consult the drawing. Consider deleting the entice collision tep
Most crash reports contain oruy the last ~d most obvious links of the collision cause chain. Police who ace investigating collisions do not have the time or, in mmy cases, the training to seek links farther up the chain. Police often use favorite standbys such as •driver inattention• when other causes ace obscure. Most analysts believe driver and witness reports ace virrually worthless for providing crcdlble information on collision causes. Indeed, it is very difficult for even highly trained investigators with sufficient time to construct m accurate chain. Thus, in the case of a singlevehicle, run-off-road collision, the object that was struck on .the roadside will be prominent in the report:. Analysts may tend to seek treatments for this "cause" while ignoring driver, vehicle and highway factors that preceded the vehicle leaving the road. That tendency can be resisted by searching the collision n:p<~rt for other clues to links on the chain that are not prominent. Another way to attack the collision-cause problem is to malyze smaller samples of more detailed collision data when designing ways to break the chain. ' 3.5.5 limit4tions ofthe Report Form Collision r~port forms have resulted from balancing competing interests. One pf the major interestS is the police; whose time is·required to complete the report. The report form thC:CCfore is likdy to be missing dements that traffic engineers would like to use. Usually, traffic engineers consider infor1nation on the highway to be insufficiem on collision report forrns. Information on the vehicle is also lacking for some analyses, especially analyses involving trucks and other large vehicles and their contents. Driver and passenger information on collision report forms is usually more than adequate for traffic engineering uses. Analysts should check the standard form md the coded collision files to ensure the informacion needed has been collected before launching a study.

4.0 ANALYSIS OF COLLISION DATA The previous sections of this chapter dealt with how to select and prepare collision data for m analyois. After those tasks, the analyst is ready to make some type of inference about the •saretf of the highways or drivers of interest. In the following sea:ion, four different types of collision analyses are described: summarizing numbers of collisions and uend.s, identifying haurdous locations, selecting collision countermeasures and evaluatins existing countermeasures. Those four types of studies encompass most of the current uses of collision data by practicing traffic engineers. Other types of studies ace possible and useful but typically fall within the researcher's domain. The focus in the following sections is on identifying and correcting crash-prone locations, because that is the primary task of most transportation engineers in agencies who conduct collision studies. Those interested in idemil)ing and Traffic Collision Studies • 351

correcting crash-prone drivers or vehicles can, in most cases, use the same methods while substicuting the words "driver" or "vehiden for "location.~ Engineers arc typically more interested in focusing on countermeasures that can reduce the likelihood of future collisions in the physid environment (for <:xample, sight lines, grades, speed limits, ere.) and nor collisions focused on things char cannot necessarily be controlled (such as state of mind, cell phone use, etc.) by typical coumerrneasures. In addition, countermeasures focused on the physical surroundings are more easily anal}'led because collision records typically focus on causes in the physical environment.

4.1 Number of Collisions and Trends Engineers, policy-makers, the news media and ochers often wane ro know the number of collisions of some type that occurred ar a location or set oflocations during some time period. Such summaries are useful for comparing highway safety with ocher competitors for funds, for noting trends with time (chat is, annual summaries), or for grasping che magnirude of chc problem. The numbers arc usually very easy to obtain, either by manual coums or by manipulation of a database through written functions, once che file has been prepared. Chapter 3 and Appendix D provide information on how to prepare graphics, text and presentations. Analysts preparing summaries of the numbers of collisions or trends should provide the: audience with more than just a number. The main tasks facing an analyst presenting numbers or trends arc to put them in perspective and alert chc audience to any suspicions regarding accuracy. The audience might need to know che number of collisions is a random variable that flucruates through time and many collisions are unreported, for = pie. In displaying a trend of collisions chrough time, che analyst should note for the audience any major faaors char may have in£1uenced the trend. For example, the reporting threshold may have changed, inHation may be significant (it takes less d,amage to require US$1,000 of repair chan. it used to), rravel habits may have changed, groWth in the area may be taking place, or economic e£feas may have reduced the number of vehicles on the roadway (exposure). Summaries of collision data must emphasize for the audience whether statistics presented are "injury collisions" and "&tal collisions," or the "number of injuries" and the "number of &talities." Either type of measure could be important to the aims of a particular srudy. Analysts rate the severiry of a particular collision based on che most severe injury experienced by anyone in the collision. Thus, if in a collision at least one person experienced a B injury but no For A injuries occurred, an analyst would label the collision a B-injury collision. There could have been several B injuries and several C injuries recorded for the B-injury collision. The severity rating of a collision thus says little about the numbers of persons injured or killed. If none of the participants in a collision reported injuries, analysts refer to the collision as a PDQ collision. Involvements account for the number of individual vehicles in a crash. A three-car collision indudes three involvementS; a single-vehicle collision includes one involvement. Analysts comparing different classes of vehicles or drivers must not confuse collisions with involvements when looking at multiple-vehicle collisions. For example, consider a location where older drivers trulke up 30 percent of the driver population and collision data show at least one older driver is involved in 50 percent of the two-vehicle collisions. Many analysts would conclude older drivers are overceprcsemed and start designing countermeasures for them. However, a dose look at involvementS rather than collisions shows older drivers are not overrepresented in this case. If older drivers are no more likely to be involved in collisions chan other drivers, the fuUowing staremcnts are true. • The probability chat an older driver has collided with another older driver is 0.3 x 0.3 = 0.09. • The_probability that an older driver has collided with a younger driver is 0.3x 0.7 =0.21. • The probability chat a younger drive r has collided with an older driver is 0.7 x 0.3 ="0.21. • The probability that a younger driver has collided with another younger driver is 0.7 x 0.7 "0.49. Therefore, the total probability that a collision involves one or more older drivers is 0.09 + 0.21 + 0.21 • 0.51. Since 0.51 is so dose to che aaual proportion of older drivers involved in collisions (0.50), older drivers are most likely not overrepresented. Analysts comparing driver gender, vehicle types and other similar faaors must separate inw/vmzmts and coUisions.


4.2 Identifying Hazardous locations-"Network Screening" . Due co limited budgecs, it is essential agencies making highway safecy improvements direct resource s to real prob..Jem locations. Good litigation risk management also demands agencies idencif}t collision-prone locations through a logical process. Thus engineers have developed procedures co identifY collision-prone locations using performance measures based on collision data, many of which are descri bed in detail in the HSM (2010). Although che I-ISM describes a doun or more methods, this section presen tS the mosc commonly used of those procedures. A discussion of which procedure a parcic~lar study might use and the drawbacks of each follow che presentation of all che procedures. The procedures not described in chis manual arc left out because they are complex, not used frequently and/or well understood through a basic d escription. Some of the more complex identification methods use Empirical Bayes (EB) merhodologies co accounr for regression-co-the-mean (RTM). RTM occurs when locations wirh high collision counts during one time period experience more normal counrs during che no:t rime period even if no causacive factor changes. Although many pracci cioners do noc cypically use the more compla methods, chcy should know RTM is important and work toward using methods co identifY hazardous locations that account for this phenomenon. Lastly, FHWA developed a software package named SafityAnaiyst, which can be used by Various agencies.

4.2.1 Spot Maps A spot map is a simple and effective way to determine collision-prone locations qualitatively. Engineers create spor maps by marking the location of each relevant collision on a map. For a small area or a limited number of collisions, a map hung on a wall with pins to mark collisions is useful. Computer graphics such as in a GIS easily allow spot maps of large areas or with large numbers of collisions co be updated and displayed on a monitor. AnalystS can use different colors and siz.es of pins or graphical symbols to represent different cypes or severities of coUisions, or to represent "multipliers" of co.llisions (for instance, one large pin/symbol equals 10 collisions or red equals a fatal collision). Spor maps are useful forspecialized situations such as pedestrian collisions or parked-car collisions.

4.2.2 Co/JisUm Frequency Some agencies identifY locations through lists of locations (spots, sections, intersections, etc.) ranked by rhe coc:U number of reported collisions, by the number of collisions of a particular type, or by the number of severe collisio.ns. The primary virtues of this approach arc chat ir is simple and makes intuitive sense. If the agency goal is to minimize total collisions, arcacking the locations wich the most collisions seems logical. However, 'the analyst must undcrsra.nd this merhod of site selection almost always chooses heavily traveled sections of roadway (higher vehicular exposure w ill have a higher probability of being in a collision), and therefore ignores less busy sites that may have significant problems chat need addressing. In addition, this method does not account for RTM. However, using longer time periods does help negate some of these effects.

4.2.3 Co/Jisitm RAtes Agencies have also identified collision-prone locations through lisrs of locations ranked by collision rate. Agencies usually compute highway stction collision races in terms of collisions per 1 million vehicle miles using Equation 17-1.

RSEC

= 1,000,000 * A

365 * T• V •L

Equation 17-1

where:

RSEC

= collision rate for the section

A

= total number of reponed collisions

T

• time frame of the analysis, years

V

= AADT, vehicles per day

L

= length of che section, miles

Traffic Collision Studies • 3sP 9

For highway sections, collision rates arc typically expressed in collisions per 100 million vehicle miles. In this instance, RSEC should be calculated using rhe constant 100,000,000 instead of 1,000,000. For pot:s, agencies usually calculate the collision rate in terms of collisions per million emcring vehicles using Equa· tion 17-2.

Equation 17-2 where:

RSP • collision rare for the spot A

= total number of reported collisions

T

=

V

= AAOT, vehicles per day (for intersections, this is the sum of the average daily approa.ch volumes)

time frame of the analysis, years

An example calculation is provided in Example 17-4 (sec page 366), which references Exhibit 17-8, where the rates arc calculated for each intersection ("spm" location). Obviously, ranlcing locations by collision race requires traffic volume dau. The time period of the volume qau should match the time period of the collision data being analyzed. Analysts may use volume dara somewhat outside the collision data time period if they adjust the volumes for temporal variation (growth) and seisorulity (ste Chapter 4). For high growth areas, analysts should use volume data that arc 5 years or more removed from the time period of the: collision data with extreme caution, if at all. Like the frequency method, RTM could cause a problem with th.is method as well. Analysts should note the n.te method of identifying hazardous sites is likely biased in &.vor of sites with low exposure (the opposite bias of the frequency method) because only a couple of "unlucky" collisions on a spot or section with low exposure will produce a relatively high rare. l..astly, there is no apparent threshold value to determine if the site is indeed luz.ardous; however, the HSM docs provide a methodology for calculating a "critical rate" which can be used to provide a threshold.

4.24 kcountingfor Severity Unng Equntdmt ~erty D~~mAge Only An_alysts can adjust collision frequencies or collision rates to reflect the greater costs of injury and &.tal collisions. One co=on method of caking severity into account before ranking locations is to compute the number of equivalent property damage only (EPDO) crashes (NCHRP. 1986). The method uses a weigh ring factor, which is the number of PDO crashes that would have to take place that society would deem them "equivalent" to one &.tal collision or injury collision. The weighting factor, w, is calculated as follows:

cc,

W=-' CC,00

Equation 17-3

where:

weighting factor based on crash severiry y

W1

"'

cc;.

= ·crash cost for crash severityy

CC,00

•

crash cost for PDO crash severity

The EPDO rating is calculated as follows:

EPDO Rating • W~c(/0 + W,(J) + WpDo(PDO)

,.." • MMIIIlll f"'~ TRAN<;PC"lRTATlC"lN FNCiiNFFRINCi '\TIJDIES. 2ND EDITION

Equation 17-4

, where: ; WA: I. PDO"' weighing factors for each crash type

K

= number offatal crashes

I

• number ofA, B and C injury crashes

PDQ

• number of PDO crashes

Weighting factors are typica11y based on crash cost estimates developed and upda~d by FHWA (FHWA, 200 5) or regionally calibrated estimates (such as those collected at the state level) based on specific region.allevd collision and C!JSt data. For this reason, different agencies may use different equivalency factors. For example, in 2008 North Carolina used wcighcing factors of76.8 PDO collisions per fatal or A-injury collision and 8.4 PDQ collisions per B-injury or Cinjwy collision. The arialystshould note some agencies may separate collision severity types differently than others; therefore, the equation can be easily manipulated. Analysts should keep in mind that chis methOd is generally biased towards locations that' have a higher proportion of severe crashes (such as low volume rural locations). RTM is again problematic using this method because sit es identified as having a high EPDO rate would likely rerum to a lower rate even if no coumermeasure were implemented. Last, there is no obvious threshold value to determine if the site is indeed hazardous. EXAMPLE 17-ls Application of the EPDO Method State •A• has provided the crash cost estimates by severity in Exhibit 17-3. Determine the EPDO scores for the foU! signalized intersections describ~ in Exhibit 17-4. ·

s~ · lntaaudon

Total

K

A

8:

c

0

EPDO

Traffic Collision Studies • 361

Solution: The weight factor and EPDO rates will need to be calculated for each of the four sites. Calculations of weights for fatal and injury collisions are:

$97,500

»-:t .. $9,300

=

10

The EPDO value for Intersection 1, in equivalent property damage only collisions over the time period of interest, is calcubted as:

EPDO Rating = 274(2) + 10(12) + 1(13) = 681 Exhibit 17-4 shows the EPDO values for lnrersecrions 2-4. Intersection 2 has the highest EPDO rate of 1,498 equivalent property damage only collisions, more: than double that of any other site. Note that if the: sites were ranked by frequency instead ofEPDO, Intersection 2 would have been last on the list. The high EDPO rate is primarily due to the. large number of fatal crashes at this site:. The rdacive severity index (RSI) method is another (similar) method that may be useful for analyst.

4.25 RelAtive &verity Index , The RSI is a simple identification method that uses crash costs developed for specilic crash types to determine if a site is in need of further. review (HS¥, 2010). Societal crash costs are typically determined at the jurisdiction or sratc: level to rdlect local conditions; however, FHWA also provides crash cost estimates by qash type at the federallevd. The RSI is calculated as follows.

tRSI1 RSl = J.:!___

Equation 17-5

NO.y

J'

where: RS~

• average RSI costs for the segment or spot y

RS~

= RSI cost for each crash type j

No., • number of observed crashes at site y The RSI for the segment or spot is determined to be hazardous or in need of funher revkw if the: RSI is greater than the average RSI for the populatiofi. The average RSI of the population is calculated as follows.

tRSI,

RSI, =-7---

Equation 17-6

LNo,, ,..

where: RS~

= average RSI costs for the reference population

RS~

• total RSI cost at site y

N0.r = total number of observed crashes at site y Analysts should keep in mind this method is generally biased toward locations that have a higher proportion of severe crashes. RTM remains problematic since sites that deem themselves as hazardous based on high RSis in one year would likdy return to their ?atural average the following year. 362 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

EXAMPLE 17-2: Application of the RSI Method Determine the R.Sl for rhe following four signalized intersecrions in the local jurisdiction. Exhibit 17-5 provideS the crash data. The scare has currenr crash cost esrimares by crash rype, provided in Exhibir 17-6.

Signaliz.ed Intersection I Rear End

'

<'

~

•

. ,•·

.i!":~~~~,-~. . :~F • -;._ •· • ··$· . .

r . . .,~o · · -~ ,

.

~

,

"'

.

·

:

;£~1· .

... ."1:, .. ,

__

:~· -.,rifi•r.;.,..£......-..-:--;r-

•· ...~

.·,A~tj~'f~'f'\n'11t\~~~-~~~~i~i;~

Crasb Type aod Costs Sig. lnc:x

Rear End

Sideswipe

$ 12,000

$9,500

Faed

Angie

Pcd.Bicycle

Head On

Object

Other

$74,000 '

$115,000

$72,000

$83,000

$47,800

~·

Total

RSl _

"'ol'ti'!"'"''"'" .,. .

,. ,.;•ji

~-~s~~~.t · --~~~--~

I

$60,000

$19,000

$740,000

$230,000

$72,000

$249,000

$239,000

$1,609.000

$57,46~

2

$168,000

$47,500

$296,000

$0

$332,000

$191,200

$1,034,700

$33.377~

3

$72,000

$0

$370,000

so so

$216,000

$166,000

$334,600

$1,158,600

$50.374

4

$132,000

$9.500

$148,000

$1 15.000

$0

$498,000

$143,400

$1,045,900

$43.579

Tow RSI •

$4,848,200

Avecage·RSJ •

$45,738

-

Solution: Calculations of the R.Sl for Intersection 1 and the RSI fo.r the reference population are shown below, wich the solutions 'for the R.Sls for Intersect.ions 2-4 shown in Exhibit 17-6.

Rear End: 5 x $12,000 = $60,000

Head On = 1 x $72,000 = $72,000

Sideswipe= 2 X $9,500 • $19,000

Fixed Object= 3 x $83,000 • $249,000

Angle= 10 x $~4,000 =$740,000

Ocher= 5 x $47,800 = $239,000

Ped./Bicycle z 2 x $115,000 = $230,000

RS!t •

$1,609,000 =$57,464 28

RSI • $4, 848• 200 = $45 738

,

. 106

,

Exhibit 17-6 shows the R.Sls for Intersections I and 3 are greater than the average RSI and should b; consideredfc:?r further evaluation. ' Traffic Collision Studies • 3& 3

4.2.6Rtlu Qwdity Control MnhoJ The rate quality conuol {RQC) method wes a statistical test to determine whether the traffic collision rate for a particular intersection or roadway segment is abnormally rugh when compa~d with the rate for other locations with similu characteristics (Stokes and Mutabazi, 1996). A number of agencies have used RQC for several yeus. When using thls method, the engineer should know the following assumptions apply. • It applies only to collision rates, not frequencies. • It assumes the number of collisions at a sec oflocations follows a Poisson disuibution. This is a wellaccepted assumption in the safety field, and analysts can verify it using collision data from a representative sample of sites. • It compares the race of a particular location to the mean race at similar locations rather than at all locations.

RQC can apply to spots or sections. For spots, analysts use collisions per million vehicles. For stctions, analysts use collisions per million vehicle miles (mvm) or per 100 mvm. The RQC method Bags a location as hazardous if it satisfies the following inequality.

xs)o.s 1 ( v, +-2V,-

· Equation 17-7

OBRI >XS+K where

OBR; .. crash rate observed at location i

XS

= mean crash race for locations with characteristics simil2r co those oflocation i

K

z

~

• volume of traffic at location i, in the same units as the crash rates are given

constant corresponding to a levd of confidence in the finding

Agencies commonly use 90, 95 and 99 percent levels of confidence, which correspond to K values of l.282, 1.645 and 2.327, respectively. The question of which locations ue similar enough to include in the computation of XS is a difficult one. Generally, agencies have wed relatively broad definitions of similarity to compute XS. For example, one agency used statewide average rates based only on intersection type (for instance, arterial meeting collector in an urban area) and traffic volume to provide XS. EXAMPLE 17-3: Application of the Rate Quality Control Method

Roadway Section AA had 40 reported collisions in 3 years, and the agency responsible for the ~ection estimated travel on the section was 19 mvm during that time. The mean collision rate for all sections in the jurisdiction similar to Section AA was 140 per 100 mvm. Should an analyst fbg Section AA as haza.tdous with 95 percent confidence?

So/mum: The RQC method requires the same units for each variable in the inequ.ality, so the analyst should convert XS to 1.4 eollisions per mvm to be consistent with the units given for V, OBR is 40 collisions/19 mvm s 2.1 collision.s per mvm. For these daca, the rate qu.ality concrol inequality {Equation 17-7) above holds crue as follows.

XS}o.s I

OBR> XS+ ·K ( - +-

V, .

2~

4)o.s 1 1. +-2(19) 2.1>1.4+1.645(19 2.1 > 1.9 Therefore, the agency should consider Section AA hazardous with 95 percent confidence using the rate quality concrol

method. '\11:4 •

Mt.l\ll l t. l ()F T Rt.N<;P()RTt.TI()N Fl\l(;INHRII\l(; <;TIIniF<; ?Mn

FOITI(\~1

4.2.7 "Sites with Promise" The methods of identifying hazardous locations for treatment described in the sections above have important flaws. Bias is evident in some methods with a tendency to choose higher or lower exposure sites. Other methodologies make 'assumptions of distributions of collisions which are inherently random occurrences and are hard to predi ct. Ratebased methods require volume data that are expensive to collect and may have large errors. The method of"Sices with Promise" (SWP) was developed to overcome some of these flaws and provide some logical and defensible basis for recommendations (Hauer, 1996). The SWP method seeks to find fixable sites, nor necessarily sites that are the most hazardous, so that safety funds ace optimized. This method reduces the data collection burden to the analyst because the overall frequency is the primary building block, with less emphasis on rate calculations requiring exposure data. In addition, this method is more proactive by identify up-and-coming hazardous locations and taking advantage of proven countermeasures. The SWP method uses five conditions (A- E) for choosing promising sites. A;! though five conditions are presented, it is not necessary to usc each one. Instead, the analyst should only usc conditions applicable for his or her study. Following application.of d)e conditions, engineering experience and judgment is necessary to choose which sites should be selected for potential coiintermeasure employment. The five conditions are outlined below. Condition A: "Do I have a good countermeasure I would like to implement in my jurisdiction?" Often, engineers can have success in obtaining funds for a targeted countermeasure program for which there will be some certainty of good collision savings. In addition, it is often easier to implement a targeted countermeasure program, doing one thing at many locations, than individual treatments at scattered locations. Sites should only be considered if the counterm~ure would make sense at that location. The frequency (F) of collisions is used for the collision type being targeted as needing some treatment. Only 2-3 years of crash data are necessary. The condition should not be.C()nsidered iffunds are not available for countermeasure implementation. Condition 8: "Ate there newly co~structed or rebuilt sites that may have some deficiencies present?" Only sites . that arc recendy constructed or rebuilt would be considered in this calculation. The frequency (F) at each individuallocation is compared to the mean frequency (F,) of similar sites using Equation 17-8. ~-F..,

Equation 17-8

(}'F

This condition should be checked shorrly after new or rebuilt sites are constructed to correct deficiencies. Usc of all the available collision data possible in the after period is important. Typically, very shott periods of time are all an analyst has in identifying a sire; however, whole years of data· are best·for proper usc of this method. If shorter time periods (< 1 year) are all the analyst has available, it is advisable the collisions be multiplied by a time period ratio for proper comparison. For instance, ifonly 4 months of collision data were available, a ratio of 3 (there are three 4-month periods in a year) would be used. If 10 collisio~ were reported during this 4-mont:h period, the proper comparison would be 30 collisions (10 colJision.s x 3 =30 collisions/year). It is important to note using this ratio method has inherent flaws (the 4 monthS of data may be during the worst collision period of the year); therefore, if the newly constructed site is considered deficient, use caution and good engineering judgment. I..asdy, it should be noted forming groups could be difficult if sites were not previously defined. Condition C: ·Ate there sites that have rapidly deteriorated in recent years?" All sites should be checked for deterioration, not just a subser. The analyst should look for spikes in the collision frequency (F) on a regular basis, say every year. Usc as much crash data as reasonably possible, preferably 10 or more years. Condition D: ;.Are there potentially hazardous sites we might be missing because they are low-volume! lowcollision sites?" Thj.s criterion accounts for exposure and makes sure all sites have an opportunity to be identified (not just large, high-volume sites). Only2-3 years of collision data are necessary; however, exposure data in the form of traffic volumes will be necessary to calculate a rate. This check should be done on each site every )-10 years. Refer co tate calculations for spots and segmentS in Section 4.2.3. Condition E: •Are there newly constructed or rebuilt sires that may have some deficiencies present but could be missed because they are low-volume, low-collision sites?" This criterion is similar to Condition B but instead looks at <;ollision rates. Only sites that are recendy constructed or rebuilt would be considered in this calculation. The rate (R) at each individual location is compared to the mean rate (R.) of similar sites throl!gh the·caJtula; cion bdow: Tr~Ui ...

r n.llidnn C:.tllriiDC:

•

~f:.lt

Exhibit 17-9

R, - R., (JR

This criterion should be checked shordy after new or rebuilt sites are consuucred ro correct deficiencies. The same basic principles applied in Conrlilion B can be used for short lime periods of collision dara. Ir should be noted forming groups could be difficult if sires were not previously defined. Analysts should keep in mind the SWP method does not account for KI'M. Its basic premise is that it uses the combined strengths of the frequency and rare methods to determine sires where countermeasures could be most productive. EXAMPLE 17-4: Application of"Sites with Promise" In an attempt to decrease overall collisions, a city has decided to screen for hazardous signaliied intersection locations by looking at angle, left-rurn, rear-end and total collisions. The list of intersections shown in Exhibit 17-7 was pulled from a larger database of collisions the city populates. The two newly constructed sires include a full year's worth of data. The city has acquired safety funds foe use in fixing hazardous locations. Although the city is considering any possible countermeasure solutions for decreasing crasheS, they are considering insraUing additional red light running cameras (RLCs) as a part of the program they started 7 years ago. Which sires the city should consider further evaluating and what other measures shoufd be taken once the sires are determined haz.ardous? 11B'f.il~

-<-;

..,._

. ...

.. ...

··~

-.-~...::

..

~--·--

...,,

•, -~-

.....

. .....

..::-

~·

-

-~b~V;fr]t...'JC"'~o.4-""'~~~..rt-~j-c""-"~~ ¥"~ : ~~ .: ~'J~i ~~~ •• '-!~ .• -~~:~~--.,~~--:

Annual Aftrage 2008-2010

Site

Site Type

Angle Collisions

30

Left Tam Collisions

Rear-End Collisions

10 7 Site rebuilt in summer 2009

I

Sig.lna..

2

Sig. In a..

3

Sig. Ina.

16

22

27

4

Sig. Ina.

10

5 G

Sig. Ina..

18 12

12 27

Sig.lntx.

9

8

9

11

7

Sig. Ina..

25 18 18

8

Sig. Ina..

24

9 10

Sig. Ina. S"tg. Ina.

II

11

Sig. Ina..

l1

17 16

7 Site constructed in lace 2009 13

I

15

II

I

8

24

17

I

9

!

I

ADT

39 32

13

Total CoUisioos

Other Collisions

I

2008

2009

2010

42,000

27

36

26

60,000 19,000

..

.

50

37 27

34

28,000

33 21

36,000 31,000

24 18

30

32,000 26,000

11

16

24 27 17

17 20 14

28

18

25,000

-

.

24

24

35,000

17

19

23

15

~1,000__

18

16


L_ _

15

'

-~

.

-=

~~~~.tllit'~'i~~ffili.i'€m#-1i1't":."-~~4ir'~~~·· ..... {~:~:~ - ,_,_•. ,.,:it_'flti: A .• · ~- •• • h!!.. . .. ;+: _ .. ·-~ >o"u _ !f-=.~~- • _:.: :__ • o_ ·.w .. _( < o~~ -~'il~-,~~.~i-:~~M.,-..~~':S£''"1' .:.:; ·~·,~ - ~-~·,J~::f~~~;,::;t,-.;:.'·--~~

Rate per mlllloo vebid~

Site

Angle+ Left

Frequency per yeu

1

55

29.7

1.94

50.0

2.28

2 3

38

34.7

5.00

4

28

24.0

2.35

5

37

27.0

2.05

6

27

17.3

1.53

7

27

18.0

1.54

8

32

17.7

1.86

9

-

24.0

2.63

lO

24

19.7

1.54

11

26

16.3

2.13

22.7

2.2

6.5

1.1

Mean (si.mila.r sites) • Standard Dev. {si.mila.r sires) •

----

-

_,___

,___

-

Solution: The solutions f?r each of the five conditions are shown bdow and refer to calculations shown in Exhibit 17'-7·

cicr

Condition k Since the iS considering installing additional RLCs in the jurisdiction, this condition will loOk specifically at collision types associated with running a red light. Typically, red lighc running involves angle or · left-rurning vehicles (such as chose sitting at a permitted signal). Exhibit 17-8 sums the frequency of those cW0 coUi.sion types. Site 1looks promising with an average of 55 collisions per year ftom 2008-2010. . Condition B: Compare n~y constructed sites 2 and 9 with comparable sites using total collisions. Since ~I other sites are ·signalized intecsections" all the remaining sites can be used as a comparison group using the D'le~~ and standard deviation for each of the frequencies of the other nine sites (22.7 ± 6.5 collisions). Refer to Exbil71t 17-8 for frequency/year calculations along with standard deviations.

S.iiU 50-22.7 - 4.2

sm..2

6:5 Both sites are above average because their resulrs are greater than ttro. Site 2looks most promising. Condition C: Looking at "Total Collisions" in Exhibir 17-7, Site 8 shows there was a rapid deterioration d.v.J:'ing the 3-year period, with nearly 3 times the nll;!llber of collisions in the third year compared to the first yc;~· Although the site appears to have rapidly deteriorated, it would be a good idea to look at the long-term frequet1 cY of collisions (say 10 years or more). Condition D: Excluding both new sites and looking only at collision rates for the other nine intersections, Sire .3 stands out with a rate of five collisions per million vehicles, shown in Exhibit 17-8. Note that rates arc calcuJace~ for "spors" and not •segments,• and that total collisions are used in the calculation of"R.are per million vdticles • Condition E: Compare Sires 2 and 9 with similar sites. Since all other sites are "signalized intersections," ~ comparison should be made using the mean and standard deviations of the rates for each of the other nine sire& · Again, total collisions are used. Refer to Exhibit 17-8.

Traffic Collision Studies • 36-:;:;;iiii'

SitU 2.28-2.2 = 0.73 1.1

S&.2

2.63-2.2-0.39 l.l

Both sites are above average because their results are greater than u:ro. However, neither site is promising since their value is not greater than 1 (one standard deviation). Summary: Sire 1 appears promising for implementation of a Rl..C. Sites 2, 3 and 8 appear to be good sites for implementation of some form of countermeasure; however, the exact countermeasure(s) has yet to be determined. Since Site 8 appeared to rapidly deteriorate in 201 0, it would be a good idea to look at the long-term history to confirm this excessive increase in collisions was not random. Lastly, a condition and collision diagram for each site would be useful for determining crash causes (discussed further in Sections 4.3.1.1 and 4.3.1.2).

4.2.8 Empirical Bayes. Many researchers worry that previously discussed identification methods do not flag truly hazardous sites often enough. They eire the fact that these methods are unable to combine informacion from previous srudies or information about the location characteristics with current collision information during an analysis. In response to those concerns, researchers ~ve devdoped methods of identifying collision-prone locations based on Bayesian statistics. The HSM provides a method for identifying hazardous sites based on the EB methodology (HSM, 2010). The strength of the EB methodology is that it allows the analyst to determine the cpmet{ number of collisions at a site based on the predicted and observed collisions. The observed collision frequency is based on the crash histOry at . the site. The predicted collision ,frequency is found using a safety performance function (SPF). An SPF pfuvides a prediction of the expected crash frequency at a site for a base set of conditions (lane width, shoulder width, etc.) and is calculated for a segment using. typically, only AADT and segment length. Most likdy, SPFs were devdoped from srudies conducted in ocher regions, so a calibration factor, C, can be applied to adjust the SPF to local conditions. Since the SPF predicts collisions for base conditions, accident modification factors (AMFs) are provided to "adjust" the SPF for conditions that vary from the standard condition. These are discussed in more detail in Section 4.3.3.1. Therefore, the predicted number of collisions at site "A~~ take the basic form

N pnJJatd = SPF6.,. • C • A.MF; • A.MF1 •

...

A.MF/t

Equation 17-1 0

Using the predicted and observed crash frequency, the expected number of collisions could be calculated. The basic premise of this methodology uses a weighting factor based on the overdispersion factor of the SPF. The more reliable the SPF used to predict the number of collisions, the higher the overdispersion factor, and thus the more reliance given to the predicted collisions versus the observed collisions. The HSM provides methodology for determining the expected collision frequency, N~The expected number of collisions would then be used for ranking in lieu of the actual collision frequency (Section 4.2.2). The expectation is that RTM is no longer an issue because of the reliance on predicted collision frequency. The EB methodology is a much more rigorous method for determining hazardous sites; however, the downfall is that it is muCh more cumbersome. In addition, SPFs are nor available for all facility types. The first edition of the HSM includes SPFs for rural two-lane two-way coads, rural multilane highways and urban and suburban arterials. It's hoped SPFs for other facil.icy types will be added to the HSM as research is completed.

4.2.!1 Cboonng a Method No single method to idenwy collision-prone locations is universally superior. The best approach for an analyst is to select a particular method foe a particular analysis, or to use several methods foe large srudies when adequate resources are available. Smaller agencies and srudies with limited resources will tend to choose frequency and rate methods. Both methods have serious flaws. Using frequencies results in identification of too many high-volume urban locations, since a primary factor related to collision occurrence is traffic volume. These high-volume locations also may be places where the search for realistic and effective countermeasures is especially difli.cult. Using rates results in identification of too many low-volume rural and local street sires; because a chance occurrence of a collision or two divided by a low volume results in a high rate. Usins frequencies and rates together hdps to mitigate these biases somewhat. Many agencies therefore rank by rate those locations that have experienced some minimum frequency of collisions.

(Methods to account for severity can supplement ocher methods and reveal locations chat experience extreme numbers·of ~ere collisions. However, severity methods such as EPDO and RSI introduce yet another arbitrary judgment a nd vola1pte source ofvariation into the analysis. Also, since underrcporting levels vaty by severity, the choice oflogical EPDO values is difficult. Therefore, severity methods should n()( serve as the only means of identifying locations for further review. The RQC method is generally superior to simple frequency and rate in that it identifies truly l=ardouslocations correctly more often. However, the rate quality conuol method n:quires many mon: resources than ocher, simpler methods because agencies need average rates for different classes oflocations. If an agency has a rdiable source for avenge rates or sufficient resources to collect such data. RQC could be effective. The SWP method looks for sites where countermeasures could be most useful, tries to be pro:lCtive, is more efficient in its usc of data and looks at potential sites that may be able to take advantage of proven treatments, chw usi ng safecy funds in a manner that takes advantage of proven countermeasures. The method is simple since it builds on the basic building block of collision. frequency, with some emphasis on collision rates co account for exposure. The EB techniques .o f identifying collision-prone locations hold great promise and agencies should consider them..At this time, SPFs provided in'the HSM only include three facility types (rural two-lane ~way roads, rural multilane highways and urban and suburban arterials}. Futun: facility types will be added as n:scarch is completed in other areas.

4.3 Selecting Countermeasures Once agencies identify a sire as hazardous or promising, they need to conduct an investigation and determine whether the site could benefit from a countermeasure and, ifso, which type. This stage ofsafety analysis is much more detailed than the hazardous site identilicati
4.3.1.1 ullisWn Diagrams Once an agency has identi£ed a location as •collision-prone• or otherwise worthy of investigation, a search for affordable and cffCctive countermeasures begins. The lint stage in sdecting councermeasures is to look foe overrepresented dusters of particular kinds of collisions. No mathematical procedun: to find overn:prescnted dusters has gained widespread we. Collision ~ are the f!Uin tools wed by agencies to identify these dusters. A collision diagram iS a schematic, not-to-scale, gr.~pbical n:prescntation of the collision pattern at a partia.liar location. Collision diagrams can be useful tOr many types oflocations but are most often used at spots such as intcrscaions. Collisio!l diagrams can quickly show analysts where concentrations (or •dwccrs") of collisions are located, the types of collisions dut prtdominace and other useful i.nformarion. Each collision is usually plotted scparatdy, on the approach and near the place where the first harmful ~nt is said to have occurred. An arrow in the direction of travel represents each vehicle involved in a collision. Symbols wed with the arrows, as shown with the sample collision diagram in Exhibit 17-9, describe vehicle types, intended vehicle movements, collision severities and collision types. Again, th.e intended vehicle movement can be more valuable during the analysis than the collision type. Some police repocrs describe vehicles that hdpcd cailsc but were not damaged in a collision. Arrows with a diffuent type ofline than that used for jnvolved vehicles can n:prescnt these noninvolved vehicles. The type of fiud objcas struck may also be useful and"tn:~y be shown. T ...

u: . r-u; ..;.. "

C+··"'l""'

..

.,.c.a

~·

.

ounD

~ Road#l

· -r

. lttttSectlon Colltstcn Ofa..,.

...

,.-~ , ~. IIJiliiDl

~t:,Cttod:

~••01..-m~

D·D.,U&f* M-Dwt,NoUII'!t

·-- ·"""" 1-CNsk

P.. rto'l

If the diagram becomes too crowded when each collision is shown separately, analysts can use symbols to represent sets of collisions of a particular kind. A bold arrow representing 10 collisions of a particular kind is common, for ex· ample. The diagn.ro loses interesting derails by summarizing the data in this way, however. Twenty to 30 collisions fit comfortably on an 8.5- by 11-in. sheet, and analysts can usc oversize sheets for spots with more collisions or longer periods. Labels on the arrows could indicate the dace of the collision, the day of the week, the time of the collision, the road surface condition at the time of the collision and the lighting condition at the time of the collision. Collision diagrams should show military time for clarity and brevity. Abbreviations such as "OD" and "WN~ can represent "dry, daY and "wet, night without lighting.• for example, w show important information concisdy. Light condition and road condition are two of the most important variables used to idel'ttify overrepresenteq dusters of collisions. Day and night conditiotls are assumed to be correctly entered by the officer that produced the report. A cluster of night collisions, for example, might indicate the need for countermeasures such as reflectors, delineators and street lighting. Light conditiotl is a reasonably accur.i.te data item analysts can check against the time of the collision to find errots. Analysts often check the night-to-day collision ratio to ~uge wltether a location or duster of collisions is overrepresented at night or during the day. Analysts must he cautious, however, because the night-to-day ratio will fluctuate a great deal With low numhen of collisions. Most collision report forms include codes for "dawn" or •dusk" light condition at the. time of the collision. These codes present a dilemma since the samples of dawn or dusk collisions are often too small co analyze separately. It is customary in many agencies to eliminate dawn and dusk collisions when computing night-to-day ratios and when necessary during other analyses. FmalJy, since the amount of daylight per day varies throughout the year, light condition measures, such as night-eo-day ratio, will also vary throughout the year even if other faaors are constant. Analysts making comparisons using light condition should therefore use the same months of the year for all locations being compared. CoUision diagrams contain very few details about the particular location. Street names, outlines of the edges of the pavement and a direction orientation are all that are necessary besides the symbols for collisions and fixed ohjeets. Sometimes, collision diagn.ros contain a table or supplementary diagrams for summarizing the collisions by type, se370 • MANUAL OF TRANSPORTATION EN GINEERING STUDIES. 2ND EDITION

~eriry, light condition, or road condition. This can be particularly hdpful when a collision diagram beco mes crowded

~nd clusters are hard to identifY. Computer programs are available to help prepare collision diagrams, and an atrial 'i.mage is nice for overlaying che curb lines. The programs are often flexible and allow the user to specifY which co\Jisions to include and which variables to show on the drawing. The flexibiliry of computer collision diagrams is a major advantage. For example, analysts can quickly obtain separate diagrams of night collisions, injury collisions, or :lilY subset of the total reported collision picture. The output from computer-generated collision diagrains can be ploned or shown on the moniror and computer plots are usually easier ro read chan manual drawings. Use caution co ensure rhe information on computer-generated collision diagrams is complete and accurate-some collision derai ls may have to be added manually.

4.3. 1.2 Condition Diagram Analysr:s often use condition diagrams wich collision diagrams to generate countermeasure ideas. Condit ion diagtams are scale drawings of the location of interest that show the layout, lane and roadway widths, grades, view obstructions, TCDs, crosswalks, parking practices, light standarru, major roaruide fixed objects and other potentially imporlllnt and notable safery features. Condition diagrams are usually completed using one of rwo possible methods. The first is the traditional method: Tile diagram is drawn as a rough sketch in che field and later drawn to scale in che office. The second uses an aerial image printed prior to visiting che site; points of interest are sketched in ·me field and can be overlaid in che office following che site visit. Condition diagrams must reflect che same time period as the collision diagrams they accompany. A rypical condition diagram is shown in Exhibit 17-10.

t6

N

"""'"""

\>1 01

,....

..

~~. ~

.@

.

•

~";·~

:::~;., ~~t

.,,<· ·:··

~ :~ ~; · ~.

s::-o

:I:

I

1

I

:

.Jil.

I

:

:

~:

I

u•:u· I ..

:

:::c

~~~

:iu· l~ :

1l... :: ! !

0

N

'-1• 1 oil

· o~

. ,. __

Road Ill

!~ !Iii

----~-----~-;:----~-------

- -----~--~;------r--------

--------------··r··------..,._.s;____. . .__-:::~'-~:..;:..::.....-:

-'-----------------u;---------------· r

--------, \

o.. str.~~on

t.•SQI&on

t:rRM<•I'l

f*Tin« ll

I I I

Hottt:

•Sidew.lb IRptthnt:Mtkh •ldt ofuchffttt txc.,t for ~·-.o•ut •httot1t01dl2 o.I!IKh
I

' u•:tl' I I

•SIP\IIq'd~tlc~lti,OMOIUII

I

.,.II~Unstt"tlctdii'ISW,.,...tstiN't -o.d...,JVIIf:IC'i.t,.nsdll"''tfMdrt~ on

I I

41Mrdn..ar ~ n.••o•tHdu/d.,. ol4t412AOI' • 7,010vtl'lltlu/W.,

Gos stllllon(4Pumps) & COnvenience store

I I

I I

!


The site diagram should be approxi.matdy to scale and include all the physical features of the study site (for example, sidewalks, crosswalks, bicycle lanes, vegetation, drivewa~, embankments, signs, traffic signals, markings, roa~way shoulders, abutting land uses, bus stops and other characteristics or conditions that may affect visibility and/or sight distances). All streets or highways should be labeled by their official names and/or numbers. The diagram should indicate estimated measurements for roadway widths, shoulder widths and lane widths. A checklist of items commonly required for a complete site description is shown in Exhibit 17-11. All these items may not be required for every study.

4.3.2 Gener11ting 11 List ofPossibk CtnmtnmetUUres Collisions occur as a result of contributing circumstances such as human, vehicle, or roadway factors. A crash typically has two or more contributing factors associated with the event, sometimes called a "causal chain." Traffic engineers will typically focus on the human and roadway factors to break this chain since this is the domain in which they have the most expertise and control. Vehicular causes should be considered by the automotive designers and engineers, and some human factors, such as speeding and driving under the influence, may need to be addressed by an appropriate enforcement agency in conjunction with the transportation engineer. After an anal~t has identified the predominant clusters of collisions at a location and the likdy causes of those collisions, the next stage is to generate a list of possible countermeasures. A countermeasure is a roadway strategy intended to reduce collision frequency or severity, or both, at a site. Prior to choosing countermeasures, it is imperative the anal~ have completed necessary studies such as: • detailed inv~tigations of individual collisions Oook.ing especially at collision cause cha.ins); • reviews of site plans;

• site visits; and • other transportation engineering studies, such as spot speed, intersection sight distance, traffic volume, etc.

\•,

Countermeasure selection is typically done based on the contributing factors associated witb crashes and the e:xp<:rience of the analyst. For instance, an intersection may have a high number of angle and rear-end collisions on an approach that rounds a curve. Using the collision and condition diagram, along with collision summaries by type and "severity, the analyst identilied in the likdy causal ciWn a sight distance problem at the site during all hours of the day. The analyst recommended landscaping be removed bea.use it was encroaching on the ability of the drivers to see the imersection signal displays as they rounded the curve. Highway loca cions vary gready, tke state-of-the-art in collision countermeasures changes rapidly and there are limits on the quantity md quality of collision data at many locations. Therefore, generating lists of possible countermeasures for particular sites is, in some ways, more art than science. Ideas for countermeasures come from e:xperience and a thorough knowledge of the technical literature more than other f.tctors. The major sowce for countermeasure ideas is the technicalliterarure. Many references arc available to the eng.ineer to aid in countermeasure selection for certain highway situations. A few primary references used by traffic engineers are discussed bdow. • Exhibit 17:12 provides lists of countermeasureS for several typical collision patte.rns and contributing facto.rs (FHWA, 1991). This countermeasure list is certainly not exhaustive and does not include newer technologies. However, the collision patterns and probable a.uses are summarized nieely for the user and most of the traditional countermeasures that inspire confidence are listed in the exhibit. • Another good source of countermeasure information is NCHRP Report 500: Guidance for Implnnml4rion ofthe AASHTO Strrz~ Highway Safrty Plan (NCHRP 2003,2004: Volumes 3-10 and ll-14). The volumes focus on various highway crashes and contributing f.tctors, and aid analysts in treatment and/or prevention of these crashes through recommended coumermeaswe ideas.

of

• The HSM provid~ a list crash contributing f.tctors for various emphasis areas in "Chapter 6: Sdect Coumermeaswest however, the concurrent lists of potential countermeasures are nor provided (TR.B, 20 10). HSM chapters pertaining to individual types of roadway features have countermeasure ideas and accident modification f.tctor (AMF) estimates for each.

Traffic Collision Stvdies • 3 73

·-·_.

.,,.

'~w&:trr~'~·:>'·,..,~ ~· ""·-~~~-~;~r -~~·:·~~-.:.-·
..

-. . -. .... .

. .

'. '

-

Type oflmprovement to Construction Classification Intersection and craffic control

.

. '..

..

-

Indexed Percent Reduction in Accident Cost of Rates aher Improvements (%) Evaluated (millions) Fatal lnjury Fatal + Injury $562.3

37

15

15

..

' *' . -

~·

.

~

;;oo:. ~

--~

....

Cost per Accident Reduced (thousands) Fatal $3445

Benefit! Fatal + Injury Cost Rlttio $18.7

4 2.8

Channelization rurning lanes

$297.7

48

23

24

$510.9

$19.7

Sight distance improvements

$7.8

44•

31

32

$371.4

$22.8

3.6

Traffic signs

$19.6

34

3

4

$59.3

$15.7

20.9

Pavement markin&$ and/or delineatOrs

$~3.7

15

(I )•

(I)'

$751.8

-

1.6 10.3

Illumination

$,13.2

45

8

9

$122.6

$16.8

Traffic signals upgraded

$63.1

40

22

22

$412. 1

$8.6

4

Traffic signals, new

$127.3

49

21

21

$344.5

$10.5

5.1

$307.7

50

28

29

$752.2

$92.1

1.7

Bridge widened or modilied

$79.8

49

22

23

$103.8

1.2

Bridge replacement

$156.9

72

47

49

$1,201.6

$159.3

1.1

New bridge construction

$26.2

77•

40

43

$1,637.8

$223.2

•0.8

Minor srrucrure repbcemeru/improvul

$39.0

36

20

21

$277. 1

$39.8

4.5

Upgraded bridge rail

$5.6

n·

41

45

$189.5

$39.4

6.5

Roadway and roadside

$1,971.3

31

13

13

$722.2

$54.0

1.8

Widened travel way

$511.0

9"

7

7

$4,041.3

$174.2

0.4

Lanes added

$212.9

-2

13

13

-

$66.8

0.1

Median scrip ro sep=re roadway

$56.8

73

17

19

$382.4

$72.8

3.2

Shoulder widening or improvement

$88.3

28

11

12

$497.9

$37.9

2.6

Roadway realignment

$329.9

61

32

34

$1,193.2

$111.1

1.1

Skid resistant overlay

$468.4

26

18

19

$837.0

$30.3

1.8

Pavement grooving

$12.6

34•

15

15

$377.3

$14.5

3.8

Upgraded guardrail

$149.7

42

8

9

$151.7

$31.5

8.1

StmctuJ'CS

$1,0n.9

' '

Upgrulal mcdim barrier

$7.4

45*

28

29

$192.8

$10.8

7

New median barrier

$58.9

62

o·

3*

$224.7

$213.6

5.4

Impact attenuators

$10.7

31*

36

36

$390.0

$8.0

4

Flatten side slopes/regrading

$36.5

(2.5)'

9

8*

-

$102.2

-

Bridge appco.acb guardrail cransition

$4.8

61 *

44

45

$200.9

$25.8

6.3

$15.0

49

22

23

$202.5

$16.0

6.4

$331.2

89

63

67

$570.0

$114.2

2.2

New Hashing lights

$55.0

91

74

77

$551.4

$103.5

2.2

New Bashing lights and gates

$138.1

n

84

86

$591.8

$115.3

2.1

80

$432.8

$96.7

2.8

Obstacle removal

Rallioad-highway crossings

New gates only

$63.2

90

78

Note: Numbers in parentheses indicate increased crash races • No significant change ar the 95% confidence level Source.: U.S. DOT, 1989. 374 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITJON

i

, 4.3.3 Choosing Among Countumeasur~ AUI!T1Iatives \The final stage of countermeasure selection is to narrow the range of possibilities to one or more m easures co be 'implememed. The analyst will use many of the same strategies during this stage as outlined above for generating list of possibilities. At this third stage, however, potential crash reductio n, available budget and co untermeasure cost-effecriveness become important. Various sources of information are available ro analysts regard ing this scage, and a few of these sources are described in the following seccions.

a

4.3.3. I Crash Reduction The potential safecy effect of a countermeasure installation is an important piece of information w hen choosing a ueaunent. Crash reduction is accounted for by AMPs, also referred to as crash reduction factors (CRFs) . .AM'Fs are based on quantitative results from safecy research swdies of variow treatments indicacing the exp ected percent reduction of crashes following the installation of a countermeasure. AMPs are quick and weful tools for estimating the effectS of various countermeasures and ace becoming increasingly popular among analysts. AMPs are provided for many countermeasures on assorted roadway cypes in variow sources ofliteracure; howeve r, the analyst mwt ~e these data cauciously as some arc; based on outdated or simple safety srudies that do nor account for various effects such as seasonalicy,or RTM. AMPs can be found from varioJJS sources, su ch as FH~A Desktop Rtfonnct for Crash Reduction Factorr (FHWA, 2008) which provides l2 tables of countermeasure altema.cives for inrerseccion, pedestrian and roadway departure crashes based on various level.s of research. This desktop reference does not provide good informacion on the strength of the study the AMF was based on, and efforts are underway by FHWA to provide a Web-based clearinghouse of CRF/AMP clara to provide that information in a moce concise format. Another potencial source is NCHRP Report 500: Guidllnce for lmpkmmtation ofthe AASffTO Strategic Saftty Pum, which provides a series of succinct documents developed to help agencies reduce collision frequency and severity for specific focus areas (NCHRP 2003, 2004: Volumes 3-10 and ll-14). Lastly, some state department of transportation (pOT) Web sites include helpful AMF information, such as the one maintained by the NC DOT (NCDOT, 2007), which provides estimated AMFs for many countermeasures. AMFs rhat are based on more sophisticated srudy methods, such as those referenced in the HSM, are likely to provide results the analyst can have more confidence in because they are based on more robust srudies. However, even though the study methodologies are more stout, the analyst should be aware that all AMFs or CRFs will have some uncertain!)' because they are based on collisions from the region or area in which the study took place (HSM, 2010). It is recommended analysts use the high-quality HSM AMFs if at all possible, and only use-wid'l caution-other AMF or CRF resources when a good H SM AMF is not available. One minor difference between a.n AMP and CRF is the AMF is expressed as' a decimal, whereas a CRF is expressed as a percent. Therefore, an AMF of 0.75 would mean the expected crashes following a treatment would be 75 percent of what would have been predicted had the ueaunent not taken place. In contrast, the correspondiP& CRF would equal 25 percent, meaning the expected crashes following a treatment would be 25 percent less than predicted h·ad·che treatment not taken place. AMFs are calculated ja.s: AMF . _E_xp:...e.:_ted_a_verag~:..e_c_ras.:_h_fieq.......;:..u_en_cy.:.._WJ.:..'th:;:_con=d.:..il..:.ioo.:...::..'a' Expected a~ aash fiequency with cooditioa 'b'

Equacion 17-1 1

The corresponding CRF would be calculated as: CRF • (I - AMF) x I 00 percent

Equation 17-1-:2-

The basic method for applying AMPs a.nd CRFs is the same, and they can be used interchangeably as long as all cb. te analyst undetSt.ands this cli1ference in calculation a.nd meaning. An AMF of 1.0 (CRF =0 percent) would be the "basG condition: which essentially means there would be no cha.oge in the expected number of collisions because ther~ would be no change in the conditions related to collision causation. Often, the standard error is provided with an AM.F. The standard eccor can be used to provide the analyst an estimace of the effective range of collision change, as shown il! Equation 17-13:

Traffic Collision Studies • 37!$0

Cl (X"/o) -

Equation 17-13

AMF, :1: SE, • MSE

when::

CI (X pe.rccnt)

= confidence in.terval, which would snre mar one is X percent sure the AMF value falls within the

inrerval provided

.

AMP

.

= accident modification factor for condition x

.

SE

• S.tandud error of AMP

MSE

&

multiple of.the s.tanda.rd error provided in Exhibit 17-13

AMFs cannot be applied directly when the original condition is nor the same as the AMF base case. AJ an example, consider lane width, when: the base condition AMF (AMF = 1.0) would be for a 12-ft. (3.7 m) lane. U an c:xisting roadway has 9-ft.(2.7 m) lanes (AMF = 1.50) and the agency is considering expanding to 11-ft. (3.4 m) lanes (AMF = 1.05), an analyst would need to take the ntio of the updated condition tO the existing condition AMFs, or 1.05/1.50 to get an AMF for the treatment of 0.70.

Source: Higl,way Saftty Manu4 2010.

One major ftaw with applying AMFs or CRFs is the method assumes countermeasures act independendy of one another. At this time, the combined effects of multiple tn:a.tmenrs an: generally not known and further research is needed to fully understand how multiple treatments interact. A.t this point, two methods can be wed co make an estimate, if multiple treatments an: being considered for a roadway facility. Firs.t, the analyst could try to sepa.nre the collisions into those types that would be affected by a treatment and those types that would not be affected. For instance, an intersection has been ide.ntified that has high numbers ofleft-turn and right-rum crashes. The analyse may be considering cwo ueacmenrs: 1) changing protected/permissive left rums to protected only and 2) adding channdized right-rum lanes. In this case, it would be a good idea to separate collisions by movement for left-rurning vehicles and right-turning vehicles and applying each AMF direaly to its rdated crashes instead of the entire intersection. The second mecbod should only be considered if it i.s not possible to separate collisions, such as on a roadway with no shoulder and small lane widths. If the analyse was considering increasing both shoulder and lane widths, it would be nearly impossible ro determine how collisions should be separated for those treatments. lns.tcad, the analyst could multiply the AMFs together to get the combined effect. & mentioned earlier, one mus.t we caution and unders.tand the limitations of this second method. A second 8aw in applying AMFs is the estimate of the number of collisions expected under the ·do nothint condition. The recommended method is tO use the long-term average number of collisions as the base levd because it is simple and easy to understand. Usually, 4 years or more of collision data are necessa.ry to reduce spiking tendencies. The AMF is then applied din:aly to the long-term average number of collisions to determine the apeaed collision savings. Shorter durations of time could seriously jeopardize the estimate if the collisions spiked up or down during those shorter time periods. This method iJ flawed because it doesn't account for RTM. A better method would be to use an SPF ro predict the long-run expected number of collisions without the treatment. However, this method is fhwed because it assumes the predicted collision frequency is similar to that of the Location the SPF was based upon. Therefore, the best method would be to calculate the expected number of collisions using an EB methodology which is based on a long-term observed crash history and a prediction from an SPF. See the HSM for a thorough explanation of these methods.

~.,e

a

iAII~IIt/\ 1

()CTDA~Ic"nl"'\nTATII"'\t.l

,..,,,.. ,,,,.,..""',...

,....,,,"'•.rr ,.,.,,.. ---•-••••

JPCAMPLE 17-5: Using AMPs for CountermeasW"e Selection A. 4-mile section of rural two-lane road~y has been idenrified as hazardous based on an c:valuacion of the existing rbadway network. The responsible agency has idemified chat the road~y has numerous run-off-the-road collisions a1d has determined the lack of raised pavement markers (RPMs) and the 9-ft. (2.7 m) lane widths are the mos t likely treatable causes. This section of roadway has had an average of eight collisions per year for the last 6 years. The agency is considering RPMs and/or 11-ft. (3.4 m) lanes. What would be the expected colli5ion reductions and numbers of collisions for each treatment separately and for both treatments combined? Assume the AMFs for the RPMs and lane widths are: AMFRIM = 0.85 AMF,.r.t.... • 1.5 AMF 11 ·" ' - .. 1.05 Solution: I. lfRPMs were installed by themselves, one would apect a collision reduction ofl5 percent {{1-0.85) x 100}, or a reduction of 1.2 collisions per year {8 x (1-0.85)}.

2. If lane widths were increased to 11-ft., one wolild first need to calculate the AMF from ia emling 9-ft. lane width condition. The AMF i5 calculated as: AMFu-/t 1.05 AMFrreacm~mt = AMF9-/t = Ls = 0.70 The expeCted collision reduction would be 30 percent {(1-0.70) x 100}, or a reduction of2.4 collisions per year {{8 x (1-0.70)}. · 3. If both treatments were· installed, it would be nearly impossible to separate collisions related to each treatment. Lacking a trustworthy SPF, the analyst would therefore need to use the combination method described above. The AMFs for RPMs and 11-ft.lanes are 0.85 and 0.70, as noted above. The comp utation for the combined AMF i5 AMF....w...~ "' 0.85 x 0.70 = 0.60

Assuming the two treatments act independendy, the expecred colli5ion reduction from installing both treatments would be 40 percent {(1-0.60) x 100}, or a reduction of3.2 collisions per year ({8 x (1-0.60)}.

4.3.3.2 Countc1Masure Cort-Effictivmess Evaluating the cost-effectiveness of a particular treatment i5 an important step in countermeasure selection. The most popular method of conducting a cost-effectiveness srudy is the benefit-cost (BO method. ABC ratio greater.. than one m~ the treatment was cost-effective. The basic method for conducting a BC srudy i5 oudined in the following paragraphs. i . The first step in calculating the BC ratio i5 determining the ner btnelit of installing each countermeasure based on crash cost savings. States often produce their own crash cost estimates; however, FHWA provides crash cost estimates chat are updated periodically (FHWA. 2005). The predicted increase or decrease of collision types for a given time period (usually 1 year) following installation of a treatment i5 then multiplied by the crash cost estimate. The sums of the crash costs estimates (positive and n~tive) represent the overall benefit for the given time period. The calculation of the estimated bene~t i5 shown in Equation 17-14.

Benefit = .

i:. (CC 1•1

1•

N1 )

Equation 17-14

where:

CC1

m

N, •

crash cost associated with collision type or severity, i increase or decrease in collisions for crash type or severity. i

Next, the annwl cosrs must be identified for each countermeasure. Annual eosr is based on the annualized installation costS, ·

AG;, and the annual maintenance cosrs, C,. The calculation of the estimated acrual COSts is shown in Equati~n 17-15.

Equation 17-15

Annual Cost= AC; + C,

The annualiud installation costs, Ac;, are based on the total installation cost, C? discoum rate, R, and the expected service life in years, N. The di.lcount rate is cypically becween 3 percem and 7 percem and can usually be obtained from the local agency. The calculation is shown in Equation 17-16.

C,* R

Equation 17-16

AC, = 1 -(l +R)'N

EXAMPLE 17-6: Calculating Benefit-Cost for Countermeasure Selection

State "B• wanes to determine the benefit-cost ratios associated with the installation of RPMs along four similar 2-mile long cwo-l.ane rural roads. Crash costs and crashes by cype are provided in Exhibit 17-14. Assume the following: AMFRI'M.h.u.... ~

0.50

AMFmf.,.,-.ff.-J~

0.75

The discount rate obtained from your agency for RPMs is 5 percent with a service life of I0 years. The total installation cost is $100,000 per site. The maintenance cost for RPMs is $5,000 per site per year.

Solutiom Exhibit 17-15 shows the estimated benefits, annual costs and the benefit-costs for each AMF and segment.

378 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES. 2ND EDITION

The benefit-cost calcul~tion for Segmenc I is shown below. ·._N _

_=

( N..,.. • ( I- AMF))

N - . = (I • (I - O.S)= 0.5]

N_..-_, =

[N_.....,• (1-AMF))

N..,_ _ _ = [3 • (I- 0.75)] • 0.75

Benefit.... = (S72,000 • 0.5) + ($83,000 • 0.75) • $98,250 The benefit to cost ratio associated with installing RPMs is $98,250/$17,950, or 5.5; since the ratio is over 1.0, RPMs would be a good iqvesunent on Segment 1.

4.4 Countermeasure Evaluation Agencies frequendy use collision data to evaluate highway improveme.nts, whether or not the improvements were installed with enhanced safety~ a goal. Indeed, a program of regular countenneasure evaluation is esse ntial for incc:lligent future countermeasure selection. Analysts typically evaluate counterm=es using one of che e:xperirnen cal techniques described in Appendix A, with collisions as the me~ure ofeffectiveness. The most imporrmt point regarding countermeasure evaluation usi11g collision data is mentioned here and in Appendix A: Before-and-after srudies are often misused or misunderstood. · ·

As currendy conducted by many practitioners, before-and-after srudies using collision data suffer from two serioUS fu.ws and provide inconect and misleading results. The flaws are 1) che failure to control for the effects of changi.ng conditions during the lengthy time periods required ro amass before-and-after collision statistics (seasonality, cime period, historical effects, etc.) and 2) the failure to correct for RTM. An example of seaso~cy and/or time period may be evaluation of~ countermeasure where 5 complete years of data are available; however, only 5 months of data ~re available after implementation. If this aftet period were during a particularly harsh winter, collisions would naturally be higher because of the harsh winter and do not have anything to do with the coun~rme~ure. To account for ti-lls seasonality or time period factor the analyst could usc a comparison group, where the comparison sites would natUrally fluctuate in a similar manner to treatment sites. These similar trends would negate most of the effects caused by the harsh winter, allowing the analyst to pull only the effeets of the treatment. Although co~parison groups are commonly used, the most rigoro~ way to overcome these two flaws is to tond~ct an experiment using randomly selected control sires where the agency measures collisions but does n ot install rbe improvements. However, rmdom sdection of conrrol sites is rardy done because agencies typically install treauneo.cs at the sites in some predetermined manner. For this reason, it is more common to conduct observational before-afcer studies such~ Comparison Group and EB methods (Hauer, 1997). Appendix A describes how to conduct a beforeand-after experiment with control sites and how to conduce other common countermeasure evaluation srudies. If" a rigorous safecy study (such~ Comparison Group or EB) is needed to determine the effect of a particular counterme ::asure, the analyst mar, need to consider hiring an evaluation expert..

5.0 REFERENCES American Association ofS12te Highway and Transportation Officials. Highway Safny MAnUAl. Washington, DC: AASHrO,

2010.

o/

Antonucci, N. D, K. K. Hanly, K. L. Slack, R. Pfcfer and T. R. Newnan. NCHRP Report 500: GWlanu for lmplnnmllllion tiN AASHTO Snaugic Highway Safoty Plan, V6h.mu 12: A Guilk for Reducing CoOisions I# Sigruzljud lntn1tctions. Washington, DC: Transporoi.tion Res=ch Board, National Research Council, 2004.


~

.

~

Balm, G., er al. Desktop &fmnu for Crash &duction Facton, FHWNSA-08/011. Washington DC: U.S. Deparunem of Transportation, Federal Highway Adminismtion, 2008. Box, P. C. and J. C. Oppcnlandcr. Manwd ofTnzffic Engillming Stu.ti.Us, 4th ed. Washington, DC: Institute ofTransportation Engineers, 1976: p. 49.

Federal Highway Administration. Charat:UriJtia ofEmnging Road and Tnzil Usm and Thnr Safny. Washington, DC: U.S. Department ofTransportation, 2004 . Federal Highway Administration. Crttsh Cott Estimarn by Maximum Polia-&porud lnj11ry &vniiJ within &lrckd Cnt~h ~. FHWA-HRT-05-051. Washington, DC: Federal Highway Administration, 2005. Federal Highway Administration. Highw.t] S4fitJ Enginming SluJin Proud~~ntl G!IUk. Washington, DC: U.S. Department of Transportation, Federal Highway Administration, 1991. Harl
Rrst~~rrh

Inscirutc ofTran.sportation Engineers. Sl4tittiuJ EvaJuation in Tnzffic SafotJ StudiD. Washington DCi ITE, 1989. KnipUng. R. R., P. Walle:, R. C. Peck. R. Pfcfer, T. R Neuman, K. L Slack and K. K. Hardy. NCHRP Report 500, Volume 13: A Glluu for kldrming CDIIiMns lnHiving Htllll] Trueks. Washington, DC: Transportation R.tsearch Boatd, National Research Council, 2003.

Lacy, K., Srinivasan et.al. NCHRP R.tport 500, Volume 8: A O..Uk for .AJJJraring CDI/iMns /nll()/ving UliBIJ Polts. Washington, DC: Transportation R.tsearch Board, National R.tsearch Council, 2004.

Mak. K. K. ct al. "Automated Analysis of High-Collision Location.s.• TnlnJ!Drtaiion Rnt~~rch ~ }OIIrnal oftiN TrrmspDnatiDn Rnearrh BoJUd 1068 (1986): 59--64. McGee, H., S. Taori and B. Persaud. NCHRP R.cport 491: Crttsh Experimet Warrantfor Trttjfic SigMI.t. Washington, DC: Transportacion Rtsea,rch Board, National Resca:ch Council, 2003. Miller, T. ct al. Snuilivity ofRrstnlrrt A1IDauUm MH!tls 10 Dismmz RAu IfNi Un"JDrutl CDIJiril11U. FHWA./RD-35/092. McLean, VA:. Federal Highway Admini.stracion, 1985. Nacional Coopencive Highway Reseuch Program. NCHRP R.tport 128: Mttho.ls for Idmtif;ing fUwtrdow Highwa, Eltmmt.J. Washington, DC: Transpo112tion Reseatch Board, National R.csearch Council, 1986. National Highway Traffic Safety Adminisrncion. Tnzffic S4fitJ Factt 2008.Washin.gton, DC: NHTSA,, 2008. National Safccy CounciL Estimllltd Com ofTraf!U CDI/iMns. 1990. Chicago: National Safety Council, 1991. ,.,.,.., .,

a • A• n l At

r\,. TD At.IC"Q('\CTATU'"\~l t•lr..IMI:COIMt: (TIInll: ~

iMn f:nJTt(lN

~

Neuman, T. R. er al. NCHRP RA:pon )00, Volume 3: A Guide for Atid=ring ColliMns with Trtes in HUArdous lAcatiom. Wa.shington, DC, National Cooperative Highway R=tch Program, Transponation RA:search Board, 2003. ·Neuman, T. Ret al. NCHRP RA:pon 500, Volume 4: A Guide for Addrtning Head-On ColliJions. Washington, DC: Transportation RA:seatcb Board, National R=tch Council, 2003. Neuman, T. R et al. NCHRP RA:pon 500, Volume 5: A Guide for Addressing UnsignaliuJ lntnumon Co/Jisions. Wa.shingron, DC: Transportation R.csearcb Boatd~. National R=rcb Council, 2003. Neuman, T. R, ec al. NCHRP RA:pon 500, Volume 6: Guidtmct for lmplmrmtllrion ofthe MSHTO Stra~ Highway Saftty Plan, A Guide for Atldrasing Run-Off-Road Co/Usions. Washington, DC: Transporwion R=rcb Board, National Research Council, 200.3. North Carolina Department ofTransportation. North Carolina Project Development Crash RA:duction Factor Information. www.naloLorg/doblpreconsuucdua.ffidSafety/ RA:sources/ projcct_guidelrcgionalf:a.ctors.pdf. 2007. Pons, I., J. Scuta, R. Pfefer, T. R Neuman, K. L. Slack, and: K. K. Hardy. NCHRP RA:pon 500:

Guiti4~t for lmplmuntation

Tran.~portacion

·

4 the AASHTO Smiugir HigbwiiJ SD.fny Pum. Vo/111111 9: A GuidLfor .&dudng Collisioru With 0/dn Driwr1. Washington, DC: RA:search Board, National R=rcb Council, 2004.

Swb, T. TIN Poclttt T~ kn.lmt RtcwutruciUm Guide: A Cqmpflu TNJfic Alxidmt Rtftrmct Ha.ntihoolt. 6th cd. Tucson, AZ: Lawyers and Judges Publishing. 2008. Stok.es, R W. and M. I. Mutabci. •Race Quality Control Method ofldencifying fuzacdous Road Locations.• TrrwportaiWn Rtsearrh &am/:]011rnai oftht Trruuportati4n &searrh &arri, Issue 1542 (1996).

:r.

Scum. J., R. Knipling, R. Pfefer, R. Neuman, K. L. Slade and K. K. Hardy. NCHRP RA:port 500, Volume 14: A G uidLfor Rtt!udng Crashes Involving DroUISJ and D~t! Drivm. Wa.shington, DC: Transportation Rescacch Board, National Research C!>uncil, 2005. Tocbic, D. J. et al. NCHRP Report 500, Volume 7: A Guide for &Judng ColliMns on Horizontal CUTWS. Washington, DC: Tran.~ponation RA:scacch Board, National Research Council, 2004. U.S.. Department ofTcansportation. The 1989 Ann114i &pon on Highway Stzftty lmprovmu:nt Pro:rrzms. Wa.shington, DC: U.S. Department ofTransportation, 1989. Zegccr, C. V. et al. NCHRP RA:pon 500: Guidance for lmplmrmtalion oftht AAS/1(0 Strategic Highway Saftty Plan. Volumt 10: .A Guide for &Juring Co/Jisions Involving Pttfatrimu. Washington, DC: Transportation ReSearch Board, National Research Council, 2004.


Chapter 18 . .. .... .. .. .. .......... ... ........ ... ......... .. ......... . .. ...... ..... ... ......... .... ...

Alternative Safety Studies Origitud By: joseph E. H~~mmer, Pb.D., P.E. Edited By: Chriswpher M. Cunningham, MCE, P.E. 1.0 INTRODUCTION

383

2.0 ROAD SAFETY AUDITS

384

2.1 Introduction

384

2.2 Study Design

384


385


388

3.0

TRAFFIC CONFLICT STUD!ES

390

3.1 Introduction

390

3.2 Study Design

390


394

3.4 Data Reduction and Analysis 4.0 ADVISORY SPEEDS

405

406

4.1 Introduction

406

4.2 Study Prepara tion

406

4.3 Data Reduction and Analysis.

409

5.0 REFERENCES

412

1.0 INTRODUCTION ypically, safety studies are conducted at sites wbe.rc large numbers of collisions take place and the need fo.r a potential countermeasure may be necessary to reduce collisions. These studies, presented in Chapter 17, a.s:-e reactive in nature because they require collisions take place in order to make a decision on what improveme.n.. -cs are needed at the site. This chapter looks at three alternative safety s.tudies that are regularly conducted in lieu of co 1lision studies as p7ri4Ctive measures for identifying safety issues th:u may need remediation. These studies are typicallY conducted as separate studies; however, many times they c:m be considered during ua.ffic impact studies.

T

This chapter first covers road safety audits (RSAs) in derail. RSAs have been conducted since the 1980s in the United Kingdom and since the 1990s in New Zealand and Australia. Since the fitst road safety audits in the United States we&e conducted in New York, Pennsylvania and Iowa in the lace 1990s, the RSA has become used increasingly as a proactiV'"C: safety tool. This safety study can be conducted at any time during the life of a roadway and is highly recommcode as a proactive measure during the construction and design process. Secondly, collision surrogate srudies, often aile .d conHict studies, are described. Traffic conllicts are an alternative to coUision studies when linle-to-no collision data ar-c: available (s~ch as a newly constructed or rural site), or a quick study is needed to determine safety problerps.la.stl~· the chapter discusses advisoty speed studies based on recent updates to guidelines provided by ITE. Advisoty speeif...s

a

Alternative Safety Studies • 38.:::;S3

"·}

have been appHed in many different ways by different stare and local agencies. Recently the ManUAl on Uniform Traffic ContrrJ! Dnticn (MUTCD) has recommended all advisory speeds be updated by 2020. The methods for posting appropriate advisory speeds are updated in an attempt to bring some consensus among users.

2.0 ROAD SAFETY AUDITS 2.1 Introduction RSAs are becoming increasingly popular in many state and local agencies as a tool for decreasing collisions. An RSA is a formal safety evaluation of a future or existing roadway by an independent audit team. The RSA guidelines presented Later in this section arc based on two primary documents provided by the Federal Highway Administration (FHWA), Road Safety Audit GuUklirm (FHWA, 2006) and Pttkstrum Road Safny Audit Guitklirm and Prompt Lists (FHWA, 2007). The guiddines describe a formal eight-step process for completing an RSA, the team member qualifications and necessary tools for completing the audit. It should be noted an RSA is not a "road safety review.» Road safety reviews arc conducted by an in-house team, are often motivated by high frequency and/or severity of crashes and arc consequently reactive in nature. Instead, RSAs arc more proactive reviews of a design or finished product that attempts to identify potential conBicts and crashes, with the goal of preventing them before they take place. In addition, the audit team should be ari independenc team (instead of an in-house team) that is not familiar with the road(s) being audited so there is no conflict or bias when visiting the site.

RSAs can be a very useful tool for the engineer if used correctly. State and local agencies have many ~ompeting interests indud.ing environmental, right-of-way, socioeconomic, roadway opacity and political interests. With all these interests competing for limited funds, safety is sometimes overlooked. RSAs inject safety into the mix of ocher interests by cxpUcidy identifying the safety impHcations of road management decisions. In addition, there arc other advantages to conducting RSA.s. • Collision reports do not always identify many road safety rdated issues. • Roadway designs need to anticipate and accommodate common driver errors which can be uncovered during a comprehensive safety audit. • It is much easier to design and build safer roads to begin with than to modify existing roads or entrenched behaviors of many drivers.

Because RSAs arc newer in the United States, they arc typically conducted on existing roads; however, applying R.SAs early in the design process offers the greatest opporrunity to influence the acrual design. Changes that improve safety performance typically become more difficult, costly and time-consuming as a design progresses. Internationally, RSAs are being mandated by the United Kingdom on all new roadways, while Australia, N~ Zealand and other countries recognize the RSA as a valuahk tool in reducing collisions. International RSA sources are available onllne, including the United Kingdom's National Roads Authority Design Manualfor Roads and Bridga- ~lumt 5 (NRA, 2009), Ausualia's Austroads Road Safoy Audit GuUklirm (Au.suoads, 2002), and New Zealand's Tmnsfond Road Saftty Awiit Promium for Projecr (Transfund New Zealand, 2004).

2.2 Study Design FHWA provides a formal eight-step process for conducting RSAs, shown in Exhibit 18-1.

This process lays the basic framework for conducting a successful audit. The steps provide information on the work to be done and by whom. A basic summary of the steps is provided below, with more detailed information provided in the foUowing sections. Sups 1-2: hkntifJ the ro/Uh to t:ndit-" tk mulit tum. The party responsible for assembling the audit team i.s the road agency. Good ~didates for preconstruction audits are safety-oriented projects, high-profile projects usually • • •

-· • • ., ,,. ... .,.,. , .,...

r~t U'" \J r r

"'UIII"'' t:"ITlf"U\1

.(

RSATeam

Source: &ad S.Zfoy A.udit GuitJLiina. FHWA. 2006.

audit~ at the request of politicians or the public, and complex roadway designs. Good post-consaucrion candidate projects include high-collision sires, high-profile projecrs, or sites where traffic characteristics have changed (or are expected to change) due to long-term construction detour routes or new devdopments in the area. St~s 3-6: OmJuct the -"it tmd report (11J firulinp. This is the backbone of the RSA. The aCCIW RSA procedt.~re takes place during these steps and ~dudes the start-up meeting, site visit, audit analysis and presentation of findings. The roadway agency is only present during the beginning and end of this series of steps ro exchange information. These steps will be the primary focus in Section 2.3, Data Collection Procedures.

Steps 7-8: FoiJDw up (11J the mulit finJinp. The local agency follows up on the findings of the audit by issuing a formal response co each of the safety issues and making improvements wherever feasible. These steps will be the focus

of Section 2.4: Data Reduction and Analysis.

2.3 Data Collection Procedures 2.3.1 PenMirul Ntttls RSA teams are chosen by the roadway agency early in the process and are typically hired consultants unfamiliar with the area under srudy. In some instances, RSA teams have been forme~ by local agencies outside the jurisdiction. These teams are referred to as •c:xchange staff; and state and/or local jutisd.ictions set up various agreements to conduct RSAs for each other's agencies. Whether consultants or prearranged teams are brought on to conduct the RSA, the key is the audit ream be an independent group with no prior experience or knowiedge of the area and no reliance on the host agency for other funding. Ifexchange staffi are used for auditing purposes, each staff member should be formally trained in the process of RSAs before taking pan in conducting the srudy. The audit team is typically composed of a minimum of three and no more than five engineers who have experience with roadway design (for example, American Association ofState Highway and Transportation Officials (AASHTO) Policy on Geometric Design ofHighways anti StrtetJ-•Green Book"), operations (for example, Transportation Research Board (TRB) Higlrway C11pacily Manual (HCM), and safety (for example, MSHTO Highway Si1fity Design anJ Opmtions Guil:k-"Yellow Book"). Supplementary auditors may be necessary for some audits to complement the team by adding c:xpertisc: in enforcement, maintenance, 6re/rescue, signing, bridges, pedestrians (especially Americans with Disabilities Act (ADA) issues) and cyclists. The local agency should provide information to the audit team early in the process regarding whether other c:xpertise may be necessary. Once the team has been established, it is ~good idea to appoint a secmary and pbotograpb.er. This will allow the team to be systematic in its data collection approach as the site visit progresses along the corridor. The audit team snould cemembc1: to remain unbiased and consider the safety of the roadway from the standpoint of all road ,llSCIS including passenger cars, pedestrians, pedal cyclists, motorcyclists, large trucks, buses, police, firelighters, maintenance vehicles and older drivers. Alternative Safety Studies • 385

2.3.2 Project Dot:tlments During Seep 3, rhe preaudit meeting, me roadway agency should provide materials mar may be helpful w me audi r ream. All rde,•ant information should be requested by rhe audit ream in advance, with supplementary information provided by me roadway agency as needed. During me kickoff meeting, me roadway agency should explain all rhe materials provided. Useful icems generally requesrM by the audit ream include typical sections of me roadway, planl profile sheers, roadway design exception criteria, aerial images, projecr history documentation, condition diagrams, summarized collision data and collision diagrams. Multiple plan drawings and/or aerial images are useful tool$ for recording various observations directly onro an overhead view during, or just after, the site visic. 2.3.3 Equipment Needs During Step 4, the site visit will require certain equipment be available so the team can r=rd safety issues in a safe and efficient manner. First and foremost, each member of che ream should wear a high visibiliry safety·vest. Although some road agencies do nor require their staff ro wear vests, they are highly recommended, especially during nighrcime visits. Another important item is a camera, which is essential to record on-sire conditions and observations for later reference. Photos are useful for auditors to refer co in the office and are invaluable for inclusion in che RSA repott. A video camera may also be useful for a drive-through or to record specific notable events during the walk through. A measuring wheel and stopwarch may be useful if elements such as lane widths, inretsection clearance distances, signal timings (such as yellow and aU red intervals or phasing) and vehicle speeds. Other items that should be considered are clipboards for all team members ro write various comments that the secretary might forget to record, water or other fluids especially during long visits, and walking shoes.

2.3. 4 Prompt List During Steps 4 and 5, a prompt list may be useful. A prompt list can provide suucrure co a sire visit or analysis, by enumerating feacures such as signs, signals, roadway geomecry and roadside conditions. Ic is a way co ensure the RSA team gets the most out of the site visic and analysis, and aids the ream so it does not forget specific feacures during the review process. This is especially uue to a new RSA team or one chat has nor performed an audic for some time. Even for seasoned safery audicors, prompt lists can be used at the end of che RSA or at a meecing following the field visit co help organize observed problems. A prompt lise is general and identifies broader level items. For example, a prompt lisr may include a reminder to observe specific pMescrian elements (cracked sidewalks, audible pedescrian push buttons, pedescrian beads, marked crossings, etc.). An example prompc list from the P~demian &ad Saftry Audit Guideunn and Prompt Lisa (FHWA, 2007) is provided in Exhibit 18-2. The example prompt list is provided for the analyse co highlight pedestrian focus areas to look for during various stages of a project.


•rR~<;.;c-~@@~~:~"1~ ~~"'.?.~1-.~£~~~~--~~ 0 .··" ~~- !.. · f:~..c;·~~·.- ~-·. '~. . ., + '.:_ • .• • : . .. :~~·.~.--~~-k~~ ·. ...:~~~~· .t-~ .,;t,~·~·!,'t>.~.>!il"5d1?::;n• ..~- ~~~~·.j ::'~!~· RSAStages

Master Promp!

Detailed Prompt

A.2 Quality,

Conditions, and Obstructions

construction

post· construction

Are sidewalks provided along .d1e meet?

./

./

./

./

A.l.2

If no sidewalk is present, is mere a walkablc shoulder (one wide enough oo accommodate cyclists/pedestrians) on the road or other p:uhwayhnil nearby?

./

./

./

./

A.l.3

Are shoulders/sidCW2lks provided on both sides of bridges?

./

./

./

./

A.l.4

Is me sidewalk widm a.Clequate for pedestrian volumes?

./

./

./

./

A.l.5

Is there adequate separation distance ~tween vehicular rraffic and pedestrians?

./

./

./

./

A.l.6

Aze sidewalklsu~t boundaries discernable to people wim visual impai.rmcna?

./

./

./

A. l.7·

Are runps provided as alterna~ves to stain?

./

./

./

./

A.2.1

Will snow sro~e disrupt pcdesuia.n access or visibility?

./

./

,;

./

A.2.2

Is me path dear from both temporary and permanent obstructions?

./

./

./

./

A.2.3

Is the wallting surf.oce roo steep?

./

./

./

./

A.2.4

Is me walking surface adequate and well-maintained?

./

./

./

A.3.1

Arc sidewalks/walkable shoulders continuow and on both sides of me succt?

./

./

./

A.3.2

Aze measures needed to direct pedestrians to safe crossing points and pedestrians access ways?

./

./

./

A.4.1

Is the sidewalk adcq uatdy lit?

./

./

./

./

./

./

./

./

./

./

./

./

./

./

./

./

./

A.3 Continuity and

Connectivity

A.4 ugbting A.4.2

A.5 Vtsibility

design

.>\.IJ

A. I Presence, Design,

and Placement ·

planning

Does sueet lighting improve pedestrian visibility at night?

A.5.1

Arc pedestrians adequately visible when they are walking along the sidewalkfshoulder?

A.6.1

Ate the conditions at driveways interseaing $ldewalks endangering pedestrians?

A.6.2

Does the num~r of driveways make the route undenirable for pedestrian travel?

A.6 Driveways

-

_,

-1

./

./

Source: Pdatritm &ad Saftry Amiit Guitklina tmd fumpt Lim. FHWA. 2007.

..

-

./

. Alternative Safety Studies • lS:;;;iii'

·~

2.3.5 Fuld Rntiew

The field review is the primary focus ofSrep 4 and leads to the findings reported in Step 5. Prior to the site visit, auditors review any available crash data and collision diagrams if evaluating an existing site. This review will likely suggest where, when, to whom, or during what events crashes are taking place and provide a good starting point for looking at potential conflictS or problem areas in the field. In addition, it may be useful to take prompt lists along, especially if the team is new or has not conducred an audit in some time.

It is a good idea for the audit team to travel together to and from the site to encourage discussion among team members. As mentioned earlier, it is a good idea to designate a secretary and photographer to doaunem any observations during the visit. If possible, the phocographer should sit in the front passenger seat to document problem areas from the driver's point of view. This can be done using a still or video camera. It is a good idea to do a drive-through of the site along the mainline and on aU the approaches and movements during both daytime and nighttime conditions, and even during various peak traffic conditions. A nighttime visit is especially good for looking at street lighting and sign rcrroreflecrivity; Conducting multiple visits gives the team an appreciation for the driving tasks and drivers' workloads under aU rypes of conditions. An example could be a side-street permitted right-rum moVWtent with prohibited right-rum-<~n-red. Driving this approach may reveal a signal compliance issue with the green ball indication, with drivers not yielding to pedestrians during the permitted green phase. VISits to larger sites with multiple intersections may restrict the ability to drive each corridor. In this scenario, a video and/or log of the points of interest should be taken while walking the site.

If the RSA is conducred on an existing site, it is wise to walk the site. There is no substitute for getting out of the caJ and walking the site, especially if the site includes pedestrians. RDadside and pavement conditions are much more apparent when the auditor is not in the vehicle. Push buttons, skid marks at the stop bar, roadside hazards such as wood poles, multiple driveway enccances, broken curb cuts, protruding $idewalk tracks, pavement markings at the crosswalks, signs and sight obstructions at driveway and side Street openings are all a:unples of things that ace much easier to see out of the vehicle.

During the drive-through or walk-through, it is ideal to record observations direcrly on an aerial image or plan drawing of the site. However, the drawings may be too large. Another good method for capturing information is to take pictures in an orderly manner as the auditor walks or drives (at, for aamp!e, intersection signs). When taking pictures, auditors should malcc sure to capture the intersection signs as they go. so they are aware of where the picture was t:aken. In addition, the photographer can provide a piaure number to the secretary to include in the notes during the drive or walk.

2.4 -Data Reduction and Analysis 2.4.1 Ctnuluaing th. A1utly.N The RSA analysis and repon preparation are completed in Step 5 of the RSA process. The analysis should take place shortly after site visits so information is fresh in the minds of the audit team. The analysis is done in a workshop atrno· sphere over one or two short sessions. A review of aU the informacion takes place, scatting with the materials provided in the starrup meeting followed by the drive- and/or walk-through of the site. The picrures, videos and any supporting ma.rerials (such as marked aerial images or plan drawings documented by the secretary and photographer) are used to come to conclusions about the site being audited. The audit team then prioritizes and provides possible mitigation measures for all safety issues identified during the site visit. This is the main focus of the RSA. A good method for prioritizing safety issues is to determine the relative safety risk, which is based on frequency and severity. Frequency refers to the likelihood of a collision happening. Severity refers to the type of.crashes and expected speeds of vehicles likely to be involved. Four basic classifications are given for frequency and risk, including: l,ikcJjhood F~uent: ~

5 crashes/year

Occasional: I to 5 crashes/year Infrequent:< 1 crash/year

Rare: < 1 crash/5 years

.

'

~ '

~gh: Fatality likely '.

Medium: Severe injury likely Low: Injury likely

Negligible: PC9perty damage only The matrix provided in Exhibit 18-3 is a good method for determining the relative risk when prioritizing safery issues. A low racing of•A" corresponds with a safety issue the RSA team would consider to luve low risk and •p" correspon4s to expected frequent fatalities. If this method is used, the audit team would prioritize the issues from high risk (F) to low risk (A). .

Accident Frequency Category Source: &ttd Saftty Audit Guilklinn. FHWA; 2006.

Once the analys.is is complete, a report of the findings is prepared fur the local agency. This report should include background information related to the project being audited, the RsA team members and their expertise and materials provided to the team, all site visit dates and times, important notes (such as weather or unusual uaflic events), prioriciucion method used, the overall site observations made during the visits and a list of the safety issues and suggested mitigation (usually one per issue). 2. 4.2 Presnumg FnuJn.11 Once the issues are prioritized (typically by risk and severity), a preliminary findings meeting is hdd with the local road agency. This is Step 6 of the RSA process. Unlike the preliminary audit meeting, this is more a roundtable discussion. The idea is for the local road agency to have an opportunity to ask qtlescions and seek clarification of the RSA findings. The meeting also provides the agency an opportunity co make additional suggestions for improvements in conjunction with the RSA team. However, the meeting is not an opportunity for the local agency to persuade the audit team to eliminate any concerns. Any and all safety concerns should be documented by the team in the RSA final report; the local road agency may provide explanations in their formal response letter. The discussion items in this meeting will be useful and can be included in the RSA report. 2.4.3 Prqlaring 11 FontJIIl Re4p0tue

The local agency is ~ponsible for fullowing up on the issues revealed in the final reporr. This is done in the form of a formal response letter and is Step 7 in the RSA process. The response letter is relatively simple; fur each audit issue, the agency should identify wlut action will (or will not) be taken, with a brief explanation. The response letter is provi-ded to the RSA team and becomes part of the project record with the RSA report. 2.4.4 lru:orporllte Fnulinp The final step in the RSA process is the responsibility of the local agency to incorporate the findings from the audit team. The form21 response provided to the audit team is added to the RSA summary report and given to the local agency for reference throughout the remainder of the construction project. The idea lx:hind an RSA is to be proactive in reducing d\e potential for collisions; therefure, the findings should be conside.red seriously fur implementation as soon as possible. Safety is the primary reason for conducting an RSA; agencies should rernemlx:r findings not imp lemented could be used against the agency ifan agency is later proven to be negligent. . ' Alternative Safety Studies • 3.89

3.0 TRAFFIC CONFLICT STUDIES

3.1 Introduction Traffic conflicts are interactions between rwo or more vehicles or road user.; when one or more vehicles or road users take evasive action, such as braking or weaving, to avoid a collision (Parker and Zegeer, 1988). Engineers use traffic conl!icts as a supplement to uaffic collision srudies in estimating the traffic collision potential at an intersection or other location. Traffic conflicts arc useful because the srudy resulrs are often available much sooner than the resu lrs of traffic collision srudies (when several years' data may be needed). Traffic conflict studies can also provide much more derailed informacion than traffic collision srudies. However, conducting traffic conflict studies is nor simple. When performed improperly, they may provide misleading information. Traffic con8ict studies require a relatively small investment of time and other resources and require no special equip· ment. Trained observers watch traffic and note on a form when a conl!ict occurs. Observers usually require a week or less of training. On a single intersection approach, one or two persons are usually needed for one-half to 3 days. Besides training the observers, engineers establish srudy guidelines and analyze results. Research sponsored by FHWA during the 1980s improved the state of the art of traffic conflict studies. Migletz. et al. (1985) demonstrated traffic conllicts predict future traffic accidenrs about as weU as collision records. Manuals by Parker and Zegeer (1988, 1989) provide excellent detail on how to conduct traffic conflict studies. Traffic confucr studies supplement traffic collision studies in several ways. The magnitude of the traffic safety problem at a particular location can be estimated from traffic conflicts. One possible result of a traffic confuct study at an inter· seCtion is a mean rate of traffic confucrs of a particular type per day. This rate may then be compared to a standard or certain percentile rate from a sample of similar inrerseetions. Trearmenrs may be needed at the location ifthe ~bserved mean rate is higher than the comparison rate. · The use of ua.ffic con8iets for estimating the magnitude of a safety problem is restricred due to the lack of good com· parison confliCt rates. It is time-consurciog to collect a database of traffic conflicr rates for comparison purposes. Glauz and Migletz. (1980) and Migletz. et al. (1985) provide comparison rates for some common types of interseCtions. However, if the intersection types for the published confuct rates do not apply to the location under study, traffic confucts may not be hdpful in estimating the magnitude of the safety problem. Traffic con.l!ict studies are very useful in determining the types of safety problems that exist at a location. Once the type of problem is known, possible countermeasures can be identified. There are, for instance, 14 basic types of traffic conftiets at imersections (Exhibirs 18-4 through 18-17). The rd ative overabundance of one of these 14 types at an intersection reveals a particular problem. Again, a database of typical rates must be available against which to compare a location's rates. Lisrs of countermeasures that may reduce the occurrence of a type of conflict are available (Parker and Zegecr, 1988). At many highway locations, it is impossible to obtain enough rdevant collision data to make such a detailed diagnosis. Traffic confucr data are often collected and analyzed to evaluate the effectiveness of a safety-related countermeasure. Countermeasure evaluation using traffic conflicrs is attractive because traffic conflict data arc available to the analyst before ttaffic collision data. In fact, a before-and-after srudy with traffic confucts may not need comparison or control sites to overcome the history and maturation threats to experiment validity (see Appendix A for details on experiment design). Using traffic conllicrs as the measure of effectiveness in a countermeasure evaluation may also eliminate the threat to experimental v:ilidity posed by regression to the mean, since countermeasures are typically chosen on the basis of collision data rather than conllicr data. To evaluate countermeasures, the conflict types being studied should be closdy rdated to the countermeasures that have been implemented. If not, the true effeCtiveness of the counter· measures will remain. unknown (Parker and Zegeer, 1988).

3.2 Study Design 3.2. I Prim4ry Cmjlicts As stated previously, traffic con£licts are interactions between rwo or more vehicles or road user.; when one or more vehicles or road users take evasive action, such as braking or weaving, to avoid a collision. Thus, aCtions taken by vehicles or road users in response to traflic control devices (TCDs), highway geometry, or weather arc not traffic conllicrs. A driv~r braking to join a queue at a red signal is not involved in a traffic confl.icr. Another driver braking to avoid a 390 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

; rear-end collision with a slow-moving vehicle during a green signal phase is involved in a traffic conAict. ObservtJS use ;,brake lights, squealing tires, or vehicle front ends chat dip or dive as indications that braking occurred and a conflict 'was possible. A collision or near miss during which no evasive actions were observed also counts as a traffic conflict. Traffic conflicts can involve motor vehicles, pedestrians, bicycles and other road users. Rates of pedestrian and motor vehicle conAiccs can be high ar intersections with appreciable pedestrian volumes. Researchers have identified 14 basic types of con!Uccs at intersections, as shown in Exhibits I B-4 through 18-17. T)l.ese conflict types apply at signalized intersections, unsignalized intersections and driveway openings. However, not all che 14 types apply at every interseccion. In mosr conflict studies, observers record only the conflict types that are related to the study purpose rather than all 14 types. Traffic conAict types are not well defined for nonimersecrion locations such as weaving sections, diverges, or merges. Preliminary observations or pilot tests are necessary to inform obserVers which conBiccs to record at nonintersection locations.

I

I

III

~~ . · I · I~L ·

L

--r1!§!1 ~ ~ ~ - -~

---

---

I I

1·

Source: Glauz, W. D. and D. J. Miglecz. Appliclllion of Traffo Confoct Analym Ill lntersmions. National Cooperative Highway Research Program Report219. TRB, 1980.

.

!

.

I

-

Source: Glauz, W. D. and D. J. Miglea. Application of Traffic Confoct A114/ysis ar lnuructions. National Coopemi~ Highway Research Program Report219. TRB, 1980.

·---.--J'l iL _j '· I IL. f!--' P JI. I' I ---

-

I

I

.

t----

--.-

§ __. -

I~

t-· - -

__· -

..

I

I I

· ·

.

I II§ I

I

Source: Glauz, W. D. and D. J. Miglecz. AppLicaJion of Traffic Conflict Arra/ym Ill lntnrmions. National Cooperative

Source: Glauz, W. D. and D. J. Miglea. Application of Traffo Conflict Analysis at lntnr~ctions. National Cooper.cive

Highway Research Program Report 219. TRB, 1980.

Highway ~earch Program R.eporr 219. TRB, 1980. Alternative Safety Studies • 39 __,

_ji

- .- -

\

.

!L ---

_j

~

---

!§~

~

Source: Glauz, W. D. a.nd D. J. Miglecz. AppliCillion of Tnrfftc ConjlUt AMlysis aJ lntn-uaions. National Cooperative Highway Research Program ~port 219. TRB, 1980.

~---

:

I!

1

I

.I

Source: Glauz, W. D. and D. J. Miglecz. Appliutilm of TrrtfJic Gmjlict Analysis aJ lntm«tions. National Cooperative Highway Research Program ~port 219. TRB, 1980. \

I.IL . L I

-·- ·-.·;f-. _· _.,

l . ·lrl 1

I .

I· I

·

Sou=: Glauz, W. D. and D. J. Miglecz. Appliutilm of TrrtfJic ConjliitAn
_j' lI i'L

- - --- ---

I

--- --________ __

l§r-

~CIJJ

~~ I·I

l . I

.

Sowce: Glauz, W. D. and D. J. Miglecz. Application of TrajJU ConjlUt AnAlysis 41/ntmutions. National Cooperative Highway Research Program ~port 219. TRB, 1980.

"L I.

~rn

I

·

Source: Glauz, W. D. and D.]. Mi¥)ea..Applkllli4n of . Traffic ConjlictAn:alysis at ln.tnw:t:ions. National Coopentive Highway Research Program Report 219. TRB. 1980.

_j - .-

.-

I

'l· ~ -·

. · I ~I I

i·· i L _j _·__ . I#. __ . -

1§11 I

.1 ·

I

I

·

·

Source: Gla111, W. D. and D. J. Miglea. .Applkllli4n of Traffic Conj/kt An4Jpis at lntnstt"tiims. Narional Cooperative Highway R=arch Program Report 219. TRB, 1980.

'L. ___j ::L I

I·

.

. ·! I

ccc-==-..-. .

~I

....;

.

\

---

·--~

- .- -

~~~~~ I, I I . I ~~ . I, So=: G!auz., W. D. Uld D.]. Migle~ Application of Tniffo unfticr Anaiym 41 lntm«tiiJns. Natioml Cooperative Highway Rese=h Program Report 219. TRB, 1980.

Source: Gla111, W. D. and D. J. Miglc~ .ApplicatUJn of Tniffo unj/ktAnalysis at lnttruaions. National Cooperative Highway Research Program Report 219. TRB, 1980.

Alternative Safety Studies • 393

JilL . rc - ~~,'<'td'~ . :_ :;:~-e, .· ~:I:UllillMt!~~ ..· 1 - _ -(:.l.lLI:

~

....'!;',~

-

...

.....

~ ;,_~

IL g;IJ

.

.

t:l---

~

~- ·-~-.

~II

I

~

~~-

---

Sou.rce: Glauz, W. D. and D. J. Migle12. Applkation of Traffic Conflict Analyris 41 lntmections. National Cooperative Highway Research Prog.ram Repon 219. TRB, 1980.

IL

I I

- -·-

!r- I !

.....,

fm-~-~~!_:·;

,

-----

I* I§ I

Source: Glauz., W. D. and 0. J. Migle12. Applioz~n of . Traffic Cr.mjlitt Analysis at lntm«tions. National Cooperative Highway Research Program Repon 219. TRB, 1980.

3.2.2 Secondary Conflicts Secondary conflicts occur because of the prior occurrence of another traffic conflict (Parker and Zcgecr, 1988). Secondary conflicts co!Wt primarily of"slow-vehicle, same..&.rection• (Exhibit 18-6) or "lane-change• (Exhibit 18-7) conflicts and involve a third vehicle in response to a conflict becwcen the lim rwo vehicles. Usually, a maximum of one secondary conflict is n:corded for ~ch nW.n conflict even if more than one secondary conll.ict occurs. For e:xample, if an entire platoon of vehicles braked in response to a lead vehicle executing a right rum, three conflicts should be recorded: a "rightrurn, same-diceaion• conflict between the first two vehicles (two primary confliru) and one secondary conB.ict.

3.2..3 Tra.f!U Events Traffic events are unusual, dangerous, or illegal nonconflict maneuvers. Typical traffic events include vehicles running red signals, executing right rums on n:d without a full stop, weaving across painted gore areas and slowing considerably in travd lanes. Traffic events are not alwa}'li considered in confl.ict studies, but can be included along with conll.ict data if they arc determined to coincide with the potential for a traffic collision. Traffic events arc defined very loosdy, and some types of traffic events have not been researched thoroughly. Engineers should be certain the traffic events under scrutiny are useful estimators of traffic collisions. Engineers should also be sure an observer has a clear idea of the actions thar constitute a traffic event. For example, an engineer could define a "running red signal" traffic event as when "the front tires of a motor vehicle that proceeds tluough the intersection without stopping cross the stop bar when the signal is red. • Pilot resting is necessary to enswe there arc no gaps in the defi.n.ition (chat is, actjons that cannot be appropriatdy coded}.

3.3 Data Collection Procedures 3.3. I Personnel Needs Personnel requirements for traffic conflict studies at intersections depend on whether turning movement counts are needed. Turning movement counts are necessary if the traffic confl.ict rote to be estimated is per vehicle rather than per unit of time (discussed in Section 3.3.3). Twning movement counts are also necessary ifstrict experiment controls are used, such as in a before-and-after experiment when the traffic volume is suspected to have changed berwecn the da~a collection periods. Turning movemem counts an: not essential (but may still be useful) for rates per unit of time with some srudy purposes, such as countermeasure generation and estimation of the magnitude of the safety problem at a location. One or rwo persons per intersection approach an: sufficient ro conduct traffic conflict studies. One person can record traffic t:onflicts on one inte.rsection approach if rurning movement counts an: not necessary. One person should also 394 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

be able w observe craflic con fliers and coum turning movements on an approach where three or fewer movements with low or moderate volumes are made. If there are more than three movements or volumes are heavy, rwo or more observers will be needed. Observers should not have to look away from the location where conflicts are b eing watched ' to record turning movements. If turning movements are to be counted at an intersection roo far down stream for an observer tO see both the approach conflict area and the intersection, a second observer counting only n uning movements is needed. A traffic conflict observer should watch only one intersection approach or one end of a weaving a rea at a time. Consequently, when it is desirable co study traffic conflicts at an entire intersection or weaving area, either larger crews of observers must be used or particular observers must stay longer at the location. For example, if traffic conflicts per vehicle are desired on all four legs of a busy intersection and a full day of observation is needed on each approach tO gather the appropriate sample, staffing options include: 1. a crew of six (four conflict observers and rwo turning movement coumers) is scheduled for 1 day; 2. a crew of three (two conflict observers and o~e turning movement counter) is scheduled for 2 days; and 3. a crew of rwo (one conflict observer and one turning movement counter) is scheduled for 4 days. A traffic conflict study requires very little equipment. Observers will need forms, clipboards, pens, watches and a place to sit (a vehicle or a folding chair). Electronic or manual turning movement recorders may easily be modified for traffic conflict studies with a template or by labeling keys so each key is associated with a conflict t)'l'e. However, traffic conflict totals tend to be smaller than turning movement counts so electronic boards are usually not required. Observers have used audiotape recorders co record data during traffic conflict studies. T he extra time necessary to listen to the tape and code the data later makes this alternative less desirable. Finally, video can be used during traffic conflict studies,.A digical video creates a permanent record so close calls can be re-evaluated. The disadvantages of video, including "the extra labor to record and view media and the technical problems associated with lighting and fields of view, usually outweigh the advantages. A well-designed data form remains the best choice for most traffic conB.ict studies.

3.3.2 Training Observers The goal of observer training is to create observers who are consistent with themselves over time, with each other and with the established definitions of conflicts and events. Without the confidence that what is called a conflict now will be called a conflict later, traffic conflict .studies degenerate into simply observing traffic. Conclusions drawn from an inconsistent traffic conflict study are misleading because co~parisons to standard rates or cares from other locations must be made. The amount of training needed by personnel learning to observe traffic conflicts is controversial. Parker and zeg~ eer (1988) recommend a 1-week training period for experienced ~affic technicians and a 2-week training period for others. However, many professionals view those training periods as economically infeasible. [n practice, che amount of training provided varies widely depending on the experience, professional status and motivation of the trainee and the quality of the instruction. Observers should train until they achieve consistent performance. Exhibit 18-18 shows the activities that make up the 2-week training period recommended by Parker and Zegeer (1988). Appropriate training activities include lectures, reading, discussion, videotape viewing, supervised practi. ce and unsupervised practice. The Observer's ManUill by Parker and Zegeer (1989) is available for training preprofessional personnel. •What if" examples are useful to provoke discussions during training. The instructor poses a h-Ypothetical traffic situ'ation verbally, on a chalkboard, on a handout, or on videotape, and the trainees state wheth-er the situation should be classified as a conflict and, if so, what type. An experienced traffic technician or engineer can learn to observe traffic conflicts alone by reading and unsupervised practice. However, it is preferable for [V'VO or more people to learn the technique together so they can discuss key points and compare practice data.

Alternative Safety Studies • 39P S

~~r-<-'"" •

•• •

•

J

•.7.

~T••

•. ·-·- • !.•~,.D,l-ll'/

..

:r:i...

..

... 0

•

""'•

• . . . ;..>;·

. 0 •. •

Period

·

-::r.:.....~!~""!l"ff.--..- r-;:~

""";;. · .·~i~:.c -~--~:

Topic Introductory remarks Oricnration to the training program General background on tnffic safety History of tnffic conflicu

Ovuview of a traf!ic conflict survey What the survey is Day 1

How the results are used How the survey is conducted Contents of the Observer's Manual Traflic counting Turning movements Use of mechanical counting bo:uds In aoduaory field work Pcesenration of traf!ic conflict definitions General definicion

Day2

Same-directioo and oppo$ing lcft-r:urn cOn8icu Group field observations at a signalized intersection Discussion Definitions of cross-traf!ic conflias

Day3

Group field observation at an unsignalized intencction Discussion Use of videotape to illustrate conflict situations Small-group field practice

Day 4

Question-and-answer session Special con6ct types Simulated Umited conflict counts

Day5

Discussion lntcrsection.s Yt'ith unusual geometries Use of other data forms

Day6

Field colkction of other data Discussion

Day7

Simulate4 full con.llict (8-hour day) Discussion Rt:view of the concq>ts and procedures

D~y8

Analysis of day 7 dar:a

Day9

More field practice

Discussion of problem areas

rhylO

Analysis of day 9 dara

More fidd praa:ice

Source: .Parker. M. R. and C.'(. Zcgcer. Trrz/M Cowjlia Ttdmil:puJfor SafnJ tmd OpmaioN: Enginm's G..Uk, FHWAlP-8S-026. FHWA. 1988. 'lK • MANI IAI (lf TRAN<;PORTATiflN

FI\II.II\I~~~~IMr. <;TI U"'IC<; 1~1fl cnoTir"'M

'

· .·

Observer consistency can be estimated by having two or more observers record conflicts independently at the same location and time. If an observer at a particular type oflocation produces results that are consistent with other observ_ers, especially with experienced traffic conflict observers, his or her uaining may be considered complete for th at type ·oflocation. Engineers setting up such a consistency test must make sure the observers see the same ponion of roadway and are not influencing each other. An informal and usually sufficient method for judging consistency is simply to . look over the completed data forms from different observers. A variety of statiscical testS are available for formally estimating consistency. A correlation coefficient, a Z-test for proportions, or a paitcd comparison t-test can be used to estimate the difference between two observers. A group standard deviation, a chi-square test, or an analysis of variance can be used to estimate consistency among a group of three or more observers. An F-rese can be used co compare the variances between two groups of observers or between two trials with the same group of observers. Appendix: A contains a discussion of experiment design (a formal comparison between observers is an experiment) and Appendix C contains a discussion of common statistical tests. There are few standards for observer consistency to indicate whether more uaining is needed. Based on total conflicts at an intersection approach in 15-min. periods, Parker and Zegeer (1988) stare a correlation coefficient of 0 .95 be-

tween two observers is desirable. They also recommend observers whose conflict counts consistendy fall one or more standard deviations above or below the group mean should be singled out for more training. However, if the gro.up standard deviations are small, the group is ready for unsupervised data collection and·singling out particular trainees for further training would waste time.

3.3.3 Sample Siu Requinments The sample siu requirements for a uaffic con.tlict study depend on the type of confuct rate to be analyud. Engineers usc two basic types of rates: rates per unit of time, and rates per vehicle observed. Often, an agency customarily uses only one type of rate. If there is no customary type ofconflict rate, the choice should be based on the relative advantages of each type. Rates per unit of time arc advantageous because tables of such rates are available fur interSections that arc convenient for comparisons. Exhibit 18-19 shows some typical rates per unit of time; Glauz and Migleu (1980) and Migleo: et al. (1985) have provided others. Since existing data arc necessary to judge the magnitude of the safety problem at :a location or to generate lists of countermeasures, those studies generally use rates per unit of time. Also, rates per unit of time are advantageous because turning movement data may not be needed. The major advantage of rates per vehicle is the amount of time spent collecting data to achieve a given level of precision is usually lower, especially when confucts are relatively rare. This is because rates per vehicle are treated as proportions (each vehicle observed either docs or docs not instigate a conflict), whereas rates per unit of time arc ueated as frequencies. Another imponant advantage of a conflict rate per vehicle is no previous knowledge of the variance of the mean conftict rate is necessary to estimate the needed sample si?.e.

'fiL

Note: Basic ~terscctlon conflict types not shown had mean hourly rates less than 0.5. Statistics are based on sample counts conducted in· the Kansas City metropolitan area on all four approaches of signalized incccsccrions and on the approaches with the right-of-way at unsignaliz.ed intersections. Counts were taken during the daylight, in dry weather, and do not include secondary conflicts. a. "All same direction~ includes left-tum same direction, slow vehicle, lane change, and righr-turn same direction conffiet types. "Through cross traffic~ includes cross traffic from left and cross traffic from right conflict types. b Not available. Source: Gbuz, W D. and D. J. Migletz. Appliemion ofTraffic Conflict Artal]sis at lntn1mions. Natio.W Cooperative Highway Reseatch Program Report 219. TRB, 1980; ~liglea, ct al. Rrlationships bmutm Traffic Conflias andA«iJmiJ. FHWARD-84-042, 1985. 398 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

.

·:&
.

---·"'

'

·•

. ... . . . .

... - - .. :::r.· _.,

Consta.nt,

t

•

• > ;'-'-:·

,.·•. c.

... ... _r · ~ ......

~-'!.j'1.<<,~·?,-~~.r-:-·~~-i~~$.C:;:; · >:t ·

--·· -

"-.~-.... _"!",..._~----·-~:~"":"..;;-~:~~

-

Confidence Leve.l (%)

1.28

80

1.5

86.6

1.64

90

1.96

95

2

95.5

2.5

98.8

2.58

99

-

.........•~~

-

---

Once the rype of rare is chosen, the necessary sample size can be calculated. The equation for determining the salflple size needed to estimate a mean conHict rare per unit of pme, if previous estimates of the variance of the mean and. the mean arc available," is:

[(1oo• -'-) ~ PC mean 2

NT •

] •

Equation

1

18~1

where:

NT

• number of units of time that must be observed .. a constant corresponding to the desired levd of confidence, from Exhibit 18-20

PC

• permirred error in the estim~te ·of the mean conflict rate, percent (if the mean hourly conflict rate is 6 and PC = 50, the precision of the estimate is 6 t 50 percent of 6 or 3 to 9 conflicts per hoUr)

var

• expected variance of the conflict rate, from previous studies or Exhibit 18-19

mtan • expected mean of the conHict ate, from previous studies or Exhibit 18-19 Equation 18-1 can be used for conflict rates per any unit of time if prior estimates of the mean and variance of tPC rate are available. If a previous estimate of the variance of the mean is available but not rhe mean itself, the equation becomes: 1

NT •

(...!._) PQ

•

Equation 18.-2

var

.

where NT. t and varare as defined for Equation 18-1 and PQis the permirred error in the estimate of the mean coPBier rate, in confficrs per unit of time (if the levd of confidence corresponding ro tis 90 percent and PQ is 7 conffic-.:s per hour, the acrual conflict rate will be within 7 conflicts per hour of the estimated rate 90 percent of the time). EXAMPLE 18-1

For a.n approach to a signalized intersection with a total enuy volume of 35,000 vehicles per day, an engineer wan-es an estimate of the mq.n number of same-direction confficts in an hour. The engineer would like to be 90 perceot su.t:"e the estimate is within 10 percent of the actual rate. Using Equation 18-1 and Exhlbits 18-19 and 18-20,

NT=

[(too • 1.64 f] .~ 10 90 2

• 2.5 hours of observation

If conffict rates per vehicle arc desired, the sample size necessary to achieve a certain precision in the escimate of .::;;a mean conftict rate is:

Alternative Safety Studies • 3 g$lll

..:..'r

.,

;:

l

p•q• rz NV = ~~ ppz

Equation 18-3

where tis as defined for Equation 18-1, and where: NV

e

number of vehicles that mwt be obsetved

p

=

expected proponion of vehicles observed that are involved in conHias

q

• expected proportion of vehicles observed that are not involved in conHias

PP • permitted error of the estimate of the proportion ofvehicles involved in conllia:s, in a proportion~ 0 and 1.

If the levd _of confidence corresponding tot is 95 percent and PP is 0.01, the actual conHict.ratc per vehicle will be within 0.01 of the estimated rate 95 percent of the time. The sum ofp and q mwc be 1.0 in Equation 18-3. A consetvative estimate of the sample size (that is, a larger sample will be gathered than probably is necessary to achieve a given precision) can be provided &om Equation 18-3 without prior knowledge ofp and q. Ifp and q arc unknown, they are assumed as 0.5 and Equation 18-3 reduces to: 11

NV= 0.25 • pp1

Equation 18-4

Since conHict rates per vehicle are wually much closer to zero than to 0.5, use aJ). estimate of p and Equation 18-3 Hthe.r than Equation 18-4 to reduce the sample size estimate dramatically. EXAMPLE 18-2

An engineer wants an estimate of the mean rate of right-turn same-direction conflictS per approach vehicle. The engineer would like to be 95 percent sure the estimate is within 0.01 of the actual rare. Using Exhibit 18-20 and the conservative Equation 18-4,

Nv = 0.25 • 1.9621 0.01

- 9,600.

Therefore, 9,600 approach vehicles would have to be obsctved to achieve the desired precision. If the engineer is certain that the con.ll.ict race is tess than 5 pcrcenr, Equation 18-3 can be wed and the needed sample size is much smaller.

NV= 0.05 • 0.95 • 1.96 0.011 e

1,800 vehicles

Inverse sampling is another sample size formulation for proportions chat may be useful for traffic conHict studies wing races per vehicle. Inverse sampling applies when the proportion to be estimated is known co be small (for example, less than 0.10) bur no rdiable estimate of the proportion is available. The sample size formulation given in Equations 18-3 and 18-4 may require a larger sample of vehicles than is necessary for a typical estimate. Inverse sampling depends on the coefficient of variation, which is the ratio of the standard error of the proportion to the proportion being estimated. The lower the coefficient of variation, the more precise the estimatt. Cochran (1977) bas shown: CV <

\ \ \

.Jm

Equation 18-5

m-1

where CVis the coefficient of variation and m is the number of occurrences of the event to be obsetved (that is, the number of conflictS). Exhibit 18-21 is based on this inequalicy (Equation 18-5). To achieve a certain coefficient of variation, one keeps observing until m con.ll.ias arc counted. For instance, observing an approach until 27 coafficts of a particular type arc recorded. guarantees a coefficient of variation ofless than 20 percent. Exhibit 18· 21 illwuatcs how difficult it is to gee a precis~ estimate of a very low proportion. • • -- · • , ........... _,..__ , ,. .. ,. , .. ,,.,.nu.t,.- I""TI H''IICC "''MI"\ c:nmnM

3,3.4 Omtiuding the Stuily . Conflict observers sit upstream of the fearure of interest. Observers record each conR.ict that happens between their position and the feature of interest and ignore conflicts observed in other places. The distance between the observer position and the intersection or other feature depends primarily on the type of fearure, the purpose of the study, the visibility afforded by different positions md the speed of vehicles being observed. Observers are typica.l.ly positioned 100 to 300 feet {ft.) {30 to 91 meter (m}} from inte=tions in wban areas with cluttered roadsides and relatively low approach speeds. Observers are 300ft. (91 m) or more from intersections during srudies in suburban areas with uncluttered roadsides and rencivdy high approach speeds. Observer posicion is constant during repeated visits t6 the same site. The c:listanC:c beQVeen observers md fearures of interest should also be as consistent as possible during studies comparing different sites. Observers should try to be concealed from approaching traffic while keeping visible the area to be observed. Usually, w observer sitting in a vehicle, parked Lcp11y in the shoulder or just off the roadw:ay, is sufficiendy concealed. If no leg11 parking space is available, positioning observers on folding chairs behind utility poles, trees, or any fixed roadside object is adequate. Cof\llict srudies are conducted in daylight with dty weather and pavement unless the study is specifically oriented to other conditions. Weekdays between 7:00a.m. and 6:00p.m. are the usual hours for conHict studies. In fact, daily statistics, such as given in Exhibit 18-19, are based on this 11-houc period. Ducing a study, similar time periods should be used at each site. Conflict studies are scheduled to avoid periods of recucrent congestion, since conflict data collected under stop-md-go conditions are invalid. Observers should alSo avoid unwual traffic conditions such as constrUction or malnrenance. If unusual traffic conditions suddenly occuc during an observation period (a signal malfunctions, a collision occurs, a maintenance crew arrives, etc.), observers should note the time and nature of th!;. condition. Observers should stop temporarily ifit appears typical traflic conditions will be quickly restored, or observers should quit for the day ifit appe= the unusual condition will ~a long time. Observers must maintain a high levd of concentration during a traffic conflict srudy. Frequent breaks allow observers to regain concentration and allow tasks such as recording data, clearing counters and changing forms to be performed without disttacti.ng from observing. Parker and Zegeer (1988) recommend a 20- or 25-min. observation period followed by a 10- or 5-min, break ducing each half bouc of a conR.ict study. Before conflictS are observed, it is wise to record data on the physical features of the site. A condition diagr.un of the site, such as the one shown in Exhibit 17-10, may be helpful in diagnosing a safety problem from conflict data. Pllotos of the site may also be useful.

Alternative Safety Studies • 401

L INTERSECTION TRAFFIC Location _ __ _ _ __ _ _ _ _ _ _~-:-:-:-:-~::\---Date Observer(s) _ __ __ C"' Conflict

Day Ql

Ql

E

F--.

"0

.... ~

"' "'

c..

... c ::z: ......

E

~ ~

Vl= :::1

0

0

•t: Ql Ql

F

u

E :::1 0 > s:. 0

"'e

Q. Q.

Left-Turn Same Direction

c

sc

Right-Turn Same Direction

c

sc

Slow Vehicle

Lane Change

c sc

c sc

Opposing Left-Turn

c

Right-Turn From-Right

sc

c sc

I

I

Total

c +sc Daily Count Rate Per 1,000 Veh

Source: Parker, M. R. and C. V. Zegeer. Traffic Conflict uchmqun for SaftlJ and OpmuioiU: Enginm's Guitk, FHWAIP-88-026. FHWA. 1988.

402 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITION

N

CONFLICTS SUMMARY

_ _ __ L.eg

Numbl";~~h

<$)

of Reconliog "'riod

SC = Secondarv Conti' - Left-Turn Through Right-Turn Left-Tum Through Right-Turn On-Red From-Right From-Right From-Left From-Left From-Left

c sc

c sc

c sc

c sc c sc

c sc

All Same Direction

c

sc

All Through CrossTraffic

c

...
.s:: .... 0

sc

·.

..

; -

------

Alternative S
i.: .

. ~.

-•-·

..,

.. ;.

6

' lilt•

'-4-•

..

,

0

' r -.;

.....

ACfORCODES

.

t~

. . ........ .. .~~:/i'T:
.').;

ACTION CODES

. · .~.

·~

' ...~.J~~~~~~

Name: Date: Tune Peria
Tune

Actor 1

Action

Actor 2

Action

Comments

'

Source: Hwnmer ec al. An &alsumon ofuJUiing W11VS lAgging J4i Tt.m Signm Pluuing. Fi711ll &port, FHWA-IN· JHRP-89-17. Purdue Univenicy,Joim Highway Research Project, 1989.

Exhibits 18-22 and 18-23 conu.in forms for recording mffic con8ias (the forms are reproduced in Appen
\

\ \

404 • MANUAL OF TRANSPORTATION ENGINEERIN
·r.

'J

. ... ,._.

·~~<;;"""•

'

\

•

•

i

:

,...

~

~--·.

• =t -~~JJ•. · .·•

j

........ ~-

.

. .

!~

D ~

,1

lo~ •s 1l

~-

,.,.. ._

··~:.-

0

~

.~ .. ...-....•

···-~:;-~~~A

Lclt Tum Projea Conflict Data

ACfOR CODES f.

,.

·.•·

ACTION CODES

.

Name:BMS

+- " Hescir:ue on Green Alrow

Dare: 12119/09

H • Hesitare on Green Ball R ~ Ran lW

lime Period: 5:00pm-5:30pm

!:\

S • Srop D ~ Decderate

i~

A • Accelerare

lnrerscction: Arendell Srreet & Manslidd Pltwy Direction (leg with actOr 1): West Weather: Clear 4o•

B • BackUp

w .. weave

Ttme

Actor 1

Action

5:03

1

H

5:05

1

5:05

1

Actor 2

.Action

l1

s s

R

" 3

D

B

11

B

2

w

11

s

5:07

1

5:12

1

5:13

1

5:16

2

w

5<17

1

R

5:18

2

R

5:19

1

w

Comments

Driver did nothing unusual

Seemed confused abour directions

·--

Several key points apply equally to traffic conflict dara forms with one line per observation period (Exhlbir 18-22) and one line per conflict (Exhibit 18-23). Fusdy, c:xrensive resting of and practice with a. form is CS$ential. Secondly, observers must complete all •header" information or the dara on a form are useless. A critical piece of header information that is often forgotten is the direction of the approach leg being nudied. Thirdly, every form should bave a place for comments, which are important during dara analysis.

3.4 Data Reduction and Analysis Before data are reduced to an analyz.able format, the data forms must be checked for commentS or descriptions of unusual events. If an event is described that probably biased a certain section of data, analyses should omit that section. A check with the observers may be necessary to dea.r: disaepancies in the data. Traffic conBict data are reduced by summing the totals of each type of traffic conflict. If the data collection form re7 .sembles that shown in Exhibit 18-22, reduction consists of summing each column. If the form resembles tliit shown in Exhibit 18-23, the number oflines mrresponding to panicular mnHia types must be determined. Unless the srudy is a-

JllliL uemely large, this is best accomplished by manually scanning the completed data forms instead of by a computer program. A manual scan allows flexible scoring of conflicts by type and allows the analyst ro obtain a good feel for the data. During data reducrion it may also be appropriate to sum counts from individual approaches into a coral for an intersection. If conflicts per unit of rime are of interest, analysts need to adjust for unobserved rime periods. For example, suppose a particular study calls for an hourly conflict rate. The observers used 20-min. data collection blocks with 10-min. breaks becween blocks so 40 min. of data were collected per hour. Therefore, multiplying the number of conflicts in rwo adjacent blocks by 60/40 provides the needed rare per hour. Analysts should adjust for unobserved time periods with data from similar time periods that were observed, not by assuming constant conflicr rates throughout a long period. A total of 30 conflictS recorded from noon to 3:00 p.m. should not be expanded to a standard 11-hour day (7:00 a.m. to 6:00 p.m.) coral of 110, because the 3 hours observed did not covet either the a.m. or p.m. peak hours: It usually would be permissible to use a 3:00 to 4:00 p.m. count of x conflicts and a 5:00 £O 6:00 p.m. count of y confHas to esrimarc the 4:00 to 5:00 p.m. count would have been (x + y)l2 conflicts. Conflict rates per vehicle observed are produced by combining the conflict sum and the appropriate rurning movement count. An intersection or approach count should be done at the same time the conflict counts are recorded. Once the data are reduced, the purpose of the srudy should guide analysts. For many studies, mean rates of particular types of conflictS and the accompanying standard deviations are necessary. Appendix C provides descriptions of and formulas for conunon statistical procedures. If the srudy purpose is to determine the magnirude of the safety problem or to generate countermeasures, comparison conflict rates are needed. For agencies without their own comparison rates, Exhibit 18-19 provides typical rates for some types of conflicts at intersections.

4.0 ADVISORY SPEEDS

4.1 Introduction Most U.S. hlghway agencies post advisory speeds for some changes in horizontal alignment. The first known record for setting advisory speeds goes back to 1940 (Moyer). Eight years later, the MUTCD adopted the curve advisory speeds for implementation throughout the United States. Recent research sug,aesrs that methodologies commonly used for the determination of advisoty speeds for changes in horizon raJ aligrunent can be improved upon with the use of updated design criteria that are more reflective of current driving conditions. In 2010, ITE approved a new informational reporr, Methodologies for the Iktmninmion ofAdvisory Speedr describing three methodologies that may be used to determine advisory speeds including: design speed equation, traditional ball-bank indicator and the newer accelerometer method. These three techniques are very closely related and should provide the same or similar advisory speeds. It is expected many of the procedures for setting advisory speeds used by agencies will now have a standard criteria to compare agaipst, rhus eliminating many of the very conservative speeds currendy posted. Three acceptable methods bave been developed for determining advisory sp«ds along horizontal curves: design speed equation, traditional ball-bank indicator and the newer accelerometer method. The three techniques are very closely related ar1d should provide the same or similar advisory speeds. It is expected many of the procedures for setting advisory speeds used by agencies will now have a standard criteria to compare against, thus eliminating many of the very conservative speeds currently posted.

4.2 Study Preparation The choice of srudy method is dependent on the agency conducting the study. If agencies have easy access to plan drawings or have a limited number of horizontal curves which need srudying, the design equation method may be an option worth considering. Measuring superdevation and curve radii, the inputs into the design equation, may be cumbersome for larger jurisdictions. The ball-bank indicator method is the traditional method for determining advisory speeds, and thus the one most agencies are familiar with. It is simple and easy to learn. The accelerometer method is becoming more popular as it is basically the same study method as the ball-bank indicator; however, only one person is required to drive the srudy site. This section will provide informacion on equipment, training, sample sizes ~d any potential da~ items necessary for each of the three srudy methods. 406 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

· 4.2.1 Equipmmt Needs :The equipment required for an advisory speed study is dependent on the method used for analysis. The design ·equation method requires one first determine rhe radius and superelevation of a curve. If the radius can not be easily 'determined from plans or circular templates overlaid onto an aerial drawing, a field study is required. A rape measure is necessary to make measurements. If the superelevarion cannot be determined from plan drawings, a 4-ft. (1.2 rn) level is required to make the measurements in the field. It is important that a 4-ft. (I .2 m) level is used so any ruts or imperfections in the roadway do not skew measurements. Some newer levels, such as the one shown in Exhibit 18-25, will calculate the superelevarion automatically.

..___....,,

"·==. . . . --.

-- ------.-··

.::~

------

The ball-bank and accelerometer methods require a vehicle rhac is driven through the curves identified for study. The vehicle used should be a standard passenger vehicle, not a high performance vehicle, truck, van, or sports utility vehicle which may have significantly inore or less body roll than a Stlllldard vehicle. The ball-bank indicator uses a curved rube filled with liqUid. A weighted ball floats i~ the rube. The move· ment of the ball is measured in degrees of deflecrion, which is based on a combination of effects related co superelevation, lateral acceleration and body roll. A ballbank indicator is shown in Exhibit 18-26.

Source: Rieker, Inc.

The accelerometer is similar to a ball-bank indicator but measures laceral acceleration only: Some accelerometers, such as the one shown in Exhibit 18-27, will automatically correlate the laceral acceleration to a ball-bank reading. Either thdace¢ acceleration or the correlated ball-bank reading can be used to determine the po-sted advisory speed.

Source: Rieker, Inc. Alternative Safety Studies • 401""'

Both ball-bank indicators and accelerometers arc very easy to inscall. They can both be mounted permanently; however, they are cypically mounted semipermancntly with suction cups or Vdcro strips. Both unitS have methods for zeroing. or "kvding." when placed on a flar surface. Further information on how to use the equipment to collect the various data elements is discussed in Section 4.2.4.

4.22 Penomu:l anti TrAining &tpdrnnnm Conducting an advisory speed study requires minimal training and personnel. Just as the equipment needs are different for each study, !he number of people and training necessary for each is also differenc.lf radius and superdevacion data can be collected in the office, !he design equation method may require no formal training. However, if field data collection is necessary to determine the radius and superclcvation, a minimum of rwo people will be necessary to determine !he radius of a curve. Personnel will need to wear reflective orange vests at all times along !he side of !he road. It is a good idea for born persons to look out for one another nat to the roadside. It might even be necessary to have a third person as a lookout if traffic volumes are high. /.$ mentioned earlier, ball-bank and accelerometer methods are similar. However, a bali-bank indicator requires a

second person be available to manually record data while the driver drives a constant speed and pays attention to !he road. The accdc:rometer method only requires one person because the device will store data, allowing one to download the data in the office. Ball-bank and accderometer me!hods require !hac the respective devices be calibrated prior to collecting any data. The easiest method for calibration of either device is to use a low-volume undivided roulway and center the vehicle over the centerline of !he roui along a lengthy tangent section. The driver and passenger should not change places once !he device is calibrated as a shift in body roll and/or suspension may .cause a change in de8eccion. The speedometer of the vehicle should also be calibrated prior to conducting any studies. This should be done every 5 mi.lcs per hour (mph) (8 kilometers/hour [km/h)) on a periodic basis, ideally with a laser speed gun. The calibrated speed should be used when conducting the acrual ball-bank or accelerometer study. Tires should be in8ated to the proper pressure so body roll is not signiJi.cancly affected. 4.2.3 s-pk Siu R.tqui-mttnu For each of the three study methods described, great precision of the cscimate is not critical and a reasonable cstimace of the average will yield an appropriate wvisory speed. It is a good idea tO take three to five measurements for each method of analysis. If the design equation method is used and radius is needed, a minimum of three measurements along different parts of the curve will suffice. The superclcvation can be dcrennined using a levd ar different locations along the curve. For the ball-bank indicator and accelerometer methods, a minimum of three runs should be completed ac each 5 mph (8 kmlh) increment in both directions of cravd, for a coral of six runs per 5 mph (8 kmlh) increment. 4.24 Mu.tfll'ing RMiiru • s~ The design equation method requires the curve radius and superclevation. The easiest way co coUcct curve radii data is to overlay circular templates onto an aecial image. This can be done using hand-drawn or computer-generated templates that are scaled to the referenced accial image. An alternative is co use the •chord and middle ordinate" method which is slighdy more cime-ronsuming. Using this method, the equation for dcrermining the radius of a curve is:

l2

h

R=-+Bh 2 where: R ~ curve radius. (ft.)

I • chord length.(ft.) h

1.

~

midclle ordinate (ft.)


Equation 18-6

jThe chord length ({) and middle 'ordinate (h) for a horizontal curve 1are shown in Exhibit 18-28. First, 0 ~ chord of some predetermined length should be measured from the edge of pavement or outer Middle edge of the lane (if the pavement Ordinate,fi edge is not a smooth line similano the lane line). A measurement of A' 'B 50 or 100 ft. (15 to 30 m) is usually adequate and works well with Chord,£ most survey tapes. Measurements should be done near the middle of the curve to make sure any possible spiral curves are not included. The middle ordinate is the longest measurement from the chord to the edge of pavement or lane line, whichever is used. This is meaiured from the middle of the chord. Middle ordinate values will be small (in the order of inches) and a small change can affect the radius calcula:tion by a couple hundred feet. Several measwements along the curve will help alleviate some of the error and are strongly recommended. Typically, it is good practice to use the median value of three measured middle ordinate values measured roughly in the center of the curve.

ca

Superelevation is another input into the design equation. Superelevadon (l") can be determined from plan drawings, but in many cases it will need to be determined in the field. If a field calculation is needed, a carpenter's level can be used by taking one end of the level, laying it on the pavement and raising the other end until th.e bubble indicator reads true. The superelevation is measured as the vertical distance divided by the horizontal distance (the length of the level) and expressed as a percent. As indicated earlier, measurements should be taken in several locations, ideally in the center of each lane. The minimum superelevation in the curve should be used for determining the advisory speed.

4.3 ·oata Reduction and Analysis The reduction and analysis of data for setting advisory speeds is simple and is dependent on the method used and/or the equipment available. As noted in earlier sections, certain methods can be easier if plan drawings are available or the analyst has the abiliry to manipulate aerial images. The ball-bank and accelerometer methods use Exhibit 18-29 as a reference for determining the correct posted advisory speed for a horizontal curve. This exhibit will be referenced throughout the rest of this section. Data collection forms are available in Appendix E· for each of the methods shown, with a cover sheer that should be ;submitted fOr each study.

Source: Seyfried, I<. and J. Pline. "Guidelines for the Determinacion of Advisory Speeds." rrE}tnmutl, January 2009.

Alternative Safetv Studies • 409

4.3.1 Design Equation Metbotl The design equation method is based on AASHTO's Policy on G~ommic Design for Highways and Smm. Ir was originally used to determine a good starting speed for the traditional bal l-bank indica10r method; however, the three methods identified in this section have been modified so that each should give the same advisory speed. The design equation is: V = ..f15R(O.Ole +f)

Equation 18-7

where:

V • design speed (mph)

R = curve radius (ft.) t

=superc:levation ( percent), and

f • side friction factor The radius and superelevation are collected using the methods described in Section 4.2.4 of this chapter. This method is iterative because the side friction factor is not known. Therefore, one can assume the side friction factor &om Exhibit 18-29, which is denoted as the •lateral acceleration: A good method is to use the side friction factor associated with an advisory speed of25-30 mph (40-48 km/h) and check if that speed is correct. If the design speed does not fall within the bounds of25-30 mph (40-48 kmlh), compute another iteration at the nearest speed previously p.lculatcd. Design speeds should be set to the ntarm 5 mph (8 km/h) increment.

Radiw (ft..)

Sour= Seyfried, K. and J. Pline. "Guidelines for the Determination of Advisory Speeds." ffE]ournal, JanU2rf 2009.

Exhibit 18-30 shows the calculated values for standard radius and superelevacion. The data sh«t provided in Appendix E-46 is useful for recording field-collected clara and documenting the work done at thar site.

EXAMPLE 18-3: Application of the Design Equation Method What advisory speed should we post at a curve with a field-measured middle ordinate of 15 inch (using a 50 ft. (15 m] chord) and a st,~perelevation of 6 percent? Solution: In order to use the design equation method, we need to calculate a curve radius. Using Equation 18-6, the curve radii, using the chord and middle ordinate method, is 250ft. (76 m). Using the design equation, we assume the lateral acceleration is 0.24 for a speed of25--30 mph (40-48 kmlh). The calculated design speed is 33.54 mph (53.98 kmlh). Therefore, we should redo the calculation using the lateral acceleration of 0.21 for a speed of 35 mph (56 kmlb) and higher. R=Uculacing, we compute a ~go speed of31.82 mph (49.89 kmlh). & the method suuests, we set the advisory speed to the nearest 5 mph (8 km/h) increme.nt, at 30 mph (48 kmlh). Note, if the second iteration were not done, we would have set the advisory speed at 35 mph (56 kmlb).


4.3.2 Ball-Bank Indicatcr Method . The ball-bank indicator method is currently the most common method for determining advisory speeds due co its · simplicity. It is important the instructions for vehicle use and calibration of the speedometer and ball- bank are fol· towed during any advisory speed study.

Once the driver and data collector are ready, an assumed starting speed should be determined. Two simple methods for assuming a starting speed are 1) arbitrarily choosing a speed 10 mph (16 kmlh) below the posted s peed limit or 2) drive 5 mph (8 km/h) below the driver's comfortable speed. As mentioned earlier, the driver should make sure h e or she is driving the "calibrated speed" and not the speed noted on the speedometer. A minimum of three passes in each direction is necessary to obca.in a fair reading of the ball-bank indicator. This will yield a minim ttm of six tot al passes through the curve at each speed until the desired advisory speed is determined. The speed should be inccea.sed or decreased by 5 mph (8 km/h) until the average ball-bank reading is not exceeded for either direction o f travel based on Exhibit 18-29. Data collection forms are available in Appendix E-47.

EXAMPLE 18-4: Appli~tion of the Ball-Bank Indicator Method A roadway with a posted 45 mph (72 km/h) speed limit has a curve that has been identified as potentially risky. 'fhe responsible agency has been asked to determine if the curve is safe as is or if it needs an advisory speed posted. A{ter inflating the tires to the appropriate air pressures and calibrating the ball-bank indicator and speedometer, two an~ lysts make a series C?f passes through the curve starting at a speed of35 mph (56 km/h). The data are shown in .ExhibH 18-31. What should the posted advisory speed be for chis curve?

Direction of

Travd

North

13.4

South

11.4

North

ll.S

South

10.0

Solution: The ball-bank results indicate an advisory speed should be posted. Based on the results, the northbou~ d direction of travel q>nsistently yields higher ball-bank readings (the superelevation is most likely lower in th::3-t lane); therefore, this 'will be the direction used to set the advisory speed. The first set of 35 mph (56 km/h) runs i..Jl the northbound direction yielded an average ball-bank reading of 13.4 degrees, which is higher than the recoif.1· mended reading based on Exhibit 18-29. The speed was reduced by 5 mph to 30 mph (8 km/h to 48 km/h) and tb-e crew made three more runs in each direction. As expected, the northbottnd direction again yielded higher reading-S·

Alternative Safety Studies • 41 -1

The average ball-bank reading was 11.8 degrees, which falls below the maximum required reading of 12 degrees. Therefore, the agency should post an advisory speed of 30 mph (48 km/h) along this curve.

4.3.3 AccekrtmJeter The accelerometer method is becoming a more popular method for determining advisory speeds. The accelerometer is very similar to the ball-bank indicator, with only two major differences. Firstly, this method only requires one person for the study since the accelerometer will store data that can be downloaded at a later time and date. Secondly, an accelerometer measures lateral acceleration only (measwed in "g~s), and does not take into accounr body roU or superelevation. However, each of the three methods has been calibrated and should give roughly the same advisory speeds. The srudy is conducted in exactly the same manner as the ball-bank indicator method. The lateral acceleration is read from Exhibit 18-29 in lieu of ball-bank readings; however, some accelerometers will convert lateral acceleration to b
5.0 REFERENCES American Association ofState Highway and Transportation Officials. A Policy on QotMtric lksign ofHighways and Struts ("Ac.cderometer and Ball-Bank.") Washington, DC: AASHTO, 2004. Austroads. Road Saftty Audit Guuulines. Sydney, New South Wales, Australia: Austroads, 2002.

Cocliiail, W. G. Smltpling TtthniqUIJ, 3rd cd. Ne'iv York: Wiley, I9n: pp. 76-n. Federal Highway Administration. Manrilll on Uniform Traffic umrrol Dtvices for Strem and Highways. Washington, DC: FHWA, 2009. Federal Highway Administration. Ptdestrian Road Safety Audit Guwlines and Prompt Administration, 2007.

n;ts. McLean, VA:. Federal Highway

Federal Highway Admini.stration. &ad Saftty Audit Guuulirus. Mclean, VA: Federal Highway Administration, 2006. Glauz, W. D. and D.]. Miglett. •Application ofTra.ffic Conflict Analysi.s at Intersections.• National Cooperative Highway ~rch Program Report 219. Washington, DC:Transportacion Research Board, 1980. Hummer,]. E., R. E. Montgomery and K. C. Sinha. An Evalrilltion ofuadint~JmUS lAgging Lift Tum SignAl Phasing, Final Report, FHWAIIN/]HRP-89/17. West Lafayette, IN: Pusdue University, Joint Highway Research Project, 1989. Miglea, D. ]., W. D. Glauz and K. M. Bauer. &/4tionships bnw«n Traffic Conflim and Accidmtr, FHWA/RD-84/042. Washington, DC: Federal Highway Administration, 1985. Moyer, R. A and D.S. Berry, "Marking Highway Curves with Safe Speed Indications," Proceedings of the High~y Research Board, Vol. 20, 1940. National Roads Authority. Design Manualfor Roatfs and BriJgn-VoiUtM 5. Dublin, Ireland: National Roads Authority, 2009. Parker, M. R and<; V. Zt:geer. Traffic Conjlia Techniques for Safoy and Optrasions: Enginm's Guiik, FHWA-IP-88-026. McLean, VA; Federal Highway Administration, 1988. ·

Parker, M. R. and C, V. Zegecr. TrtlfJic Conjlia Ttclmiqutt for Safny and Optrasions: Obsnwri Manual, FHWA-IP-88-027. McLean, VA:. Federal Highway Administration, 1989. · Public Road Administration. Manrilll on Uniform Traffic Control Dtvim for Smm and Highways. Washington, DC: Public Road Administration, 1948: pp. 39, 53.

Seyfried, K. ar~d]. Pline. "Guidelines for the Determinacion of Advisocy Speeds.• ITE ]Du11141. Jar~uary 2009. Trarisfund. Road Saftty Asdit Promiurts for Projtm. Wellington, New Zealand: Transfund, 2004.

Chapter 19 \ '

t

t

t

t

t

t

t

•

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

I

•

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

t

•

t

t

t

t

t

t

t

t

t

t

t

•

t

t

t

t

t

t

t

t

I

t

•

•

Roadway Lighting Origitud By.L FJJis IGnr. D. Eng., P.E. EdiudBy:

DllllielJ Findley, P.E. 1.0 INTRODUCTION

413


414

2.1 Existing Conditions

414

2.2 Before-and-After Analysis

415


415

3.1 Inventory of Existing Lighting

415

3.2 Roadway Grouping

415

3.3 Roadwa:t: Pedestrian WalkWay and Bikeway Classifications 419

4.0

3.4 Area Classifications

419

3.5 Collisions

420

3.6 Traffic Volumes

423


424

4.1 Control for Normal Variation

424

4.2 Use of Night Percentage

424

4.3 Ratio of Rates

424

4.4 Improvement Evaluation

425

5.0 REFERENCES

426


426


427

5.3 Other Resources

427

1.0 INTRODUcTION he ability to dearly sec the roadway is essential for the safe and efficient Bow of ttaflic on our bighwaf3. However, in many instances, limi12tions of the hwnan eye prevent vehicle headlights alone from Cl)lllplctdy satisfying visual nighttime driving requirements. Fixed roadway lighting supplements vehicle headlights by extending the visibility range both longitudina.lly (distance along the roadway) and latcrally (distance across the roadway), thus aiding the driver by providing earlier warning of hazards on or near the roadway. Research shows the nighttime collision rare can be reduced by the provision of ad~uate roadway lighting (Box, 1971, 1972a, 1972b, 1989; Walker and Robe~, 1976). lighting defines the roadway geometries, such as the edge of pavement, curves and dead ends, and illwnirutcs obsttuctions in or near the roadway, including channelization islands, bridge pictS and parked cars. lighting allows the driver to sec a pedestrian in the road?ny beyond the headlight beam and even before 'the pedeStrian enters the road. It also aids pedestrians by illuminating obstacles on the: sidewalk and roadway in their vicinity.

T

Rr"'~that:il\t l

inktinn •

41~

Lighting raises the surrounding brightness level to which the driver's eye.s adapt and increases the driver's concrast sensitivity, resulting in an overall improvement in the driver's ability co see. Fixed roadWlly lighring also conrributes ro a more pleasant and comfortable night d.riving environment, which in turn reduces driver fatigue and improves driver efficiency. Lighting is an aid to police surveillance, and a reduction in stree.t crimes may be experienced foUowing installation of improved srreet iUuminacion. Auto theft, assault and vandalism are three of the types of night crimes most frequently cited as being reduced. While the reported impaccs of Lighting on crime are statistically inconclusive, there are srrong indications the fear of crime is reduced following increases in street lighting and chat feelings of safety are higher (lien, 1979, Sherman et al., 1997). The negative aspects of lighting include glare, collisions with light poles, initial insrallation costs and continuing maintenance and energy costs for the lighting system. Ught pollution is a concern with roadway lighting and should be considered to maximize the benefits to roadway users and minimize the unWllnted effects of sky glow, light trespass and glare (NLPIP, 2007). Light sources include (Szary, Maher, Srriz.ki and Moini, 2005): • mercury vapor (MV) • metal slide • low pressure sodium • high pressure sodium (HPS) • HPS recro white • HPS resrrike • induction lighting (Icetron, quam light [QL]) • fluorescent • comp:act fluorescent • light-emitting diode {LED) • solar

2.0 TYPES OF STUDIES The two primary types of srudie.s presented in this chapter are of existing conditions and before-and-after analysis. These types of srudies focus on lighting impactS on collisions (or on crime, if desired). An existing-condition srudy attempts to srudy lighted roadways and compare them to unlighted roadways. A before-and-after analysis compares dte same roadway for dte rime period before :llld after the implemen12tion of lighting improvements.

2.1 Existing Conditions In this type of srudy, existing conditions are determined at a specific location, along a given route, or in a defined area of the city. Existing facilities are inventoried and coUision data are collected. In terms of collisions, the tabulation may yid d the percentage of total collisions that occur at rught. Another measure is in terms of direct exposure, using the collision rate per million vehicle miles (mvm) or per 100 mvm. The rates may be calculated separatdy for day and for night collisions. The ratio of these rates, such as the nightlday ratio, may be used to identify sites that could benefit from improved lighring. More detailed discussion on collision srudies is presented in Chapter 17. Srudie.s of c:risting conditions are comparative in naCUie. They are first used co determine collision or crime &equency at locations with adequate lighting. This provides the norm to be used for comparison purposes. If the srudy concerns crime, ic must be focused on the number of total crimes of particular types, and dte number and/or percent of these at night.


·. 2.2. Before-and-After Analysis : The potential effect of an improvement in lighting sometimes can be judged by measuring the acrual changes in · ~ollision or crime conditions for routes or areas that have had an upgrading of illumination. Such srudies should be made on a routine basis to justifjr additional expenditures. The studies may show a need to modifjr ligh ting designs if substantial reductions in night collisions and/or crime are not .occurring. Chapter 5 presents more detail on beforeand·a.fter analysis with a discussion on inferential statistics.

3.0 DATA COLLECTION PROCEDURES 3.1 Inventory of Exist ing Lighting An inventory of the existing lighting should be made with tbe results shown on maps and in tabular form. The inventory should include luminaire spacing, mounting height, overhang, offset distance of pole and type of luminairc. The data may oftc!l be available from the local utility \=(lmpany, contractual records, or tbe municipal official wi}o is r_esponsible for lighting maintenance. Informacion regarding sueet width, sidewalks, median areas, bordering uees and other vegeracion should also be recorded. Refer to Chapter 15 for more information regarding inventories. · When a.nalyz.ing lighting needs, it is helpful to use the inventory clara to prepare maps that show existing lighting condit:ioru;. In general, two maps are prepared. Qne indicates the major traffic routes and, by mearu; of appropria te coding, the average illuminance and/or luminance level of each street. The second map further indicates, by appropriate color coding. the levels found on residential streetS. In most municipalities, various neighborhoods are lighced to different levels. Computer programs for calculating roadway illuminance and luminance are readily available, although certain required clara, such as mounting height, luminaire overhang and pavement classification, may require field checks (IES Design Practice <::o~ttee, 1981). Field measurements of actual illuminance and lu minance levels are highly desirable in order to verifY calculated values. Measurement techniques are well established (IES, 1983, 1987). Municipal or utility circuit maps may provide basic data that can aid in making luminance and illwninance calculatioru;. The layout, design, construction and rnaintenmce of sueet lighting systems are complex and specialaed. Guidance is available from various sources (AASHTO, 2005; IES, 1983, 1987; ITE, 2009; US oar, 1978).

3.2 Roadway Grouping The American National Standard Practice for Roadway Lighting establishes recommended illuminance and luminance levels for tbe lighting of various types of roadways, pedestrian ways and bikeways in different areas (IESN'A, 2000). Classificatioru; and recommended illuminance levels are shown in Exhibit 19-l for roadways and in Exh.i~Jt 19-2 for ~ys and bikeways. These rabies may be used when designing new lighting systems and for ~uacing the adequacy of existing syscems. Road sumce classifications for use with the illuminance recommendations of .:Exhibit 19-3 are shown in Exhibit 19-4. Any areawide or multiple-route studies of existing lighting conditions should separate the roadways into functional classifications such as those shown in Exhibit 19-4 (lES, 1983).

Roadway lighting • 4_,5

~

- ·-_ •

. ~.. ~ ~--. ~ -.. ,·.::-: •

..

• --• • . .,._.,r ~ .. : ~

:f-:c• •

'"_._

-

~ --~·,

-

... •

• •. •

---'

-.;;.

--~~-

{a) Maintained Luminance Values (!_) (Canddas per Square Meter)'

Road aod.A.ma

Average Luminance,

L

Veiliog Luminance Ratio (muimum)

Luminance Uniformity to L

L. to L

Freeway class A

0.6

3.5 to 1

6 to 1

0.3:1

Freeway class B

0.4

3.5 to l

6 to 1

0.3:1

Commercial

1

3 to 1

5 to 1

Intermediate

0.8

3 tO 1

5 10 1

R=idential

0.6

3.5 ro 1

6 to I

Corrunercial

1.2

3 to l

5 to 1

Intermediate

0.9

3 to 1

5

R=idenrial

0.6

3.5 to 1

6 co I

5 to 1

Oasaification

L to L"""

L

Expressway 0.3:1

Major

10

1

0.3:1

Collector Commercial

0.8

3 co 1

Intermediate

0.6

3.5 co 1

6 to 1

Residential

0.4

4 to I

8 co 1

0.4:1

Loc:al Commercial

0.6

6 ro·I

I 0 ro 1

Intermediate

0.5

6 to l

10 tO 1

-R=idencial

0 .3

6 to 1

10 to 1

0.4:1

(b) Average Maintained lllumlnance Values (E.,Jb

Road and Aza.

Wwninaocc Uniformity Ratio,

R1

Pnanent Cla.ssi6c:ation R2 and R3

R4

E_ toE_..

Freeway class A

6

9

8

· 3:1

Freeway class B

4

6

5

Commercial

10

14

13

Intermediate

8

12

10

Residential

6

9

8

12

17

15

lntermedb.te

9

13

ll

R=idential

6

9

8

Commercial

8

12

10

lntcrmedb.tc

6

9

8

Residential

4

6

5

CLusi.6cation

Exp_,_,3:1

Major Commercial

3:1

Collector

•••

o •••,.•••

... ,. • ..,,..,,-I'U"\1"1.,.1TI"" I I ~·1,-ni~,-MIII,.. l"""""lf"\tr-r

~··"""'

t"f"'\r'Tf,....l

4:1

Notes: I. L. • veiling luminance 2. These tables do not apply to high mast interc:hangc lighting S)'ltcm.s (e.g., mounting heighu over 20 meters).

3. The relationship between individual and respective lwninance and illuminance values is derived from general conditions for dsy paving and straight road sections. This rclacionship does not apply to averages. 4. For divided highways, where the lighting on one roadway may differ from that on the other, calculations should be made on cac:h roadway independently.

5. For freeways, the recommended values apply to both mainline ;and ramp roadways. •For approximate values in candelas per square foot, multiply by 0.1. \For approximate values in foorcandles, multiply by 0.1. Source: Reprinted by permission from the IESNA Handhoolt, 9th Edition. By the 1lluminating Engineering Society of North America.

•Crosswalla rnversing roadways in the middle oflong blocks and at suect inreneaions should be provided with additional illumination. • For approximate values in footcandles, multiple by 0.1. 'For pedestrian identification at a distance. Values u 1.8 mea:rs (6 feer) above Source: Reprinted by pc.rmission from the IESNA Handbook, 9th Edition. By the lUwninacing Engineering Society of North America.

Roadway lighting • 417

Rl

0.10

Portland cement concrete road surface; asphalt road surface with a minimum of 15% of the artificial brightenec (e.g., Synopal) aggregates labradorite,

Mostly diffuse

R2

Mixed (diffuse and specular)

R3

Slightly specular

•o•• representative mean luminance coefficient. (llluminat.ing Engineering Society of Noeth America, 1983, RP-8-83, Chap tee 16, Table 16-3, page 7.) Sou=: Reprinted by permission from the IESNA HanJhqo!, 9th &JitiDn. By the Illuminating Engineering Society of North America.

ITel:ll'

C,LAJJ,tCAtiON

·•

'

·--- -

M.Ao~o•

•ovT«

COI.I. I Cto• • •.

t.OCAt. . , ,

LANO

Ust

~ ~

e::z ---, ~

~

co••••<•~>'

...........,

..NttCM I OIA.fl

. . EX·AMPLE OF: STREET CLASSIFICATION AND LAND USE

Source: Ammcm National Prtu:tice for &adway Lighting, illuminating Engineering Society of North America, 1983, RP-8-83, FJgUre 1, page 6.


3.3 Roadway, Pe destrian Walkway and Bikeway Classifications Roadway, p~destrian walkway and bikeway classifications are listed below (lESNA, 2000). These classificarions ,Uow for comparing various study locations or combining study locations for ovenll area performance levels.

Fruway: divided major roadway with full control of access and with no crossings at grade. This definition applies to coli and noncoll roads. Expressway: divided major ro~dway for through traffic with partial control of access and generalJy with inter· changes at major crossroads. Expressways for noncommercial traffic within parks and parklike areas are gene...UY known as parkways. Major: that part of the roadway system which serves as the principal nerwork for through-rra.ffic Bow. The roo res connect areas of principal uaflic generation and important rural highways entering the ciry; also known as arceri:Js. Colkctor: distributor and collector roadways servicing traffic becv.·een major and local roadways. Th ese are road· ways used mainly for traffic movements within residential, commercial and industrial areas.

Local: roadways used primarily for direcr access co residential, commercial, indUStrial, or other abutting pro~. LO· cal roadways do not carry through traffic. Long local roadways will generally be divided into shorter sections by c0l· lector roadway systems. Additionally. local streetS within residential areas should be grouped into rwo genero t)'J'es: 1.

those serving single- or rwo-f.unily homes; and

2. those serving aparrments or condominiums.

Alley: narrow public wayS within a b~ock, generally used for vehicular access co the rear of abutting p ropertieS· . Foorway: paved or otherwise improved areas for pedestrian use, located wirhin public sueec rights o f way, wW ch also conain roadways for vehicuJar traffic. Ptdmrian walkway: public walk for pedestrian traffic not necessarily within the right of way for a vehicularcr:J'lic roadway. Included are skywal.ks (pedestrian overpasses), subwalks (pedestrian runnels), walkways giving access 1:0 parks or block interiors and midblock street crossings. Bilttway: any road, street, path, or way that is specifically designated as being open co bicycle travel, regardless of whether such facilities are designed for the exclusive use of bicycles or are co be shared with other transponaciO.n modes. There are rwo general rypes of bikeways. 1.

Typt A~signaud bicyck laM: portion of roadway or spoulder that has been designated for use bJ bicyclists. It is distinguished &om the portion of the roadway for motor vehicle traffic by a paint Sl!ip e" curb, or other similar device.

2.

Typt B-bicyck trail· separate trail or path from which motor vehicles are prohibited and which is br the exclusive use of bicyclists or the shared use of bicyclists and pedestrians. Where such a trail or puP forms a part of a highway, iris separated from the roadways for motor vehicle traffic by an open sp~ee or barrier.

3.4 Area Classifications Typical area classifications are listed below (IESNA, 2000). These classifications allow for comparing varioussud:7' locations or combining study locations for overall area performance levds.

Commm:Uzl: business or industrial area of a municipality where ordinarily there are many pedestrians duringch~ night hours. This definition applies to densely devdoped business areas outside, as well as within, the centralp.r-.ot: of a munio;ipality. The area contains land use that attracts a relatively heavy volume of nighttime vehicular aci/. or pedestrian traffic on a frequent basis. A commercial area without night pedesttian activity may .l>e classificl ~ a special land use. Roadway light ing , ll ~

lntmnediizu: those areas of a municipality often characterized by moderate nighttime pedestrian activity such as in blocks having libraties, community recreation centers, large apartment buildin~, commercial buildin~, or neighborhood retail stores.

&sidmrial: residential development, or a mixture of residential and small commercial-establishments, characterized by few pedestrians at night. This definition includes areas with single-family homes, condominiums and/or small apartment buildings.

. Probkm at?as: certain land wes, such as industrial areas, office packs, commercial parks and public parksi may be located in any of the foregoing area classifications. The classification selected should be consistent with o:pected nightt.ime pedestrian, vehicular and.Qther rdared activity. Within any given municipality there should be consistent application of the definitions.

3.5 Collisions When undertaking collision analyses as parr of a roadway lighting srudy, four general data items arc needed for each collision that occurs during the study period. Refer to Chapter 17 for more information on traffic collision studies. These data items arc the location, severity. type and time of day (day or night). The location element includes the street name and whether it occurred at a specific interscaion or at midblock. In the case of &ccWay studies, the six desirable location categories include: 1. on mainline between interchanges 2. on mainline within interchanges

3. at ramp c:xit gore from mainline

4.

at racnp

entrance to mainline

5. in racnp proper 6. at ramp inrcrseaion with crossroad Collision severity classes include fatal, injury (overall or A, Band C type), or property damage only (PDO).ln the case of fatal or injury collisions, the number of collisions involving a fatality or an injury should be wed, not the number of persons killed or injured. Even when a cost analysis is being performed, it i.s preferable to assign average unit costs to collisions by class rather than by numbers of persons involved. Due to their low frcquericy of occurrc~:~ce, fatal and injury collisions are sometimes combined for analysis PUJPOSCS· The primary collision types that arc weful in lighting studies ace: l. vehicle/pedestrian and/or bicycle

2. ·vehicle/vehicle 3. vehicle/fixed object

\In spccialiud studies, other distinctions may be useful, such as sepuating vehicle collisions into two categories: !hose ~nvolving oilier moving vehicles and those involving parked vehicles. Fixed-obj~t collisions may. be categorized by !~e of obstacle struck such as light pole, sign post, guardrail, bridge pier and so on. Pedestrian collisions may also be of particular interest as a special study. The basic tabulation is by day or night, from the light condition indica ted by rhe investigating officer. This item is generally coded on collision summary reports and shown on collision diagrams. If not, it will be n~essary to determine light conditions directly from the fully detailed collision reports. Up to 5 percent of collisions can be expected to be noted on the collision report as occwring at dusk or dawn. These collisions should not be classified arbitrarily. One option is to omit tabulation of this group. The disadvantages include reduction of sample size and incompleteness in total collision data ror comparison with other srudies, crosschecking, or usc in nonlighting-related studies. If rhis group is included, there are two methods of placing the dusk or dawn collisions into a day or night category. The simpler merhod is to assume all dawn collisions occurred during darkness and all dusk collisions during daylight. The preferred method of grouping is to assume the time of occurrence is reasonably accurate as shown on the collision repon. The local sunrise or sunset time is determined from readily available tables (see, for example, Exhibit 19-5) and corrected for daylight saving time if it is in effect on the collision date (AsUQnomical Applications Depanment of the U .$. Naval Observatory, 2008). Ifthe collision occurrod w,ithin the period &om 15 minutes (min.) after sunset to 15 min. before sunrise, it is classed as a night occurren~e, and all collisions during the remaining period are daylight events. Collision records obtained from computer files may pose some problems. The printouts may code the time of collision occurrence only to the closest hour and some tabulations do not identify exact location, object scruck, or other pi~ of information that may be viral ro an accwate analysis of collisions as related to lighting. Pasr experience has shown errors of 20-60 percent when comparing automated data versus manual tabulation from original collision repon forms (Box, 1971). If there are serious doubts regarding the validity of the computer printout data, the original collision repon forms should ~.used. These are commonly available as electronic files and are accessible to the analyst for most studies. Suggestions for lo<:atin~and possibly correcting erroneous collision data are given in Chapte r 17.

a Add I hour for Day!ighr Saving Tune, if and when in use. Sour= Source: A.stronomic:al Applications Depamnenr of c:he U.S. Naval Observatory. (February 27, 2008). Sun or Moon Rise/Set Table for One Yeu. Retrieved December 4, 2008, from U.S. Naval Ol»ervatory: hnp:/laa.wno.oavy.miUdaa/docs/ RS_OneYear.php.

The effecr of lighting on roadway collisions is tenuow and easily masked by other variables. Controls in v:~riability can be achieved by careful selection of the analysis method, provided an adequate collision sample size is available. In an e:xisting<.nndltion study, it is generally desirable to ~ 3 to 5 years of daca. In a before-and-after srudy, 2 to 3 years' data are typically wed for each srudy period. Research should be conducted to determine ifchanges other than lighting levels have occurred in the area which may inHuence nighttime coUisions. To account for any time effects, ic is advisable to include a set of unimproved comparison sites, having essentially similar characteristics as the improved sites, in the after period. More 9-etails regarding the design of before-and-after experimentS are given in Appendix A. The decision criteria discussed in Exhibit 3 of Appendix A (Exhibit A-3) may be used to determine whether a change in the number of coUisions recorded during a before-and-after srudy is significant. These criteria assume collisions are Poisson-distributed and also recogoiu the before count is a random variable. To illustrate the use of Exhibit A-3, consider an wban roadway that received a lighting improvement. A total of 12 nighttime collisions were recorded 422 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

during the 2-year before period prior co che improvement, and three occurred during che 2-year after period following ·.the improvement. This represents a decrease of (12 - 3)/12 = 75 percent, which is greater than the figure value of 67 :percent for a 95 percent confidence level. Therefore. it may be concluded with 95 percent cc:rtaincy char che reduction 'in collisions was not due to chance alone. If rhe number of collisioo.s in the before period had been considerably larger, say 26 occurrences, che same conclusion would be reached if che number of collisions in the after period were rcdueed. to 13 or fewer, since che 26/13 = 50 percent change is greater than the figure value of 46 percent.

3.6 Traffic Volumes To understand the relative volume of traffic during daylight and nighttime conditions, uaffic volume counts shol.lld be considered as pan of the study. Night traffic volume data are generally gathered by use of automatic traffic counters. Refer to Chapter 4 for more informa.tion on volume studies. Counts should be selected so as co represent traffic during various seasons of the year. When averaged, January, April, July and Ocrober counts yidd a good yearly estimate in most cases, but can be skewed by high volumes of tourist traffic, special events, or other situations. Count data shotlid also be selected to represent different types of routes, such as:

1. industrial Streets, which may show very low night volumes; 2. primary business streets, which may show high night volumes on days of evening shopping and low night volumes on days of earl:y closing; and 3. boulevards or parkways, which may show high night volumes in summer and low volumes in winter. In general, other types of routes show little daily variation. Volume counts are tabulated for boch the night periodaild for the full24 hours. Example 19- L provides a typical computation of night rraffic volume percentage.

EXAMPLE 19-1 A typical computation of night traffic volume percentage at a single location is made as follows: . Sunset: 19:05 hows + 15 min. = 19:20 hows Sunrise: 05:48 hours - 15 min. = 05:33 hours Night traffic • (19:20 to 05:33) = 4,200 vehicles 24-hour count = 17,000 vehicles Night percentage .. 4,200 vehicles I 17,000 vehicles =25 per~t The night percentages are computed separately for each seasonal count at the location. These percentages are theO. averaged to provide the typical annual percentage of night traffic by type of route. This seep is repeated for each route:~ Night traffic percentage by type of route may be applied individually when conducting lighting studies. However, the: type of abutting land use typically changes from point to point along major traffic routes. Portions of a given rouce: may carry predominandy industrial-cype craflic; ocher portions m~y carry shopper craffic; and scill others, socialrecreational traffic. To simplify the collision studies, a cicywide major route night volume percentage should be used.A separate residential night volume percentage may easily be computed by averaging four seasonal counts for one: typical screet in each of several different neighborhoods. 1. Single-family-low cost

2. Single-family-medium cost 3. Singl~-farni1y-high cost 4. Multiple-family- low rent 5. Multiple-family-medium rent Roadway lighting • 423

Past experience indicates a consistent range of 23-27 percent of total vehicle miles are traveled at night and the use of an average value of2S percent for the night portion of the 24-hourvolume appears warranted (Box, 197l).Ifit is desired to determine the actual percentage of night traffic ar a given location, hourly tabulations of volume are needed for all36S days of the year and should include the volumes in both directions of travel. These hourly data are generally available only from a limited number of automatic recording stations along freeways or other high-volume facilities.

4.0 DATA REDUCTION AND ANALYSIS 4.1 Control for Normal Variation No two routes have identical traffic, geometric, or operational characteristics. Comparative studies of similar routes are valid only to the degree that similarity acrual.ly exists. Collision rates per mvm may change with volume, and therefore miles of travel do not truly represent exposure. Thus a simple compariso11 of night collision races among different routes, even of the same functional class, cannot be expected to yield meaningful results. The ideal method of comparison is to test a facility against itself, with the basic variable being day or night condition. This can be done in two ways: by calculation of the percentage of total rught collisions that occurred along the route or by calculation of the NID ratio of rates (night collision rate per mvm divided by day collision rate per mvm). At intersections, the percentage of total collisions occurring at night is the usual comparison; however, if the NID ratio is used, it should be based on millions of vehicles entering the intersection.

4.2 Use of Night Percentage The percentage of collisions at night (%N) is directly calculated, as shown below.

%N= lOOAlf AN+AD

Equation 19-1

where AN is the number of night collisions and AD is the number of day collisions. A condition ofN that is equal to the night traffic volume percentage would typically represent a collision rate at night equal to the day rate. A night collision percentage in excess of the night traffic percentage may be evidence of inadequate illumination. For example, a major study of lighted urban freeways found an average of 32 percent of collisions occurring at night, while the unlighted freeway average was 44 percent (Box, 1971). Similar findings luve been reached in a before-andafter study of major routes (Box, 1989). Based on such research, a rule of thumb is sometimes applied, with night proportions up to 30 or 35 percent representing reasonable safc:ty, while greater percentages may indicate a need for lighting improvemenL ·

4.3 Ratio of Rates When the actual vehicle miles of day and night travel are known, the NID rate ratio can be calculated directly by dividing the night collision rate by the day collision rate. The study of freeways found an average NID collision rate ratio of 1.4 for lighted routes and 2.4 for unlighted routes, while the major route study NID collision rate ntio was 1.3S during the before st\ldy and 0.87 in the after study. The ratio of rates can also be calculared directly if the percentage of night travel is known or estimated. In such a case, data on vehicle-miles of travel are not required. This method, which gready simplifies data gathering, uses the foUowing equation. ·

R AN(l-P)

Aoi'

Equation 19-2

iwhere:

;

iJ?

= NID ratio ofcollision rates

P

= percent of travd at night

AN

= number of night collisions

AD

= number of day collisions

For instance, if Pis raken as 25 percent, the equation simplifies to:

R=3AN

AD

Equation 19-3

4.4 Improvement Evaluation A cost-effective lighting installation or improvement program requires use of both hazard identification (such as high night percentage ofcollisions or high NID rate ratio) and the annual number of night collisions (usually averaged over the last 2 years). The added cost of providing lighting must also be considered. Exhibit 19-6 shows how cwo alternative lighting projecrs could be compared. The NID rate ratio can also be used to rank improvements, where the expected number of night collisions (E,) is give!l by the following formula, using previously defined variables.

E=RAtf' N .. 1- P

Equation 19-4

EXAMPLE 19-2

Exhibit 19-6 shows how cwo alternative lighting projects could be compared. Project A is evidendy more effective, although a simple ranking using night percentage would rank project B as first.

• Based on assumed local cxpcric=c for routeS provided with recommended illwnination lcvcl. b From assumed lighting design calculations of amortized capical cO-tt, plus annual maintenance and eoccgy·chacges, minus existing lighting cosu (if any).

The ENvalue is subuacted from the existing number of night coll isions, and the ;,nnual cost per reduced collision is calculated as in the preceding example. A separate calculation is made for each project, and tbe rankings then follow in simple order, starring with the lowest annual cost per reduced collision. Since pedestrian collisions arc especially susceptible to reduction by improved lighting, rhis class of collision can be singled out for separate analysis. Priorities can be established along major routes by directly calculating the number of collisions per mile and total number at each intersection. Sound engineering judgment must be used to prevent one or rwo unusual collisions from creating unrealistic programming. Priority schedules for residential lighting may be established by comparing crime percentages (night percentage of total daily) for the various areas or by using the total number of night crimes. The nacwe of the crimes may also be a consideration. After installation of lighting improvements, an evaluation can be conducted to determine rhe actual effectiveness of the improvements. The actual effectiveness of the lighting improvements can increase the reliability of funu:e cost COfllpariso~ and analysis, as shown in Exhibit 19-6.

5.0 REFERENCES 5.1 Literature References American Association of Srate Highway and Transportation Officials. Roadway Lighting Drsign GuitU. Washington, DC: AASHTO, 2005. Asaonomical Applications Dep=mem of che U.S. Naval Observatory. "Sun or Moon Rise/Set Table for One Year." February,27, 2008. Washington, DC: U.S. Naval Observatory. www.usno.navy.mil/USNO/asaonomical-applications/data-services/rs-one·year·us. Retrieved December 4, 2008. Box, P. C. "Rdationship Between Illumination and Freeway Accidents: llluminaring Enginuring, May-June, 1971. Box, P. C. "Comparison of Accidents and lUumination." HighwiZJ &uarth Rmrd 416 (1972~): I 0. Box, P. C. "Freeway AccidentS and Illumination." HighwiZJ &uarch &cord416 (1972b). Box, P. C. "Major Road Accident Reduction by lUumination." Transportation Rnta.rch Rtcord:journAl oftht Trttnsportalion ReJtarch Board 1247 (1989). lES Design Practice Committee. "Available Lighting' Computer Programs: A Compendium and a Survey.• LD+A. (March 1981): 35. Illuminating Engineering Society of North America. Amm(ll!l National Standard Pracriu for Roadway Lighting. ANSI/IES RP8-1983. New York: Illuminating Engineering Society of North America, 1983. Uluminating Engineering Sociery of North America. "Section 14, Roadway Lighting." lES Lighting Handbook, 1987 Appli~ation VqJu~m. New York lllwninacing Engineering Society of North Amc:rica, 1987. illuminating Engineering Society of North America. Tht /ESNA Lighting Handbook: Rtftrma & Applicarum. New York: Illuminati.ng Engineering Society of Nonh America. 2000. Instituce ofTransponatio_n Engineers. Traffic Enginm-ing Handbook. 6th ed. Washington, DC: ITE, 2009. National Lighting Product Information Program. LightingAnswm. Troy, NY: National Lighting Product Informacion Program, Lighting Research Center, Rensselaer Polyrecbn.ic Lnsititute, 2007. Sherm;,n, L W., eta!. Prwmting Crimt: Wha: WOrkr. What Doem'~ Whals Promising. A Report to the United Stam Congrw. Washington, DC: U.S. Deparancnt of Jwtice, Office ofJwtice Programs, National Institute of Justice R.cscarch, 1998. 1ie~, J.. M.

"Lighting's Impact on Crime.• LD+A. (December 1979): 20.

U.S. Department ofTransponarion. RoadwiZJ Lighting H4ndboole. lmplemenation Package 78-15. Washington, DC: Federal Highway Admlnistttci.on, 1978. Walk.:r, F. W. and S. E. Robera. •Evaluation of Lighting On Accident Frequency at Highway Intersections." Traruportation &starch &cord: journal ofthe Trrrn1portation Rntarrh Board 562 (1976). 426 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

; 5.2. Online Resources (Available as of January 5, 2010) . Acuity Brands Lighting. VisuaL Acuity Brands: "'ww.visual-3d.com/Software/Overview/Defaulc.aspx?SessionlD =O. Rccric,.,d . December 9, 2008. Ascronomical Applications Departmenc of the U.S. Naval Observatory. "Sun or Moon Rise/Sec Table for One Yea r." Febru>ry 27, 2008. Washington, DC: U.S. Naval Observatory. www.usno.navy.miUUSNO/asrronomical-applications/data-services/Jlone-year-us. Retrieved December 4, 2008. Coop<:r Crouse-Hinds. Lwcicon. Cooper Crouse-Hinds: www.crouse-hinds.com/CrouseHinds/resources/luxicon2 .cfm. Retrieved December 9. 2008.

DIAL DV\Lux. DIAL: \vww.d.ial.de/CMS/English/A.rcicles/DIALux/Download!Download_d_e_fr_it_es_cn.hcm l. Retrievc4 June 10, 2009. luxAn. Micro/ux Light. LuxA.rt: www.luxarc.com/ProdMiu.xLighc.htm. Retrieved December 9, 2008.

5.3 Other Resources Inscirute of justice. College Park, Maryland: Universiry of Maryland at CoUege Park, Department of Criminology and Crimioal Justice. www.ncjrs.org/workslindex.htm. Texas Deparrrnent ofTrarupomtion. Highway Illumination ManuaL Austin, TX: Texas Department ofTransportation, 2004. Transportation Research Board. NCHRP 05-19: Guidelitm for Roadway Lighting Based on Saftty Ben<}its and Costs. Washington, DC: TRB, 2008. Transportation Resmch Board NCHRP Synthesi~ '71: MaMging Selmed TrtWpmafiqn Ane~; Sign4/i, Lighting, Signs, Pavemmt Markingr. Culverts, and Sidewalks. Washington, DC: TRB, 2007.

Roadway Lighting • 42 -:J

Chapter 20

Transportation Planning Data Origiruzl by: Dtmna C Nelson, Ph.D., P.E.

EJiudby: Rohert S. Foyk, P.E. 1.0 INTRODUCTION

2.0

430

TYPES OF STUDIES

430

2.1 Defining Study Areas

431

2.2 Inventories

432

2.3 Origin-Destination Surveys

437

2.4 Swvey Participant lf!centives

438

3.0 DATA COLLECTION PROCEDURES ·

438

3.1 External Surveys

438

3.2 Internal Studies

446


5.0

429

1.1 Objective of This Chapter

448

4.1 Presenting 0-D Data Results

448

4.2 Checking Survey Accuracy

448

4.3 Additional Sources of Data

450

REFERENCES

450

1.0 INTRODUCTION ransportation planning is~ complex process that involves the evaluation and selection of highway or transit hcilities to serve present and future travel demand. The decision-making process involves state DOTs, mettopolitan plwrung organizations (MPOs), CUJ'al planning organizations, larger municipalities and sometimes regional effota as wdl (Mc:yer and Miller, 200 1). Studies ue conducted to gather information for transportation madding dfons or to evaluate the potential impactS of specific programs and projects. In almost all cases, softwue programs ue wed c:xtensivdy in the model,ing and evaluation process. Tcavd demand forecasting models, in particular, require extenSive dam for creating mulrimod;U tranSportation netwooo that ue interconnected with land use and socioeconomic ~ta foruscrs.

T

~ta needed for these models and studies usually include inform~tion about population and rr~vd characteristics and tranSportation facilities. The ~ta ue often organized and analyt.ed to:

The

economic utiviry,

1. identify the scale of present system inadequacies; 2. provide basis to foreost future land use and rravd; 3. derive population, land use and u~vd relationships; and 4. calibrate travd demand models. T .... ...,,..,..._,... .,..;~ ..., Dl"nni,.,n

n .,.t"i\

A

4?Q

The data needed for a planning study vary according co the scope and nature of the project, past planning activities, cost of data collection and so on. Generally, rhe transportation planning process requires that data from a wide variety of sources be complied, organized and analrzed. Data collection efforts can include most of the major operational surveys described in this book, including volume, speed, travd time, transit, goods movement studies and invemories ofland use, traffic control, geometries and other characteristics. Refer to those chapters for derailed discussions of the design and implementation of these studies. ln addition, Appendix B describes the process of designing, testing and implementing surveys. Derailed explanations of transportation modeling processes are beyond the scope of this book; refer ro the lTE Tranrportn.tion Planning Handbook for further information on this topic (ITE, 2009).

1.1 Objective of this Chapter This chapter focuses on the data needed ro support decision-making for planning rransportat.ion alternatives. Of specific importance are the data needed for a standard four-step trip-based travel demand forecasting model, the most common type of travel demand model in current practice. This type of model is a critical tool for the practitioner in both forecasting future travel demand and in resting multimodal alternatives to support an increased demand for travel both within and through a srudy area. There are MPOs across the United States and agencies throughout the world that are developing 11ewer models un· der the general descriptor of tour- or activity-based models. These models address some of the shortcomings of the standard four-step trip-based models. Some agencies are also working on ocher specific category models, li.kc freight models and special generator models. All of these alternative models require additional details for each household individual along with trip behavior for proper calibration and validation of these models. Additionally, simulationbased travel demand forecasting and analysis packages are becoming available for u$e by transportation profesSionals. Descriptions of tour- and activity-based models and their data needs, along with microsimulation models, can be found in recem publications (ITE, 2009, TRB, 2008, 2007). ·

2.0 TYPES OF STUDIES Planning srudies may be categorized into three general time horizons: short-, medium- and long-term (Stopher and Meyburg, 1975). Short-lmn or projmplanning activities focus on sdecr projects that ean be implemented within a 1- to 3-year period. The short-term planning process includes systems monitoring (collection of performance data), identification of system deficiencies, formulation of process goals and objectives, identification of strategies, development of performance criteria, evaluation of strategies and ranking of strategies using a cost-effectiveness approach. The end product of the process is often a list of recommended projectS or programs designed to provide better management of existing ncilities by making them more efficient (ASCE, 1986). Examples of short-term planning projecr.s include: • planning for selected groups, (for example elderly or persons with disabilities); • goods movement srudies; • land development projecr.s; • bikeway planning; • energy contingency planning; and • impact srudies (for example, traffic. safety). The rransportation plan for each metropolitan uea muse include a rransportation systems management element that covers the short-term transportation needs of the urbariized uea. M~dium-tmn plans usually focus on efforts such as transportation systems management (TSM) srudies, air quality

management planning and corridor srudies. ProjectS commonly categorized as medium-term extend beyond the time span and narrow focus of short-term plans but are more special.ized and immediate than the long-term planning efforrs described below. TSM d ements identify traffic engineering, public rransportation, rcgularory, pricing. m~e ment and operational-type improvements to the transportation system. 430 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Long-range plans project transportacion needs of an area and identifY the projectS to be constructed o ver a 20-year (or 25-year} period to meet these needs. Long-range transportation plans (LRTP) rend to be capital imensivc with relatively long implementation cimes. Although the problem-solving process is similar for the long- and shon-cerm · planning processes, the LRTP planning process requires more sophisticated methods for forecasting future transportation demand than are required for shore-term plaMing projects.

2.1 Defining Study Areas Most planning studies begin with delineation of the survey or study area. The exterior limit of the study area is called the cordon line. For a comprehensive urban survey, this srudy area may include the emire urbanized area p lus a portion of the outer fringes where future growth is expected. For local area studies, the cordon area is established ro encompass the area of interest and ro minimize the data colleccion effort. Internal zones can also be defined to permit daca co be summarized for a reasonably small area and add more detail co the informacion on crip movements within lhe scudy area.

2.1.1 Establishing the Co~n The cordon is, in effect, the boundary around the study area for which trip movements arc requited. T he purpose of the study suggests the extent of the survey area. The survey objectives and the conscraintS on the study should be considered in establishing the exact cordon line. To the extent praccical, the cordon line should follow physical boundaries limiting movement, such as hills, rivers, fre~ys and rail lines. Chapter 4 provides more derails on defining rhe cordon and on performing cordon counts. 2.1.2 Establishing Trajfic.ANzlysis Zones Once the outer boundaries ot.lim.its of the study are set, the study area can be
Transportation Planning Data • 4~ -1

Source: Box and Oppcnlander, 1976.

2.2 Inventories Once the scope and scale of the planning effort have been determined and the boundaries of the srudy area established, the usual next nep is to assemble the data required. The information most commonly needed for planning purposes includes information on land use and popularion, the chanaetisrics of the transport:trion system and information on travel demand. Inventories, surveys and studies are made to determine traffic volumes, land uses, origins and destinations of aavdetS, population, employment and economic activity. Collecting completely new data for a study may not be necessary. Before 6dd collection ofdata begins, existing sources of data should be identi1ied and the data aamined to see if they are useful for the study in question. The information gathered is summarized for each traffic zone and for the aisting highway and tnnsit systetn. 2.2.11Arul iJ,e mul. PopulAtion Data Land use inventories' provide important data on land characteristics and activities including current land use and vacant land. ln general, land use data are often colkcted to provide:

• a basis to derive trip generation fActors and trip generation forecasts; data for coordinating tnnsportation facilities with other uses; • a ~universe" of dwelling units from which a sample can be drawn for the home interview plwe of the crave! survey; and • data useful for the day-to-day planning activities of various agencies. ..... • " " ""'II'

n~TIIAM<;~:>f'IFITAT1nM

I'N(;INFFRIN\, <;TllniF.S. 2NO

EDITION

\A wide range of information on land use is useful in the tr.10Sportation planning process. Thi.s includes: I

i

• development trends over time; • topography and physical constraints; • current land use; • vacant land by ownership; • location of major travd generators; • identification of community neighborhood boundaries; • existing land use controls, including zoning, official maps and subdivision regulations; and • identification of redevdopment areas. Many planning agencies maintain current inventories of land W(S, vacant land and economic activities. The: dc:v.elopment of geographic information systems has really revolutionized the archiva.l and reuieval of land use data. The land use component of comprehensive plans typically contains information on zoning, activity systems, land values, environmenal conditions and aesthetic fea.rurc:s of the local area.. The Urban Transportation Planning Package of the U.S. Census also contains information useful in est2blishing relationships between land use and uips. Population disuibution, density, average income, type of dwdling and car ownership data are used for traVel models. Historic patterns of distribution, migration, density and growth trends combined with present conditions are used to prepare population forecasts. for simple studies or projects, data and relationships drawn from previous studies may be adopted {Homburger et a.J.; 1985). Information on population projections may be available from the agencies responsible for planning and economic devdopment. For larger projects, current, more detailed data may be required. Surveys may be conducted to check the accuracy and validity of available data or collect new data.. As part of a. .major planning study, the ua.nsportation professional will need to work with various committees providing policy, technical and citizen input inro the process. These committees will have distinct approval and advisory roles that will influence the final modd depicting how growth is c:xpected to take place within the study area.

2.2.2 Collecting LArul Cke D~t14

Several types of land use surveys can be conducted to provide new data, or to update and supplement available information. The decision a.s to the type of survey must be based on its purpose(s) and the resources available. Land use data may be recorded in the fidd using several different techniques. One of the simplest methods is to recor~ the d:~ta di.reetly on maps or aerial photognphs. This •m4p-r«C ofland use inventory may be appropriate in small towns oi: cities where the only da.ca required are simple use cla.s:&ilications. Typical black line prints (or plots) a.t a. scale of 1 inch (in.) =400 feet (ft.) or larger may be used. Complementary maps and land data are available dectronically and are used co check cxi.sti.ng data and maps in helping to identify gaps or errors for update effons. A second technique, fold li.rting, is the most common method for transportation planning. Traditionally, this method has been carried out using a fidd form similar to that shown in Exhibit 20..2.

Transportation Planning Data • 433

0•-o,.~....:don

S

1I

O•ogr..,hlolo••tlon X

Y

•

$

I I

"'-;r'"~ w-

";:::'

ITT TTl chi\

I I

,.,....,.

Na-=;4:!' ....

I ll ,.,..UI •

rn

~~

_y_____

-~~-

,.loor ':"::!,

Pl•Vi'...;:::=:,::.,u~~-

II IIIIIIIIIIIII ....._

~••••-•"'"

,._,., ...!:_i=::!.'='=~::..

"'•~ct•AC•tvt141

......... ~=.:.~-

._••.._.t\'1"

"-oall.-='~~:.~.._

::::-

F~r

:,.liM,

A~

$

flro" • Wl4hh HMOf'~GIAO N-::.~-tl•t_.

"::" .

· ~ IT r I r I OJ Wl4ft'l

~ ::: ~-;.;:.:.

,.~

lcrth

H~Ot'b~ltlg • Non::,.~.WeAll•~-·

-~~·

,._w-e ...,.

iJj 0 ~: ::: ~;;:..-;, I I I I I I I I I I I I I I I

..y.....w.,J....L,~-1--+..LJ..,..w_..L+.~I I I I I ...

n._, ;::=,

OJ OJ 0

TTl [0

S

Blodt.

G . .DO'..,hlc:''"-tlol'l

....!:!::~•no

I I I I I

8 14do

O•ogr•...,oto..-t!OI'I

~:~no

I I I I I

••_..

f I

iJ5

0::::: ~~-:. I I I I I I I I I I I I I I I

CD [TI 0 ~::: ~;::. I I I I I I I I I I I I I I I

A sepame form is usually reserved for each block, with a line for each pucd or land we and dwelling unk Observations may be made &om the vehicle, or on foot if parking or traffic congestion presents a problem. Data from the forms arc input directly imo a computer database. Surveys of existing land use can be classified as "inspection" or "interview• surveys, depending on whether the dwellings and other places must be entered. Inspection surveys (often called "windshield surveys") are accomplished without entering the building. Combined inspcction~intcrvicw studies arc needed when exterior inspection docs not yield enough informacion. Windshield surveys arc frequendy sufficient for transportation and traffic needs. The field listing technique may be applied easily wing a laptop computer and either database or spreadsheet ~oftwuc (depending upon the size and complexity of the :~.pplicacion). Data collected in the field can, if necessary, be ported to another computer system in the office. [f the land we informacion is co be incegl7.ted with a number of other datab= and will be retained and rewed, :1. geographical information system (GIS) may be appropriate. GISs integrate the map {or geographically oriented system) with the database, providing a very powerful tool for cracking and analyzing spatially oriented information. Land use dat:~. arc traditionally presented on a map showing the land we by general ategory of we (residential, commercial, industrial, institutional, parking and recreation, transportation, utiliries, agriculture and water). Each ategory is given a different color tO provide visual differentiation. Statistical summuics may be prepared to show the cotalland area devoted to each category of use, and may be broken down into subareas, or traffic zones. Land usc data collected for one study are valuable for later studies. Inconsistencies in categories of land we activities and dat:~. formats may make it difficult to compare results with subsequent studies. To minimize these problems in the United Scates, a standard system for identifying wd coding land we activities was devdoped (Urban Renewal Administration and Bure:~.u of Public Roads, 1965). The coding system provides four levels of detail on land use activity. Each Levd is subdivided into one-, two-, three- and four-digit categories. ·

2.2.3 Tr~ SJ1mms D4111 Transportation facility inventories provide the basis for establishing the necwocks that will be studied to decermirie presenr and IUcurc traflic flows. Data cw be otg2Jlized around travel facilities; parking; and ttip and travel data. [n many cases, some dat:l will be available from existing records of city, county, or scate offices. Because d:1.ta coUeccion can be apensive and some dara may be more essential chan others, d:1.ta needs must be evaluued cardUlly before wy study is undertaken. 2.2.3.1 Traw/ Facilities Inventory Travel facilities inventory consists oflocating and describing each link of the transporCltion system. The description of each link includes a measure of the present capacity and the current we of the link, and a statement of its performance ch:1.17.Cterisrics. This inventory includes: • classification of streets and highways; • geometric characteristics such as link length, pavement width and number of lanes; • traffic controls such as signili, green time, puking controls w d speed limits.; • free· flow travel times over nctwerk links routes; and 434 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

• existing capacity and levels of service. 'Classification of all appropriate streeu and highways in the nerwork encompassed by the study area should be i:ompleted in the early stages of the study. Jurisdictional classifications identify each participant and their respective responsibilities in regard to each segmem of che overall transportation system. The functional classification of facilities is determined by the relative importance of the mobility and access functions assigned to the m, as shown in Exhibit 20-3.

In the hierarchy of highway facilities shown in the c:xarople in Exhibit 20-4, freeways, expressways and major arterials constitute the major (principal arteria!) highway system, while collector and local streeu compose the local street system. The number ofjurisdictions invoJ..:ed in a study area varies depending on the siz.e and complexity ofthe area under srudY· Functional and jurisdictional classificatio~ are done in conjunction with the geometric inventory of che streets.

Transportation Planni ng Data • 43 .!i

.,.

-~-.:~~·

~~

'""~~:--

STREET CLASSIFICATION AND . IIHERSfCTION TRAFFIC · CONTROL DEVICES

~'·

t..ICIND

et••o cu.aa•••c••,••

~-

-

-. -

......."".

•nueu*'

....._,..c

.. - .,.....•• - ·· - · - ;!;:~ ;: ~.. ';';" •• 0 ·-·-··..

...-

._ .....

-- ~---- ~ ·il


Facility characteristics m3.y be obWr!cd from existing m.ffic conuol device (TCD) wd geometries inventories. Speed wd tnvel time ma.y also be needed for selected ro3.dw3.y wd tr.lllSit links. These d3.ta ue collected 3.5 described in Chapters 5 and 9. Existing opacity and kvds ofservice can be determined using the methods described in the Highway Capacity. Manwzl (HCM) (TRB, 2000). The d3.ta required for signaliz.ed intcncctions include basic geometries, signal timing informlu.ion, existing volumes, 3.re3. type and driver populacion (commuter or oilier).

2.2.4 Pmlting DtZta Puking inventories cOnsist pcim3.rily of informacion pertaining to the loacion, apa.cicy, time limits wd other chara.cteri.stics of existing parking sp3.ccs both along the curb wd in off-srreet including those in alleys w
=·

• number of p:uking spa.ces;

• time limits and hours of operation; • ownership (for example, public, private, restricted); rates charged (if any) and method of collection; • cutb space regulations; and •

type of facility (street, lot, or garage parking).

2.2.5 Trip mul TrAvel DAtA Trip and travel data are obtained from 0-D studies. These studies are designed to gather data on number and type of uips in an area, including movements of vehicles (including trucks) and passengers, from various zones of origin to various zones of destination. Informacion often collected includes the origin and destination of trips, the purpose, travel time and length of trip, the mode of travel and the land use at the points of origin and destination. The data are analyzed to define travel behavior and travel patterns within the area by time of day, mode of travel and purpose of trip. Depending on study objectives, travel behavior may be studied for an average weekrl2y or weekend, as well as for peak season travel (for reson areas). Travel demand for i:he base year is projeaed into the future to determine whether the current transportation infrastructure is adequate to meet future demands, to define ihe need for new facilities or improvements to the present system and to evaluate different growth rates and patterns. 0 -D data are used to plan and program major street systems, street improvements, new street locations, freeway location and design, interchange location, public transit networks and coverage and terminal facilities (bus, truck and off-street parking).

2.3 Origin-Destination Surveys Comprehensive 0 -D surveys ~e generally the basis for preparation of overall comprehensive transportation plans for an area. Because comprehensive plans are long range and slow in implementation, and because transportation facilities must be built for many years of usefulness, the 0-D data must generally be projected to provide data on future transportation demands. The scale of 0-D studies varies widely. Large transportation planning studies may conduct home interviews (homebased survey) to establish patterns for all uips made during a typical day throughout a large area. Large-scale planning studies may consist of a combination of several complementary surveys. Complete studies are very expensive and time-consuming and are rarely conducted. More frequendy, limited 0 -D studies may be conducted to supplement and update existing 0-D data. Major 0 -D data are projected to a planning horizon or design year (usually, 15 to 25 years in the future) based on anticipated future economic and population growth, vehicle ownership and usage, transit availability and patronage, land use changes and other factors. The scope of survey questions may be narrowed to a single trip or encompass all uips made during a 24-hour period. Small-scale studies may focus on a specific rteigli~ borhood, an indtvidual trip generator (or group of generators), or;limited sections of roadway. Studies may also be performed for limited sections of freeway to determine weaving and merging patterns, or to develop alternate routes (Roess, P.wsas and McShane, 2004). Depending on the purpose of the 0-D survey, studies focus on the basic "when• (rime of day), "where• (origin and destination), "how (modal choice} and •why (trip purpose) people travel. Tune-of-day data are used to establish potential peak periods and to estimate travel demand throughout the day. Trip purpose is used to establish trip patterns by trip purpose. Categories of data needed vary among srudies; however, commonly selected categories include home-co-work, homt-to-shopping, home-to-business, home-to-social/recreational and other trip purposes. The trip origin is the beginning point of the trip unless that trip begins or ends at home. The origin of any home-based trip (for instance, home-based work trip) is classified as "home• regardless of the direction of the trip. Trip types are also classified in their relationship to the study areas (for example, whether the origin and/or destination are within the study area). 1. Extemal-c:nernal or through uips are those uips with neither origin nor destination in the study area. 'f.nvelers make no stops within the study area.

2. External trips, funher classified as imernal·external or external-imernal trips, have either their origin or destination outside rhe study area.

3. lmernal-incernal rrips have both their origin and destination within the study area.

2.4 Survey Participant Incentives Incentives provided to participants can increase the response rate of a survey. Examples of incentives include a free week's Merro card, a savings bond, cash (perhaps as high as $100), or a gift such as an MP3 player. For large participant sample sizes, incentive programs may end up costing several thousands of dolla.rs. Cost out the incentive program and weigh the pros and cons prior to iQlplementarion.


3.1 EXTERNAL SURVEYS Cordon surve~ may be used to study external travel (that is, trips that either pass through the wne or have one trip end outside the zone). The type of external survey to be conducted depends on the information to be collected as well as the size of the study area. Common types of studies include: o

roadside interviews

o

license plate surveys

o

postcard/mail-back surv~

o

vehicle intercept method

o

rag-on-vehicle method

o

lights-on study

The study area may be a central business district (CBD) or other major activity center. Dara collected usually pertain to auto and truck travel, although there is an increasing need for pedestrian and bicycle trips as well. Separate surveys of rail, bus, raxi and air travel are made as desired to obtain additional travel information. Practitioners can have success with conducting surv~ over the Internet. In some contextS, an electronic survey actually may have a higher response rate than whar could be expected through tnditional tecliniques. ·

3.1.1 Esuzb/Ubing CoriUms Cordons should follow the boundaries of the study area, for example, an entire urbanized area, a transportation study area, a ciry, a CBD, a neighborhood, or an indusrrial area (ITE, 2009). Adjustments are made to the line to minimize the number of road~ crossed and to cross roadwa~ at midblock locations (to minimize the problem of vehicles at interseCtions). Stations are established at the cordon line on all intercepted streetS; however, stations may not need to be placed on local streets known to carry negligible traffic volumes. Cordons must be large enough to define the area of interest, yet small enough to define an area useful for planning purposes. Cordons may define areas of similar land use.

3.1.2 RoiU/sUU Intnviews In comprehensive studies conducted for a large area or a metropolitan area, the roadside interview may be used for obtaining external travel information. The method is an integral part of the comprehensive 0 -D survey, with the interview stations beirig located along the external cordon boundary. This method has the advantage of permitting the observer to ask the motorist the purpose of the trip as well as the destination and origin. Interview srations are established at all major roads and most other roads crossing the cordon line encompassing the study area. Stations are located to intercept at least 95 percent of the crossing tnffic and to minimize safety and congestion impacts. If the survey is concerned only with trip data on a single isolated route, driver interviews taken at a single midpoint location might suffice. If data are desired on all traffic entering and leaving a small ciry; it is necessary to select interview locations on all routes radiating from the city. Check with appropriate authorities on acceptability of using this technique. 438 • MANUAL OF TRANSPORTATION ENGINEERI"!G STUDIES, 2ND EDITION

TYPICAL

LAYOUT OF INTERVIEW . STATION

METROPOLITAN

~

i I

AR.EA TRAFFIC SURYEY

~

0

1NT£AV.~WI!ft6

&

Reo f..IANTE.RfitS f'"LARit RCO Ft...AG

e ..

i-


Exhibit 20-5 illustrates a typical layout of a roadside interview station for a moderate-volume, two-lane, two-way road. The location should be fairly level and have a sight distance of over 800ft. (244 m). Where only part of the driv~ can be interviewed, one or more bypass lanes are needed co avoid congestion. A paved shoulder of a rural route or a section of curb lane (cleared of parked CaJ's).may ~e used for the interview, leaving the regular traffic lanes for use as a brpass. M'ost stations are operated for 16 hours a day (6:00a.m to 10:00 p.m.), Some stations are operated 24 hours on major routes. Additionally, 24-hour volume counts must be taken concurrently with roadside interviews. Data from the z4hour count are used to expand survey data and to account for the sample of vehicles interviewed. Since freeways :iJ-Ce normally too congested to permit stoppages for interviews, other techniques must be used to obtain travel inforro.a tion. Several alternatives are discussed bdow. Stopping drivers usually requires the assistance of a uniformed police officer for traffic conuol, slows traffic considerably and may antagoniz..e the public unless skillfully handled. Generally, a large ponable sign explaining the projeCt prepares the motorist for the dday and often enlists cooperation in answering the questions more readily. (Check wi -ch appropriate authorities on signs necessary to meet traffic control laws.) It can be useful to involve the news media in explaining the need for the study. Every effort should be made to avoid congestion, not only for safecy and for mai.Ptaining good public rdations, but also because congestion may cause local drive..S to detour around the interview sr::ation and thus distort the traffic flow patterns. Not every vehicle can'be stopped on a high-volume route; usually, ofl.lY a sample of drivers is interviewed. The required sample size iS determined using the methods described in Appendix :13. Samples usually consist of20-50 percent of the traffic. For a 50 percent sample, it is satisfactocy to interviewdri~rs of three vehicles and pass the next three vehicles. A typical crew at a location carrying 3,000 to 5,000 vehicles per day in both directions includes a survey chief, [\IV'O recorders, six interviewers and one or two uniformed police officers (and other individuals as required by local regui:;ations). The number Qf inrerviewers should equal the number of vehicles stopped in each group. Recorders count a.r:t-d record all traffic (by type and direction). A police officer may be used to hand each driver a card explaining the purpo~e of the survey as each car is stopped. Exhibit 20-6 shows a form for recording 0-D informacion. Column 5 of tb-JS form may be used to record the purpose of the trip. The form should be modified to meet individual study needs. TPe sample data are then expanded to represent the total traffic volume and to obtain total estimated origins and desrin~ tions. This method has several advantages and disadvantages. The most complete and accurate informacion is usual.l.Y obtained when personal contact is made between respondent and interviewer. The response rate is greater (rdative co the voluntary return technique), thereby minimizing the survey bias, and samples can easily be chosen from a craff.i.c stream to satisfy planned statistical standards. The roadside interview technique is more expensive than several other techniques b~use more personnel are required. On high-volume facilities, there may be some traffic delays durir.Jt..g the survey, especially during peak travel periods. This technique is often dangerous, especially on higl)-volwne Facil..i.ties, because survey personnel must operate on the highway and interfere with the regular flow of traffic. Transportation Planning Data • 43 ~

Orlgln- Oe•tJn•tJon Sludy Fl•ld Sheet \.OCMion

8laeiOn ...,..,...

Time: e.o~n:

~

and;

~

T

""""---:.

OOOiiiiOiJOO

---

AOUiiOMd

4

l5iiiOi"

~--------------------------


A vmation of the roadside interview is often used on comprehensive downtown parking studies to interview drivers who have just pa.rlced their cars. An interviewer is responsible for interviewing a sample of the parkc;n in a certain &cilicy. Origin, destination, loation parla:d, time of day, purpose of trip and parking duration information are most often obtained. Details on this procedure are given in Ch2pter 16.

3.1.3 Pomani Struliu Returnable: postcards can be handed to drivers at the intercept stations where traffic is heavy and it is not possible to dday vdlldes long enough to complete an interview. A survey may rdy entirdy on postcards; however, cards are often used in conjunction with interview studies. The poscpaid, addressed cards are precodcd with station idenci.6cation and time. Drivers are asked to list the origin and destination of the trip and drop the card in any mailbox. The questions on the postcard should be simple. Normally, cards contain five to seven questions. Exhibit 20-7 shows a postc:ud questionna4re. As in roadside interView studies, 24-hour volume counts (by direction) should be takt:n concurrendy. Data from the 24-hour count are used to expand survey data and ro account for the sample of vehicles interviewed. A 20-40 percent return i, common from this cype of survey (ITE, 2009). A 30 percent return is considered acctient; a n:rurn of at least 20 percent is needed to maintain the accuracy of this type of study.

~· ~\.M7 &.• . ,. .• - · • ll7 '-•• Cit...J'

er caa.••••· e--c., et

rtf.ASf ANsw•a OUfSfiOHS .AHO 0•011 IH MAlt OAJID

c:•• ....._.... •'

•ox · ...

J11.S..a• t.•

NO SfAA4111' •tUllllt6D

aao.

c ........ ft4 &a....... _ , V••• ur ..~, o. ~•· 'l'rt.p .. I r •-• • Pl•••• -=-••il: ..... •

J"lM-•

~t..r••'

w•..,.••

.... ,...,. a · T•

a-...

0

;: ::::... 8

Clt.y

... ..•-•r• ,........... z...ile

:.

Dr£.••

~

co. ................~ :tor\. . . . , hp•• . . _.l' 0Al'M'•t. -:r C:b.i.cac• &a.l'..-.:r

·-·

b•••..

8Cl

8

g

· .,,.., ••~•• ~.11 r .. ••• r • r

.,....

~··•

,...... o r r ....

~•T

a:...•-.;1••• .......

-r••ak

t ... •••r ""'•" ~ or

1'•- C:••.-... u.. ...

~ - --~'";,- ;; -.-;;.; ;;----

Sow= Box and Opptttbndcr, 1976.

The postcard questionnaire technique is relatively inexpensive. Traffic delay is less than for direct interview, and untrained personnel on be used for handing out cards. A major disadvantage is possible sample bw due co better cooperation by some drivers. Care is required in location of disuibution points to intercept a representative cross-section of uips. Through uucks and passenger vehicles will not provide a high percentage rerum. It may be difficult co include all important vehicle movements, especially in largt cities, and, like the interview method, it requires stopping traffic.

3.1.4 Lknue Plate Strulks The basic concept is simple. As a vehicle passes each station, a portion (or all) of the licen.se number is recorded, which permits vehicles co be tracko:l through the study area. Stations may be established on the boundaries of the study area and also at intermediate locations within the :uea. Plate numbers and the time are recorded for shon periods of rime (such as 1 min.). For the purpose of this study, the origin is the place where the vehicle is jim observed, and the . destination is where it was last observed. · Observation points must be located carefully. If only informacion on through traffic is being collected, at least rwo stations are needed: one representing the origin (for one direction) and the other the destination. To uace vehicle movementS through an area, every reasonably heavy bypass point must be surveyed. In addition, intermediate Stations may be used to establish routes or to give more: detailed information on origins and destinations. The reduction of data requires considetable labor co match the license plate numbers listed on the field sheets of each 0 -D station against nearby 0 -D stations to uace the route: origin and destination of each vehicle. Simple computer programs and spreadsheets on be used in this process. Often, not more chan 60 percent of the license numbers on be traced through this study. Trips may bCgin and end within the study area and may never encounter a station. Errors in tranSCribing license numbers also reduce the number of matches made. The time between obscrvacions of a vehicle indicates fiairly accUrately whether stops are made in the business disuict, but knowledge of a time gap docs not make it possible to ascenain the purpose of the scop. license plate obsecw.tions must be matched at a minimum of twO locations. If SO percent of all license plates were recorded at one location and SO percent at another, it could generally be expected that only 2S percent of all license places were observed at both locations. Sampling rates at various locations will differ widely depending on vantage poincs, volume _and skill of observers. Video cameras can be used to get a 100 percent sample at all locations, bu.t manual data cxuaction ftom video can be time-<:ansum.ing.

T... ...... _ ....... . :'""'

nl~ "' "i"'"'

n .... t ,

•

AA 1

.·~ - - ~-·»~~~

'

./~ ~'\a>

W,/

®\,_~ __ , ,,_

<3)

..

- .....

~ ~

.~-- -~~~~:~~X~-::;;i~it~~~?~~~-~~1~ -~;~~W~1~~

Destln.otlon Senti o n

J

2 3 4

CoiSum T1 Volume Vi

Ori&ln Station 2

1

so

8

10 IS

13 88

210

6S

3

4

20 21 38 18

17

12 14 97 99 __2~ - - -32S ---

10

Row S unlT_.

VoiVJ

95 106

~0

3 10 200 37S

80 87

IS

42 84

400 -~-

----

-~~

Source: Rocss, Prassa3 and McShane. Traffic Enginuring. ITE, 1990.

Rocss, Prassas and McShane (2004) suggest a method for expanding license plare counts to estimate total volume between poincs in this type of study. For this method, the cocal volume at all observation pointS must be recorded for the period of interest. Sample data :ue presented in Exhibit 20-8. The variables for count expaiiSion :ue defined as follows:

r.N=

TU(N-1) •

(5!!L)

Equation 20-1

2

11

f; - 7;

\.')

F1 - 1j Where

TiiH

= number of trips from Station ito Station j

Tlj(N, IJ •

after theN"' iteration of the data (trips)

number of trips from Station ito Station j afi:er the (N - I)"' iteration of the data trips

r;

• sum of marched trips fi:om Station i (trips)

~

.. sum of matched trips to Station j (trips)

V,

= observed total volume at Station i (vehs)

\j

= observed coral volume at Station j .(vehs)

F;

= adjustment factor for Origin i

~

=

adjustment factor for Destination j

The 0 -D volumes of Exhibit 20-8 :ue ~ded using these factors. Results :ue rounded to the nearest vehicle. The results of several iterations :ue shown in Exhibit 20-9.


f.~ .t$~~J_fu,'@i-t?J~~·~?(¥P-it;.• ,.i.c;i~~~~~ilc~~~"·~'l';i·~"-~;,"_j6,.l',l~i','i'\if., ·~· ·•"-~ ~v, ·. -~ ,:-~·,-:_ ~~-.- ~ · · · '• ....~_t; ;"\-t; ·-·-,;.:; -~~····1! · -~-1-. _-:'· · ~ - . .- • _...:..-.,

,

__ . _

-

Destination Station

1

2 3 4

T.

•

,..

•

•...

"' •. - ! ... _ _

' -..

1 2 3 4

1 50 10 IS 13 88 210

2 8 65

3 20 21

4 17 10

12

38 18 97 325 3.35

15 42 84 400 4.76

14

T

v

---

95 106 80 87 368

250 310 200 375

F. _

2.63 .--

2.92--2.50 4.31-

----

1135

Origin Station

1

2 19 161

3 60 66

27

Ill

44

69 306

345

325 1.06

400 t.l6

v

125 27 37 44 234 210

251 200

p

0.90

0.80

T

.:!.

..

Origin Station

99 v, 200 2.02 F 2.39 (a) Field Dar.a and Factors for Iteration 0 Destination Station

- " '- -. ,, • -: 1

4 63 38 54 190

T 267 292 229

347 1135

v

F

250 310 200 375

0.93

1 2

1135

3 4 T

v F

Origin Station

1 115 26 33 44 218 210 0.96

2 16 150 23 41 230 200

0.87

3 60 70 107 74 311 325 1.05

4 66 42

55 213

376 ; 400 1.06

T 257 . 288 218 3n 1135

----

1.06 --0.87 1.08

(b) Initial Expansion of 0 -D Mauix (Iteration 0) Destination Station

---

v.

F.

250

0.97 1.08 0.92 1.0!

310 200 375 1135

(c) First Iteration of 0-D Matrix Destination Station

Origin Station

1

1

Ill

2 3 4 T

27 31 43 212 210 0.99

v F

2 15 146 21 39 221 200 0.90

3 61 75 105 76 317 325 1.03

(d) Second Iteration ofO-D Matrix

4 67 45 54 220 386 400 1.04

T

v.

F

254 293 211 378 1136

250 310 200 375

0.98 1.60

0.91 0.99

--

1135

.

Source: Roess, Pras= and McShane (2004}, Table 8.14. Transportation Planning Data • ~

As an example, T(1•11 in Exhibit 20-9(b) is calculated as:

2 39 2 63

1(1.1) =50-( · ; · )= 125 vehicles The procedure should be iterated until the second-column 0-D volumes are within t 10 percent agreement with observed volumes {that is, all factors are within 0.90 to 1.1 0). For the example in Exhibit 20-9, two additional iterations are needed beyond the initial expansion to achieve agreement at this rate. This method does not work well in locations where a large number of trips begin and end between observation points. Exhibit 20-10 shows a form that may be used to record license plates. In small surveys, only the last three digits of the license numbers need be recorded. Place numbers may also be recorded on audiotape with a time queue. Data are later transcribed to paper forms or directly into a computer file. Video cameras may be used to record license plate data. Vide<;> has the advantage of providing a more permanent record of the data that may be played and replayed to obtain information. Most video recorders also allow the time {and date) to be recorded directly on the video image. Cameras must be positioned carefully to record the plates of a wide variety of vehicles. Under certain lighting conditions, plates may be difficult to read. Reducing the data from video recording is time-consuming and tedious if done manually. Newer systems based on machine vision technology, such as automatic number plate recognition (ANPR), allow license plate data to be reduced automatically from digital recordings. ANPR technology is also used in redlight-running safety programs and automated toll collection operations.

\

lJ_ .

~-:_7*~·-

License Plate Study Field Sheet t..ocaUon

Time: Begin: End:

UCense

Time

Number

TNc:k

or Sus

OUt-

Ucense

of

Number

?

~-------------------

Time

Truc:k or

I

o.A·of

Bus?

~·---------------------------

Source: Box and Oppc:nlander, 1976.

The method is especially adaptable to locations where ttaffic is too .hea.vy to be stopped for driver interview. Licerue plate studies a.re also very useful in sma.ll-sca.le, limited 0-D studies, or where the destination is known (CBDs. shopping centers, airportS) (Rocs.s, Pcassa.s and McShane, 2004). Tracing license plate numbers over ala.rge number of exits and entrances is difficult. The method has the adV1111tage of a.llowing the observers to obtain daa without depending on the cooperation of individual drivers, as is required by interview methods. The IW:lihood of a bWed sample because of poor driver cooperation is less with this method than with the methods described ea.rlier. This method does not work wdl for large a.rea.s because of the enensive labor required, but is pa.nicula.rly adaptable to srudies ofsingle routes or fa.cilities. Each srudy must be completed in one day, and it must be continuous. This method does not produce any information on the purpose of the trips, nor does it produce information on tranSit trips or vehicle pa.rking.

Tr.anc;:nl\rt~tinn Pl~nninn n~t~

•

445

3.1.5 Vehicl~ Registrations An alternate and far more detailed license plate technique involves recording the full license numbers, identifying vehicle ownership from registration records and sending a mail-back questionnaire to each owner. This technique has che same advanrages as the voluntary return postcard technique, of which it is a variation. It is less disruptive because traffic is not StOpped. If video techniques are used to record license plate numbers, a smaller number of field personnel will be needed. Research indicates the response nte to mailed-out questionnaires may be higher than chat of the voluntary return-postcard technique. The disadvanrages include the fact that personal conracc is not made with respondents. Fewer questions can be asked. Ic is difficult to use economically and efficiently unless the agency responsible for motor vehicle registrations will provide addresses of drivers co whom questionnaires may be mailed. 3.1.6 Vehicle Intercept Method The vehicle intercept method can be used for small study areas. Stations are established at all entrances and exits to the study area. Each entering vehicle is stopped and a preceded or colored card is handed co the driver with instrucdons lO surrender the card upon exidng the area. Exiting vehicles are stopped and the cards collecced, or the notation that they had riot received a card is made. Variations of this procedure are to use colored rape affixed to the bumper of the entering vehicle or co rape the colored card to the windshield. These variations eliminate the need for stopping vehicles at the exits from the area and permit the collection of data at intermediate locations (ITE, 2009). 3.1.7 T11g-On- Vehicle Method The rag-on-vehicle method is a variation that does not depend on the complete cooperation of drivers. It may be used where traffic is too heavy for effective use of driver inrecview and where limited staff makes the use of the license plate recording methods impractical. A coded card is handed to the driver or fastened to the vehicle as it enters the route or area of study. The driver is informed of the nature of the survey and cold the card will be picked up wh~ leaving the route or survey area. 'When the vehicle leaves the route or area, the time, station, direction of travel and any other readily observed inform a· cion are recorded on the preceded card. If traffic is too heavy, the cards may simply be bundled into groups according to the time intervals and the time wrircen on the top card of each bundle. This method has many of the same advantages and disadvanrages as the license plate method but requires some driver cooperation. !J.l.B Lights-On Studies In headlight or lights-on studies, individual vehicles are rraced from one or rwo origin points to a maximum of rwo or three destination points, generally within one-half to 1 mile of each other. These studies were particululy useful for studying compla weaving areas since one approach leg could ask drivers to turn headlights on and then those vehicles were traced to their exit. With many vehicles now using daytime running lighcs, this type of methodology is no longer viable (Roess, Prassas and McShane, 2004).

3.2 Internal Studies Internal studies P'ovide information on trips .nude by residents of an area. These trips cypically comprise the bulk of travel within an area. Variations of the home interview survey are a commonly used form of internal survey. Several other forms of internal surveys have been used, including controlled postcard surveys, multiple cordon surve~ and television as a replacement for the home interview (McGrath and Guinn, 1970). In addition, the license plate survey techniques discussed above can be adapted for internal studies. The types of studies discussed below include: • dwe!Jing-unit interviews; • vehicle owner mail questionnaires; • interviews at workplaces and special generators; • transit ro ute passenger questionnaires (on-board tranSit surveys); and • truck and taxi surveys. Internal studies are comprehensive in scope, seeking to obtain information on all cypes of trips by residents in an area, including aavd by public tranSit vehicles, trucks, taXis and private vehicles. They are commonly made as part of a 446 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITION

comprehensive metropolitan area 0-D study. Manuals of detailed instructions and guiddines on these s tudies should be consulted for additional details beyond the scope of this manual. The basic concepts are described below. · 3.2.1 Dwelling-Unit Interviews The study area is first divided into analysis zones that are as homogeneous as possible. Every dwdling uni t in each zone is identified. Sample sizes are selected for each zone based on housing density, and the dwelling units to be surveyed are systematically selected within the zones. Each dwelling unit selected is initially contacted by mail to alert its residents to the impending studies and advise them an interviewer will be contacting them in person. Th e mailing also indicates what information will be req uested and approximately when the interview will be conducted . A uip dia ry (or travel diary) is included so respondentS can record information themselves. The interviewer can then confirm and supplemenc the self-recorded data.

Interviewers are hired, t.raincd and assigned to conracr personally the residentS of the selected dwelling units. They are inscrucced to interview oQly the sdecred unitS (that is, they cannot substitute another dwelling unit if no one is at the selecred unit or it is vacant at the time of the interview). A minimum of three repeat visits are made to residents who are not home duriqg th~ initial contact attempt.lmervi~ ace usually conducred during the early evening hours (6:00 to 10:00 p.m.), when there is greater likelihood of people being at home. The information collecred includes social and economic data on all of the household residents as well as da1:1 on all trips made by each resident age 5 or older during the preceding weekday. Dwelling-unit data consist o f the number of people living at the unit, the number of vehicles owned and/or operated by unjt residents and derails (e.g. gender, age, etc.) on each person. Travel information is collecred for each person living in the unit for trips made during the preceding weekday. Information includes. time of trip, origin and destination, mode of travel, purpose of trip and parking information for each link in each trip by each person. Interviews are not generally conducted on Mond ays because preceding weekday (F~iday) trips are easily forgotten over the weekend. Once the interviews have been complet~, the data on travel trips are coded by origin and destination. Each trip c:nd (origin or destination) is coded ro the zone in which it occurred. The data arc then expanded to the full sample us ing a zonal faeroe that is calculated by dividing the total number of dwelling unitS in the zone by the actual number of interviews completed suocessfully. Other facrors are used to convert the data to an average weekday. The survey rcsul ts are compared to the screenline counts (see Chapter 4) and correction facrors are developed as described lat.er in che chapter. 3.2.2 Vehicle Owner Mail Q!ustionnaires This method involves mailing questionnaires or return-addressed, stamped postcards co owners of motor vehicles f.iving within the survey area. This survey may be combined with driver interviews or postcard questionnaire surveys of traffic entering and leaving the study areas. The elements not obtained include taxi and ttansit patterns. T he ~pie~t is asked to .record all trips made by motor vehicles on the day after the card is received, normally a weekday. A poceJ'lcial source of bias comes &om the possibility of getting lower response from vehicle owners who don't appreciate t:.he benefits of such a survey or who don't have good reading and writing skills. ~ a result, the travel patterns of th e:se populations are under-represented in the survey results. Different-colored cards may be used for private passcngc:r vehlcles and trucks.

Names and addresses of vehicle owners are obtained from State motor vehicle registration 6.les. Cards are marked :;~.C cord.ing to the address zone before mailing, ~d tabulations are made of the number of cards for each vehicle classsc:..,.t co each zone. This enables proper estimation of the total ttips based on the completed cards returned, as compared with the total vehicles listed for each zone. Truck Beet owners (three or more crudes) may be contacted petsonally, co ensure better accuracy for this relatively small but important group. Some state highway departments have developed d~tailed procedural manuals for this type of survey. An inexpensive variation of this method that has been used s~C cessfully is tO Send the questionnaire tO the motorist along with the annual vehicle license certificate. 3.2.3 Intentiews at WOrkplaces and Spedal Gnum#ors Travd questionnaires may be distributed co all employees of an acrivity center, such as a large industrial plam or a group of office buildings, or to people visiting an acrivity center such as an airport, shopping center, transit rennin~• or other special generator. For employees, completed forms are picked up the same day they are disuibutecl Data 0' .61 where employees live, how they get ro work. times of arrival and departure, parking information and trip cosu ca-:J:l all be obcained for auto drivers, auto passengers and transit and taXi passengers. It is important to encotirige agw ..d response rate by keeping the questionnaire short. Preparation of questionnaires is reviewed in Appendix B. Transportation Planning Data • 44:::;:7

This method is most dfeaive when thc:ce :ue a few largtt employ= involved; however, it will work under other cin:umstances if the employers fed the survey is important. After obtaining the cooperation of mmagement, individual firms will often distribute and pick up the questionnaires within their own organizations. This type of study often yidds information of direct value to the employers and thus hdps .!CCUI'e their cooperation. It is important to =rd the total number offorms distributed to c:adt fum as wdl as the number of employees in c:adt 6rm so nip dan for each mmpany can be expanded properly. Similar surveys can be conducted at other special generators, such as aiipom and shopping ccncers. Depending on the circumsances, parrons can be intervicwal or be given cards to complete and mail back. 3.24 Trtmsil Rouu P~ Quutitmtl4iru This study (:also called an on-board transit survey) is confined to the ultimate origins and destinations of passengers using a particular transit route. It is used primarily for planning improvements to the routes or schedules of transit units. One or rwo survey persons ride each transit vehicle and distribute a questionnaire card (and pencil) to each passenger who boards the vehicle. Passengers may be askd to complete the card and return ic to ndd survey personnd when leaving the vehicle, or it may be mailed back. This study is best suited to lighdy travdedlines where passengers are all seated. It is difficult for sanding passengers co .6.ll in the questionnaire cards. The results are expanded to represent 100 percent of passengers, as based on the ratio o( the total riders to the number of cards .6.lled in.

Rerum-addressed postcards may be used when transit loa.ds make it impossible to conduct this study properly on moving vehicles. The cards are handed co passengers as they board or alight, to be complered and returned u a later time. The return rate may be low, as with other postcard studies. Considerable care muse be exercised in the analysis to make certain the results are not biased. There may be a tendency for only the regular commuters or only those subject to most crowding. or other discomfort, to show interest in returning cards. Transit surveys are discussed in more derail in Clupter 13.

3.2.5 Tn4k 11rul TIIXl Surveys The data on commerci:a.l trips (cruck, taXi and transit nips) are collected in separare surveys. Truck and caxi data are obtained by desk studies at the dispatch offices of these vehicles because most of the data are available from office =rds. In some I'CCI:Ilt studies, installation of truck-mounted global positioning system (GPS) unitS and regular polling of unitS from a ccntr:a.llocation has been used successfully to collect &eight trips in New Zealand. Studies of goods m

4.1 Presenting 0-0 Data Results A number ofstandard tabulations are recommended to summarize the basic trip data and to determine the numbers and percentages of total trips made by car, transit and tni, as wdl as the numbers and percentages of trips made for various purposes. Other tables may be co~tructed to show the numbers of trips between zones by mode and purpose. Following the 0-D survey, a set of trip tables is prepared that shows the number of trips between each zone in the study area. These tables can be subdivided by trip purpose, truck trips and taXi trips. Tables are also prepared that list socioeconomic characteristics for each zone and the uavd time between zones. Data presented in these tables may be graphed or plotted on maps for easy interpreracion.

4.2 Checking Survey Accuracy Thc:ce arc several methods used to check the accuraey of all or portions of the comprehensive 0-D survey. Some of these are listed below. 1. NOTE: When routes of trips have been obtained in the interviews, two or three control points are selected. These should be important viaducts, bridges, or other well-known pointS of traffic Bow constriction. Information obtained in the internal and c:nemal surveys, as expanded to represent 100 percent of traffic passing such pointS, is compared with accual field counts.

2. A natural burier such as a rivet ot a railroad track i.! selected as a screenline thac divides the incemal area into cwo parts. The estimated 100 percent count of traffic crossing the saeenline as derived from the internal and atemal surveys is compared with actual field countS.

3. A cordon line comparison involves deriving traffic ar a cordon line station from the internal survey and comparing this with similar trips of area residents as derived from the c:xternal survey. 4. Public uansit riding derived from the internal survey can be compared with total riding observed in the fidd or obtained from transit company records. 5. The accuracy of trip reponing may be estimated by sdecting a wne in an area of high employment for which employment 6~ are available and comparing the number of persons employed in that wne with the number of work trips into the woes, as determined from the expanded interview data. In this comparison, work trips by all modes of travd must be included. Proper allowance must be made for those persons walking and bicycling to work, average absenteeism and the likelihood of employees making midday shopping or business trips to and from the area. If these comparisons reveal siuble discrepancies, adjustments should be made in the survey daca to correct them. Exhibit 20-11 shows a comparison of screenline and 0-D data.

I I

I

Source: Box and Oppcnlander, 1976.

4.3 Additional Sources of Data In addition to primary studies, where data arc collected direcdy, there are a number of secondary data sources. The U.S. Census has proven the best secondary source of residential uip·production data (Dickey, 1983). More detailed information on household characteristics is available from the Urban Transporr:ation Planning Package of the Bureau of Census. Traffic z.ones are often designed to follow census traer boundaries. In those areas where rraffic tones differ, the Bureau of Census, for a small charge, will tabulate household characteristics using a block-to..z.one aggregation cable. Updates from the base year are sometimes necessary. Shon of a survey, local records or ciry directories arc the best source. Automobile registrations for a number of areas are commercially available. A few regions tabulate land use data regularly from local sources such as assessors' records, utility companies and building or occupancy permits.

5.0 REFERENCES American Society of Civil Engineers. American Socirty ofCivil Engineers Urban Planning Guide, ASCE ~uals and Repot!S on Engineering Practice No. 49. New York: ASCE, 1986.

Kraft, W H., W S. Homburger and J. L. Pline. Tmffic Engineering Handbook, 6th ed. Washington, DC: Institute of Transportation Engineers, 2009. Institute ofTransportation Engineers. Transportation Planning Handbook, 3rd ed. Washington, DC: ITE, 2009. McGrath, W. and C. Guinn. •simulated Home ln~rvicw by Tdevi.sion: Origins and Destination Techniques and Evalu;.rion." Transponati~tn &search &cord:}rnmuzl oftiN TMnsponation &search Board41 (1970): n~3. Meyer, M.D. and E. J. Miller. Urban Transportation Planning, A Derision-OrimtedApproach. New York: McGraw-Hill, 2001. Nelson, D . C. Manual ofTransponation Engineering Srudier. Washington, DC: Institute ofTransporration Engineers, 1994. Roess, R. P.., E. S. Prassas and W. R. McShane. Traffic Engineering. Upper Saddle River, NJ: Pearson Prentice Hall, 2004. Stopher, P. and A. Meyburg, Urban Transponatiqn M~tfkling and Planning. Lexington, MA: Lexington Books, 1975. Transportadon Research Board. Conftrmu Procttdinf1 42, lnnowz.tions in Traut/ Dmurrul Mofkling, Volume 2: Papm. Washington, DC: TRB, 2008. Transportation Research Board. Highway Capacity ManuaL' Washington, DC: TRB, 2000. Transportation Research Board Special Report 288: Mnropolitllll Travel Forrcasring, Currmr Practiu and Furun Direction. Washington, DC: TRB, 2007. Urban Renewal Administnrion and Bureau of Public Roads. Statu/ani UmJ Ute Coding Mamuzl. Washington, DC: U.S. Government Printing Office, 1965.

450 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

Chapt er21 ••

• •

• •

• •

• •

• •

• •

•

• • •

•

a •

•

•

• •

• •

•

•

• • •

•

• •

•

• •

• •

• •

• • •

• •

•

• •

• • • •

•

• •

• •

• •

• •

• •

• •

• • •

.. • • •

•

•

•

• • • • •

"' •

Environmental Impacts of Transportation Projects Original By: Donna C Nelson, Ph.D., P.E.

EJiutl By: Dani~IJ Fitu!ley, P.E. 1.0

INTRODUCTION

451

2.0

HIGHWAY NOISE IMPACT STUDIES

452

2. 1 Noise Impacts and Analysis

453

2.2 Determination of Existing Noise Levels

454

2.3 Noise Measurement Procedures

455

2.4 Prediction of Noise fmpacts

456

3.0 AIR QUALITY IMPACT STUDIES

457

3. 1 Air Quality Impacts

458

3.2 Emission Rates

459

3.3 Air Pollutant Dispersion Models for Impact Anafysis

459

3.4 fnput Data Requirements

459

3.5 Construction Impacts

459

3.6 Air Quality Reports

460

4.0 REFERENCES

461


461

4.2 Online References

461

4.3 Other Resources

462

1.0 INTRODUCTION ransportation systems affect the environment in many ways. The environmental impactS of ttanspornuion proj ccts are coming unde• close scrutiny because of priorities across the United States and around the world. Alon.g with safety and. efficiency, environmental responsibility is a primary goal of transportation agencies. The impact transportation and oilier industries exen on the environment can result in climate change. Transportation accoun t::S for approxiinarely one-third of greenhouse gas emissions in the United States, with 72 peccent of ttansponation.'s impactS coming from road use (AASHTO, 2008). Climate change has potential impacts on transportation througl'l increases in weather/climate extremes and Boeding of coastal transportation infrasrructUie (TRB, 2008). For inst:wcc=: , more frequent intense precipitation could lead to increased disruptions to operations at transportation facilities, e~ sion of road base foundations and scouring of bridge supporrs and pipes.

T

In the United Stares, governmental regulations require that an environmental impact Statement (EIS) be prepared fo .X any transportation project that will affect me human and physical environment. For fedeml-aid transportation projecn, th.~ EIS muse indu
• natural resources, including prime and unique farmlands, wetlands, threatened and endangered species, natural land forms, groundwater resources and energy requirements; • rdocation of individuals and families, including the number of households displaced, neighborhood disruption, avUlable housing, number of businesses displaced or affected, documentation of public pmicipacion and any unusU21 circumstances; • air quality studies, including microscale impacts, mesoscale and mobiie stationary impacts, analysis methodology and a description of how consistent the project is with the state implementation plan (SIP); • highway noise studies, including the identification of sensitive receptors (such as schools and hospitals); comparison of future noise levels with Federal Highway Administration (FHWA) criteria and existing noise levels; noise abatement measures; and noise problems with no reasonable solution; • wetlands and coastal zones' studies to document the analyses, practical measures to minimize lwm and that there are no pn.ctical alternatives; • social and economic impact studies of impacts on lifestyle, uavd patterns, school districts, churches, recreation, businesses, minorities and ethnic groups, urban quality and secondary impacts; • water quality issues, such as erosion, sedimentation., use of deicing and weed control products, chemical spills, groundwater contamination, stream modifications, impoundment, as wdl as impacts on fish and wildlife; • flood hazard studies including impacts on beneficial floodplain values, incompatible devdopmem, measures to minimize flood risks and the evalU2tion of alrernatives; and · • construction effects including impacts of the construction process on air quality, noise, water, traffic detours; and the impacr of spoil and borrow (the need to dispose of or find soil and other materials). A detailed discussion of each of these studies is beyond the soope of this chapter. The
2.0 HIGHWAY NOISE IMPACT STUDIES Highway noise studies ace conducted to help determine the additional noise generated by the use of tn.nSponacion systems in the community. To do this, the noise levd generated by a new or improved highway facility is estimated and compared tO existing and future •No Al;tion· noise levels in the community in which the fxility is proposed. The

cl=aaeristics of environmental noise that ace of particular concern are (Cohn and McVoy, 1982): • magnitude of the sound; • frequency of the sound; • temporal distribution of~ sound; and • time variance of the sound. The 1114gnitude of sound is perceived by the human ear as a shon-duration Huccuation in atmospheric pressure. The levd o~ sound pressure or ~e magnitude of a specific sound (or ambient sounds) is expressed in decibds (dB). The

!decibel scale is logaridunic; therefore, an increase of I dB reflectS a tenfold increase in the sound pressure level. Sound jpressure levels are generally adjusted to one of three scales: A. B, or C. The A-weighted sound-level scale is used co 1measure the magnitude of traffic noise because it most closely reflects the response of the human ear to transportation hoise. The A scale (referred to as dBA) de-emphasizes lower-frequency sounds. The foqumcy if> of a sound is determined by the number of times per second the sound pressure fluctuates between posirive and negative values on a sinusoidal wave. Noise &om transportation sources are not usually pure tones but are broadbanded sounds with a wide frequency range. Typically, the range of human hearing is 20 hem to 20 kilohertz.

Tnnporal distribution of noise is important because the time of day, day of week and month or season affect the perception and effect of noise on the receiver. Noise levds are usually higher in the daytime than at night. However, nighttime noises may be perceived as more offensive. Noise levels acceptable at one time of day may not be acceprable at another time. TirM varittnc~ of sound refers to the fact that enviionmencal noise is rarely stationary; that is, the magnitude often varies over very shore periods of time. The passage of an automobile or airplane might cause the noise level in a quiet neighborhood to rise by 10 to 20 dBA for a short period of time.

Several noise descriptors can be used to account for the above characteristics. • L50 is the sound level exceeded 50 percent of the time (that is, the median sou.nd level). • Ll 0 is the sound level exceeded 10 percent of the time. • L90 is the sound level c:xcecded 90 percent of the time.

• Leq is the equivalent s~und level. • Ldn is the day-night sound level.

2.1 Noise Impacts and Analysis Although measurements of noise emissions from individual vehicles are commonly performed by traffic and environmental regulatory bodies, they do not generally fall within the sphere of interest of the traffic engineer. Two aspectS of areawide noise levels are of concern: actual !evds of traffic noise and perceived annoyance of traffic noise. Interference with speech (including 1V listening) and sleep are the most common complaints concerning uansport:~tion-rdated noise. There is no completely satisfactory measure of the subjective effects of noise. However, for the United Stares, FHWA has devdoped maximum permissible noise-level criteria for~ federally funded highways. These design goa!$ require noise evaluations for new or expanded roadways (Type but not for existing highways (Type ID which can be mitigated at the discretion of the individual srate. Evaluations must be performed and any impactS reported during the location planning and design phases. Exhibit 21-1 presents a summary ofFHWA noise abatement criteria for various land use descriptions. The design levd comparison can be selected &om either LIO (defined above) or Leq, which is the equivalent steady-state sound level which contains the same acoustic energy as the time-varying sound level of equal time periods (23 CFR. 2009). Local areas, municipalities, or States may have additional noise criteria. A traffic noise impact occurs when either the projected noise levels approach or c:xcecd the noise abatement criteria (Exhibit 21-1) or the projected tnffic noise levels substantially exceed the existing noise levels in the area (FHWA, 2006).

n.

'

Environmental Impacts of Transportation Projects • 453

'~ .. · ~~~~'J.lo~li~~~~t·~~~~~r~'t;~;;·t~t~~!;.)l~tt.~t~~~T,(·~'!,Jit(~?.L'"''-' -· '0 -~ -JJ -~ . :!'... 0: ._..,..,••.• "'.• - ... ~ .•r:.~ •:....:,,.,.,. -~ ...:>~~~ ......, •. :c.~ .,.{1:~•. ":.--. ~-\¢t~t~.~·-:~t:~t:tl>,-~ Design Level, Design Level, Land Usc Category L (h) L 10(h) Description of Land u~ Ottegory ,<;

57 (Exterior)

GO (Exterioc)

Lands on which sereniry and quiet arc of extraordinary significance and serve an important public need and where the preservation of those qualities is essential if the area is co continue co fulfill its imended purpose

67 (Exterior)

70 (Exterior)

c

72 (Exterior)

75 (Exterior)

D

None

None

Picnic areas, recreation areas, playgrounds, active spom areas, parks, residences, morels, hotels, schools, chutchcs, libraries and hospitals Developed lands, properties, or activities not included in Categories A orB above Undeve.loped lands

A

B

Residences, mote.ls, hon~ls. public meeting rooms, schools, churches, libraries, hospitals and auditoriums Source: Highway Trajfo Noiu Anttlysi! and Abasmzmc Policy and Guida net. U.S. DOT.IFHWA/Office of Environment and Planning. Table 5, page 7, 1995. E

52 (Interior)

55 (Interior)

Traffic noise analysis includes the following steps for each alternative of a highway project: 1. identification of existing activities, developed lands and undeveloped lands for which development is planned, designed and programmed, which may be affected by traffic noise from the highway; 2. determination of existing noise levels; 3. prediction of traffic noise levels; 4. determination of traffic noise impactS; and 5. examination and evaluation of alternative noise abatement measures for reducing or eliminating the traffic noise impaets. (FHWA, 2006)

2.2 Determination of Existing Noise Levels Existing noise is composed of all natural and human-made noise sources that can be considered as part of rhe acoustical environment of the general area. Sources of ambient noise include aircraft and airporu, railroad tracks, fire stations, emergency medical facilities {sirens), schoolyards, other roadw.tys and recreation mas; and areas where birds, crickets, or other noise-making wildlife convegare. The difference berween predicted noise levels and existing ambient noise, together with information on the area itSelf, give an indication of the impact of the highway on the area. The methods and equipment described here are restricted to the type of field measurementS necessary for highWf!Y noise studies.

2.2.1 MeiZSII1't1M1U Sites Noise levels cannot be measured at every point in the srudy area.. For projeCts covering large land areas, the study area is divided into representative sections enclosing similar noise environments. Particularly noise-sensitive locations are identified and studied separately. These include residential neighborhoods, churches, parks, libraries, schools and hospitals. Other representative sections could include indu.scrial parks, office complexes and shopping centers. Exterior !cations are typically-sampled unless ouaide activities are unlikely for that use. The exterior loations could be taken at or near the highway right of way, buildings, or an area in between the right of way and a building where outside activity is common. Noise rneasurtmena should be conducted at representative sites in the affected area. If an area is too large to be described by a single noise level or a small range oflevcl.s, it should be subdivided into appropriate: smaller areas. For example, a large residential area near an existing traffic artery may be broken down into three groups of sites: one group !oared at the edge of the area adjacent to the existing highway, where the ambient noise is clearly due to the highway 454 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDmON

traffic; a second group located toward rhe interior of rhe residential area, where the arterial traffic is still a major factor '.in establishing the noise bur where orher noises of the community are beginning to make significant conrribucions; ·.~nd a third group deep enough imo the community that the only noise measured is from the community itsdf.

2.3 Noise Measurement Procedures A sound level meter (SLM) should be used to quanrif}r noise levels, in accordance with ANSI S1.4-1 983 (FHW'A. 1996). A gtaphiclevel recorder can be used for visual displays of noise levels and co veril}r the acoustic in tegrity of individual events. Frequency characteristics can be studied with a one-third octave band analyzer. For further derails about equipment, calibration and manufacturers refer co FHWA's M~asuf(mtnt ofHighway-Re!aud Noise (FHWA, 1996): The operations manual for the specific equipment should be read carefully. Generally, the microphone or soundlevel meter is mounted on a tripod so the person raking the readings has both hands free. In addition, handheld microphones will bias the data. The microphone is U.Sll;ally positioned 5 feet (ft.) (1.5 meters (m)) above the ground for ground-floor ambient noise measurements. The microphone may be supported outside upper-level windows but as far as possible from the exterior wall of the building (at lease 3 to 4 ft.{9 to 1.2 m)) for upper-level measurements. The SLM should be calibrated in the field before and after each measurement session. Iris a good precaution to calibrate sound meters periodically in the laboratory as well. Calibrators are standardized, stable sources that genera ce a predefined sound presswe level. When placed on the microphone, the meter defleaion can be adj usted to a fiJced value. The manufacrurer's instructions must be foUowed canefuUy. A device that cannot be properly calibrated shou.l d not be used. If a sound recorder is used, its frequency response should be measured in the laboratory periodically. The gain should be set permanently at a suitable value i:o avoid errors when the recording is played back. The sound recorder may a.Jso be' calibrated using the techniques described above. The calibration tone should be recorded for approximately .30 seconds (sec), and the position of the sound meter attenuator should be noted, preferably on the recorded media. Th-e list of equipment used, including serial numbers, should also be recorded verbally. The recording should be played back on the sanne machine that was used to record it. Noise measurements should produce the highest hourly noise level generated from the rep.resentacive noise sources for that area (FHWA, 1995). It is important to realize the peak hour of traffic volume might not yield the highest noise levels; times with higher speeds or heavy vehicle traffic mix might produce higher levels of noise. Other factors co consider .include the time of the day, day of the week, weekends, work days, tourist season$ and representativeness of the noise. Guidelines for sampling period arc presented in Exhibit 21-2. Typical sannple periods range from 2 rninllres (min.) to 30 min. (FHWA, 1996). Local acoustic assessment standards might dictate which criteria should be used for analysis and srudy. Under sceady conditions, a minimum of three repetitions is recommended, while six repetitions ace preferred. Under some circumstances a 24-hour srudy period may be desirable.

• A minimum of three repetitions is cco>mmended, with 6 repetitions being preferred. Source: Federal Highway Adminisuation, Mea.surmzmt ofHighwtty-&laud Noiu, 1996.

Environmental Impacts of Transportation Projects • 45~

The noise level measun:mem analysis should involve the foUowing steps (FHWA, 1996}: 1. Adjust mea.~ured levels for calibration drift.

2. Adjust measured levels for ambient sound. 3. Compute the mean sound level for each receiver by arithmetically averaging the levels from individual sampling periods. 4. Perform an assessment of the ave.raged sound levels based on study objectives.

2.4 Prediction of Noise Impacts Numerous computer software programs arc available for the prediction of traffic noise levels. FHWA's Traffic Noise Model (TNM) is a commonly wed software in the United States. The TNM contains the following componentS ~K~:

.

• modeling of five standard vehicle types, including automobiles, medium trucks, heavy trucks, buses and motorcycles, as well a.1 user-defined vehicles • modeling of both constant-How and interrupted-flow uaffic using a 1994/1995 field-measured database • modeling of the effects of different pavement types, a.1 wdl as the dfccts of graded roadways • sound level computations based on a one-third ocave-band dambase and algorithms • graphically interactive noise barrier design and optimization • attenuation over/through rows of buildings and dense vegetation • multiple d.ifiTaction analysis • parallel barrier analysis • contour analysis, including sound level contours, barrier insertion loss contours and sound-level difference contours For manual calculations of noise levels, NCHRP Report 174 (TRB, 1976) includes a comprehensive design guide for highway noise computations. An c:xam.ple of noise contours from a study is presented in Exhibit 21-3.

20ft

I

o.,.,........

I

f

Source: Kugler, B. A. et al. NCHRP Report 174. Highwtty Noist: A Dmgn GuUU for Prtdictitm 111111 CtmtroL Copyright, Nacional Academy of Sciences, Washington, DC, 1976. Reproduced with permission of the Truuportacion Research Board.

3.0 AIR QUALITY IMPACT STUDIES Deterioc;ating air quality is a severe problem in many wban and subwban areas with impacts across all areas. State and local agencies have the responsibility for monitoring ai.r quality which typically includes five pollutams (Forkenbroclc, 2004).

• Ozone (0 ) • Particulate maner (PM) • Carbon monoxide (CO) • Nitrogen dioxide (NO) • Sulfur dioxide (SO)

Each region is measured for the five criteria pollutants to determine if the area is classified as •attainment• o r "nonattainment" (Forkcnbrock, 2004). •Maintenance" status is used to reference specific pollutantS when an area moves from an attainment to nonattainment level. Exhibit 21-4 shows the national air quality standards that must be met to achieve attainment statuS. Responsible state agencies and metropolitan planning organizations (MPOs) within nonanainment rcgiops mUSt prepare state implementation plans (SIPs) that ensure no mnsponation projectS or policies will inpease regional emissions or cause a pollutant violation. When mnsportation improvement programs arc consistent with the corresponding SIP, transportation conformity is achieved. Nonanainment or maintenance status can result in rules that a transportation project must not cause an increase in a specific pollurani:, or the implementation of more stringent analY'i.s procedures. The impact of transportation projeas on greenhouse gas (GHG) emissions is also an imponant consideration. In 2003, tranSportation accounted for 27 percent of the GHG emissions in the United StattS (EPA, 2006). The prim.ary human generated componentS of GHG emissions include carbon dioxide (C01), chloroBuorocarbons (CFCs), hy.d!oBuorocarbons (HFCs), perfluorocarbons (PFCs) and sulfur hc:xafluoride (SFJ. GHG emissions are typically reponed in terms of C01 equivalent unics.

Environmental Impacts of Transportation Projects • 457

Parciculate (PM,.)"' Particulate (PMu)•..

Sulfur dioxide (S01) • Parenthetical value i.s an approximately equivalent concentration. ""' Particles with aerodynamic diameters of I0 micrometers or less. - * Panicles with aerodynamic diameters of2.5 micrometers or less.

So=: Forkcnbrock, David J. and Jason Sheeley. NqiRP, Report 532, Ejfoaive Mnhods for Environmmtll/justiu Assessmmt. Copyright, National .Academy of Sciences, Washington, DC, 2004. Reproduced with permission of the Transportation Research Board.

3.1 Air Quality Impacts Air quality impacts arc determined in respect to ambient concentration standards. The impacts of a proposed transportation project are usually analyzed by (Cohn and McVoy. 1982): 1. projecting the amount of traffic expected co result from the project; 2. calculating the quantity of pollutants that will be emitted by the projected rcaffic; 3. estimating the resultant concentration of the pollutants of interest for a particular receptor site, wing a dispersion model or some other analysis tool; 4. adding the traffic-generated pollutant concentration to an expected background concentration generated by other pollutant sources; and 5. comparing the results to the ambient standard for various project alternatives. Air quality impact analysis for a highway is commonly conducted to (Cohn and McVoy, 1982): • determine whether or not the proposed project is likely ro cause a violation of the ambient air quality standards; • compare the relative impacts of the various project alternatives, including the null (or do nothing) alternative; • make an informal determination regarding consistency with the SIP; and • plan to defend the walysis against criticism from professionals with a vested interest in opposing (or suppoccing) the project. 458 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

The microscale CO analysis is the most common air quality assessment. This analysis focuses on ground-levd or "hot·•. spot" impacts (EPA, 2006). This process includes combining the microscale effects and demographics of affected ~reas. : Microscale analyses generally employ a dispersion model to predict concentrations at critical recepcor sites. Jncersec·. tions are studie
3.2 Emission Rates The determinacion of emission races is an important step in the air quality impact analysis process. Models are typically used to estimate emission factors for vehicles which is the rate of pollutant emissions by an average vehicle, expreSsed in grams/ mile for moving vehicles or grams/how for idling vehicles (EPA, 2006). The parameters of an emission model should be customized to the area of study, which include vehicle age; mileage by vehicle type; inspection srudy year and maintenance programs; and specific fuel characteristics. The most widely used method for determining veh.i_cle emissions is the MOBILE6 model developed by the U.S. Environmental Protection Agency (EPA) (for automobile, diesel and gasoline trucks, etc. under various conditions). The Emissions Factors model is used in California to generate emission factors. Research efforts have studied the emission levels for real-world microscopic speed profiles and the impact during high-emissions events that coincide with high acceleration and speed (Frey, Rouphail, Zhai, 2006). These microscopic emissions mod.els are incr~ingly used in traffic simulation tools for e.ovironmencal studies (see Chapter 11). ·

3.3 Air Pollutant Dispersion Models for Impact Analysis If the amount of pollutant emissions is known or can be estimated, dispersion models can be used to predict con· cenrracions at variow locations. There are several models available that model the dispersion of pollutants. Some of the more common programs available for the highway environment that are preferred by EPA are the AERMOD and CALPUFF modeling systems. Other models include BLP, CAL3QHC/CAL3QHCR, CTDMPLUS, OCD ~d CALINE-4 (EPA, 2009).

3.41nput Data Requirements Ce.rtain basic types of data are required to assess potential air quality impacts. Models most commonly require information on vehicular emi$$ions, meteorology, traffic flow characteristics, ambient air quality, topography, land ~se, region, climate and geometric configuration of roadway. Other parameters of interest might include current and fu~e population, current and future employment, current and future traffic conditions and future transit operating polic .i.~s (Forkenbrock, 2004). Ambient air temperature and the fraction of vehicles in warmed-up or cold operation con ell· cion are used to define an average composite emission factor for each specific situation. The procedwes are upda-c:4::d periodica11y, and projections for future-year emission are based on current federal automotive emissions scanduds as prescribed by the Clean Air Act Amendments.

3.5 Construction Impacts The air quality impactS &om highway consrruction activities come from three basic sowces: dust, rraflic con~c?n due to construction and emissions of construction equipment. When land is cleared of vegetation and the soil dri. e> out, fine-grained soil material can be picked up by the wind. Some types of soils are more susceptible than others -!CO this problem. The U.S. Depaccmenc of Agriculcwe, Natural Resources Conservation Service has mapped much of t:f:;?e nation's soil and describe
tion can be wed to estimate the magnitude of the dust problem and to recommend mitigation techniques, including minimization of the amount ofland cleared, replanting, use of dust suppression methods and others. These cypes of controls are part of the standard specifications for most transportation projects.

Air qualicy impaets due to traffic congestion occur when access to existing highways must· be restricted during construction. The impaCts of these diversions can be assessed in much the same manner as any other highway project; however, due to their temporacy nature they are usually determined to be insignificant. ImpaCts associated with construction also involve emissions from construction equipment. Emission rates for construction equipment arc published by EPA and impaets can be analyzed using conventional modeling techniques. In most states, constmccion impaCts are considered temporacy and/or insignificant. Their discussion in air qualicy impact reports is usually a commitment to undertake all appropriate mitigation measures.

3.6 Air Quality Reports Air quality reports should be written for professional reviewers. They should contain all the information nec.cssary for the reviewer to assess the air quality impaCts of the proposed project and should contain sufficient information for another professional to reconstruct the entire analysis. Every detail does not have to be included; however, the information regarding the model inputs, receptor locations, 1- and 8-hour conversion and so on should be made available simply in a convenient format. Chapter 3 and Appendix D contain more details on communicating data to the public. Reports can contain analysis resulrs showing intersection volumes or level of service, where higher volumes and lower level of service values represent higher potencial for localized air quality effects {Forkenbrock, 2004). A report should discuss arrainment statuS and air qualicy guidance. A map of protected populations and nearby sources of air pollution is a u.seful presentation aid (Exhibit 215). Tables containing emissions-level details and a corresponding map of receptor locations is useful for reviewers.

;"'t,:o:·A'..~.-..o~~~

·tt<,_

Source: Forkcnbrock, David J. and Jason Sheeley. NCH.RP, Report 532, Effiaiw Mahods for Environmml41f~ Auasmnu. Copyright, National Academy of Sciences, Washington, DC, 2004. ReprodllCCd with pamission of the Transportation Research Board.

460 • MANIIAf nF TRAN<;Pr'\RTIITinM cw:,.ICCDI•Ir ,.,., ' " ' ' '

.4.0 REFERENCES i4.1 Literature References American Association of State Highway and Transportation Officials. Primu on Transportation and Clitn4te Changt:. Washington, DC. AASHTO, 2008. Code of Federal Regulations. Proudures for Abatnnmt ofHighway Traffic Noise and Constructwn Noise. Title 23 : Highways, Part 772. Washington, DC: U.S. Government Printing Office, 2009. Cohn, L. F. and G. R. McVoy. EnvironmmtalA114lysis ofTransport4tion Systmu. New York: Wiley, 1982. Crossett, J. et al. Synthais ofDatil Ntttis for EA and EJS Docummtation- A Blueprint for NEPA Docummt Con1mt. National Coc!perative Highway Racardt Program Projca 25-25, Task 1. Washington, DC: Transportation Research Board, 2005. Federal Highway A4rninistration. Highway TrajJic Noise Analyris andAbatnnt:nt Policy and GuidAnce. Washington, DC: F~deral Highway Adminisuation, 1995. Federal Highway Administration. Measuremmt ofHighway-&Iaud Noise. Washington, DC: Federal Highway Administration, 1996. Federal Highway Administration. Traffic Noist: Modt:l Vmion 2.5. Gainesville, FL: University of Florida for the Federal Highway Administration, 2004. Federal Highway Administration. Highway Traffic Noist: in tht: United St4tn-Problnn and Raponst:. Washington, DC: Federal Highway Administration, 2006. Fork,enbroclc, D.}. and J. Sheeley. National Coc!pcrative Highway Research Program Report 532: Effictiw Mt:thodsfor Environmmtai justice Assmmmt. Washington, DC: Transportation Research Board, 2004.

Frey, C. H., N. M. Roupluil, and Z. Haibo. ·speed- and Facility-Specific Emission Estimates for On-Road LightDuty Vehicles on the Basis of Real-World Speed Profiles.• Tramport4tion &search &cord: jour1141 ofthe Trllnsportlltion &search Board 1987 (2006): 128-137. Soil S~ey Staff. W(b Soil Survey, Washington, DC: Narural Resowces Conservation Service, U.S. Department of Agriculture. http://websoilsurvey.nrcs.usda.gov/app/HomePage.htm. Transportation Racardt Board. National Coope.rarive Highway Research Program Report 174: Highway Noise: .A Dtrign Guide for Prediction and Contrr~L Washington, DC: TRB, 19!6. Transportation Research Board. Potenti4llmpam ofClimate Change on U.S. Transportlllion. Special Report 290. Washington, DC: TRB. 2008. U.S. Department of Transportation. Environmmtill A$1emnmt Notebook Snits. Washington, DC: U.S. DOT, 1975. U.S. Environmental Protection Agency. Grrt:nhouse Gas EmissWns ftom the U.S. Transportiltion Washington, DC: O~ce ofTransportation and Air Quality. U.S. EPA, 2006.

S«fqr, 19!}()-2003.

U.S. Environmental Protection Agency. Prrftrredl&comrnt:nlkdMOtkls. Washington, DC: TechnologyTransfer NetWOrk. Support Center for Regulatory Aunospheric Modeling, U.S. EPA. 2009. epa.gov/S0'2tn001/dispersion_prefrcc.htm.

4.2 Online References AASHTO Practitioner's Handbook-Using the Transportation Planning Process to Support the NEPA Process: www.cnvironmenc.cransportation.org/pdf/programs/practitioncrs_handbooktO.pd£ California Environmental Protection Agency Air Resources Board-Modding Software: www.arb.~.gov/HTMU Soft. hem.

EPA MOVES (Mocor Vehicle Emission Simulator): www.epa.gov/otaq/models/moves. European Commission-Strategic Environmental Assessment: http://ec.europa.eu/environment!eio/ home.hcm. FHWA Environm~nral Guidebook: www.environment.fhwa.dor.gov/guidebooklindex.asp. FHWA Traffic Noise Model: hrrp://fuwa.dot.gov/environ mem/noise/mm/indc:x.htm. U.S. Environmenral Protection Agency-Preferred/Recommended Pollucant Dispersion Models. www.epa.gov/ scramOO 1/dispersion_prefrec.htm.

4.3 Other Resources. McDonald, P., D. Geraghty, I. Humphr~ and S. Farrell. "Assessing Environmenrallmpacc ofTransporc Noise with Wireless Sensor Networks." Transportation Rmarch Record: Journal ofthe Transportation Research Board 2058 (2008): 133-139. Transportation Research Board. "Environmertt and Energy 2008. • Transportation Research &cord: joumai ofthe Transportation &uarch Board2058 (2008). Transportation Research Board. NCHRP Report 466: Desk R.ef=nu for Esrirruzting the JnJimt Effict.s of Proposed Trrznspomuion Projects. Washington, DC: TRB, 2001. Transportation Research Board. NCHRP Report 532: Effictive Methods for Environmental ]U$tice Arsesmzmt. Washington, DC: TRB, 2004. Transportation Research Board. NCHRP Report 54 1: Consideration ofEnvironmmtaLFacton in Transportation Systems PlAnning. Washington, DC: TRB, 2005.

Zegras, P., D. Guruswamy and H. Rojas. *Transportation Modeling for Energy and Environment: U.S. Experience and Relevance to the Developing World." TrrznsportiJtion &sea~h & cord: ]ou1714i ofthe TransportiJtion Research Board 1487 (1995): 41-48.

462 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDrTION

Chap ter 22 ................... ........ .. ............... ....... ................................

Traffic Access and Impact Studies Original By: Dom1a C Nelso11, Ph.D., P.E. Edited By: Daniel]. Findley, P.E. 1.0 INTRODUGION 1.1 Purpose of Studies

464

1.2 Preparation and Design of Studies

464

1.3 Need for Studies

464

1.4 Study Timing

465

1.5 Study Components

466


3.0

4.0

463

473

2.1 Site Traffic Forecasts

473

2.2 Nonsite Traffic Forecasts

477

DATA REDUGION AND ANALYSIS

478

3.1 Total Traffic Estimates

479

3.2 Capacity

479

3.3 On-Site Circulation

480

3.4 Site Access and Off-Site Improvements

480

3.5 Transportation Demand Management

481

3.6 Residential Neighborhoods

481

SUMMARY

482

4.1 Presentation

482

5.0 REFERENCES

483

1.0 INTRODUCTION ypically, traffic~= and impact studies are conducted to assess the transportation impacts of proposed dtvc::J.opments and other land use changes. Studies might also be required for transportation infrastructure cbange5 • such a.s a new roadway or a road widening. The proposed development could be a new office building, subdiv.i.sion, &ccory, or shopping center. Proposed changes in land use might include the redevelopment of an existing are: ::a into an area that includes a mix of uses. Traffic impact studies project future transportation demands, assess the impac;;: 1: of changes in demand and suggest ways for mitigalling the adverse effects of land use changes in defined geognphi.:. areas. For these studies, mmportation dnnand is defined a.s the need for movement of people and goods by all f«ro.. .::S of transportation, including autos, car pools, transit, ta:ri, trucks and bicycles; and movement of pedesuians in th. &= vicinity of a proposed development-

T

Traffic Access and Impact Studies • 46~

The design and implementation of meaningful impact studies are complex processes. The results and recommendations of the studies are heavily dependent on the experience and knowledge of the persons conducting them, as well as those reviewing them. This chapter provides an oucline of the analyses required with an emphasis on the collection and organization of materials and data for traffic access and impact analyses. A complete study draws upon data collection and analysis procedures described in a number of chapters in this book as well as several other references. These include the most recent issue of Transportation Impact Analyse. for Site Dl!llewpmnit (ITE). Trip Gm"ation (ITE) and the Highway Capacity Manual (HCM) (TRB). Local and/or state guidelines must also be followed when conducting a traffic access and impact study. Computer-based methods are used for many of the analysis steps required for traffic access and impact analysis. For example, computer-based methods are typically used for level ofservice (LOS) and capacity analysis, as well as for most steps in trip distribution, traffic assignment and modal choice modeling. Available software packages will also estimate trip generation and parking generation for various land uses.

1.1 Purpose of Studies Traffic impact studies are conducted to evaluate the impacts of proposed land developments on an existing transportation network, and to assist professional staff in making decisions on the allowance or disallowance of major land use changes and new developments. These studies evaluate changes in traffic attributable to the proposed changes and translate these changes into transportation impaCtS in the vicinity of a development. Most studies also define the onsite and off-site transportation system improvements needed to accommodate the additional traffic generated by the development. The recommended road improvements and other transportation improvements are termed mitigation measum. Often traffic impact studies are conducted within the context of a larger mvironmmral impact rrport (EIR). State and federal law mandate these studies for all projeCtS conducted under the auspices of a public jurisdiction that may have significant impacts on the narural and human environment within a reasonable area surrounding the project. The traffic impact study is incorporated into an environmental impact statement (EIS) or an EIR. An EIS/EIR has two primary purposes: determining and disclosing all significant environmental impaCtS of a proposed project and identifying mitigation measures that could reduce or otherwise compensate for environmental disruptions.

1.2 Preparation and Design of Studies Site traffic access and impact studies should be prepared under the supervision of a qualified and experienced person who has specific training in traffic and transportation engineering with several years of experience related to preparing traffic studies for existing or proposed developments. Also, traffic access and impact study reviews should be conducted by properly trained transportacion engineers and/or transportation planners. All jurisdictions impacted by the proposed changes should also be offered an opportunity to review and comment on the study. Currently, some agencies require a registered professional engineer to sign and seal these studies. The road authority and any agency affectc·d by proposed changes should adopt an approval process and variance procedure to diminish future litigation concerns.

1.3 Need for Studies Warrants for traffic access and impact studies are addressed in a variety of ways. There is no general consensus on when a study should be done. State or local law may diccate when a traffic impact study is required. In the absence of legal guidance, the need for a traffic impact study is usually determined by professional judgment or accepted local practice. However, data collecced by ITE (ITE, 2005) indicates the need to conduct a traffic access/impact study, which is commonly determined by the following conditions: • when a new devdopment will generate (add) more than a specified number of peak-hour trips • when a development will generate more than a specified number of daily trips •· when more than a .specified amount of acreage is being rezoned

• when the devdopment contains more than a specified number of dwelling units or square foo!'a&e • when financial assessments arc required and the extent of impact must be determined • when the development will require a significant amount of transportation improvements • when a previous c.ransportation impact analysis for a site has been deemed out of dace • at the judgment or discretion of public agency staff • when the devdopmcnc is in a sensitive area

Transportaritm lmptUt Analyses for Sit~ Dnt~~pmmt (ITE, 201 O) recommends that in lieu of other locally establish~ thresholds, a traffic access/impact study should be conducted whenever a proposed devdopment will generate 100 or more added (new) peak direction trips to or from the site during the adjacent roadway's peak hours or the devdopment's peak hours. The ra'tionale supporting this recommendation is that: • 100 vehicles per hour are of a magnitude that C:an change the LOS of an c:xistin~ intersection approach.

• Left- or right-tum lanes may be needed co accommodate site traffic satisfactorily without adversdy affecting through (nonsitc) traffic. Virtually any m2jor cra.ffic generator (which may include approved or anticipated developments) must be considered as a potential candidate for traffic impact analysis. Examples include high-density residential areas, offices, retail/commercial hotels, business parks, hospitals/medical offices, schools, industrial facilities and scadiumsfcoliseums.

1A Study Timing Transportation needs should be a major consideration for new or expanding developments throughout the planning stages, including site sdcction. Detailed formal studies, however, may only be required ac specific development planning stages. Under normal circumstances there are several stages in the development process where traffic access/ impact studies are potentially appropriate. • Zoning and rezoning applications • Redevdopment • Land subdivision applications • Erivironmental assessment • Site plan approval • Building permit application • Formation of a special-purpose district • Development agreements • Amendmcn~ to comprehensive plans • Pennits for major driveways • Annexations

• Signal warrants • All-way scop warrants • · Consiruction staging

Separate srudies are not needed at each development stage. However, studies completed very early in the development process may ne.ed to be updated to include additional detail as the sire plans become specific. In some cases, the planning process will result in a substantial reformulation of che development program and plan, resulting in a need for re-analysis and re-examination of study findings and conclusions. The initial study should be reviewed at each phase of the development co ensure consistency with the current development plan or to indicate the need for additional srudy because of substantial changes in impact over those predicted initially. For scaged developments and projects, me original studies should be checked ar each stage and updated as appropriate.

1.5 Study Components The major components of traffic access/impact studies include: • definicion of the scope and extent of the study; • collection of data on existing conditions; • site traffic forecasts; • nonsite traffic forecasts; • traffic assignment a.nd mode split; • analysis; • recommendations; a.nd • funding for improvements. The precise components and level of derail of an individual srudy will vary depending on the size, the rype of land use and complexity of me multiple use development, the existing conditions of the local network and the requirements of the approving agencies. JS.I Scope ofStudy The first step in the process is to identify che issues and needs of the particular study. lr is critical to discuss the project with the reviewing agency's staff at an early stage in the planning process. The agency is a potential source of data for me scudy. The agency may be aware of projects approved for construction in the area, proposed changes in geommics at key intersections, future roadway additions and other factors that will a.ffeq traffic patterns and the requirements for the srudy. Consrruction impacts might need to be considered as part of the scope due to construction traffic; or a mixture of construction a.nd user traffic due co the development's consrruction staging plan and work.zone traffic control plan. Shore-term construction impacts can include drainage issues, storm water runoff, or sidewalk obstructions.

An 2pproved study methodology should be reached with the approving agency on the scope of study and appropriate assumptions for the analysis prior to the commencement of the study. Examples of issues that may be discussed with me reviewing agency include (ITE, 2010): • What components of a full site transportation impact srudy are needed to address issues associated with rhe site, proposed development and nearby transportation system? • How detailed. a.n analysis is needed for the trip generation forecast? Should standard equations and rates be used, or is a special study needed? Should modal split be considered? Should pass-by and/or captured traffic be analyzed? Is an internal!external analysis warranted and if so, how detailed should it be? • How large should the study area be? What is the area of influence of the project? · • Aif: traffic counts needed? Which da}'$ and hours should be countedl What growth rate is appropriate for background traffic?

• How should adjacent developments be considered in the study? How should areawide growth estimates and future traffic assignments should be used? Should existing zoning be used to generate future traffic for · undeveloped parcels? 466 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDmON

• How should planned or programmed transportation improvements be accounted for? • Should the various stages of multiphased development be analyud individually? Which horizo n yeaN should be used? • Which traffic dimibULion and assignment merhods sho.uld be used? How dccailed should traffic distribution and assignment be? • Which roadway sections ~d intersections and driveways should be analyzed? What proposed a·oadway and transit improvements should be considered? • Which capaciry analysis technique should be used? How many iterations of capaciry analysis should be performed? • To what extent wiU nonauromobile modes of travel, such as walking, bicycling and transit, be affected>\'(Till the site generate sufficient nonauromobile traffic to warrant off-site improvements? WiU the au tomobile traffic generated by the site adversely affect the level of service for nonautomobile mod6? • Are other analyses needed, such as crash, sight distance, weaving, gap and ~ueuing analyses? • What rypes of improvements should be considered? • How derailed should the recommendations be? How should improvement phasing and timing be addressed? • What are potential funding sources to implement the recommendations? 1.5.2 Stutly Horizons . Suggested srudy horizons are shown in Exhibit 22·1. Some agencies have specific requirements for study horizons; 0 dJe.rwise, the srudy horizons should be discussed and agreed co by the reviewing agency. Commonly, the target year of the idlpacr study is ar fuU build-our and occupancy of the project, or the horizon ~of the planning studies for a mcuopoli o;tft area. The latter may provide a greater database co aid in the evaluation; however, some menopoliWI plan horizon yeaiS have_not been extended fu enough into the future co encompass the major build-out of all components of the projeCt·

Large single-phase development (> 1,000 peak-hour trips)

Moderare or brge multiple-phase development

No.c-.: Peak-hour trips based on ITE Trip Gnuration. Source: Tnmspo114rion lmpattAMiystJfarSiu lkwlopmmt: an /TE hco11U11nUkd Pramc~. ITE. Table 3-1, page 15, 7010.


1.5.3 Study Amz Dauz The study should incorporate all uansponacion and land devdopmenc information char is considered current for the area, including aisting and prc:vious traffic studies in the area. Suggcsced background data are shown in Exhibit 222. As suggested by ITE (2010), specific data requirements vary according co the complexity and sco~ of the study. Studies will frequently include some or all of the following:

• peak-period turning movements for site and stcecc, and for surrounding major intersections and interchanges if pare of the project sco~; • vehicle classification counts; • adjustment faaors co rdatc count dara to design pe.riod; • machine counts to verify peaking characteristics;

pri~acy traffic control devices; • traffic signs and pavement markings;

• signal phasing and ciming; • roadway configurations, geometric features and lane usage; • sidewalk and crosswalk information; parking regulations; sucec lighting; • posted speeds; • driveways across from or adjacent co site; • transit stop locations; • adjacent land usc and zoning; and • pedestrian volumes on adjacent succts and crosswalks. The assembly and organization of available data should be accompanied by detailed reconnaissance (physical inventory).of the project sire, area roadways and the succounding vicinity. Inventory all relevant characteristics and information needed for the analysis, recording number of lanes, lane assignment, turn poem lengths, signal timing and observe existing traffic conditions while noting any queuing problems. Only those data needed to address both shortand long-term issues to be studied must be collected. Only current data should be used. In areas undergoing change, data should be less than 1 year old. In more stable areas, older data may be used provided studies are conduaed to verify current conditions are reflected. Any adjustment factors applied to survey data in the report must be described and justified. The report should include data representing conditions appropriate for the analysis, such as. average, design day, or seasonal peak traffic counts, and surveys factored to represent all members of a surveyed population. All procedures and factors should be summarized in the report. lmpaa studies may be subject to I~ challenge, therefore care must be taken to fully document the source of all data used and to explain the choice of data sources. Methods of presenting these data are d.i.scu.ssed in the following section.

i

I\

Traffic volumes

Land use

Dcmogr:ophics

Transportation S}'ltem

Other transportation data

I Crash history (3 years ifa:vaila.ble) adjacent to site and at nearby major intersection.! if hazardous condition has been identified

Source: TTrliUJortafiqn lmpactAnlliyusforSi~ Ikw/4pmmt: an ITE Rmlmmnuled Practiu. ITE. Table 3-2, page 17, 2010.

1.5.4 Stwly ArM JNforiiUnt Define the study arc:i to include all portions of the tnnsporation aerwork that may be affected by the proposed devdopment: Focus the analy$is on the segments of the surrounding transportation system when: users are likdy to perceive a change in the existing lcvd of service. Local guidelines can recommend a radius around the proposed project or until a specific level of impact is no longer perceivable &om the proposed dcvdopment. Roadway intersections :are often major areas of impact. Include all known major intersections that will be a.ffi:cted by the proposed development. & an example, the key intersections and project location an: dearly idcntilied for a srudy as shown in Exhibit 22-3.

The location and siu: of all approved projects in the vicinity of the project site should be shown and can be presented as in Exhibit 22-4.

~::~~~:~~-:~~~:~~~J~.~-:z.=~~~

ROUTE

~ • -·

NOTTO.c.A.L...8

.,..., c:,.._,.

eo

fWt

Pt:tl\.EY

....V. LeMA

•

8'1'1.CJT-·~·--

& ......... --.

maq

S'Tl.JDY AREA AND PAO.JEC'T LOCATlO N

Source: Mohle, Grovtr & Associates.

~__) + : ~ r---------------~

SMD,

~.....-.-·~·

-·---

ttt • Q K$F

£ !

(APP. P"AO.JECTG!

-· - P-~

lAPP. f'Ao.Jl!!OT <41 SWD'IIIH•63KSF"

l ~I"AA..JecT 8 . . ·-

1B tCSF• .. Cl.'1WNG

~

and

Ls;oetP

17 tCSr

f'RO'CCT

[Af'P:-~01 CAM'OiltJit PRo.I:CT 1IH •

(APP;..~e!

7!

.503 IC$F

maq

.

~1

SW0 • $JofCDt WJ.Itntt WH • WIJit£HOUS£, ttV W •

• •

WAN.FAC1\IItlloiG. KSF

• $l\JOY IOTIRSI'CliONI

APPROVED PROJECTS LOCATION

Source: Mohle, Grover & As5ociates.

470 • MANUAl OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITION

:)

~~ .II!

I I

I I~ .

olfmt

cma-

-~~'f., • 2 .. KS"

1.5.5 EXisting Transportation System :The existing transporration system in the area of influence should be described as shown in Exhibit 22-5. This e.xhibit ·should show the existing roadway system serving the sire, including all major strcers, minor meets uljacem co rhe si re 'and site boundaries. The exhibit can show all transit, bicycle and major pedestrian routes (if applicable), as wdl as right-of-way widths, signal locations and other traffic control devices (TCDs). For sires in developed areas, consider using several exhibirs or one per exhibit per mode.

~

I

..... •

N<71'TO...,...... Motllha.O..-- ..

....,. loUR.;. \.OMA

-+ ..............

~....

maq

$1UO"f' watrCTQNI

- - ......cllt01Uilu . , _ . . . . 14 .,aw ¥OIJUCI (IIOlH

.

.

Pmi...EY

~

EXISTING TRANSPORTATION SYSTE M

Source: Moble, Grover & .A$sociates.

1.5.6Selection ofAMlysis Periodr In general, che critical time period for a given project is directly associated with me peaking characteristics ofboch. projecc-rdated travd and areawide cransporration system. The peak period of the adjacent roadway and the generat:OC' under consideracion are equivalent in some situations and different in other situations. The peak period of the adja· cent roadway is typically =ined even if it docs not correspond to the peak p~riod of the generator because there can be significant impact from che additional vehicles co che craflic ~cam. The peaking charaetedscics of the adjaceo c street and highway system are determined by analyzing me traffic COUnt data for the area and projecting pe.ak dcmaod periods for che devdopmem. Typical peak traffic Bow hours for selected land uses are shown in Exhibit 22-6. In general, observed peak periods typically occur during weekday morning (7:00 to 9:00a.m.) and evening (4:00 t:O 6:00p.m.) hours, although local area characteristics may result in other peaks. Some land uses, such as schools an. d. entertainment facilities, often have schedules that do not conform to the "normal" 9;00-to-5:00 day; peak volumes for shopping centers and recreational uses may occur on weekends. If these land uses constitute a rdatively large propo r tion of the craflic genqated in the area, they may have a significant impact on the peaking charaeteristics. Additionally, peaking characteristics may change over time, especially in growi.ng areas. Trip generation and general craffic levds alsO vary daily and seasonally. Land uses such as shopping centers, banks and restaurants exhibit different daily pattern s For example, large shopping centers (over 400,000 square feet (sq. ft.]) should be analyzed for the period lxrwee~ Thanksgiving and ChristmaS, uadicionally the busiest shopping season of the year, if che maximum craflic demand iS desired for che study. Nearby schools should also be considered when choosing an atUlysis period, co include school traffic and avoid counrs during breaks. To establish peaking characteristics in che vicinity of the site, hourly bidirectional counts muse be obtained. The tim~ period(s) chat provides the highest cumulative directional traffic demands should be used co assess the impact of sire:_ traffic on the adjacent suect system and to ddi.ne the roadway configurations and aaffic control measure ~ needed in che study area. intersection volume data are collected during chis rime period. For the count data plotlJTraffic Access and Impact Studies • 471

in Exhibit 22-7, traffic demand peaks b~cween 5:00 and 7:00 p.m. for both directions. Turning should be conducted during this period.

movem~nt

counts

Fifteen-minute rum.ing mov~mrot counts should be obtained for intersections and driveways during the period. The peak hour can be identified more closely by selecting the four co~cutive 15-minute (min.) periods that contain the highest total counts by direction. Existing directiona.l volumes can be presented as shown in Exhibit 22-8. Orher methods for presenting turning movement volumes include tables, graphic intersection summary diagrams and intersection flow diagrams. Examples of these are given in Chapter 2. These data will be used for the LOS analysis for these locations.

Residential

Regional shopping center

Office

•Hours may vary bucd on local conditions. ~Period of maximum Wttkday traffic impact. Sowc.e: Trip Gmnwlion, 8th Etiititm. ITE, ZOOS.

_____ .. ... .,.._

300

ccunM11:t

I

.i j

250

200 1!!10

100

!50

0

--

------- " ........ -..,.,

12 t

rnc:aq

II

•

•

10 H

-

12 1

2

lJ

<4

II

Ill

..

7

HOUR aiDING

HOURLY TAAF'F1C VOLUME VARIATION FELSP.AR BTAEeT 8/0 OALENA S'l'l'W!E1

Sow= Mohle, Grover & As-sociates.

8

•

SO IS

-~/W

~

--~ .....................

• . -i..

CCUCT CM'IE: ~

-S1UOY

lf~S

~-1 COUNTS, 'IIOOCUS 1t~ GA.~ MMsoAY 31\.$ lQ «15 ,......

I1'1(Jq Source: Mohlc, Grover & ASsociates.


2.1 Site Traffic Forecasts The potential traffic impaa:s of a planned development are forecast for the projeaed conditions in the horizon yc:ar(s) of the project. The steps in the process include aip generation, modal split, trip distribution and traflic assignment. 2.1.1 Trip Gnurlllitm The trip generation process provides an estimate of the number of trips that will be generated due to the n~ development. Trip generation rates are applied to the various land uses within the development. It is usually not necessary to carry out the·c:nensivc: databased trip generation analysis described in,Chapter 20. The number of aips generated by the devc:lopment and approvc:d projects can be estimated in a variety of ways. These include:

• determining trip generation rates for similar developments in the area for the same time of day and multiplying the race per unit area on a proraca basis (FHWA, 1985); • using trip generation rates &om a similar area and land use; • obtaining trip generation rates and equations from ITE's publication Trip Gmmztion (TfE, 2008) or ocher accepted pubLiShed sources; • using the tc:chniques available on computer sofrware packages; and • using information coUccted from household tcavd surveys.

Published nati-onal rates oc local rates acceptable in the cc:sponsible jurisdiction may be used. Many state, ~onal and local agencies have: established their own aip generation database for sites within their boundaries. ContaCt the appropriate agencies to determine ifsuch a database exists and ifit is apprQpriau for the eutmlt stUdy. Narional sources can be used as starring points in estimating the amount of traffic that may be generated by a specific building or land use. Whenever possible, these national rates should be adjusted to rdiea local or forecasted conditions. These national sources should not be used without the application of sound judgment and a dear understanding of the stati.stical significance of the proposed rates. · Traffic Access and Impact Studies • 473

If existing local data samples are limited, collect additional local data to provide a credible sample size on which to base the trip generation estimate. Local trip generation data should be collected at sites that exhibit similar characteristics to the development being s!Udied and that are self-contained, with adequate parking not shared by other activities. Developments with Wlique demand or parking management schemes should be considered carefully for inclusion in the study. The following guidelines are suggested for collecting data on a similar site (ITE, 2005). • To obtain daily machine counts, select a generator where automatic counts can be made wirhour doublecounting turning vehicles and without counting through traffic. Directional counts should be in 15-min. periods. Daily counts should be made for a full24-hour period at the minimum, although a 48-hour period would be preferable. If feasible, a full7-day period should be counted.

• If peak traffic hours of the development are unknown, conduce automatic counts during a typical week of the year to provide data concerning the weekday and weekend peak hours. • For uses that do nor demonstrate substantial weekly or seasonal variations, select average days for the analysis. • For developments chat exhibit major seasonal variations, tksign days (approximating the 30~ highest hour) should be selected.

Location

• If only peak-hour data are needed, conduct manual counts for s~ hours on a typical weekday to record inbound and outbound vehicular traffic and compare these values with corresponding automatic counts at the same location co determine a counter factor for adjusting the raw automatic counts. • Vehicle classifications and occupancies may also be counted if relevant co the analysis. • Weekend counts may be needed to cover developments with peak activity on Saturdays or Soodays.

Independent variables

Other data

If there is reason to expect travel characteristics for the proposed development will be unique, manual counts or controlled interviews are needed. to determine average weekday person trip ends by mode, the number of trips actually generated by the site and the number of trips acrcacted to the site from traffic passing the site on the adjacent street. Information on sire/development characteristics of the survey generator may be obtained through an interview with the site owner or manager, telephone conversations, mail-back questionnaires and/or measurements as necessary. Obtain informacion on as many variables as pos· sible to determine which is the most closely correlated to trip generation. The data used to apply ITE trip generation rates are listed in Exhibit 22-9.


.2.1.2 Trip Distribution and Assignment ·.O nce estimated, the generl!.ted trips muse be distributed co geographic origins or destinations and assign ed to specific ~ections of the transporracion necwork. The origination of traffic co a site can be affected by the cype of d evelopment, area, competing developments, size of the proposed development, surrounding land uses, surrounding population and roadway conditions (ITE, 201 0). Three commonly accepted trip distribution mechods are ana/Qgy, gravity mvdels and surrogate data. The analogy method uses existing trip patterns from similar developments near the proposed sire. Traffic generate d at the project's entrances is distributed based on the directional split of current traffic on the roadways and nearby similar developments. At each intersection, project traffic is assigned in the appropriate proportion £O exis ting curning movemencs. Examples of the trip distribution for a development and nearby approved projects are shown in Exhibits 22-10 and 22-11.

~ •

sruov "'TERstenONs IIH80UNO

our aOUNO

Source: Mohle, Grover & Associates.


~

.....,_ ..

.. __ .. --our..,._

·~ ·

..._,_

~-~ · -~~ ....,.CTS l.OCA'ftOH (JQJM .,_5)

McJq Source: Mohlc, Grover & Associates.

Trip distribution models can also be applied to determine appropriate trip distributions. A common trip distribution modd is a gravity model which biases che number of trips on the distance from the source. Another type of model that can be employed for large-scale projectS is an areawide travel demand mode,!. Use of cravd demand modds is most appropriate when substantial changes in land use, transportation f.tcilities, or both are anticipated during the analysis period. Most modeling effortS are based in part on origin-
One method of estimating modal split is by analyzing existing modal splits for similar developments in similarly located sites with similar transit service levds. Local transit agencies can often provide historical data on transit use in the project area. Assuming a constant level of transit service into the future, these modal split data can chen be applied to the prospective trips generated by new development. Areawide models can be useful for projecting modal split for planned transit services. They can also project the effectS of transit's modal share on major changes in the overall transportation and land use patterns of a community. Only limited cia~ are available for estimating the percentage of trips made using modes ocher chan auto and transit. MPOs occasionally have survey.data on prevailing use of walking and cycling for some trip purposes. A survey should be

I

\ \

.47£ e

lA~MIIAI (')!=

TQAM<;Of'"\OTI\Tlf"'\M 1:1\U: IMCI:Illll'r

f'TIIr'ltr'r

'"'"~ "'"''~""'"

I ''

~

' '

. considered if it appears such modes account for more chan 5 percent of all trips generated. In the absence of informal cion, mathematical models are used co forecast the probability that a trip will include uansit. These models are often ·. described in terms of a disutility function that accounts for the generalized costs associared wirh modal choice.

2.2 Nonsite Traffic Forecasts ln addition to traffic generared by the site and approved projects, changes may occw outside the study area that will affect the transportation system in the horizon years. Estimares of nonsite traffic are required for a complete analysis of horizon-year conditions. · These estimates represent the "base" conditions: that is, before the site has been developed (or redeveloped). NonSite traffic consists of two components: (1) "through" traffic, consisting of all movements through the srudy area without an origin or destination in the srudy area; and (2} traffic generated by all other developments in the study a rea, with an origin or dmination in the st1.1dy area. Ptojecrs tha~ have been propo.scd and approved {but not yet built) ~y be considered in the analysis. Furure traffic demand estimates are developed by summing the conuibutions from the site, all approved (or potential) developments in i:he area and from current traffic volumes adjusred for general growth in the area. There are three principal methods of projecting off-sire traffic: build-up, use of area transportation plan or modeled volumes and trnuis or growth rates. Each method has its appropriate use and is based on the data thar may be available <>r generared as part of the site traffic access/impact study. The three methods are not murually exclusive; for instance, 'che build-up method can be used to update travel demand model forecasts or improve growth rate projections. The advantages and concerns of each technique are summarized in Exhibit 22-12.

Build-up

Transportation plan

Growth races

2.2.1 Build-up Method A popular technique for estimating the cumulative impacrs ofdevelopment projectS and area growth is to estimate the nip generation of approved and potential projects as described above. These trips are added to base craffic volumes to produce the cumulacive impact of known projects on traffic service levels. Trips a.Ueady made in the area street system should be accounted for to avoid double counting. As pan of the process, all transportation system changes (physical and operational) that are programmed, committed, or hlghly likely during the forecast or study period, according to local agencies, and projected probable changes in ccavd patterns are identified. 2.2.2 ~ 1'rtJwi.DenumJ Motkl Fflr«llltl

Most metropolitan areas maintain forecastS of fucure travel demand. These forecasts are based on computer-assisted models which reflect all officially anticipated land use and tranSportation network changes. Travel' demand model

forecasu cypically have only ~ few defined horizon years, which might not match up wi th the forecast years of the proposed development. Therefore, imerpolation techniques must be employed to determine the relevant volumes for the study. These forecasts can be used in a cumulative analysis, with the followi ng precautions: 1. The projecrion may be out of date and not reflect changes in local land usc or highway plans subsequent to the uaflic forccasr.

2. Regional traffic forecasts may only be available on a daily basis. If so, there is no straightforward way to derive the peak-hour volumes required for a detailed uaflic impact analysis.

3. The land use and transportation network components of travel demand models are frequendy too coarse in scale for the level of detail needed in traffic impact studies. 2.2..3 Trends or Grtnvth RAu Method Average growth rates can be used to project nonsite traffic. The growth r~te method is one of the simplest methods to use; however, it often resulu in inaccurate projections, as it assumes the patterns of growth rates in ua.ffic volumes will continue tluough the study target year, or will change predictably. Moreover, use of such unconstrained growth could easily result in unrealistic projected future uaflic levels. Growth in through traffic may be estimated by recent growth trends in traffic volumes in or near the study area, or from traffic volume projections used in che area transportation plan. If area transportation plan volumes are used, all site volumes associated with che development chat have already been projected must be subtracted. If the study year is not che same as the transportation plan projection year, Straightline or variable growth rates should be used to interpolate between eurtem volumes and transportation plan forecast year volumes. The growth rate selected should be justified in che study report. Tlte growth rates applied should be appropriate for the situations. For example, a 2-3 percent annual growth rate may be applied co the existing traffic in well-developed urban areas to reflect possible changes in land use. For a newly developing area, a 5-7 percent annual growth rare may be applied to reflect the pace and type of development raking place.

This method requires that recent development trends and population growth rates will continue ar approximately the same rare or at a rate char is predicable. If the study horizon is greater than I 0 years, or growth rates are expected to change, another method should be used. This method is the least accurate because growth is never U!liform tluoughout a community and its transportation necwork. Past traffic growth may not be indicative of future growth.

3.0 DATA REDUCTION AND ANALYSIS The·analyst can begin to assess the impacu of the proposed project once baseline data have been gathered and the key assumptions for trip generation, distribution and assignment and modal split have been determined. In addition to the analyses of rra.flic facility capa~ity, a number of other factors should be considered (ITE, 2010). These include:

• safery; • circulation patterns; • traffic conuol needs; • transit needs or impaets; • transportation demand management; • neighborhood impacts; • on-site adequacy and off-site parking facilities (if any are to be used for site generated parking); • pedestrian and bicycle movements; • service and delivery vehicle access; • driveway location and operation; and • air quality and no_jse impact. 478 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

The impact analysis and the development of improvement plans are conducted in an iterative process for each rime horiwn and key location. The intent is to show the relationsltip becween operations and geometry and to assess deli · . ciencies, as well as to identify alternatives for further consideration. Analyses should lx conducted for existing condi"tions, predcvelopment conditions and for conditions with the proposed project to gauge the incremental impactS of the project and the incremental needs it generates. k. stated in the introduction to this chapter, a comple te description of all the analyses required for a complete, comprehensive study is beyond the chapter's scope. Import ant topics are discussed briefly below. Air quality and noise impacts are covered in Chapter 21.

3.1 Total Traffic Estimates For each analysis period being studied, a projected total traffic volume mwr be established for each segment of (he roadway system being analyzed. These projected total traffic volumes (consisting of site and nonsite traffic) arc used to determine the ability of the transportation system to handle the increased demand. A capacity analysis should be performed for all key intersections and roadway sections. Projected cumulative traffic volumes can be shown as in Exhibit 22-13. .

~ •

-f•.•.

~

.

S"l\JOY

Oflr:RSI:CTlONS

TUAN~Ci WOW:..OfT \t'Q..lNES , \t:HK:ILS PER HOUR

OM.I\.A.n\11!: lltN'FlC • OOSTlHC

+

GlltOWTH FAC'I'OR

APPitOWO

+

~CTS

~C1

nv.rnc.

+

......... a - - .............

maq

TRAFFIC VOLUMES

HOUf'

Source: Mohlc, Giover & Associates.

3.2 Capacity Several methods, including the HCM (TRB, 2000), intersection capacity utilization and critical lane technique, c~ be wed ro assess the adequacy of key intersections to handle additional demand generated by the new developme$lt (lTE, 2008). Apply the same techniques for the existing situation and to the projected alternatives. For exampLe, projected volumes at signalized intersections should be analyzed wing the same capacity analysis techniques as tho.se used to characterize existing conditions. All unsignalized intersections should be analyzed wing the same unsignali~ d analysis technique, unless signaliz.ation is warranted and planned. Traffic signal warrants, such as those contained j.n the M1Z11wzi on Uniform Traffic Control Dmces (MtiTCD) (FHWA. 2009), may lx used to determine whether si~ nalization is desirable, particularly for the peak period. If the analyses indicate mitigation measures needed i.nduc3e improvementS co the geometries at the srudy intersections, improvements in LOS and intersection capacity utilizarico- n can be represented as shown in Exhibit 22-14. ' Traffic Access and Impact Studies • 47' 9

£)QS'J9.10 +OR.P'ACY'. +P'ROJ[CT I)Q'S"'MM

OECUitntcs

I

C\IWUIILAJJW TRATF'IC

ocu - o.eo l<>nAY • os - c I ocu ·-

---¥'

GALE NA ST. AND FELSPARST.

"'

' •1H flllOAD ~DY0o1011S

o.,. loo.Av •os -

•

~LIE_

~[llF.

GALENA ST.

AND BAOOKHOLLOW C1R.

GALENA ST.

AND RlJTLE ST. ~

LfCENO

_ _ _ _ , _ _ UIC

w

·---¥

*

,M::r. •

- -- ~ tllWUIC ('t.). _.......,.._ _ _. . . . , . ....,. ('8).

maq

- . r.wrr. •...,.,.. tAmOI -..&ll.A.liW ~ - .,..,... • - .

No.ccr + ,.,.,.,

_LI:If- PMW~Cr

(I)..

LEVEL OF SERVICE AND MITlGATION MEASURES

Source: Mohle, G!lmr & AssociateS.

If signalization is proposed for access to a development along existing anerial ro~dways, it is important to assess the impact of the new traffic demand on operation of the through street especiilly if a new signal will be pan of the sueet's progression plan. The HCM defines anerial roadway levels of service based on operating speeds and type of faciliry. The required analysis m:ty consist of :1. microsimul:~.cion :tnalysis to show that a proposed signal will, or will not, disrupt the platoon Bows operating along the :I.CtCrial.

3.3. On-Site Circulation The assessment of internal circulation illusuates the rd:~.tionship between external access points and building :1.ccess loacions, drop-off points, delivery points and parking loations. Internal circu.l;.tion should be efficient to minimize indirect travd while ut.iliz.ing the least unount of lmd space and generating the fewest conflicts with pedestrims and other vehicles. Locations of internal roadways, number of travd lmes md rum lmes on internal roadways, maneuverability in parking ;.reas md accessibility to buildings ue ill factors that must be considered in a successful project. Adcqu;.te puking or ;.ppropriate traffic and parking management should be provided co meet site-generated demands md be <:onsisrent with applicable local policies, which rn;.y be included in traffic dcrn;.nd mm;.gcmenc programs. To ensure adequ;.te parking is provided, a parking accumulation srudy should be conducted. Specific dimensions, parking mgles md parking ratio requirements arc addressed in a variety of documents md arc addressed in derail in Chapter 16 md the associated references.

3.4 Site Access.and Off-Site Improvements Access points ue intersections md should be designed with the same perspective as legs of any other intersection.

D.rivcway placement is critical to the overill function of the street necwork operations. Driveways of adjacent parcels should be combined when possible or spaced as far from each other to minim.i.z.c the traffic conBict points. Sjghc distance is also a key component when conducting driveway design. Study recommendations md conclusions should provide safe and efficient movement of traffic to and from, within md past, the proposed development, while minimizing the impact co nonsite trips. Sice access objectives arc co serve abutting properties, preserve roadway capacity, maintain efficient traffic Bow md maintain safety. Among the factors co be considered ue the following: 480 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

• site access for service vehicles should be examined based on size and operating characteristics; • adequacy of site driveways and the internal site circulation scheme must be studied; and • the design and location of driveways for the amount and type of traffic that will be using both the adjacent street and the driveway must be analyzed. Estimated driveway volumes, by turning movement, can be shown as in Exhibit 22-15 and used in the analys is of access points (driveways) and inrernaJ traffic circulation. ITE (2010) deals with topics related to site access and off-site improvements.

_B!_

'I ~ ...... .............

Ill

-.ttiT--

~ •

S'R.OV~S

..

.._

~ Source: Mohle, Grover & Associates.

3.5 Transportation Demand Management Site traffic can often be reduced or spread over a longer period through a variety of techniques. Each of these techniques has the potencial to reduce site uaffic for certain types of development under certain conditions. These techniques include increasing transit service to areas of trip origins, carpool and van pool prognm.s, reducing parking availability below normal demand or substantial increases in parking costs, signilicant uansit subsidies, creation of a high-quality pedestrian environment on-site, modified work schedules and a wide range of other techniques (ITE, 201 0). However, befo1e offering these techniques as acceptable mitigation, developer long-term commitments for the provision of these techniques must be considered by the study.

3.6 Residential Neighborhoods A variety of studies have found the safety, neighborhood amenities and overall livability oflocal or residential sueets can be severely degraded by rraffic volumes well before the physical capacities of such streets are reached. Many rrafficcalming tec:hniques can be used to improve the livability of a street. Consequently, if a project will add significant traffic to residential Streets, the volumes should be quantified. &sidmtilll Stmt Daign tmd Traffic Control(])~ et

Traffic Access and Impact Studies • 481

al., 1989) and Livable Streer.s (Appleyard, 1981) suggest guidelines for measuring such impaccs, and concain strategies and techniques for controlling the intrusion of nonlocal traffic into neighborhood streets.

4.0SUMMARY The purpose of a sire traffic access and impact study is to assess the effects a particular development will have on the surrounding transportation network, to determine what provisions are needed for safe and efficient site access and traffic Sow and to address related mobility issues. Other transportation objectives could include jurisdiction-specific goals such as the promotion of transit. The study report documen ts the purpose, procedures, assumptions, findings, conclusions and recommendations of the study. Common uses for these reporu include: • providing developers or designers with recommendations on site selection, sire transportation planning and traffic impacts; • assisting public agencies in reviewing the artribuces of proposed developmencs in conjunction with requescs for annexation, land subdivision, zoning changes, building permits, or other development review; • establishing or negotiating mitigation requirements where off·site impacts require improvemenrs beyond those otherwise needed; and • as a basis for levying impact fees or assessing developer contributions to roadway facility improvements.

4.1 Presentation The sample table of contentS below provides a framework for organizing the report in a maightforward and logical sequence (ITE, 2010). Some studies are easily documented using this oudine; however, additional sections may be warranted because of speci..Gc issues to be addressed, local study requirements and the results of the study. Inapplicable sections may also be deleted from the report. Chapter 3 and Appendix D provide useful information on communicar· ing data to the public and presentation techniques, respectively. The documentation for a traffic access and impact study should include at a minimum: 1. Study purpose and objectives; 2.

Description of the site and srudy area;

3. Existing conditions in the area of development;

4. Anticipated nearby development;

5.

Trip generation;

6. Trip distribution; 7. Modal split; 8. Traffic assignment resulting from the development; 9. Projected future traffic volumes; 10. Assessment of the change in roadway operating conditions resulting from the devdopmem traf6c; and 11. Recommendations for site access and cra.nspon:ation improvements needed to maintain traffic ll.ow to, from, within and past the site at an acceptable and safe LOS. The report should lead the reader step by seep through the various stages of the process'and to the resulting conclusions and recommendations. The report may have several different audiences. Sufficient technical detail must be included to allow the scalf of the reviewing agency to follow the path and methodology of the analysis. The report must also be


understandable to nomechnical decisionmakers a.n d interested citizens. The following suggestions are o ffered by ITE (2010). I.

Clearly document assumptions. Assumptions based on published sources should be specifically referenced. If less-available sources are used, a more derailed explanation may be necessary.

2. Avoid technical jargon (or ac a minimum, clearly define ic).

3. Whenever possible, pre5enc data in cables, graphs, maps and diagrams rather than narrative tex t. This enhances clarity and ease of review.

4. Discuss findings and recommendations with the reviewer prior to submitral of the final repor t . 5. Do not include political views or statements in the report; it should be an objective, technical analysis. 6.

Inadequate reports should be returned co the preparer by the reviewing agency for completion or modification as needed.

7. Include dectronic copies of intersection analyses if required by the reviewing agency. Refer to Chapter 3 and Appendix D for useful information on communicating data m the public and presentation techniques.

5.0 REFERENCES Appleyard, D. Livabk Struts. Berkeley, CA: University of California Press, 1981. Deakin, E., E C. BosselmiliUI, D. T. Smith Jr., W. S. Hom burger and B. Beukers. &sidmtia/Sm,r D'sign and Traffic Control. Englewood Cliffs, NJ: Institute ofTransportation Engineers and Prentice Hall, 1989. Federal Highw:~.y Administracion. D~eiiJpmmt andApplication ofTrip Gmeration Rates, HHP-22. W:tShington, DC: U.S. Department ofTransponation, Federal Highway Adminisuacion, 1985. Federal Highway Administration. Manual on Uniform Tmffic Control Devim. Washington, DC: Federal Highway Administration, 2009. h.stituce ofTranspottation Engineers. Gui&lints for Driveway Location and Design. Washington, DC: ITE, 1986. Institute ofTransportation Engineers. Trip Generation, 8th ed. Washington, DC: ITE, 2008. Institute ofTransportacion Engineers. Tmnsportation Impact Analymfor Site DeveiiJpmmf: An ITE Proposed Recommentied Practice. W:tShington, DC: ITE, 2010. Transportation Research Board. Highway Capacity Manual. W:tShingron, DC: TRB, 2000.


.

Appendix A

I

l • •••••••• • •••••••••••••••••••••••••••••••••••••••••••• • • • •••••••••• • • • ••••••••••••

'

Experiment Design OriginAl By: joseph E. Ihmmn, Ph.D., P.E. UptiAuJBy: &stilm]. Schroetln; Ph.J?.

1.0 INTRODUCTIO.N

485

2.0 GENERAL CONCEPTS

486

2.1 Definitions

486

2.2 Objectives

486

2.3 Statisticallnference

486

2.4 Random Assignment

487

3.0 SIMPLE COMPARISONS

..

3.1 Unpaired Comparisons 3.2 Paired Comparisons

4.0 BEFORE-AND-AFTER EXPERIMENTS

5.0

487

487 488 489

4.1 Drawbacks to Before-and-After Experiments

489

4.2 Overcoming Before-and-After Drawbacks

490

4.3 Analyzing a Before-and-After Experiment

492

4.4 Before and After With Control Experiments

492

4.5 Before and After with Comparison Experiments

494

FACTORIAl DESIGNS

6.0 REFERENCES

494 ' 498

1.0 INTRODUCTION xpe.riments are comparisons between two or more conditions chat are manipulated by. or arc under the control of. the experimenter. Experimena are conducted in a synematic and scientific manner so infuena::s can be drawn about g~cral populations from measuremena taken from samples of the population. For example, one could consider a seria of twO spot-speed studies to determine the effects of a change in the speed limit on the 85th percencile speed to be an experiment. However, a spot speed study conducted to dctennine the 85th pcrcencile vehicle speed on a highway would not be an experiment since the analyst makes no comparison and docs not manipulate the conditions during the study.

E

Expcrimena are one of the major means by which transportation engineers gain an understanding of the transportation sptcms they design and operate. Experiments arc also one of the major uses of the data collection methods described in this !IWlual. However, a poorly designed experiment can provide invalid rcsula (that is, results from which no inferences ourside the entities acrually measured can be made), or may mili ine.fficient use of data collection resources. This appendix provides a brief overview of the fa.cers of experiment design that are wed _m~t often by

A ............ ..a: .. A

-

AO~

transportation engineers, so these undesirable outcomes may be avoided. General concepts and terms are presented first, followed by discussions of several types of experiments. Experiment design reaches far beyond the relatively simple concepts provided in this appendix. Readers seeking details on more advanced topics are encouraged to consult textS such as Cochran :u~d Cox (1957), Anderson and Mclean (1974), Hicks (1982) and Montgomery ( 1984). The advice of professional statisticians is essential for the design of advanced experimentS with significant costs.

2.0 GENERAL CONCEPTS 2.1 Definitions Statisticians use several important terms in discussions of experiment design. Units or subjuts are Lhe entities that are selected for participation in an experiment, mtasum oft/fictivmm are rhe traits that are measured during an experiment, focron are the variables being manipulated in an experiment and lrotls are the particular states or conditions of factors. ln the spot-speed experiment mentioned above, the unitS were the vehicles, the measure of effectivesness (MOE) was vehicle speed, the factor was the speed limit and the levels were the values of the old and new speed limits. Other terms important during discussions of experiment design include trratmmts and replications. A treatment is a particular combination oflevels of various fuctors. For example, an experiment on observance of traffic control devices (TCDs} had two fuctors (sign type and sign size), each at two levels (sign types were "stop" and "yield~ and sign sizes were small and large). Thus !here were four treatments: small stop sign, small yield sign, large stop and large yield sign. The number of repucations is Lhe number of unitS to which each treatment is applied. In the aperiment on observance ofTCDs, ifthere were 28 different intersection approaches (units) tested, and the experimenter applied the trbatments evenly, seven replications were made.

2.2 Objectives The process of experiment design begins with recognition of a problem. Next, the engineer muse determine whether an experiment is needed ro solve the problem. This determinacion is often made with a vety rough estimare in mind of the resources required to conduct an experiment, and the potential benefitS to be gained. Better estimates of the resource requirements of an experiment wiU be available after the design is complete, which will guide the decision of whether to proceed. In some cases, modifications of the experimental design may be considered at this time to condense the data collection effon co the most pertinent data d ements. Oftentimes, there are a number of variables of potential inrerest, that may be rank-ordered based on priority and need of the particular element for analysis purposes. The central tasks of experiment design are to define the MOE(s), list the units to be studied and determine a treatment for each unit. The experiment design is complete when an analysis plan is developed.

2.3 St at istica l Inference One other important task of the experiment designer is co specify the inftrmC( space of the experiment. Experiment results should apply ourside the narrow range of the particular unirs measured. "How widely do the results apply?" is the important question that fuces the designer. The answer depends primarily on the number of factors and levels studied and ~e population of units from which a sample is drawn. If the ocperiment draws units from a vase population and includes a wide ~ge of fuctors and levels, a large inference space is likely. Engineers often use the results of an experiment to accept or reject a hypothesis about the relative effectS of the treatments applied. Typiq.lly, rwo hypotheses are constructed: (1) that the mean value of the MOE associated with one treatment is not different from the mean value associated with the other treatment, and (2) that the mean values are different. Statisticians call the former the null hypothesis. Rejecting the null hypothesis when it should be accepted (that is, concluding there was a difference when there really was not) is called a typ( I nror. Confoimu kwls, such as the "95 percent confidence level" used so often, refer to the probabiliry that a type I error will not be made. Experimenters make type II nron when they accept a null hypothesis that should have been rejected (that is, conclude there was not a difference when there really was). Mosr experimenters estimate sample sizes (that is, the number of replications) and 486 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

. construct confidence intervals based on a low probability of type I errors, because those are usually more critical. Most '; r::ypc If errors merely preserve the status quo. Sample-size formulas are available for experiment design t hac mini{l'li ze :borh rype I and rype 11 errors. More discussion is offered in Appendix C, and specific sample size examples were given 'in Chapter ).

2.4 Random Assignment The experiment designee must address several potential problems, including the presence of factors that affect the MOE bur are not accounted for in the experiment, results that are unrelated to the project objectives, treatments f~r one unit that affect the responses of other units (that is, treatments are not independent) and ochers. Many expe:Jmenc designs fail due to nonrandom assignment of tceatments co units. Nonrandom assignment of treatments 10 Ufllts is sometimes difficult co overcome and could lead to biased results in a number of ways. Nonrandom assignmeflC of treatments to units might occur, for example, in an experiment to evaluate the elfecr of an educational progmm for persons convicted of reckless driving on future collision rates. Jf rhe program is offered on a voluryrary basis during the experiment, the volunteers may have some unmeasured characteristic, such as :uucious spouses, that distinguish chem from the general population of reckless drivers. That characteristic may influence the experiment results; a lower future collision rate may be due to the anxious spouses rather chan the educational program. Most experimenters usin_g nonrandom procedures to assign treatments to units acknowledge a lower degree of confidence in their cesulcs chan tf units arc chosen for treatment at random. A random number table or computer program that generates random numbers is useful in ensuring uni ts to be gi-"en a particular treatment are randomly selected. For example, if the experiment designer is to select a sample of 30 buses from a fleet of)OO, she could assign each bus a number from 001 to 500. Then she could list the first 30 unique threedigit numbers between 001 and 5QO in a column of random numbers. The buses corresponding to the numbers on this list would be treated in the experlment. The requiremenc chat the nuntbers be unique is necessary to enmre the sairie bus is not sampled cwic.e. This requirement does not unduly influence the results from samples of large populations.

3.0 SIMPLE COMPARISONS 3.1 Unpaired Comparisons The simplest ex.pcrimenc designs involve the comparison of two levds of a single factor, or cwo treatments. The 5peed limit example mentioned earlier is an example of chis simple experiment. This design requires cwo random samples from the population of units. The experimenter applies one treatment co one sample and the other treatrncm (or ~ 0 aeacment) to .the other sample. With large samples of more than 30 replications a two-sided t-test is an appropria"C:e analysis meth.od for comparing the mean values of the samples. To Use the Hest on small samples, analySts mu:st ~ sucne the distribution of the MOE value in the population is nocmal (bell-shaped). If char assumption is poor, t:Pe r-eese outcome may be misleading and nonparametric tests should be applied. With larger samples, no such noan~ icy assumption is necessary to compare mean values. T-tests, nonparametric tests and other useful statistical an.tysis methods are described in Appendix C. Experimental comparisons between more than two levels of one factor are common. Such comparisons are dt:sigPed in a very similar way tp comparisons between two treatments described above. One-way analysis of variance (ANOV'_.A.-) is a convenient and uSually valid method for comparing more than two levels of one factor that is described in mos-t: standard statistical reference books such as Washington et al. (2003). ANOVA is valid only if the data meet cmai~ conditions. Forrunacely, tests are available to see whether a data set meets those conditions. The resultS of an ANOV' .A will show whether there are significant differences in the MOEs among the levels of the factor tested. A means ~es-.:=' such as Ncwman-l
Appendix A • 48-;;?'

3.2 Paired Comparisons One common vuiation on the simple experiment design described above for comparing two treatments is called a paired compariJon. Paired comparisons are made by bloclcing, which is forming groups of units that have similar characteristics. After the blocks are formed, the experimenter chooses two units from e:~.ch bl.ock for srudy. One unit receives one treatment, while the second unit rcccives the second treatment. As usual, units must respond independently to tre:~.tments. Experimenters malyu: the differences in MOEs for each pair using the t-tesr. Usc of paired comparisons or blocking helps filter out variations among units, allowing a more focused analysis of the effect of the treatment itself (Bhartacharyya. and Johnson, 1977).

An example of m effective paired comparison design is in testing the a.ccuracy of traffic conflict data collectors after training. A newly trained collector md a teacher together spend one-hill hour at each of 15 different intersection approaches, and indcpcndendy estimate the number of confficrs observed. Each intersection approach represents a block generating a pair of measurements: one for the student and one for the teacher. The hypothesis tested is that there is no difference between the srudent's and the teacher's estimated numbers of conflicts at approaches. Exhibit A-1 provides sample data and calculations that show the hypothesis is accepted. The experiment design is effective because the variation from different intersections is removed. By contrast, it would be possible with random matching of treatments and units foe the student to obsc.rve more higher-volume intersections than the teacher.

To test null hypothesis that Averaged" 0, computet I =

A.vg(d)

StdDev

rn

= 0.133 = 0.08 6.413

.Jl5

For 14 degrees off=dom,I(0.025levd) 2.145. Since 1 computed i.s ~than c(0.0251evcl) ac.ccpt null hypothesis ar 95% con6dence levd. 5

a •

t • •• I l

I

,..#"

..,.~

•

l>tl'"r.f"'\nT•Ttl"'\•t ,.,,,-.,,,.f"l"\ta ,,- ,....,.., tnlf"f'"

~t.lf"'oo

l"'t"\ITII'''\~1

4.0 BEFORE-AND-AFTER EXPERIMENTS Th~ most common form of experiment in transportation engineering. the before-and-after tXperiment, is a sped al

typ~ of paired comparison. In a before-and-after experiment, measurements are taken at the same location r:wice: on ce

before a change and once afterward. Before-and-after conditions are sometimes also referred to as "pre" and "post" conditions. The treatments in a before-and-after experiment are "not changed" and "changed," the units are points in time and space (that is, a location at a particular rime), and the blocks are the locations. Before-and-after experiments are attractive for scatistical and practical reasons. In some cases, che unh can also be a human subject who participates in the experiment during boch treatment conditions. Statistically, a before-and-after tXperiment alto~ a paired comparison to be performed, removing from consideration variation between locations. Practically, before-and-after experiments can be condu!=ted during improvement programs and require measurements at fewer locations than ocher experiments. Before-and-after experiments are easily understood by engineers and nontechnical readers and make intuitive sense. .

.

4.1 Drawbacks to Before-and-After Experiments Engineers thinking of conducting a before-and-after experiment must consider seven serious drawbacks of that exp eci-· ment type. 1. A before--and-after experiment may require a longer rime between che decision to conduct an experiment and the achievement of a conclusion than ocher types of tXperiments. The engineer conducting a beforeand-after tXperiment muse wait through both the before and the after periods. 2. Before-and-after experiments are ver:y difficult to design while uearrnents are being implemented or after treatments have been implemented. It is difficult to obtain data for the before period later (or "post hoc") ftom routine sources. 3. Units may not react instantaneously to a treatment. Drivers encountering a TCD soon after its placement, for example, may not know how to react and may exrubit unusual behaviors thar bias any experiment data being collected. If the experimenter allo~ a few weeks to elapse before collecting data, the novdry effect . may be much smaller. 4. Units may react to the uearrnenr in an urtstable or random f.uhion (termed instability). 5. History (see discussion below) 6. Maturation (see discussion below) 7. Regression to the mean (see discussion below) People often confuse the fifth and sinh serious drawbacks to before-and-after experiments. The 6fth, history, refers to changes in MOE values through the before and after periods ca.md by faaon oth" than tht trtatmmt (Council er al., 1980). For instance, an experiment measuring che number of &tal collisions on rural freeways in the United States through the 1980s could be affected by the change of speed limit on many of those roads in 1987 and 1988. .Experiments that compare a before measurement ftom one season of the year to an after measurement ftom another season are likdy to suffer from a history bias. Mizlunztion, the sinh drawback. refers to trends in MOE values with time (Council et al., 1980). For instance, &tal collision cares have been falling in many devdoped countries for years due to a number of f.lctors. That trend can affect experiments measuring those rates. Statisticians call the seventh drawback to before-and-after aperiments regression to the mean (Council et a!., 1980). RegrwUm to tht TM1111 refers to the tendency for a Huccu.ating characteristic of an entity to return to a typical value in the time period after an c:nraordinary value has been observed. Engineers have observed this tendency in many databases. F.xh.ibir A-2 shows regression to the mean in a typical collision database. The ex!Ubit groups intersections by the number of collisions observed in the first year, ranging from uro to 10 or more. It then computes the average number of collisioiU in the second year of analysis. Note that i:n the second year all collision groups wicb one or more collisions in year 1 regressed toward the overall mean of approximately 0.7 collisions per section in year 2. On the conuary, the group with uro collisions in year 1 showed an apparent increase in collisions. At first glance, che coll~ion reduction. Appendix A • 48'

... ::...

...,,,, ~-~~.~-- -~,!:~\1~::~ _J'i~---..~u_:·:# -:i~~?;·;~~~~i?.!!:r#f.~~~~~t

:~:~,~~ll''~l>"''?jf'1Pf!~1f€),'-;;-~,-,..,•~~·:&l'f;l

I•~~

~~~-·-~; __• •:.. ~-

· ··- ~

CoUisions per Section'

Number of Sections in Group

Fiest Year

12,859 4,457 1,884 791

_,.o..... o. .-. •. •

Second Year

Change {o/o)

0

OA04

Increase

I

0,832

-16.8

2

1.301

-35.0

3

L841

-38.6

I

374

4

2.361

-41.0

160

5

3.206

-35.9

95

6

3.695

-38.4

62

7

4.968

-29.0

33

8

4.818

·39.8

14

9

6.930

-23.0

33

~~~

10.390

-22.0

'9venl.l first. and second-yeu average collisions per section is approximately 0.7. \Average =13.33. Source: l{auer and Persaud, 1982.

for most intersections may lead the analyse to claim a reduction in collisions across the network, when in fact the overall mean number of collisions per inte.rsection stayed largely unchanged. Regression to the mean affectS before-and-after aperimencs whenever experimental units are chosen on the basis of a high or low MOE. For instance, treating the highesc-coUision locations in a jurisdiction with a program ofintense traffic law enforcement and evaluating the results on collision rates with a before-and-after experiment is a classic candidate for regression to the mean bias. The high-collision locations would probably have experienced a decline in collisions in the after period anyway, regardless of the treaanem.

4.2 Overcoming Before-and-After Drawbacks Some engineers doubt that a before-and-after experiment is ever valid, because so many previous efforts have suffered from one or more of the drawbacks described above. This is an unfortunate perception because a before-and-after experiment does produce valid results ~onomi~y Wlder certain circumstances if the drawbacks are considered and (if possible) treated. In chis section we describe ways to overcome the dnwbacks described previously. If a serious drawback js identified that cannot be treated, another experiment type should be used. The first dnwback listed above, that before-and-after experiments require more time than other experiments, is not easy to overcome. Often, analyses and their sponsors must accept the longer time required for the before-and-after experiment. Ifthc longer rime cannot be tolerated but an analyst still wants to perform a before-and-after experiment, he or she can change MOEs and use a surrogate measure. For example, if an analyse cannor wait years for the outcome of a before-and-after study using collisions, traffic cooJlicu (see Chapter 18) may be an acceptable alternative that will deliver results within months. There is also no simple way to overcome the difficulty in consuucting pose-hoc c:xperiments. Good recordkeepiog and inv~otories may help an analyst assemble data on before periods during or after the after periods. However, even the best records are likely to be missing key elements occasionally. and an analyse cannot simply measure a missing data element in the field in a post-hoc analysis. The decision to conduct a before-and-after experiment should be made prior to the before period.

490 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDffiON

The third drawback co before-and-after designs is of concern when adaptation co the trearmenc by rhe unit.s is nor instantaneous. In chis situation, experimenters use wnnn-up periods to allow the unit to adapt. For eJCa.mple, expcr~ menters should wait for one to several months after installing a TCD to coll~cr after data on motorist reaction. Etpe n·. menrers should also take care not to unwittingly use data from the period when the treatmen t was being constructed. Experimenters interested in the "novelty" effect of a rrc::aunent do not use warm-up periods. Designers overcome the fourth drawback, inmbility, by using a sample of sufficient size and by using stariscical af\alysis methods to draw conclusions from the data. The main point co remember regarding instability is a count of before events is a random variable, rather than a constant, and must therefore be treated with statistical techniques. Designers can deal with history and maruration, the fili:h and sixth drawbacks to a before-and-after experiment. by shortening the rime span of the srudy. Designers changing from 3-year before and after periods co !-month periods greatly reduce the chance of large history or maturation biases. Again, it is helpful to use a surrogate measure that can be collected quickly. Experimenters also artempt to overcome history and maturation by literature revi ews that identify and quantify biases. Manual adjustments for possible history and maturation biases rely heavily on judgment and are often open to ~eba~e, however. . 4.2.1 Overcoming &gression to the Mean · · Regression to the mean is the last of the serious drawbacks to before-and-after experiments. Experimente.rs can avoid it while selecting units for ueaunent or can overcome it during anal~is. During unit selection, experimenters can avoid regression to the mean by randomly choosing uni.ts for treaunenr from the entire list of units. For instance, in eva}t.lating the effectS of a program of school woe speed enforcement, every school zone in the srudy area must be a candidate for enforcement and data collection. T he major problem with this approach is the units most in need of treaun.em (that is, school wnes with higher speeds) are not necessarily chosen for treaunent. The experiment results do noc sO. ow. whether enforcement in higher-speed w oes is effective. Similarly, a municipality wanting to test a new intersection safery treatment would have to randpmly select treatment and control sites, which may leave the worst inrersecrion U!lueated while the experiment is ongoing.

Experimenters can overcome regression to the mean bias during anal~i3 if the experimental data are Poisson-disccibuted. D ata are Poisson-distributed if they meet certain conditions (Steen, 1982). L

They are suucrured as a number of events that happen during a particular interval of time (or space or other dimension).

2. The underlying race of event occurrence does not change through the time studied. 3. The occurrence or nonoccurrence of an event in one time segment is unrelated to the probability of eveP t occurrence in subsequent time segments. 4. More than one occurrence in a very short time is unlikelf: Traffic collision and trafli.c conHicr are examples of data that are usually Poisson-distributed. For data that are Poisson-distributed, Hauer and Persaud (1982) provide a useful method to adjust the nwnber of events experienced in the before period for regression to the mean. The expected number of after events at a locac.i on that had It events before is estimated by the number of events that occucred in the before period at locations with k + 1 events divided by the number of locations with It events in the before period. For example, consider in Exhibit A-- 2 the 1,884 roadway ~ons with two collisions each in the before period. An estimate of collisions in the after period adjusted for regression to the mean is (3 x 791)/1,884 c 1.3 collisions per section. If an experimental treatment h~d been applied to these sections and records had indicated an average of 1.5 collisions per section in the after period.. a naive anal~t would have concluded the treatment had been effective in reducing collisions when actually the trearme:t:'lt had probably been harmfuL The major problems with this adjustment for regression to the ~ean bias are that it only applies co Poisson-distribure::d data and that it requires a large database. Hauer and Persaud (1982) analyzed a collision data set from 82 sices 3.0- d reponed estimation using the adjusunent method above was very poor. However, when the data were examined ~ mulacively (~flat is, the anal~ts c:xamined sections with four or more collisions together instead of analyzing scccio~ with four collisions separately from sections with five collisions, six collisions, etc.), the adjustment m~thod produce::' a more accurate results than did unadjusted before-and-after comparisons. Appendix A • 49>

1

Hauer and Lovell (1986) have demonstrated another procedure for removing regression co che mean bias during che an~ysis of before-and-afte.r experiments. The proccdu.re involves rdativdy advanced scarislical techniques chac are beyond the scope of chis appendix but should be less data intensive chan the adj ustment described above.

4.3 Analyzing a Before-and-After Experiment Experimenters analyze a before-and-after design 100: a standard paired compa.rison test if the data are normally distributed and mean ~ues are of interest. This is the case in many experiments, such as studies of uavd times due co revised sign~ timing plans, speeds in response co enforcement programs and transit vehicle ridersh,ip in response to fare changes. The difference between the mean ~ue of the MOE in the before period and the mean ~ue of the MOE in the after period is an~yzed using statistical tests (see Appendix C). The analyse assumes the difference observed in MOEs between the periods is due entirely to the treatment unless she adjusts for history, maturation, or other known biases. For before-and-after experiments with collision, traffic conflict, or other data that arc Poisson-distributed, the usual analysis methods are inappropriate. Before-and-after experiment data that are Poisson-distributed should be an~yz.ed using the modifoJ biMmial ust. Exhibit A-3 shows criteria for rejecting the null hypothesis (that the before and after rates are the same) based on the modified binomi~ test if the data arc Poisson-distributed. The c:xh.ibit should not be used for before-after analysis of normally distributed data such as speeds. An example use of Exhibit A-3 follows. Assuming observers noted 14 traffic confliCts at an ioce.rseccion controlled only by stop signs during a typical weekday. Flashing red sign~s were then installed to supplement the stop signs. After a .sufficient time dapsed for driver ac· dirnation to the signals, observers recorded seven traffic conHiccs on a typical weekday. Were che signals effective in reducing traffic confliCts at the 95 percent confidence level? 100

··-__ -

10

... .. -. . ........,.,__ I

I

~ ,..... .,.,. .. ._. ...,....,..., _prillo_lho_llftll.. _

10

,. \

...... _ _ bf!lw

f--

\

-- ,,\r\'

Solution: Exhibit A-3 shows that for 14 before events an analyst needs a 60 percent reduction to reject the null hypothesis. Since a 50 percent reduction was experien ced (chat is, (7/14) x 100 percent) the an~ysc cannot reject the null hypothesis at the 95 percent levd.

10

d ltnot ..

._.

10I

40I

• I

4.4 Before and After with Control Experiments

/'; ' - -

'r-·

i\

~~ ·

I' ~

Ill

~ t-..

-

•,

o . m •

10

--

10

*

m

~

~

* *

Source: Weed, R. M., "Rmscd Dcl::i$ion Critcm for Bcfo~-and-.Afur Analyses," Transportalitm hmmh hcorrll068, Transporcacion Research &ard, National Research Council, Wa.sbingron, DC, 1986, p. 1.1.

Before-and-after with control is the strongest experiment design commonly usd in cransporcacion engineering. Before-and-a.ficr with control experiments overcome three of the major drawbacks of simple before-andafter experiments: history, maturation and regression to the mean. With a soun~ scatistical design to counter instability and an adequate warm-up period to let units -adjusc to the treatment, this aperimenr design is difficult to discredit. Experimenters sdea control units a.r random from the population of units when treatment units are selected. Experimenters

l,llea5Ur~

control units during th~ b~forc and after periods like ueatment units, but do not alter control units dwing

t~e experiment.

Exhibit A-4 illustrates the control unit concept. Mean values from a before-and-after with control experiment could be tested with a statistical Hest. A typical null hypothesis is that the mean difference berwecn the experimental units b~fore and after treatment is the same as the mean difference betW:een the control units in the before and after periods.

r

Con..ructl edjustm . on a nd

•ntperloci

Legend

Control ehe(8)

-- ....

Control atteta•

~I

....-

_ .J _

.... -----

l"rojectalte

..,.,,. •MOE -

Mea~uro

•

AWJ. proj..:t lllte MO!•

A 0

A\ig . G<>ntrol alta MOE

E>cp~ed MOE

--. e"~~:J Actual

"11m• Ab<

of Etfectlvenesa

Source: FHWA, 1981.

Before-and-after with control c:xperim~nts are strong but require more resources than do other e:xperimem types. Like a simple b~fore-and-after experiment, the before-and-after with control experiment may require a rclativdy long time for data collection, and the randomly chose.n units may be widely dispersed. However, With control units the amount of data collection will increase ovc.r a simple before-and-after c:xpcriment. In addition, it is essential before-and-after with control experiment be set up early, since control units must be chosen randomly from all possible unics.

a

Some peopl~ believe a before-and-after with control experiment is nqt feasible wh~ the treatment is to be implemented across an entire jurisdiction or agency. However, Council ct al. (1980) suggest an experimenter can take advantage of the fact that implementation of a treatment is not instantaneous, due ro budgeting and personnd constraints, to establish control units. Suppose guardrails arc to be improved over a 2-year period on all freeways in a jurisdiction. An. experimenter may be able to select at random some freeway sections to be improved in the first year and some to be improved in the second year. Then the sections selected for second-year improvement can serve as control units during the first year. The key to this plan is random control and treatment unit sdection. If the project engineer does not permit random selecti;on, no true •control" is established and a bias IIUY affect the results. A ~fore-and-after with control experiment may be politically unpopular and open to liability questions if high-collision or high-priority unics arc untreated. If an. experimenter is faced with this criticism, Council eta!. (1980) suggc.u that the treatment is still unproven (that is why an experiment is needed), so withholding it is not necessarily harmful. In addition, the experimenter may promote the idea of an experiment with a promise chat high-priority units not treated during the experiment will be treated immediatdy after the final set of experiment measwements, regacdfcss of the experiment outcome.

.cl.nn.. nni• I:J. • 4!13

4.5 Before and After with Comparison Experiments Before-and-after with comparison experiments are a compromise between the drawbacks of the simple before-andafter experiment and the rigid rules of the before-and-a.frer with control experiment. As with control units, experimenters measure comparison WlitS during rhe before and after periods and do nor alter oomparison unit.s during the experiment. However, experimenters select comparison units deliberatdy rather than at random. This is an important difference. Comparison units may help overcome the history, maturation and regression tO the mean drawbacks of the simple before-and-after experiment. The abiliry of comparison units to overcome these drawbacks depends on how closely they compare to experimental units, and proving mathematically the drawbacks are alleviated is impossible. To alleviate history biases, the same events must affect the experimental and comparison units. To alleviate maturation biases, the same trends must affect both. To alleviate regression to the mean, the experimental and comparison units must have the same group mean of the characteristic being measured in the before period and overall. This is the most difficult constraint on comparison units and means there is little chance of finding units to compare to a set of high-prioriry units scheduled for treatment. An experimenter with a large database can employ an adjustment method such as Hauer's and Persaud's (presented earlier for simple before-and-after experiments) to estimate the direction and magnitude of regression to the mean. However, most experimenters simply tty to choose comparison units similar enough to experimental units so readers and sponsors believe the regression effects are similar. Like the other before-and-after experiment designs, a before-and-after with comparison experime~t requires a relativdy long time undl results are available, a warm-up period and a statistical procedure to account for instability. Comparison units mean more data collection effort than for a simple before-and-after experiment. Comparison units should not be as scattered as control units. · Locating enough high-qualiry comparison units is a challenging task in many experiments. The number of comparison units needed is estimated from standard stadstical sample-size formulas. A common method for choosing comparison units is to choose :t unit for treatment and then use the unit next to it for comparison. For example, in evaluating a newTCD at minor inrcrsectiow, an experimenter randomlysdeas the intersection of Main Street and First Street for treatment. He would then use the next minor intersection along Main Street (that is, Main Street and Second Street) as a comparison unit. He should not use rhis method if operatiow at the sites are dependent, however. If drivers can sec the TCD at Main and First &om Main and Second, for instance, the larter is probably not an appropri:tre comparison unit. Comparison units are often much easier to locate when experiments are designed prior to or during the before period rather than during or after the after period. Reassembling data &om several prior years to locate adequate comparison units is extremely difficult, even in agencies with outstanding recordkecping systems.

5.0 FACTORIAL DESIGNS The experiment designs described earlier in this appendix primarily involved comparisons of one treatment to another. These: m«hod.s work wdl for determining the effeas of a few treatmcnt'i, bur the sample sius required for valid comparisow of more than a few treatments quickly prove onerous. Forrunatdy, f.lctorial experiment designs arc available to allow engineers to make valid comparisons between many f.lctors and levels using minimum samples. In addition to efficiency, one of the prime benefits of many factorial designs is the abiliry to estimate the effects of intcracdow between factors. The interaction is the combined effect of two or more factors. For example, consider a11 experiment examining the relatiowhip of traffic volume and the presence of a turn bay to delay. The experimenter found that both factors working alone (that is, the main effects) were related to delay, but that informacion was nor very usefuJ. She also found the interaction between volume and the presence of a rurn bay was related to dday. That information was very usefuJ: She then knew there were some levels of volume where turn bays would be more beneficial and could recommend designs based on the breakpoint in the rdationship. Two-way i!lteraccions often provide the mosr uscfuJ information from an experiment. Three-way interactioru can be useful but are sometimes difficult to interpret. Interactions among four or more filctors arc not generally useful in transportation engineering experiments. AN OVA is the usual method of analyzing factorial experiments. Experimenters must verify several assumptions about their clara before trusting AN OVA results. Experimenters must also specify :t modd of the relationship between the tilemrs, the interactions and the MOE to use ANOVA. Most standard compure.r sraristical pacltages contain ANOVA and quickly generate estimates of the significance levds of any number of main effects and i.nteraetions. Exhibit A-5 shows 494 • MANUAL Of TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

',ANOVA resul(S from an experiment with four factors (L, T, P and S), where the model included only main effects and :two-way interactions. Analysrs use means testS to find rhe particular levels of significant factors that deviate from other '\~vcls. Exhibit A-6 illustrates the application of a means tesr ro the main effects of the four-factor experime nt.

The notation L''T for example means the interaction between the lefi-turn volume factor (L) and the through volume factor (1').

So~n;e: Hummer et al., 1989.

Mean delay Number of Factor

Level

Observations

(seconds per vehicle)

Source: Humm~ eta!., 1989.

Appendix A • 49~

Complerdy randomized fa.ctorial designs have one or more observations in every cell of the matrix of factors and levels. Exhibit A-7 illustrates such a design with rwo facton, each at four levels, and two replications. The units, numbered 1 through 32, were assigned ar random to a cell in the matrix. Completdy randomi-zed factorial designs must be at least pa.rtially replicated (that is, at least a few cdls muse have more than one observation) to allow the significance of the main effects and all the interactions to be estimated without additional assurnptio?s.

Unit numbers

I__

Levels of factor •traffic control"

I

None

I

Signs only

I

Markrngs only

Signs end markings

None

I

Two patrols per hour

I

Four patrols per hour

I

Contlnuous patrols

31 19

2 28

18

6

32 . 21

1

22

4

26

27

I .

3 7

10 12

8

15

13

I

30 14

29 24

17 25

5

23 '1 1

16

20 9

Many types of advanced factorial designs arc available. Bltxlt designs utilize groups of similar units, as described in the section ·Paired Comparisons" above, co remove extraneous sources of variation efficiently. Block designs are useful when.treacing some units one day and other units another day, for example. The possible effect of the day created can then be scpararcd from the effect of the treatment. Within each block, experimenters assign units to a treatment at random. Experimenters use nmetltlesigns when one or more factors are embedded within other factors so it is impossible to obtain an interaction berween the factors. For example, nested designs are convenient when an experimenter identifies several jurisdictions and then sdecrs for treatment several households within each of those jurisdictions. The c:xperimenter can separate the effect of the jurisdiction from the effect of the household (Anderson and Mclean, 1~~. ' Other advanced factorial designs called ftacticnaJ factorials allow the experimenter to gain full information on main effects and partial information on interutions while running only a portion of a replication. Fractional factorials arc very difficult to design bur can be very efficient, especially when treatments are expensive. Experimenters oftm use fractional factorials in exploratory work to determine which variables should be examined in depth. Exhibit A-8 shows the design of an experiment with eight factors, each at two levds, that would have required 256 units for a full replication. The c:xperi.menters designed a one-<[uarter fractional factorial experiment that required only 64 units (labeled •run number• in Exhibit A-8) and provided complete informacion on all eight main effects and all interactions between two factors. LAtin SifU4rt and Grruco-LAtin sqtunT designs are very special..iz.ed fractional factorial designs that have been applied successfully in transportation engineering experiments. Designers interested in block, nested, fractional, or other advanced fa.ctori21 designs are encouraged to consult a professional statistician or a text on statistical c:xperiment design such as by Cochran and Cox (1957), Anderson and Mclean (1974), Hicks (1982), or Montgomery (1984).

'\

Run Nu:mb#lr

1 2 3 4· 5 8 7 8

-

16 16 17 18 19 20 21 22 23 24

31 32

33 34

.. ...

Variable Combination

BCOEF ACOEF BEF

315. 36 37 36 '39 40 41 42

CE

'14

28 29 30 ·

I AB

ABCO GH ABGH COGH ABCOGH DE ABOE

9

· 21

Run Numbe~

co

10 11 12 13

25 26

Variable Combination

AEF

BCDEFGH ACDEfGH BEFGH AEFGH BCF ACF BOF AOF BCFGH ACFGH BOFGH AOFGH BCOEH ACDEH BEH AEH BCOEG ACOEG BEG AEG BCH ACH

43

ABCE OEGH ABOEGH CEGH ABCEGH FH ABFH COFH ABCOFH FG ABFG COFG ABCOFG OEFH ABOEFH . CEFH ABCEFH OEFG ABDEFG . CEFG ABCEFG

44

46 46 47 46 49 50 51

52 53 54 55

56 57. !58 59 60 .

BOH

AOH BCG

81 e2

. ACG

BOG ADG

63 54

'--rKey Level With No

U.tter A B

c

0 E

F G H I

V:attable Th~ghvolurne

Opposing Volume Left·tum volume Alght-turJ'> volume Mean vehicle speed Upstream signal distance ~ signal dlatance

.,..._nt

Units

t.stter

VPH VPH VPH VPH MPH Miles Mil-

300

-

Bypass Jane Variables A through H e ll at "no letter'.' ' - '

300 20 20 35 0 .5

0.5

No

Csvel With A LAtiM

700 700 50 30 30

1.5

1.5 Yes

Sowcc: Bruce, E. Land J. E. Hummer, "Dday Alleviated by Left-Tum Bn>ass Lanes." TransportatitJn &uardl &rord 1299. Transportation Research Board, National Resea.rch Council, Washington, DC, 1991.

6.0 REFERENCES Anderson, V L. and R. A. Mcl ean. Design of Experiments, A Realistic Approach. New York: Marcel Dekker, 1974. Bhattacharyya, G. W. and R. A. Johnson. Statistical Conctpts and Methods. New York; Wiley, 1977. Bruce, E. L. and J. E. Hummer. "Delay Alleviated by Left-Turn Bypass Lanes." Tra•uportation Research &cord: journal ofrhe Transporuztion &search Board 1299 (1991). Cochran, W. G. and G. M. Cox. Experimental Designs. New York: Wiley, 19)7. Council, F. M., ec a!. Accidmt Research Manual FHWNRD-80/016. Wasb.ington, DC: U.S. Department ofTcansportation, Federal Highway Administration, 1980. Federal Highway Administration. Highway Saftty Evaluation Procedural Guide. Washington, DC: U.S. Deparcmenc of Transportation, Federal Highway Administration, 1981. · Hauer, E. and J. Lovell. "New Directions for teaming about the Safety Effect of Measures." Transporuztion &search &cord: Jountal ofthe Transportation &uarch Board 1068 (1986}. Hauer, E. and B. Persaud. "Common Bias in Befo~d-After Accidents Comparisons and Irs Elimination." Transportation Rmarrh &cord: jountal ofthe Transportation &search Board 905 ( 1982). Hicks, C. It Fundammtal Concepts in the Design ofExpmmenr:s. New York: Holt, Rinehart and Winston, 1982.

Hummer, J. E., It E. Montgomery and K C. Sinha. An Evaluation ofLeading venus LaggingLtft Tum Signal Phasing. FHWN INIJHRP-89/17.1ndianapolis, IN: Federal Highway Admini.mation and Indiana Deparonent ofTcansportation, 1989. Montgomery, D. C. The Design andAnalytis ofExpmmmts, 2nd ed. New York: Wtley, 1984. Steen, F. H. Elemmts ofProbability and Mathematical Suztistics. Boston, MA: Duxbury Press, 1992. Washington, S. P., M.G. Kar!aftis and F. L. Mannering. Statistical and Econometric M~thods for Transportation DaUI Analysis. Boca Raton, FL: Chapman & Hall, CRC Press LLC, 2003. Weed, It M. "Revised Decision Criteria for Before-and-After Analyses.• Transportation &s~arrh &cord: ]ountal ofthe Transporuztion &search Board I068 (1986) .


Appendix B •••• •

•

••• •

0

••

0

••

•

••••

0

•••••••• •

•

•

0

•

•

•

•

•

••

•

••

•

•

•

••

•• •

0

••

•

•

0

•

•••• •

•

0

0

0. 0

••••

•

•• 0

.

.

..

..

0

Survey Design Original by: joseph E. Hummer, Ph.D., P.E. Edited by:


2.0

INTRODUCTION

499

1.1 Articufating Study Objectives

500

1.2 Why Survey?

500

METHODS

500

3.0 SAMPLE SELECTION

3.1 Random Sampling

501

3.2 Nonrandom Sampling

504

4.0. COMPOSING QUESTIONS

505

'4.1 Number of Questions

505

4.2 Question Order

505

4.3 Types of Questions

506

4.4 General Tips

507

5.0 EXAMPLE SURVEY FORM 6.0

501

PROTECTING RESPONDENTS

7.0 TRAINING INTERVIEWERS

508 509 510

8.0 PRETESTS

: 510

9.0 SURVEY ADMINISTRATION

. 510

10.0 SOURCES OF ERROR

512

11.0 REFERENCES

514

1.0 INTROOUCT,ION ransportation engineers have used surveys for many years, particularly in tranSportation planning. H owever, man~ engineers assigned to design and conduct surveys have not had training or experience in survey preparation. Thos..::::. engineers are often swprised to learn even small surveys are very difficult to design and can fail ifdesigned poorlrThis appendix serves as a swting point for beginners and provides a reference for those already experienced in designin§5 surveys. There is no substitute for experience in conducting surveys. AD. engineer inaperienced in surveys who is conducting a survey that is c:xpcnsive or important should seek help from an experienced professional.

T

Surveys ace essentially systematic, structured conversations. Surveys range from written questionnaires where eacJ::::1 respondent answers the same questions with the same set of possible answers to rambling interviews held cogeche.)IIC. by a central theme. Most survey results ace coded and analyzed using standard statistical techniques (diseussed jr.:::a. Appendix C). Appendix B • 49~

This appendix will cover methods for conducting surveys, selection of a relevant and significant sample, how to compose appropriate questions to get at the stated objective(s), preparing the interviewers and administering the survey. Whether new or experienced with conducting surveys, this appendix should serve as a good resource for all who wish to learn more about possible surveying options.

1.1 Articulating Study Objectives Before analysts decide to conduct a survey they must carefully define the srudy problem and articulate the srudy objectives. If possible, the study objectives should be wrin:cn dearly in one shore paragraph. Clear objecrives are important since many subsequent decisions depend on them. If analysts cannot dearly articulate srudy objectives, they should suspend further work on the study.

1.2 Why Survey? Surveys are an dfecrive srudy technique for several reasons. If conducred properly, engineers can collect a large and valid data set relatively cheaply. Surveys hav.e flexible formats and allow several types of answers. Surveys are often the only way to gather feedback from users of a system if data on market responses (such as revenue, passenger volume, etc.) are not available. Surveys are also the primary way to measure attirudes and determine why people act a certain way. Fmally, based on the author's experience, people generally try hard to answer well-
2.0 METHODS Survey data are collected from the verbal or wrin:en responses of a sample of people. Common methods of collecting written responses include a m.ail~ut and mail-back form, a hand~ut and mail-back form and a hand~tit and handback form. Common methods of collecting verbal responses include telephone interviews a,nd personal interviews. Personal interviews are conducted at the reSpondent's home, at the respondent's workplace, at the survey analyse's office, at some public place, at an event, or almost anywhere. There are also hybrid methods that combine the foregoing techniques. Each method has advantages and disadvantages. Exhibit B-1 shows some of the relative advantages of three common survey methods: personal interviews, telephone interviews and mail~ut and mail-back questionnaires. The major advantage of the personal interview method is the ability to probe responses in depth and the high probability of obtaining responses from the desired sample. By contraSt, m.ail~ut and mail-back questionnaires offer lower costs and minimize interviewer bias; however, response rates decline. Telephone interviews usually take less time than the other methods shown in Exhibit B-1 but do not allow the presentation of visual' stimuli. Also, with the advent of caller ID, telephones are not answered as often as in previous decades. Tdephone interviews generally have uaits that are between the acremes of the personal interview and the mail~ut and mail-back questionnaire, making them an attractive choice for many surveys. Interview surveys have improved in recent with advances in computers. Interviewers lising handheld computers in the 6eld or personal computers in the office can immediately code and score responses. Computers also help interviewers follow complex •skips,• where some respondents are not asked some questions.

years

Hand~ut and hand-back or hand-out and mail-back questionnaires are also common in transportation srudies. They have many of the advantages and disadvantages of mail-out and mail-back questionnaires but can be Jess cosdy. Costs are especially low if workers_ (bus drivers, for example) distribute and/or collect questionnaires while performing their •Aft

-

l.AAMII A. f t"'\1:::

TDI\~IC:DI"\OTAT11"\M

Ct..lt":lt..ICI:Dit..lt':. C:Titntr:C

")Mn tnrTII"'\t..l

AJrcc.k, R A. and R. B. Scttk. TM SID'W) Rnurrh H~. Homewood, ll.: R.iclwd D. Irwin, Inc., 1985; Sudman, S. and N. M. Bradburn_ Asking QJusti4N. Sm franci=, CA:. Jo=y-Bass Publications, 1982.

regular duties. Hand-
3.0 SAMPLE SELECTION If the survey being conducted is relatively small, it is possible to use.:the entire population (called a •census"). Because the population is typically much larger, a major part of the work of survey design is the selection of a sample of people who will respond to the survey. The sample reflects the characteristics of the population from which it is drawn. Sampling is necessary because it is usually too costly and roo slow to survey the entire population of interest. Several common sampling techniques are discussed here. Creswell (2009), Rea and Parker (2005), Cochran (1977), urnun and Man.ski (l976) and others provide more details and descriptions of other teehniques.

3.1 Random Sampling With Tlllllfqm Sllmpling each member of the population of interest has some established probability of being sdeaed as part of the sample co be surveyed. Random sampling is common and has several advantages over nonrandom sampling. which is discussed later. Random sampling req~ the an:alyst to regard the pcpultztian under srudy as a collection of sampling unils. The population under srudy can be quite spec:ili.c and must be related to study objectives. For example, an engineer might limit the population for a survey on the environmental impactS of a new highway to residents living and employees working within 1,000 feet (ft.) (305 meters [m]) of the hl~y right ofway. Sampling units are the non-overlapping eniities in the population that will be surveyed. In most surveys, sampling units are individual people, although =portation surveys have also used farnilies, neighborhood groups and other entities. Once me designer establishes the sampling unit, she must obcain a complete list ofsampling units in the population, called afrrzmr. A reasonably accurate

Appendix B • 501

and complete frame is nece:ss:uy for mosc random sampling techniques. Compiling a frame is often very difficult because directories, tax rolls and ocher common framing materials are often our of dare or error-prone. 3.1.1 Simple !Un1dom Sampling

Random Sampling is the purest form of probability sampling. With chis method, each sampling unit in the population has an equal chance of being chosen. The analyse chooses units to be sampled by assigning a number co each unit in the population and then drawing random numbers. Simple random sampling can be accomplished with "f'lacemmt, when sampling unirs arc d igible co be drawn more than once, or without replacemmt, when units cannot be drawn a second time. Some formulas for analyring samples with replacement are simpler, but for most surveys there is no difference in the analysis. Therefore, to remove the possibility that a person is asked the same questions twice, sampling without replacement is standard. Simple random sampling has a few advantages over ocher sampling methods, bur for most surveys ir is inferior. Simple random sampling is easy to understand, is commonly used (making ir easier to compare results wich ocher surveys) and ic allows a designer co select a sample with no population data besides a &arne. However, simple random sampling is usually inconvenient, since survey p~onnel must track down widely dispersed units. Worse still, simple random sampling is almost always less precise chan sampling methods that utilize more infor1112tion about the population. With the same size sample, ocher sampling schemes will usually produce bcuer estimates of means and proportions than will simple random sampling. Exhibit B-2 provides formulas for estimating mean values, variances and needed sample sizes for a simple random sample. The sample-size formula foe continuous (nor proportional) data requires an estimate of the variance of the mean for the entire population or a prior estimate of the coefficient of variation, which is the variance divided, by the f!lean. Analysts can obcain chis prior estimate from an educated guess, the results of a prior survey, or a pilot study of the cucrent population. The sample-size formulas are based on the desire to estimate the mean value of some characteristic within a specified tolerance from the true mean value for the entire population. For example, an analyst may want to estimate mean household income within $1,000 of the true mean.

3.1.2 StriUified Random Sampling In chis scheme, analysts divide the population into non-overlapping groups of units called rtmta. A simple random sample is then drawn from each stratum. Estimates of means, proportions and other desired statistics are based on the samples in the ruata. Analysts should select strata so the characteristic of interest in the survey differs between them. For example, in a survey on the rravd habits of households, it may be desirable to stratify by the number of automo-biles owned by a household since auto ownership is strongly related to travel patterns. Stratified random sampling almost always provides more precise estimates of a quantity than does simple random sampling. This is especially true for populations where the quantity to be esti!ll2ted is similar within strata but differs greatly between strata. Stratified random sampling is also advantageous because estimates for the suara can be made. In addition, Cochran (1977) points out stratified random sampling may be convenient, such as during surveys to be administered by field offices, where each office can survey a srrarum. The main disadvantage of scracified random sampling in comparison to simple random sampling is the extra work needed to select rcasoll2.ble strata. The analyst muse know the variable(s) on which the srrata are based before stratification. The analyst must also know the rtumber of units in each srrarum before producing estimates, but if a complete frame is available, chis is not a major obsracle. The formulas for sample-size estimation and for estimating means and variances are more complex for stratified random sampling than for simple random sampling and are not reproduced in chis appendix. Consult Creswell (2009), fua and Parker(2005), Cochran (1977), or another statistic or sampling text for those formulas. 3.1..3 Cluster Smnpling

This random sampling scheme is best explained duough an e:xample. Instead of choosing individll21 people as the sampling units for a rravel survey of the residents of a town, the analyst chooses neighborhoods as the sampling units. Then the individual people are known as the sampling subunits, and each subunit in a sampled neighborhood would be asked to respond to the 5U1'Ve)': This sampling plan is known as cluster urmpling beau.se the sampling unit (a neigbbochood) consists of a cluster of subunits. Ouster sampling is a very popular way to conduct surveys in rransportarion and many ocher 6dds. I.cwer cost is the lll2in advantage of duster sampling c;yver other sampling methods. The rravd survey mentioned above saves monc:y by sending survey ~nne! to a relatively small nwnber of neighbochoods rather than to dwdl.i~ scattered across town. In addition, duster sampling does not require a &arne of all the subunits in the population. The rravd survey mentioned above only needed liscs of neighborhoods in the town and of residents of the sampled neighborhoods. The major 502 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDffiON

~'51-"?'!'

t..~~

· 'Yif!?" ~~

Quanricy to be .r:. Mean

. y=

-

Vaciahle npfi,

Formula•

E

IY;

Individual observation

-

i: Observation number n: To cal number of observations

l •_l_

n Variance of Mean

2

n

L:(yi- y) s2

_.:.::i =,_,_l~-

(n-1) Sample Siz.e

t: A consW!t corresponding to !he.desired level of confidence,'

n=(~J

a percenc (i.e., me statistic fur the normal distribu tion; see table 12-2)

IS: Standard deviation of the I?: Population Mean

1mean

d: i?-yl should not be greater than d more than (l 00-a) percent of !he time

n=(~~J

r: i?-y! should nor be greater than rY more than (100-a) percent of !he rime

a n

-

a: Number of units in the sample answering "yes"

--

p: Proportion of units in sample answering "yeS'

-

Proportional Data Proportion

-

p=Variance of Proportion

var(p) = pq

n-1

Sample Size

2

n= t pq

q: Proportion of units in sample answering "no" d: i?-yi should not be greater than d more than (l 00-a) peroenr of the rime

d2

-

or

t2q

n =-2--

---

r: i?-yishould not be greater than r? more than (100-a)

peroent of the rime

rp

a) Formulas for sampling with replacement or for sampling without replaoement when less than 5% of the units in the population sampled

-

-

Source: Cochran, 1977.

Appendix B • s~~

disadvantage of duster sampling is convenient units and subunits must be available. Ouster sampling is impossible if reasonable dusters cannot be defined. Another disadvantage of duster sampling. cited by Fmk and Kosa:off (1985), is a sa of relatively cnmplc:x formulas for estimating sample sizes, means and variances. Those formulas differ depending on the exact duster sampling technique used and are not teproduc.ed in this appendix. 3.1.4 Systematic Sampling Systematic sampling is often used instead of nndom sampling. lt is also called an N• name selection technique. Aher the required sample size has been calculated, every Noh record is selected from a list of population members. As long as the list dOC$ not conuin any hidden order, this sampling method is as good as the nndom sampling method, Its only advantage over the nndom sampling technique is simplicir:y. Systematic sampling is frequently used to select a specified number of records from a computer file. 3.1.5 Multistage Sampling The four methods covered so IU-simplc, stratified, duster and systematic-are the simplest random sampling stratc· gies. In most real applied social research, the analyst would use sampling methods that arc considerably more complc:x than these simple variations. The most important principle here is the survey designer can combine the simple methods described earlier in a vuiety of useful ways that help us addteSS the sampling needs in the most efficient and effective manner possible. This is called multistage sampling (CteSWcll, 2009).

For example, consider the idea ofsampling Oregon state residents for face-to-face interviews. Clearly. the survey designer would want to do some type of cluster sampling as the first stage of the process. He rni.ght sample townships or census tracts throughout the stue. But in cluster sampling he would then go on to measuce ~Min the dusters he selects. Even if the analyst is sampling census tracts he may not be able to measure everyone who is in the census tract. So, he might set up a stratified sampling process withln the dusters. In this case, he would have a rwo-stage'~ampling process with stratified samples within duster samples. By c.ombining different sampling methods the analyst is able co achieve a rich variety of probabilistic sampling methods that can be u.sod in a wide range of social research contexts.

3.2 Nonrandom Sampling NfmranMm sampling refers to schemes wherein each unit docs not have an established probabilir:y of being selected. Engineers usually draw nonnndom samples from the most readily accessible units in the population in a haphazard way. An aample of a nonnndom sampling plan would be an engineer surveying pedestrian habits in an area by stopping me first 100 people walking along a particular sidewalk. Although not the best method, most surveys by transportation engineers usc r;~onnndom sampling because there arc typically not enough funds available to do a more complex candom survey.. Nonnndom samples are appropriate for many studies but have several disadvantages. Nonrandom samples have an advantage over nndom samples in that they are much more convenient for the designer. There is no need to draw a random sample. Indeed, if one of the steps involved in drawing a random sample cannot be performed, a nonrandom sample may still be sufficient to achieve projca objectives. The major disadvantage of nonrandom samples is one cannot prove the validir:y of the findings mathematically. With nonrandom sampling. the sampling error (the error in the quantity being estimated attributed to the fact thac one particular sample was drawn of the many possible samples in the entire population) remains unknown, so engineers cannot easily generalize their findings from the sample to the entire population. A common m.istake when conducting surveys with nonrandom samples is to use me analysis techniques of random sampling and thereby mislead readers of their teports. Another disadvantage of nonrandom sampling is there is no theoretically sound way to estimate the needed sample size. Finally, survey workers may be susceptible to bias in the selection of a nonrandom sample. 3.2.1 Omvmiene• &tmplmg Convenience sampling is used in aploratory research where the researcher is interested in getting an inapcnsivc approximation of the truth. As the name implies, the samples arc selected because they arc convenient. This nonprobabilir:y method is often used during preliminary research efforts to gee a gross estimate of the results, without incuning the cost or time requited to select a nndom sample. 3.2.2 Expert ,.]wlgment &nnplmg Judgment sampling is a common nonprobabilicy method. The researcher seleru the sample based on judgment. This is usually an extension of convenience sampling. For example, a researcher may decide to draw the entire sample from 504 • MANUAL OF TRANSPORTATION ENGINEERING

STUOIF~

INn

Fnmnr..~

pnc "represemadvc~ city, even though the population includes all cities. When using this method, the researcher mlist ~c confident the chosen sample is truly representative of the entire population.

3,2...3 Quol4 Stnnpling Quota sampling is the nonprobability equivalent of suatified sampling. Like suarified sampling. the rcseaccher first identifies the suata and their proportions as they are represented in the population. Then convenience or judgment sampling is used to select the required number of subjects from each suatum. This differs from suatified sampling, where the strata are filled by random sampling. 3.2.4 SMwha/J Sampling Snowball sampling is a special nonprobability method used when the desired sample characteristic is rare. It may be extremely difficult or cost prohibitive to locate respondents in these situations. Snowball sampling relies on referrals from initial subjects to generate additional subjects. While this technique can dramatically lower search costs, it comes at the expense of introduciJ).g bias because the technique itself reduces the likelihood the sample will represent a good cross-section of the population.

4.0 COMPOSING QUESTIONS Writing good survey questions is very difficult. Consequently, many surveys have questions with no feasible responses, questions with too many feasible responses, too many questions, vague instructions and other flaws. The best sampling plan and the most exhaustive statistical analysis arc meaningless unless valid and reliable responses are obtained from the persons sampled. This section provides guidance in composing questions so major · pitfalls may be avoided.

4.1 Number of Questions Several factors aft'cct the length of a survey. Firsdy, the study objectives will aft'ect the survey length. Compla a.nd multiple objectives require longer surveys, w:bile simple objeaives require short surveys. Designers should include as many questions as needed to accomplish study objectives and should make swe each question relates to the objectives. One way to ensure each question is related is co state in writing its unique contribution to accomplishing the objectives. Secondly, the topic affects the survey length. ~pondents will tolerate longer surveys on topics that are interesting or important to them. Sudman and Bradburn ( L982) advise that on topics important to respondents, home interviews can last up to L5 hours and mail-backquescionnaires can be up to 16 pages long. Mail-back questionnaires on topics unimportant to the respondent should be limited to two to three pages. It is important to remember shorter can be better! Th.irdly, survey length depends on the format chosen. Analysts can~ more questions per respondent" with personal or telephone interviews than with mail-back surveys. ):inally. there is a trade-off with a fixed budget between the number of questions to be asked and the sample siz.e, c:sPccially for personal and cdephone interviews.

4.2 Question Order Begin a survey with an introduction explaining who is conducting the survey, the purpose of the survey, the subject matter to be covered bf the survey and i!lformation on the respondent's rights and privileges. After the lase question has been asked, leave nme or space for respondent comments. The comments may prove wcful and the opponunicy ro make a comment hdps the respondent feel that her opinion is important. F'mal.ly, the respondent should always be thanked and may be offered a summary of the survey results. ~ has shown responses to a question can differ depending on placement of the question in relation to other questions (Alccck and Settle, l985). There are no rules on question orde.r, but several general suggestions for obtaini.ng higher-qualicy responses include:

• placing easier questions early in the survey to avoid discouraging respondents; • placing questions that may be threatening (that is, embarrassing or intimidating} to respondents in the

middle: of the survey (Sudman and Bradburn, 1982);

• placing demographic questions at the end of the survey; and • avoiding response set bias (a series of questions in a row with the same response).

In addition, designers should move questions likely to bias responses to later questions. For example, do not follow a question on traffic congestion wim a question asking for "the worst suburban problems of our time" with "traffic congestion" as one possible response. Some SUJ:Yeys contain skips, where certain respondents are directed co skip certain questions. If a survey contains skips, designers muse ensure questions pertaining co all respondents are not skipped. Provide clear directions fo r skips, especially for mail-back surveys when respondents cannot ask a survey worker for help.

4.3 Types of Questions There are two basic rypes of questions. Clcsed questions present a fixed set of alternate answers to me respondent. Open qr«stions allow the respondent co create their own answers. A skilled interviewer can use open questions effectively, but for most surveys closed questions are preferred for several reasons. Firsdy, answers to a closed question are often ready for analysis, whereas answers to an open question must be quantified or classified before analysis. Secondly, closed questions allow respondents a common frame of reference (Converse and Presser, 1986). For example, a question asking for a list of retail establishments respondents visited the previous day may be confusing if left open,. because respondents are unsure whemer hair care, shoe repair and other services are "retail." Thirdly, closed questions aid respondent recall. Fourthly, Converse and Presser (1986) point out mat open questions are more vulnerable to question order and other biases. Fifthly, closed questions reduce the amount ofwriting required of respondents or intbviewers, which is especially important if the survey is conducted in moving vehicles or ocher difficult locations. FinaUy, closed questions are less taxing mentally, so respondents will provide higher-qualiry answers with a higher respc;>nse rate (AIreck and Settle, 1985). Closed questions have a few disadvantages relative to open questions, but these disadvantages are easily mitigated. Closed questions are more difficult to create because a list of response possibilities must be established. However, asking an open question during pretests may reveal a list of possibilities to be offered Iacer with a closed question. Some have also criticized closed questions because they do not allow respondents to say why they chose a particular response. To mitigate this, closed questions may be followed with probing open questions (Converse and Presser, 1986). Closed questions take many forms. "Yes or no" questions are common and simple to score. However, because "yes or no" questions are absolute, respondents may feel more intimidated by them and are more likdy to misinterpret them (Fink and Kosecoff. 1985). Distribute likdy "yes'' and "no" responses randomly on the survey form so respondents do not discern a pattern.

Checklist qr«stions provide respondents with a list of possible responses, from which they may choose one or more

items. The keys to constructing unbiased checklist questions include (Alreck and Settle, 1985): • The lise must include all possible answers. • The items must be mutually exclusive. • There should be more variation between items man within individual items. "No opinio~" or "don't know" options should be offered on most checklists (and with most "yes or no" questions) to include all possible answers and to avoid bias (Converse and Presser, 1986). Closed survey quesci9ns may provide scales to me respondents so they can choose positions on a continuous spectrum of possibilities. Scales are easier to construct than checklists since designers must simply fix the extremes and establish an increment on the scale. In comparison to other forms of closed questions, scales may save space on a written survey or save time in an interview. Once a scale has been presented to or learned by the respondent, it may not have to be repeated while different questions are asked. Scales also lead to more possibilities for numerical analysis than "yes or no" and checklist questions which are usually analyzed only as proportions (Aireck and Senle, 1985).


Among the most common types of scales are Likert, ordinal and ranking. Likert scales measure the deg ree ro which respondents agree with a scacemenr, and usually run from I = "strongly agree" rhrough 3 = "neutral" or "no opinion" w 5 ="strongly disagree." Likert scales are very common but there are several good reasons for caution about cheir use. Respondents generally wane co agree with the analyst more often rhan disagree, so biased responses are more likely with Likert scales (Converse and Presser, 1986). Also, Likert scales sometimes confound extreme and intense vieWs on an issue. For example, a person "strongly agreeing" with a statement char "More highways must be buik to solve traffic congestion" may be expressing a strong desire to hllve two highways built (an intense view) or a w~er desire co have 100 highways built (an extreme view). Rewording questions can help overcome this problem with Likert scales. In an ordinal scalt, items ace in some logical sequence. For example, for a question on when people prefer co perform errands, a convenient (and oversimplified) ordinal scale may be 1 ="before work," 2 ="during lunch" and 3 = •after work." Ordinal scales provide good data on where items are relative co other items, but provide less information for the analysis rhan do checklist questions.

Ranking st:Alts require the respondent to put a lise of items in order using some criterion. Ranking scales are useful for simulating real-life situations when choices must be made. However, data on rhe absolute standing of eaCh item on the lisr are unavailable from a ranking scale. Care should be taken in carefully writing instrUctions for ranking scales. For inscancc;, a survey may ask a respondent to rank five possible retail locations in order from 1 co 5, with 1 being the sire the respondent is most likely to visit and 5 the least likely. Confusing wording in ranking questions such as this may lead the respondent to rank each retail site on a scale of 1 to 5, instead of putting them in acrual rank order. Consult Creswell (2009), Rea a.nd Parker (2005), or other references if further details on the abovementioned scales and descriptions arc needed.

4.4 General Tips

.

Regardless of question format, in every effort to compose questions designers should consider some general tips. The main rule on composing questions is to keep the study objectives in mind. Every question should help satisfy those objeCtives. Effective questions should also: • Be as shon as possible. Use standard demographic questions and terms whenever possible (Sud man and Bradburn, 1982). For example, srandard U.S. Census Bureau occupational category definitions are available (Exhibit B-3) that will allow easy comparisons of results to other sucveys, among other advantages. In fact, borrowing successful questions from other analysts (after obtaining permission and promising to give proper credit to the author) is an acceptable, inexpensive way to constrUct many surveys quickly. • Avoid technical jargon and slang. Use words simple enough to be undersrood by the leasr sophisticated respondent. A survey of the U.S. general public should not exceed a grade 7 to 9 reading levd. For comparison, this chapter is written for a college-educated audience at a grade 14 co 16 reading level. Many word processing and giammar checking computer programs judge the reading level required to understand text. • Define terms carefully. Respondents sometimes misinterpret very simple terms. The word "you, • for example, can mean the respondent or the respondent's family, depending on the content of previous questions and other factors (Sud.man and Bradburn, 1982). • Avoid controversial and other words that evoke strong responses. Respondents associate such words as rnm, liberal and traffic congtstion with strong images that can bias the response to an otherwise mundane question. • Avoid racial, eth.n.ic, sexist, or otherwise biased language. Look for the presence of biased language during pretests (discussed later). • Avoid leading questions. For instance, "How clear is it to you that mass transit saves fuel . .. ?" • Avoid loaded questions. Loaded quescion.s usually associate one aspect of the issue under question wirh some desirable trait. Alreck and Settle (1985) provide the following example of a loaded question: "Do you advocate a lower speed limit to save human ~ves?" • Avoid double negatives. For instance, •oo you agree or disagree that highways should not. .. ?» • Avoid "double-barreled" questions. In double-barrded questions, respondents may answer one part one wa:Y" Appendix B • 50~

and another pan another way. Double-barreled questions usually result from uying to do too much in one question and are tteaced by creating separate questions from the pans. • Be very careful about prompting respondencs' recall. Mentioning an example of a possible answer during the question may bias the response in favor of the item mentioned. • IWiponses to threatening questions may be biased. The number of respondents claiming to use seat bela, ride public transportation and perform other socially desirable actions will probably be overestimated unless techniques (see Sud.man and Bradburn, 1982, for example) are employed that account for these biases. For instance, this could include a comparison group survey of a similar population which does not have a treatment installed. • Avoid hypothetical questions. Ifhypothetical questions cannot be avoided, append them onto one or more questions on real-life cxpcrieno:s. Then use additional questions to find out what respondents were thinking while answering the hypothetical question (Converse and Pl=er, 1986). For aample. a designer may want to ask whether a respondent would use an clearonic map that shows traffic congestion in her automobile. The designer might lead the hypothetical question by a question on whether the respondent listens to traffic reportS on the radio and might follow the hypothetical question with a checklist of the desirable fi:atures ofan electronic map.

• Ask respondents to recall events in the roost immediate time frame feasible. The choice of an appropriate time frame avoids ukscoping, the tendency to underestimate the passage of time since an event occurred. Longer time frames are possible for events that were very important to respondents. • Construaing high-quality swvey questions

is ti.me~nsuming. Even after several sages of writing and revising, it is likdy some questions will prove flawed during administration of the survey. Thus, the survey should ask fur informarion that is absolutely centtal to the srudy objectives ar least twice, in diiferent formats, while survey questions providing supplemental information can be asUd once. This redundancy is tolerable considering the risk in putting swvey objectives on the line in a single question.

JOB CATEGORIES A. I!AIIUFACTVR- Of 'f'RAMIPOffllmOM EQUif'tiiEifl I . OTHER MAHUMCT\Jft. C. AGRICUL1URE, FO~ AH0 RSHIERY D. MlfClNO E. BUSINESS SERVlCESAHO REPAIR SERVICES

F. PROfESSIONAl. NCO REUQ'ED HRVICEI G. WHOLI!SAU OR REUUL TRADE H. FINAIWCI!. IU!A&. Ulllll'~ OR aaoRAHCI! . I. 1liAH8PORlllnON, COMMUIIIICimONS, UlU.ITIES J. COIOISTJWC110N M. ~ORMCIWiiiON SOMCES 1..130¥~

5.0 EXAMPLE SURVEY FORM

I

\ l

The example survey form in Exhibit B-4 illustrates several interesting traics in survey design. The survey purpose was to collect origin and destination, demographic and other information about current bus patrons. Survey workers distributed the form on buses. Passengers could complete the form on the bus and return it to the worker, or passengers could return the completed form by mail. A cover sheet introduced the survey and promised a pass for free bus rides for returned survey forms. The survey primarily required respondents to check boxes, which is appropriate for on-board completion. However, questions 4 and 6 are open, requiring respondents to write ad~. and require fairly advanced reading comprehension. The survey designers were aw-are of those limitations. Take-home surveys were therefore also given to a limited sample of respondents. •n.a _..

•.til. Ml I A I 1"\1: TO A

~IC:Df'\DTAT11"'\~I

.

M. OTHER t""'- o.-tbe)

OCCUMnON TYPES A. I'ROfUaiUI.CIIl ~ a. FMMlR OR IMAM MMM1!1R c. I'MIIII L.IIIICIIIeR OR M11M I'Oia!IIM

ll. . , . , . LAIIORI!JI E. MAICA8I!II. OPI'ICIAL. CNMift 01' A BUSINESS

F.

CURICALMD - . v i WORDRS

H.

~ CJill"'lffUUIjUt MD ..-uR WOf!ICIRS

a. uua

·

IGI""-" ..auaaR Oll110111ft ftJIICU CRiiUII"'Ol .j., I'RMir'EMOIIISaaD.....,. CIIIUO. euna. nc:.) II. Qna SIIMCIE WIIMIIIl L -.rwiiiY . . cmllll~ . . . . . .

I.

Sowcc: Sheskin, I. M. a.nd P. R. Stopher. •pilot Testing of Altcma!M Adm.il!.i$cra!M Procedures and Survey Instruments.• T~ &much &crmi 886, National R=arch Council, Washington, DC: Transportation R.cscarch Board, 1982.

Ct\lt:lt.ICCOtf\11': C'Tt lf"'III:C'

~f\11"\

rr'\ITII"\111

Notice the skips in questions 2, 3 and 5. The usc of special graphics, such as bold arrows and dash-dot boxes, helps guide respondents through those difficult areas. Other features to note include: • •other· responses arc provided; • the .language used is easy to understand for the general population and assistance is available fo.r those who may be illiterate; • precise directions are given with several questions; • demographic questions arc placed at the end of the form; and • respondent comments arc encouraged. The designers could have improved the survey form by eliminating the coding spaces. Also, many on-board surveys suffer from biased samples, since those who ride more often will receive more forms. To adjust for that bias, a. question on how often the respondent rides the bus is usually i.qcludcd in all on-board $urvcys (Doxscy, 1983).

6.0 PROTECTING RESPONDENTS The United Scares and other countries have 6nnly established the right to privacy. Competent adults generally cannot be forced to answer any survey except a periodic census. Therefore, survey designers must take precautions to avoid violating respondent rights. In addition, survey personnel muse treat respondents with respect and dignity. The relatively high response rates enjoyed by some surveys may result from goodwill esca.bl.ished with respondents from previous surveys. The most common precaution widcrukcn to protect respondent rights is informing the respondent of the purpose and method of the survey before administration. In the United States, federal guidelines specify that each respondent receives {Fink and Kosecoff, 1985): • a description of the survey procedures and an explanation of the purpose of each procedure; • a description of risks and benefits; • an ·offer to answer any inquiries; and • an offer to withdraw from the survey at any time without prejudice. For many surveys, describing the survey purposes in too much decail could bias the evenrual answers. Since re.spo~ dents do not need decailed descriptions to make rational decisions ~ut whether to participate, general descriptions oriented to What the respondent needs to know arc provided. The 4cscription of risks and benefits is crucial. Rapondents must undcrscand the schedule of compensation for participation and survey wor~rs muse inform them of the slightest possibility of discomfort or embarrassment during questions. Methods for asking sensitive questions without embanassing or threatening respondents are available (Sudman and Bradburn, 1982}. Rapondents are also entitled co an accurate estimate of how long the survey will last. Although it is usual.ly a good idea, it is not always necessary to identify the survey sponsor. Some sponsors inspire such srrong images that b~ased responses or high oonresponse rates arc l.i,kely. Ifa conrrovcrsial sponsor muse be identified, ic is better to do so lite in the survey to avoid as much bias as possible (Aireck and Settle, 1985}. Survey designers should guarantee to keep res~oscs confidential whcncvcr possible. Ifa name, an address, a telephone number, a driver's license number, citizenship scarus, or other personal data must be gathered, the respondent muse be assured those dala will be used <;mly co achieve survey objeaives. Readers should never be able to identify individual respondents from published survey results. Also, sucvey designers should not require respondents to sign the survey form. Many funding sources, government agencies and universities have e.sca.blished procedures for reviewing surveys before rhey arc con~ oommonly called lnscicuriooal RevieW Boards. These reviews protea respondent J"i&hts and/or eosun: the study objccrives cannot be achieved by other means. For example, the U.S. Office of Management and B~ (0~) reviews surveys of 10 or more people conductt:d with federal funds. WJSC designers submit their SUCVC}'Yfoi review early. Universities typically require several weda, while the OMB requires an average of 60 to 75 days buc not more than 90 days.

7.0 TRAINING INTERVIEWERS Interviewer training is crucial to the success of interview surveys. Interviewers can wasre a well-written question if rhcy ask it incorrectly or if c.hey record the responses inaccurately. The goal for training imerviewers is to produce results that are consisrenc among interviewers and consistent over rim~ for individual interviewers. Hiring interviewers can be difficult for engineers untrained in personnd matters. Judge potential interviewers first for integrity and honesty, and second for any previous interviewing skills or technical knowledge. Several authors recommend that, if possible, interviewers should have demographic characteristics similar to those of the respondem population (Fink and Kosecoff, 1985, Alreck and Settle, 1985). Professional survey and market research firms may be able to provide skilled interviewers for a lower cost than agencies will incur to hire and train their own. Once hired, survey designers should introduce interviewers co the survey method, questions and answer form. If several interviewers are being crained, a lecture format may be appropriate for this part of the training. Interviewers need co learn "how" to ask questions and "why" the questions are being asked. For each question, the survey designer needs to tell che interviewers the extent to which the "script" must be followed. The craining schedule should allow plenty of time for interviewers to ask questions of the survey designer. After a general orientation to the survey, interviewers should practice asking questions and recording answers. Interviewers can practice on each other, on the survey designer, or on a sample of respondents during pretesting. Survey designers can make sure interviewers are sufficiently crained, or can spot weaknesses in interview technique and suggest corrective :tctions, by observing an interview of a respondent during a pretest.

8.0 PRETESTS During pretesting, survey conditions are simulated to scrut.inize the survey method and questions. If the survey is found lacking during the pretest, as most are, the designer can revise it before data collection. Pretests are necessary for all surveys. Some authors insist that an analyst without enough resources to pretest should not conduct a survey (Sudman wd Bradburn, 1982). While every facet of survey design can benefit from pretesting. pretests are especially useful during cwo particular stages in the process of developing a survey. Frrstly, pretests are useful while beginning to compose questions. The results of pretests during this time may prompt wholesale changes, such as revising the format, dropping an entire sequence of questions, or adopting different cypes of questions. Secondly, pretests are useful after questions have been fully developed, in a "dress rehearsal" The designer can reword questions, add closed question choices, reform scales and delete individual questions. However, larger changes are rarely warranted at this later stage. Pretesting data recording forms, data reduction methods and data analysis techniques (with made-up data, if necessary) is strongly encouraged during the second stage of pretesting. Many surveys of all sizes have &Jeered after data collection when analysts found themselves with sets of inappropriate or unusable responses. Pretest a respondent sample rhat is similar to the actUal sample, especially during the "dress rehearsal." Although drafts of questions should be circulated for comment among colleagues, pretesting on colleagues, employees, srudenrs, relatives, or other groups unlike the sample of respondents may not be helpful. Informing respondents they are part of a pretest is useful for pretests during early stages of survey development if decailed probes of the responses to questions ace then performed. When respondents ace nor informed a pretest is being conducted, the pretest is more realistic, but it will not gen,erate as much information that could be useful for diagnosing problems. The interviewers who will conduct the survey should also conduct the pretest. The survey designer should inrensdy debrief the interviewers about each respondent during :t pretest. The licerarure recommends pretest sample sizes &om 20 to 75.

9.0 SURVEY ADMINISTRATION Once a survey instrument has b«n pretested, a few final details must be attended to before data are collected. Written surveys must be printed, assembled and distributed. For effective written surveys, Alreck and Settle (1985) recom-

mend designers avoid bright-colored (pink, blue, or yellow) paper. White, off-white, beige and many pastel colors are fine for mailing. Mill-back furveys achieve highest response races with first-class scamps and lowest response rates with 510 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

1bulk mail permits, althou~h the latter is less expensive per response received. Assemble mail-out and mail-b acbur;eys 'with a cover lerter explaining the purpose of the survey, asking for parricipacion, derailing how complered sur:veys are :~o be mailed back and providing other information. The issue of respondent compensation musr be addressed before conducting the survey. Most rransportation surveys do not compensate respondents. Compensation introduces an additional expense char may restrict sample si-ze and JTial · bias results. However, designers should consider compensation for longer surveys or when respondents perform ditficult tasks. Compensarion shows appreciation, indicates goodwill, gains attention, or creates an obligation to respond (AI reck and Setde, 1985). Respondents do not have co be compensated with cash. Designers can reduce the paperwork burden of providing cash (receipts, tax statements, etc.) by providing small gifts chat respondems will value. The on-board bus survey discussed earlier in the appendix (Exhibit B-4) promised each respondent a pass for free rides and is a good example of such a small, valued, noncash gift.

·-·-·--·--.....----··---0•___, ..,,?1=:.7_-:-------·41! ---·-·--------__:__·__ -. a ____ :: :::=- - g ·=-----·--·------ ...... .---..------~=-.::..~.:=':'.:.-

·---=~==

=-·-•_ ·.. .....,._

mL;.,·-

o0

....... - --

~-------------·

""'

-·-

· -.:":"" : ·.:u . .·.__ _.. _g. -

g ::..8 - 0

--

u

- -• -

en ·-

,~....

11------------...... ··-----··---·····----o .F ~Jt--•-m=-;;:11'''' 1 ~----·--·........ I

E

.e

-

g

--c

~

:.·::=-:·.::~:.·.:..·········- - ·-

8

n0

::.-=......

g Cll -a

~

WI===--:.~

---- -- ~---· =.:.:~:::-

;;;;;;

..~ w

.,.., ..

i

I

~

ft-.r'

__ ___ _ __._ ...... ...... ..-·-- ....... ·-· ---·__............. "--.......................... ..-._.............-..-.......................--··-·-··--..........-......_.. _......

_____..

..c,,.o ..ot•••" o•o• coy•u. o•·•o••• .,,.,.••1., ew•w•

nooltn\t

"

._.,

1!1........... ____ .,:. __ , __ ....... __ --.·

• t::::::;.:=--.1 ...,_,,... ___._.... ......... ._.. .........

=:..----

:.:::=-~~::·

i2 ~

····-... . .. . .........,. ._ ......,_

:::!::...

==.::;;-.:=·~.-:::.:.

c..:

&:!

1'1

c:

__:_ I

,

--~ ..-.·

il:'='a"

,.. _.... ..

..., _.._

~

a -~:,:.,;::;-.o;r:':l;'.~-n:-~~=·.-..'l:'- •. - =:.. :.~:..,.--:_. H --::r-· D=:.-s-:::_... ~

-...__··-·. --....... ·- ..--···-·· ·-.......-.:E:-'3...:.._

"""' .....

·--·-··_ ·-·----·---· ...... ___ _

"*' _,..

-.

...

<.... - .. -----~

UL~ f :..::::::.::

,.....

§

Source: Sheskin and Stopher, 1982.

---· g

t ........ .....

...

Appendix B • 51 --'

Before tdephone interviewing begins, designers should establish procedures for handling the possible outcomes from a ca1L Exhibit B-5 lists most of the possible outcomes from an attempted tdephone call and possible responses to those outcomes. A response should be decided on beforehand for each possible outcome. Interviewers should keep a log of tdephone calls so respondents not reached with the original call are retried, and are not called twice. Survey designers need to ensure the survey is being answered by the intended respondents. For example, a survey on the effects of raised medians should be distributed to adjacent business proprietors, customers and through-tra.ffic drivers, among others. Before interviews, the qualifications of potential respondents need to be checked with a brief question or two. Unqualified persons should be thanked and · dismissed. On written surveys, designers can ask for respondent qualification at the beginning. directing unqualified respondents ro discard the survey or pass it to a qualified respondent. Analysts can also check respondent qualification on a rerumed wrircen survey and discard returns from unqualified respondents. Because certain members of a household tend to answer the telephone at certain time3 of the day, some survey designers use a response matrix like that in Exhibit B-6 to avoid biased tdephone interview samples. The interviewc:~ asks the person who answers the telephone one or two questions about household composition, consults the response matrix and requests the sdeeted household member to come to the tdephone for the interview. To ensure each adult household member has an equal chance of being interviewed, designers create several versions of the matrix with different entries. Then interviewers sdect a version at random before each call. Survey &signers should monitor interviewers regularly to reduce bias. If d.iroct observation ofinto:viewecs is impossible, the survey designer should recontact a small subsample ofrespondents for comments on the interview. Mostly, survey designers should watch the tendency of interviewers to speed delivery of questions. Respondents need time to formulate answers, so survey designers muse insist interviewus maintain a slow, constant pace during every iri.terview. Survey designers should also remind interViewers not to register emotions as respondents answer questions. Even subde movanents or noises by interviewers, such as a nod ofthe head or a brief"OK.• can bias later responses. Analysts should compare survey results from different interviewers for major discrepancies that may indicate improper technique or fraud. The number of responses received from mail-back surveys should be noted each day. That number will usually follow a skewed distribution with a long "tail.• Most responses are received within a few Wttks, but a few will dribble in months after distribution. Analyses should set a cutoff date for returns once the long tail of the distribution has been reached. Any surveys received after the cutoff date should not be analyz.cd (Alreck and Settle, 1985). Analysts should scan all rerurned surveys or interview answu sheets for missing sections, intentionally misleading data and other obvious errors before coding. Analysts can cotrcct defective forms, if it is possible co recontact the responden.t without undue bias, or can discard forms that cannot be corrected. Answers are then coded into a computer file for analysis.

10.0 SOURCES OF ERROR Like most other transportation study daca, analyses ofsurvey data use statistical processes described in Appendix C. The major difference between survey data and other transportation study daca is in the different rypes of errors c:xpecteq with survey daca. ~ (1977) identified four different types of error that are important for surveys:

• .:_....__l!t;·

...,~:~;...

,~,..__

~i?'''

1. Sequence number - - - - 2. Telephone numbe-r - - - - -

3. Flr.-t.:could you tell me how manv people 18 or otder Uve there7

0 _.......,.

m

.c.. t;tow m•nv of the.. are female?

[!] .

0 [!]

I

Terminate caU

m Sole male

Sote female

m

Voun~r maa. Sole maJe

m

m

8 or more specify _ __

III

2hd youngest 2nd yOungeat 2nd youngest

male OJder male Younger female. 2nd ?OLinge..

---•-

m...

male

2nd younc;i08't 2nd young4i~ male mala Otder tnate 2nd younQest male You1'1'Qe8t

femele

Older mete

I

4U1 oldeat

male Srd oldeet mel• 2nd olde.t nu~1e

0 1 - m• l •

e or more _ •pecll'y! _ _ _ ~..:::::.__ __;;::::.......,=--~::...J!..c:::::...._ _..:::;....a..:::::;;..._ _..:::;:o..L..:::::;;..._ _

..::::::,.__===--'

Tell the peraon on ihe phone which membet' e>f the houMhold Ja the sel-=ed repondent and attempt an Jntervlew.

&. o...,..lpt/on of finel ,.•utt: - - -- - - - - - - - - - - -- - - - - -- - - - - - - - - -

Source: Grovc:.s, R. M. and R. L. Kahn, Surveys by TtkphoN!, Academic Press, Inc., 1979.

• Sampling error. Sampling nror is due to the sdection of a sample from t4e subject population. Most samples from a population will provide results that difftt frqm other samples. Ana.lysts can control the amount ofsampling error when they use accepted proa:dwes for random sampling. • Coding and reduction errors. Analyses can detect and correct most cot/jng tmd miw:tWn before analysis. Thorough training of coders, close contact between analyses and coders and frequent double-checks of coded and reduced data are needed. • Biased Responses. Biar~d responses are very di.fficult to detect in surveys. However, once detected, there are matherna,tical procedures to estimate the effects of biases on survey results (Cochran, 1977}. In addition, Brog et aL (1982) and others have explored typical biases in transportation surveys and have made suggestions for overcoming some of them. A thorough pretest is one of the best ways to avoid biased responses. • Nonresponse. Nonresponst is a problem in surveys when the portion of the desired sample that does not respond differs from the portion of the desired sample that does respond and a bias of unknown si.z.c: is introduced. Correcting for nonresponse requires time and dfon, so smaller surveys may suffer more from noruesponse errors. Response rates from 70.percent are typical for tranSportation surveys. However, there is no staneWd response rate that should cause concern for analyses. A 20 percent response race where the r-eSpondents are similar to the nonrespondencs i.s superior to a 70 percent response rate when; the respondentS differ &om the nonrespoodents. '

The best way to treat nonresponse is through prevention: Designers should consider response rare during every stage ofsurvey development. A classification scheme for non response developed by Cochran (I 977) is helpful for designers. The classes include:

• non,om-agc people in the sample are not given the opportunity to respond to the survey. • unable to answu: people are reached but do not have sufficient information ro answer the question. • unwilling 10 answ~r: people are reached but refuse to respond.

• not-at-home: survey workers try but cannot reach some people in the sample.

Designers can prevent many noncoverage problems by using an up-to-dace frame (list of units in the population). Most "unable to answer" problems should be detected during pretesting. If respondents are frequently unable co answer a question,. the designer can change the format of. reword, or discard the question. Skillful introductions, proper inducements and polished questions will reduce the number of people unwiUing co answer. Finally, survey workers should repeatedly attempt to contact not-at-home nonrespondents. A procedure for determining the optimum number of callback attempts is available from Cochran (1977). Stopher and Sheskin (1982) describe ways tO detect and treat nonresponse bias in transportation surveys. Usually, analysrs can judge the c:nenc of nonresponse bias after a special effort to call or visit a small number of nonrespondencs. Unbiased nonresponse can be coneaed by distributing the survey to more people without violating the original sampling plan. Analyses can correct biased nonresponse mathematically using the different characteristics of the respondent and nonrespondenc subsamples.

11.0 REFERENCES Alreck. P. A. and R. B. Settle. The Surory 1/nMT'(h Handbook. Homewood, IL: Richard D. Irwin, 1985. Brog, W., E. Ed., A. H. Mcyburg and M. 1. Wermuth. "Problems ofNonreporced Trips in Surveys ofNonhome Acrivicy Patterns.• Tri11Up0rt4tion llntarrh &am:/: fou17141 ofrhe TTlliiSJ>Drtation 1/nearch Board 891 (1982). Cochran, W. G. Sampling 'ftthniquu, 3rd ed. New York: Wt!ey, 1977. Converse, 1. M. and S. Presser. Survty QJmtWm: Handcrafting the Suzndardized Questitm1111irr. Newbucy Park. CA: Sage Publications, 1986. Creswell, ]. W. llntanh Design: Qua/iuuive, Qu=riUllive, an4 Mixed Methods Approaches. Newbury Park, CA: Sage Publications, 2009. Oox.sey, L B. "Respondent Trip F~uency Bias in On-Board Surveys.• Transportation ResraTth &cord: fou17141 ofthe T~rtllrion &uarr:h &an/944 (1983). Fink, A. and 1· Kosecoff. How to CAndua Survryt: A Step-by-Step Guide. Beverly Hills, CA: Sage Publications, 1985. Groves, R. M. and R. L Kahn. Survtys by ulephone. New York: Academic Press, 1979.

Luman, S. R. and C. F. Man.ski. "Alternative Sampling Procedures for Calibrating Oi.saggregate Choice Models." Transpor14rion Research Rrcord:Jou1714l.i7fthe Trr:nsportation Res=ch &ard592 (1976). Rea, L M. and R. A. P~r. Dttigning an4 Condstding Survty ReseiZT(h: A Comprrhmsive GuU/e- Third Edition. San Francisco, CA; Jo.ssq-Bass Publications, 2005.

Shesk.in, I. M. and P. R. Stopher. "Pilot Testing of Alternative Adminiruative Procedures and Survey InsmunentS. • TramportaiWn Resrarrh Rrcord:foumall1j'the T~rt4rion llntarrh Boarr/886 (1982). Stopher, P. R. and I. M. Sbeskin. "Method for Determining and Reducing Nonresponse Bias." Transportation &search Record: jollmAI ofthe Tramportarion Reuarrh Board886 (1982). Sudman, S. and N. M. Bradburn. .AJJting Questions. San Francisco, CA; Jossey-Bass Publications, 1982. 5 14 • MAfo:lUAl OF TRANSPORTATION ENGINEERING STUDIES, lND EDITION

AppendiX C :

... .......... .. ... ............ ........... . .. ... ......... ........... ........... ....... ... .

\

Statistical Analyses Original By: L Ellis King, D. Eng., P.E. UpJAudBy: Ba.rtUm]. Scbrouler, Ph~D. 1.0 INTRODUCT19N

515

2.0 DATA REDUCTION

516

2.1 Tables and Graphs

516

2.2 Frequency Distribution

517


520

2.4 Spatial Distribution

521

3.0 DESCRIPTIVE STATISTICS

522

3. 1 Central Tendency

3.2 Variability 4.0 STATISTICAL INFERENCE

522 526 528

4.1 Estimation

528

4.2 Reliability of the Sample

529

4.3 Significance Terting

532

4.4 Nonparametric Tests

533

5.0 CALCULAnON AIDS

536

6.0

REPORTING RESULTS

536

7.0

REFERENCES

538

1.0 JNTRODUCTtON fr:er field data have been collected in a uaffic engineering study, the informacion is arranged or tabul2c~d for visual inspection and analysis. Then the application of appropriate statistical techniques aids J.n malci.ng the proper and most effective evaluation of the study results. To ensure complete and accllf~-ce knowledge of existing traffic conditions, the field study must be designed and performed properly wich due coosideradon for the following statistical analysis: If engineers collect and analyz.e data with appropria te statisti C:::~! p rocedures, they can avoid inaccurate and improper interpretations of the traffic situation. Success in determif-1 ing traffic improvements depends greatly on the reliability and correct interpretation of the informacion rb-:;;~.t describes the traffic problem.

A

This chapter describes the arrangement of data into a convenient form in the section "Data Reduction, • and discusses the subject of statistical analyses in the sections "Descriptive Statistics" and "Statistical Inference.• "Descriptiv( Sc~: tiscics" is concerned with summarUing the data collected in a traffic engineering study, while "Stacisticallnkreoce: describes the procedures for the devdopment of statistical estimates and the testing of statistical hypotheses. The 6n. ,:;;;;a) section of this chapter describes some of the calculation aids that are available. Appendix C • 51 !!!5

This ap~ndix describes very basic statistical analyses used by transportation engineers. Detailed guidance on statistical analysis principles for transportation applications can be found in Washington et a!. (2003) or other general statistics references (Meier and Brudney, 2002, Moore and McCabe, 2006, Grecnshidds and Weida, 1978, Taylor and Young, 1988). When anticipating complex analyses, engineers should seck the advice of a professional statistician prior to data coUection. Costly and frustrating analysis mistakes can be avoided with thorough planning and good professional advice beforehand. Similarly, while spreadsheet software allows for basic statistical analysis, more sophisticated tools may be necessary for spccialittd applications.

2.0 DATA REDUaiON This section describes approaches for data reduaion and display of data in the form of tables and charts, with emphasis on common distributions of data found in transportation applications. When data are arranged systematically ac-cording to the frequency of occurrence for different si.tt classifications, such as the grouping ofspeeds in a spot-speed study, the resulting tabulation may be shown as a frequency distribution or histogtarn.lf the coUcccion of information is based on the time of occurrence, such a.s the passing of vehicles in a traffic volume srudy, the array of values may be treated as a time series or temporal distribution. Data that have been recorded according to geographic loca.tion, such a.s the location of accidents on a street map, may be plotted a.s a. spacial distribution. These distributions arc commonly used in summariung and presenting the results of rransportacion engineering studies.

2.1 Tables and Graphs Data from tranSportation srudic::s genenlly can be presented in tables or gnphs. Chapter 3 and Appendix D provide suggestions on the format and appearance of tables and figures, while this discussion is focused on the content of cables and figures. Use of a computer makes data manipulation a rdativdy easy cask, and numerous tables may be generated that present various a.spcas of the data. Most handheld and roadside data coUcccion equipment provide a software incerfacc, in which data are tabulated (and sometimes analyttd) automatically. The simplest table is a straiglu-linc listing of the data in the order in which they were recorded. Exhibit C 1-a shows such a listing for 66 spot speeds coUccced on an urban arterial with a posted speed limit of 50 miles per hour (mph) (80 kilometers per hour (kmlh]). The data arc oflittlc value when presented in this form and arc more useful when ordered from lowest to highest value, a.s shown in Exhibit C 1-b. The extreme values and groups of daca are now apparem. AJccrnativcly, basic spreadsheet functions can be applied to the unordered dataset to provide descriptive statistics discussed later in this chapter.

. 49 43 46 40

44 47 49

....

~Q~~ .. ~ ·

·-

.. .

48

52

45 45

47 41

45

53 40

45 46

45 41

44 38

45 46

;~tt.,7{'1;

.

,.. -··

a) Random speed obse.rvacioos in order ob.crwd

37

35 46

. -·

44

50 48

43 42

4()

41

48

48

39 47

48

45 41

55 48

51 46

51

54

42

-. . ,..

46

41

50

35 44 52

53 52 49

47 42

42

46

44

49

41

49 45 44

b) Obsuvarioa.s ordered in increuing order

35 41

44 45 47 49 52'

35 41 44

37

38

41 44

42

45

45

46

47

4.7

47

44 46 48

49

49

49

49

50

52

'53

53

54

55

11110 • MANilA! t'IF

44

39 42

TRAN~Pt'IRTATIIiN Fl\lt;IMFFRI~Ir.

40

40

42 45

42

40 43

45 46

45 46

43 45 46

48

4.8

48

41 44 45 46 48

50

51

51

52

46 48

(TI II"\1' ( 1Mn

~r.oTor'"'

~-2

Frequency Distribution

Analysts can condense and summariu: the large number of daca poinu shown in Exhibit C I by developing a frequency table. The consuuction of a frequency table first requires the sdection of the group or class siu:. If roo few or too many groups are sdecred, detail is lost in the data reduction. The appropriate number of classes generally ranges from 10 to 20. After the field data have been collected and tabulated, the range in measurements is de~erm ined by subtracting the lowest from the highest values. The range is then divided by 10 and 20 to estimate, respectively, the maximum and minimum class si~ thar are reasonable for the observed data. A convenient class siu: is selected within these minimum and maximum values. After the analyst determines the class siu:, he or she selects class intervals to completely define the acrual sample values that are contained in each class. These limiu arc written to the same precision as the original data. Exhibit C2 shows a frequency table for the speed data in.Exhibit Cl. ~

•

·..

~.~

.

..-

.c:; ,_ -....~ ....

D

Class

Boundaries

.

..

.

. . . ._,.. .< ...

....

~·

au.

i fi!D

: ·:- ~~ ~+t&}~>~»·w1::~:!'~~~5

Cumulative Frequency

Rd&live

Midpoillt

Class Fl't<}IWICf

Frequet~cy

33-34

33.5

0

35-36

35.5

37.38

Number ·

Relative

0

0

0

2

0.030

2

0.030

37.5

2

0.030

4

0.060

39-40

39.5

4

0.060

8

0.121

41-42

41.5

9

0.136

17

0.257

43-44

43.5

8

0.121

25

0.378

45-46

45.5

15

0.227

40

. 0.605

47-48

47.5

10

0.152

50

0.757

49-50

49.5

7

0.106

57

0.863

51-52

51.5

5

0.076

62

0.939

~3- 54

53.5

3

0.046

65

0.985

55-56

55.5 .

1

O.Gl5

66

1.000

57-58

57.5

0

0

66

1.000

66

1.000

CJ.a.;s

Inu;:val

32.5 34.5 36.5

~

38.5

..

..

•

!

40.5 42.5 44.5

..

46.5 48.5 50.5 52.5 54.5 56.5 58.5 Total

.

A _ _ ...... ....,~ ..

r

-

1::'4"9

Afcer che class intervals have been recorded on che frequency cable, each field-recorded vehicle speed is placed in che appropriate class interval. Summing the number of vehicles in each class interval gives the frequency of occurrences for each of che intervals selected in che spoc-speed study. The resulting cable of occurrence in che various class intervals is known as a.frtqumcy distribution, and the sum of che class frequencies is equal co che sample size or total number of field observations. A reLuive.frequency distribution is obtained by dividing che cotal number of vehicles in each class by che total number of vehicles in che sample; che relative frequencies muse total 1.0. Relative frequency distributions provide a more convenient format for data summaries because the user does nor need co refer to che sample size. The use of relative frequencies, expressed cicher as proportions or percentages, also permi!S direct comparison of che resul!S obtained from different studies wich varying sample sizes. Exhibit C 2 shows boch frequency and relative frequency. The cumuiJltivt foqumcy dim-ibutum provides a listing of che total number of observations chat are less chan or greater chan a specilied value. A cumula.tive frequency distribution is developed beginning with the class at eicher che bo.t.tom or che .top of the frequency cable. The class frequencies, ei.ther acrual or relative, are summed in the selected direc.tion wich respect .to the class boundaries. The class boundaries are the most extreme values included in a given class and are calculated to one-half unit of greater precision chan che original data. Exhibit C-2 shows che cumulative frequency distribution developed from che frequency distribution discussed previously. If the frequency summation is from .the small to the large values of che study variable, as in Exhibit C2, .the analyst matches che higher class boundary wich the corresponding cumulative frequency. The analyst seleru che lower class boundary for this matching when he develops che cumulative frequency distribution from che large co che small measurements. The informacion contained in a frequency distribution cable may also be presented graphically as a histogram, frequency diagram, or cumulative frequency diagram. When plotted for che complete range of observations, chese diagrams provide an opportunity co evaluate the data shape wich regard to conformance to a recognized distribution, degree of symmetry and presence of irregularities. Analysrs construct a histogram by plorcing a diagram with class boundaries on che horizontal axis and che corresponding frequency or relative frequency of occurrence on the vertical axis. The resultant rec.tangle represen!S che number of observations wi.thin each class interval. Exhibit C3 is a histogram for che classes of spot speeds shown in Exhibit C2. Al.though actu2.1 or relative frequencies may be sdeaed, the use of rdacivc frequencies or percentages provides a diagram chac has general application and facilitates comparisons since it eliminates the in.8uence ofsample size. Nevertheless, the sample size should always be reported to fully document the study.

25%

::c:

2 20% Gl

E;

{;' 15% c

Cll

:I tr

! 10%

Ll. Cll

~

"'

'i

5%

0::

0%

+---..3334

3536

37 38

3940

4142

4344

45 -

47 -

49 -

51 -

53 -

55-

57-

46

48

50

52

54

56

58

Spot Speed Bin (mph) 518 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDinON

25% C'

c

...~ .e:

20%

Cll

(';' 15% c Cll

:I

cr

~ 10%

LL. CD

>

;; ~

iii

5%

0:::

0% 30

35

40

45

50

55

60

Spot Speed (mph)

The.fr(qutncy dUlgram is cons eructed by plotting a graph with the midpoim of each class on the horizonral axis and che corresponding frequency or relative frequency of me class on the vertical axis. Exhibit C-4 shows a frequency diagra.fYl for the classes of spot speeds from Exhibit C2. The points a.re plotted and then connected by straight lines. The diagram is closed by connecting the extreme points with the next class midpoints that have a frequency of zero. k wich me hiscogra.rn, actual or relative frequencies may be selected for ploning. The cumultztive frequency diagram is constructed by plotting a graph with a class boundary of each class on the hoc~ zontal axis and the corresponding cumulative frequency of the class on the vertical axis. The higher class boundary lS matched with the corresponding cumulative frequency when me frequency summation is from the small to the!a.r~e values of the study variable, and the lower class boundary is selected for chis matching if the cumulative frequency is sununed from the la.rge to me small observations. A smooch curve is used to connect the plotted points, and exccnde:G to the rwo extreme class boundaries with cumulative frequencies Q( 0 and 1. On cumulative frequency diagram.S, analysts generitly express class frequency on a relative basis. Exhibit C-5 shows a cumulative frequency diagra.rn of ci::J.e spot speed data from Exhibit C-2.

Appendix C • 51~

100% 90% ~

/.

85th Percentile Soe !d

/.

80%

c

41

e

70%

~ c

60%

!

I

41

!'

J

:J

.50%

I

II.. Cl

>

40%

;:1

• "5

E :s

(,)

30% 20%

I

/

/ v

15th Percentile Soeed

...,.. ~

0% 30

35

I I I I I I I

.i

I I I I

L

I I I

' I I

I I I I

/o

10%

I I I

I I

I I

I

t

40

45

so

55

60

Spot Speed {mph)

Cumulative frequency
Us~g percentiles of speed elimioates the imp:a.ct of outliers, while encomp:a.ssing a grc:a.tcr portion of the wnple than the mean observation. N :m c:x:m1ple, pedestri:m wallcing speed affects sign:a.l riming (duration of the B:a.shing "don't walk" pluse). If the phase duration were designed for the mean walking speed, 50 percent of pedestrians would be w:Uking too slowly to make it :a.cross the entire speed. On the other h:md, designing for the 15th percentile speed ensu= 85 percent of pedestrians luve sufficient time, while being more efficient th:m a design for the slowest observed walking speed.


=

Data that arc observed :md tabuhted with respect to the time of occurrence provide a tiTM smes distribuwm. The time interval for recording data is selected to :a.ccommoda.te the pwposc of the srudy :md m:a.y r:mge from a few seconds to several years. For example, vehicular volumes ue tabuhted in five-min. intervals for the detcrmin:a.cion of poking ch.a.nctcristics, while traffic :a.ccident scatistics are frcquendy summ.arized on :m :mnual b:a.sis. Exhibit C6 alusrrates time series distributions for traffic volumes by hour, day :md month. Observations that are collected in shore time intervals may be combined to produce summary totals for longer rime series. For aample, five minute (min.) volume countS c:m be summed to provide hourly values, which are then summed to produce the daily total. This swlum.cion c:m be extended to the largest time intc:rnl desired.

520 • MANUAl OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDITION

I ....

~

....

I:

1110

!

~ ~

t!s

!

i

::

110 100

PO

.I

I

•••

1111_1

110

B it =:

120

~

120

.. . .....

140

. ... I ... ~

.... 100

II:

§ ~

PO

-

I

Oi

J'~'i ' S'i'!'e'!'§'H'§'~

eo

I

f--

til

IIII

- ' ~·~·~·~ · ~·I!~

MONTHOf'YCA._

DAYOP"W&EM OAILVtAAII'I'IC UANA.....,. •

..aN1HLY fRAP'I'IC VAMoft.110HS

~

i

• • •

.

•

~

z ~

I

~

!

I

7

• ..:

I

r-

" 1

0

1111

n

_.

~ · :;:·~

~

n•

f1

l ' I'l'l

.,.,. ...... .• ~

.

...

·=·~:~·=·= ~

..:.

:•- = •

n

,·~·,·c·,·~·~·~·· t i ~ a ~ ~ ~

HOUIIIP2':MOD HOUIJIILY1AAP'ft0 WII.NAT'IQNS

0

_,..ST.

•

ELM ST.

2.4 Spatial Distribution Traffic information is often presented with reference to the geographical location where the event occurred or is present. For aample, vehicular volum~. rravd speeds and ddays, tnffic accidents, trip d~ires, parking suppli~ and . demands, public tnnSportation usage and inventori~ of traflic conuol devi~ (TCDs) and regulations are ofte11 summarized a.s spatial distributions. Spot maps, route logs and areawide maps are commonly wed to illwrrate data disuibutions according co geographical location. The spot map of uaflic accident locations shown in Exhibit C-7 is an aarnple of a spatial distribution.

·o-

I~ DD ~1)~ .~=:::_ CJ~c._ ·rcr J=:ktJ o o.cr o. _Lc::tC1J __ ~~ao--l:J-o-o ~ ~"~ ~·:___ s

o .

9tJ D

.

0

•

D D d~~ =-<--h>>

,.A.TA\.

OfMC.R

0 •

.

.._

~: ·oo~

0

<: -

I

Spatial distributions are analyzed using geographical information system (GIS) software. For example, the analyst may be interested in all records at or within a certain "buffer" of an intersection. Details on geospatial analysis·can be · found in the li terarure.

3.0 DESCRIPTIVE STATISTICS The purpo~ of descriptive statistics i.s to describe a collection of data by a few representative values. Descriptive Itamtics are calculated from a sample of observations with the objective to estim01.te the true (bur unknown) paramans of the overall population (of drivers). Tests of statistical significance as discussed in section 4 are used to test the confidence interval of the sample statistic or to make comparisons between two samples (for example, before-and-after speed samples following a traffic calming device installation). Descriptive statistics also permit the efficient evaluation and analysis of the traffic problem of interest. Descriptive statistics involve the central tendency, variability and shape of the data.

3.1 Central Tendency Yuious measures of an "average" are available to describe the cmrral tmdmcy of a ~t of ob~rvations. Avrnz~ i.s'a broad term that generally includes many measures of central tendency. Commonly used measures of cencral tendency for summarizing the results of transportation engineering studies include the: • arithmetic mean; • median; and • mode. The mean is generally located near the center of the data distribution, as illustrated for the frequency distribution of spot speeds in Exhibit C-8. Depending on the shape of the distribution, the vuious measures of central tendency may or may not coincide.


i

!

Speed

3.1.1 Arithmetic Mean The arithmetic mean or average is the most common measure of central tendency and is obtained by dividing the sl.l-.rn of the individual observations by the total number of observations. The general expression for the mean of unclass.ed data is: LX£

i = -

n

EquationC-1

where:

x

=

arithmetic mean

Lx, = sum of individual observations n

= total numbet of observations

If the measurements have been placed into classes as shown in Exhibit C-2, analysts use the following relationship -c:o compute the arithmetic mean.

Appendix C • SZ 3

x=

J:,ftul J:.ft

Equation C-2

where:

= arithmetic mean

x

L/; u Lfi

I

= sum of products of frequency and class midvalue for each class = sum offrequencies for all classes

Most procedtires for determining the minimum sample siu in various transportation engineering studies are based on the mean as the desired measure of central tendency. In scacistical inference, the mean is generally the most efficient estimator of the average value for the population characteristics being studied. However, in some cases the mean can be skewed by oudiers that increase or decrease the estimate. 3.1.2 Metlilln The median represents the middle value in a series of measurements that have been ranked in order of magnitude and divides the measurements into two equal parts. When the number of observations is odd, the median is the middle value in the list of ranked measurements. For an even number of ranked measurements, the median is the arithmetic mean of the two middle values. The 50th percentile value is equal to the median. The median is a useful average measure because it is less affected by extreme values than is the arithmetic mean. 3.1.3Motle The mode is that value or class midpoint that occurs with the greatest &equency in the distribution of data. The mode is not a reliable "average" for small samples because sevecal values with the same frequency can occur by chance. As the sample siu increases, both the median and the mode become more meaningful as estimates of central tendency. Depending on the data set, it is possible to observe multiple modes or peaks in the distribution. An aample would be a gap distribution of a population that includes two different driver types (aggressive and conservative). 3.1.4 E.xamplu The following aamples of computations for the various measures of central tendency are based on the observations in Exhibit C-1 for unclassed data and Exhibit C-9, which is an extension of Exhibit C-2, for classed data. When speed measurements are grouped into classes, Equation C-2 is used to calculate the mean speed. From Exhibit C-9, the summation over all classes is obtained for the product of the class midpoint and the corresponding class frequency. This value is then divided by the sum of &equencies for all classes (that is, the total number of observations) to provide the arithmetic mean, which is 3,006/66 or 45.5 mph (73.2 km/h). This procedure is similar for unclassed data, except that in accordance with Equation C-1, the individual measurements are summed in the numerator, which is then divided by the total number of observations, again giving a mean speed of 45.5 mph (73.2 kmlh).

'

-

-,..;-

•

~~~ ·~--

Class Bounduies

. .... .

-~~

.

.. . .. .... .

~

.........

~.,.

.-.,.1>..• . .• 0 .

.

..

:..~---... -!

~

~~11~~,

~-

. .

. .-

;...-l_-trf~~

Class Frequency

Oass Interval

Clau Midpoint u

(

fu

fu

33-34

33.5

0

0

0

35-36

35.5

2

71

2.521

37-38

37.5

2

75

2,813

39-40

39.5

4

158

6,241

41-42

41.5

9

374'

15,500

43-44

43.5

8

348

15,138

45-46

45.5

15

683

31,054

47-48

47.5

10

475

22,563

2

32.5 34.5 36.5 38.5

-40.5

..

42.5 44.5 46.5 I

48.5 49-50

49.5

7

347

17,152

51-52

. 51.5

5

258

13,26 1

53-54

53.5

3

161

8,587

55-56

55.5

1

56

3,080

57-58

57.5

0

0

0

Total

66

3,006

137,910

50.5 52.5 54.5 56.5 58.5

The median or 50th percentile value is found for classed data by a linear interpolation across the class in which the middle value occurs. In the sppt-speed example, this value is the mc:an of the 33rd and 34th frequencies and lies in the class with boundaries of 44.5 and 46.5. A linear interpolation for the location of33.5 in this class is accomplished by calculating: . ( 33.5 -25) median= 4-4.5 + (46.5 - 44.5) x _ = 44.5 + 1.1 = 45.6 mph 40 25

The values of25 and 40 are, respectively, the cumulative frequencies at the class boundaries of 44.5 mph and 46.5 mph (71.61unfh and 74.8 kmlh). For the unclassed data of Exhibit C-1, the median is the arithmetic mean of the twO middle values and, as mph, is calculated as: 45+46 2

median= -

= 45.5 mph Annonrliv

r •

li:,t;:

The mode is the class midpoint for that class with the largest frequency of occurrence and is equal to 45 mph (72 km/h) from Exhibit C-9. For the unclassed data of Exhibit C-1, the most frequently occurring value or mode is 45 mph (72 km/h).

3.2 Variability A statistic describing the variability, dispersion, or spread of sample data is also valuable. The following three measures of variability are widdy applied in the summary of transportation engineering data: • Range •

Standard deviation

• Standard error Although the range is easier to compute, the standard deviation is a more reliable measure of data variability. The standard error scales the standard deviation as a function of sample si1.e.

3.2.1/Umge The range is the interval berween the smallest and the largest observations. The range is markedly dependent on the size of the sample and is greatly infiuenced by outliers or erroneous measurements. This summary statistic is readily inrerprera.ble and is often used with small samples of 10 or fewer observations. As the sample size increases, the variability within the sample may increase while the range remains essentially unchanged. Due to this phenomenon, the range should not be used in any comparative evaluation, such as before-and-afrer 'analysis, that involves saniples of different sizes. However, the range is useful to give readers an idea of outlier observations that may be hidden by the average. For example, the mean estimate in a delay srudy may mask the fact that isolated drivers experience much longer wait times (while most have lower delays).

3.2.2 St4n~Uzrd Dn~ialion The more important measure of variability is the standArd tkviation, which is the positive square root of the variance. The variance is the sum of squares of the deviations of the observations, less one. The apression for sample standard deviation is wrirren in the following form for unclassed data:

.ECxt -i)2

s=

n- 1

Equation C-3

where: r

= standard deviation

X

= arichmetic mean

x, •

individual observation

n "' total number of observations The n-1 term in the denominator of Equation 3 is related ro the statistical concept of ekgrus offrmJqm, The n observations and variance have n-1 degrees of freedom since the deviations of the n observations muse sum co uro and only n-1 deviations can be used to determine the remaining one. The following equation for standard deviation is applicable to classed data.

$>::


Equation C-4

: where:

! " standard deviation u1 = class midpoint for i<~> class

J; ,.

frequency of i"' class

The standard deviation increases in value as the observations become dispersed at greater distances from the mean. Frequency distributions are shown in Exhibit C-!0 for data from two spot-speed srudies with different dispersions. The standard d~ation for the data with large dispersion would be greater than the standard d~ation for the data with small dispersion. "I)lc standard deviation suffimary statistic provides a reliable variability measure that has direct application in various Statisci.cal inferences.

I

i

J

s,-

If the shape of the data is approximately a bellshaped curve or the normal probability distribution curve, multiples of the standard deviation can be expressed on either side of the mean and represent limits that con~ various percenrages of the total values in a selected sample. Limits oft I. 2 and 3 standard deviations about the mean contain, respectively, 68.3, 95.5 and 99.7 percent of the coral observations. These relationships arc shown in Figure C-11. 3.2.3 StandArd Error The standard error of the estimate is calculated by dividing the standard deviation by the square root of the sample size.

s

SE

=..[Ti

Equation C-5

.................. _......,._:a.. ... .......... .. x-a.-·

~

61.3 • . ,......_. .......... , .. ...,..~

where:

SE

=

~

_.,.... ~ ,

standard error

.. standard deviation

n

• sample size

3.2.4 ExAmples The following illustrative calculations for range, standard deviation and standard error are based on the spot-lpCecl data shown in Exhibit C-1 and C-9, for unclassed and classed data, respectively. The range for the classified daca i .S the difference between the upper cl2ss interval for the highest speed group and the lower class interval for the lowes -c speed group and is equal to (56 - 35) or 21 mph (34 km/h). For unclassed data, the difference between the maxin:u~ and minimum observation is equal to (55 - 35) or 20 mph (32 km/h), the exact value. The application of Eqwrio .C-1 C-4 permits the computation of the standard deviation for classed data. The numerator contains the followingclas.6 summations, which are tabulated in the "total• row of Exhibit C-9.

Appendix C •

SJ.71"

• Product of class frequency and square of corresponding class midpoint • Square of product of class frequency and corresponding class midpoint • Class frequency The denominator is equal to the sum of the class frequencies less one. The standard deviation is then determined by taking the positive square root of the value that results from the indicated algebraic operations on the various summations in the numerator and the denominator of Equation C-4. The summary figures for the spot-speed example are: 2

3006 137,910-bb 66 - 1

md the squm root of this expression gives a value of3.9 mph (6.3 kmlh) as the standard deviation.· The standard deviation for the unclassed data in Exhibit C-1, calcularcd according to Equation C-3, using actual measurements rather than class midpoints and class frequencies, gives a value of 4.4 mph (7.1 krnlh). The lack of agreement between the two calculated values is an indication of the loss in accuracy that may occur when data arc grouped. The standard error is calculated by dividing the standard deviation by the positive square root of che sample size. For the sample in Exhibit C-1, the sample size is 66 observations and the resulting standard error is (~) or 0.54 mph (.87 kmlh).

4.0 STATISTICAL INFERENCE Various ~ques of Statistical inference provide valuable tools for the interpretation of the results of tra.ffic engineering studies. Statistical i.nli:rence permits the generalization of sample findings inro Statements about the populuion from which the sample was taken. Probability considerations are involved in the development of statistical inferences. Several useful procedures for statistical inference ace presented below in sections on estizmtion and on significance testing. Additional information on these twO subjects is given in readily available textbooks on statistics (for c:xample, Washington et al., 2003).

4.1 ~stimation When sample measurements of vehicle speeds, volumes, occupancy and other characteristics are made, the tabulated resultS coo.sist of a number of different values and a single value is often selected from the array co represent the data. This representative value, or average, is genccally computed as the mean, the median, or the mode. This average value is then generally taken not only to represent 'the sample of traffic observations, but also to represent the entire population of values from which the sample was taken. The resultS of many srudies are reported in the form of single-value or point estimates. The accuracy of a sample: depends on two factors. The first factor is chanc~. which is the probabk amtlUnt ofnror due: co change in the average of any sample and which can be estimated. This section describes how to calculate this estimate. The second factor affecting the accuracy of the sample is the mahod ofsampling. The sample represents only the population &om which it was drawn. In a speed srudy, if only the fastest speeds were recorded, the average: would be: representative of only the fastest dement of traffic on that street. Similarly, a speed study would commonly focus on free-Bowing vehicles that arc not inhibited by (slower) preceding vehicles. To obtain a truly unbiased or representative sample of speeds in a spot-speed study, vehicles must be selected at random in such a way that each vehicle has an equal chance of being included in the sample. The requirements for a representative sample are: • the sample must be selected without bias; • the componentS of the: sample must be complctdy independent of one another; • there should be no underlying d.i.ffcrenc;($ b~en areas from which the data are selected; and • .conditions must be .the same for all items constirucing the sample. t

•••ttt
... ,. . .,., . . . ,.- ........ nT

I 'T'll"\tl

,.,.,..111,.,.~111 ,..

1" ...11~1,.,-

~"""'~

r",..,rTII'"\11.1

; For example, jfa laser or radar gun is used, a random sample could be obtained by measuring every nm vehicle, regard\ less of its spee
4.2 Reliability of the Sample 4.2.1 M'/fU'1'9 D1Z14

Assuming the sample was chosen in an unbiased manner, it is possible to calculate the degree of error that may be due to chance by alculacing a confo/mct inttrVal {CI) estimau:. A CI of the population mean is usdUl in reportin g res~ts of various transporution srudies and is expressed by the following inequalicy:

_

tas

_

taS

..fit

.

..Jn

x--
where:

I' = mean of the populacion

X = mean of the sample s

• stand.a.rd deviation of the sampk

t~

= statistic of the 1 disuibution for (n-.1) degrees of freedom and the probabilicy defined by the subscript

n

~ total number of observations

a "" (1.0 - confidence coefficient) NOTE: The expression (:P..fil) in this inequalicy is defined as the standard error of me estimate for me mean discussed above. Exhibit C-12 presents typical values of the t statistic for various degrees of freedom and for selected levels of alpha (a). The degrees of freedom are defined as the sample size less one, and the selection of a confidence coefficient is generally confined ro the probabilicy levels of 0.90, 0.95 and 0.99. The corresponding values of ·a~ are 0.10, 0.05 and 0.01 and are listed as column headings in Exhibit C 12. A confiden~ coefficient of95 percent provides an acceptable t..s e..s interval escimate for most purposes. The terms x - {ii and i + {ii define, respeccively, lower and upper limits of the confidence interval. The interpretacion of the probabilicy statement for an interval estimate is that the confidence interval contains the population value with a probabiliry that is equal to the confidence coefficient.

Appendix C • 529

t~:m;~~4t~~·~j,.~~~~[~~it~:.~~\}1,~)~~~~~~~1'\$it;o~~·~~~$~~~~ ~. ~ J.• ~ '.• :.. ~~~'~iJII'~~ _"••;:."',;! - ~ ,!-;.~ :i'£~-t\~.:;;~.:.. ' ·"}~'1:~.~~~\t~' .. ' .•>-~f~..t"::~, .J~... }~~~;...~~~-tt~r.~~.~~~~-.~~ \:, •.a , '"11'":; ,,

t''

lA

Confidence Level

90%

95o/o

99o/o

Degrees of Freedom

(alpha • O.OS)

(alpha • 0.025)

(alpha = 0.001)

1

6.314

12.706

63.657

2

2.920

4.303

9.925

3

2.353

3.182

5.841

4

2.132

2.776

4.604

5

2.0 15

2.571

4.032

6

1.943

2.447

3.707

7

1.895

2.365

3.499

8

1.860

2.306

3.355

9

1.833

2.262

3.250

10

1.812

2.228

~. 169

11

1.796

2.201

3.106

12

1.782

2.179

3.0S5

13 14

1.n1

2.160

3.012

1.761

2.145

2.9n

15

1.753

2.131

2.947

16

1.746

2.120

2.921

17

1.740

2.110

2.898

18

1.734

2.101

2.878

19 20

1.729

2.093

2.861

1.725

2.086

2.845

21

1.721

2.080

2.831

22

1.717

2.074

2.819

23

1.714

2.069

2.807·

24

1.711

2.064

2.797

25

1.708

2.060

2.787

26

1.706

2.056

27

1.703

2.052

z.n9 z.n1

28

1.701

2.048

2.763

29

1.699

2.045

2.756

30

1.697

2.042

2.750

40

1.684

2.021

2.704

60

1.671

2.000

2.660

120

1.658

1.980

2.617

co

1.645

1.960

2.576

!530 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

•

:The following example illustrates the development of an interval estimate of the mean speed for the classed data that !u-epresented in Exhibit C-9. Sample values of mean. and standard deviation were previously calculated as 45.) mph ·

taS

~nd 3.9 mph (73.2 km/h and 6.3 km/h), respectively. The term Vn is now calculated as

2.000 • 3.9

-166 o r 0.96 mph ( 1.54 km/h) for a confidence coefficient of95 percent. This amount is subtracted from and added to rhe sample mean to determine, respectively, lower and upper limits of the interval estimate, which is written as 44.5
tas {n

4

calculated as 2.000

(.f66)

or 1.1. This results in lower and upper limits of 44.4 and 46.6,

This interval is the measure of precision of the sample, assuming the sample selection is random and unbiased. lf the sample error is too large, more observations should be obtained, making certain, of course, conditio ns have not changed since the original observacions were made. 4.2.2 Proportions When a study is made in which the result is expressed as a proponion or a percentage, the precision of the estimate is indicated by rhe following inequality:

~

~

Equation C-7

p-~~~<¢
$

= proponion of the population

p

=

q

=1.0- p

proponion of the sample

t = statistic of the t distribution for (n-1) degrees of freedom and the probabiliry defined by the subscript

n = number of observations

a = 1.0 - confidence coefficient

IPii

The term ..J-;;- is defined as the standard error of the estimate fo~ the proportion. Exhibit Cl2 provides rypic:J. values of the t statistic for the various degrees of freedom and for selected levels of a. The determination of the sample: precision for proponions or percentages is very similar in procedure and interpretation to the situation for the mea~ of a sample discussed previously. & an example of a percentage estimate, 400 drivers were randomly sampled in regard co their location of em lo en..::=

IPii

0.3. 0.7

and 120, or 30 percerit, reported they worked dowmown. The term' -In is calculated as L.960 times 400 a5 the precision of the percentage estimate for a confidence coefficient of 95 percent. That is, the estimate of 30 pero:n C:: is subject to a possible sampling error of :t4.5 percent, and the upper and lower limits of the percentage escimatear~ expressed as 25.5 «I>< 34.5 with a confidence of95 percent.

Appendix C • 531

4.3 Significance Testing Analyses ace often concerned with whether the difference in average values between rwo sets of sample data is statistically significant or merely due to chance variations that result from sampling. Significance testing is a valuable way to address those conarns. For instance, cerrain improvements might reduce the average parking duration from 40 min. to 30 min. The analyse must decide whether this reduction is really significant or is due to chance alone, and significance testing can help in that decision.

If rwo samples of data are taken from the same population, there will probably be a difference between the averages of these two samples. This d.iffercnce would be due to chance alone. A greater number of observations probably provides a smaller chance error, and hence a smaller difference berween the study averages, because the same population is being sampled. However, if rwo studies are made at different locations where conditions ace not identical, the two populations will have different mean values. This difference, added to the chana variation between the two studies, equals the to~ observed difference between the means of the two studies. Because differences between pairs of sample means from a given population occur only due to chana error, these differences are subject to the laws of probability and follow' a normal rurve. Any difference that is large enough to &II at an extreme point on this curve is not within the realm of chana error and represents a significant difference. The signifoanu urt for equality or inequality of the means of rwo populations with unequal variances is based on samples of 30 or more observations for each population. For smaller samples, nonpararneuic tests arc more appropriate as discussed in section 4.4.

4.3.1 RJng Sample D11t11 t:or raw data samples of individual observations, a test statistic for large samples (n:>30) is determined by the following formula: i l-iz

r== z 2 !1.+!1.

t=

nl

Equation C-8

Dz

where = statistic of the t distribution

X1.

= mean of first sample

x

a

s1

= SWldard deviation of fitst sample

11

•

mean of second sample

standard deviation of second sample

n1 = number of observations in first sample ~

• number of observations in second sample

The computed value oft is then compared with the aitical tvaluc (t), as obained from Exhibit Cl2, to determine the significance of the difference between the two sample means. The value oft, is selected in accordance with the specified level of significance (a). The value of 0.05, corresponding to 95 percent confidence, is often chosen as the lc:vd of significance, although an a of 0.0 1 to 0.10 is within the proper range for most evaluations of uansportation data.

If ~c computed value (either positive or negative) oft is greater than t? the d.iffercnce between the two means is considered significant and not due just to chana variations alone. When the calculated t value (either positive or negative) is less than the critical t value, the difference berween the two means is defined as nonsignificant and due to chance

variation alone.

,4.3.2 Proportions

.

~alysts p~rform significance t~ting for the difference b~tween two proportions or percentages in a similar manner bccept chat they us~ the following ~quation to compur~ the t statistic.

' Pt-P>

t=

~

1)

Equatio n C-9

-/ Poqo {f. + n2

where: t

= statistic af th~ t distribution

p1

= proportion observed in first sample

p1 = proportion observed in second sample n 1.

= number of observations in first sample

n1

•

number of observations in second sample

p0 = (p1n1 + p1 n.) I (n1 + n.) 'lo = l.O - Po Either proportional or percentage valu~ may be used in Equation 9 for p md q.

4.3,3 &mnples The following aample is presented to illusttate significance testing. Under old parking regulations, a srudy s howed 185, or 28.5 percent, of 648 vehicles were parked overtime. After new parking regulations were adopted, a similar srudy revealed that 119 of 512 vehid~ or 23.2 percent were parked overtime. Th~ weighted average, pO' of the cwo 28.5 X 648 + 23.2 X 512 percentag~ is first calculated as 648 + 512 or 26.2 percent, md 'lois equal to (100.0- 26.2) or 73.8 percent. The various valu~ are now inserted in Equation C-8 to determine the calculated value oft as (28.5 - 23.2) divided by the square root of 262 x 73 ·8 • (~ + or 2.038. The critical t value is obtained from Exhibit C 12 as 1.960 for a significance levd of 5 percent. Because t is larger chan t1 the difference between the two percentag~ is significant, and the new parking regulations appear to be effective in reducing overtime parking. An example for nonproportion data is given in section 6.

s!V

4.4 Nonparametric Tests Significance resting requires large samples sdeaed from a nonnal population and requi= numerical data. However, nor all transportation study data meet these requirements. When this is the case, various nonparametric signi6cmcc t~ts are available for evaluating the srudy r~ults. Th~c distribution-free t~ts are particularly convenient when the sample size is small md/or the data are qualit;ltive rather chan numc~cal. The following example illusttat~ the use of a nonparametric rank-order correlation ~c to see whether the speeds of veJ¥cles on two freeways with greatly differing traffic volum~ arc independent of each other. The average vehicle speeds shown in Exhibit C 13 were recorded at seven randomly ~eaed locations on each freeway. The speeds for each freeway have been ~ from 1 for the highest to 7 for the lowest. The differen~ between the rank3 have been calculated, squared and totaled. The tat Statistic is computed as follows:

Appendix C • 533

1$ = 1- 6'J;df n(n2- l)

Equation C-10

where: r, = Spearman rank-order correlation coefficient ~ =

n

difference becween samples

= number of measurement pairs

Critical values for r are given in Exhibit C- 14. Any calculated value greater than the critical value in absolute value is a statistically significant indication of independence.

Freeway 1

Freeway2

Freeway 1, X

Freeway2,

y


Difference

d =X-Y

dl

·~o.i))~~~·~t@f~-· ·. ~.~i.~?!!}""£¥~+.-':t~~~¥t:;:~ ;>-=-- ~---- .-:no:~~-~~r~~~;;::§ Significance Level N

0.100

0.050

0.010

5

0.900

1.000

1.00 0

6

0.8 29

0.886

1.00 0

7

0.714

0.786

0.929

8

0.643

0.738

0.88 1

9

o.600

0.700

0.83 3

10

0.564

0.648

0.794

11

0.536

0.6 18

0.818

12

0.497

0.591

0.780

13

0.475

0.566

0.74 5

14

0.457

0.545

0.716

15

0.441

0.525

0.689

16

0.425

0.507

0.666

17

0.412

0.490

0.64 5

18

0.399

0.476

0.62 5

19

0.388

0.462

0.60 8

20

-o.3n

0.450

0.59 1

21

0.368

0.438

0.576

22

0.359

0.428

0.562

23

0.351

0.418

0.54 9

24

0.343

0.409

0.537

25

0.336

0.400

0.526

26

0.329

0.392

0.515

27

0.323

0.385

0.505

28

0.317

0.377

0.496

29

Q.311

' 0.370

0.487

30

0.305

0.364

0.478

-

-

-

,--!

~

--

--

-

-

~

-

-

--

---

--

Substituting the example data in Exhi>r C-13 in Equation C9 gives a coefficient of. r,

=1 -

6x20

( _ ) 7 71 1

=1 -

0.36 = 0.64

The crirical r, value is obtained from Exhlbit Cl4 as 0.786 for a significance level of 5 ptrcmt and n = 7. Since c]::JL<: calculated r, is less chan the critical value, it may be concluded chat speeds on the rwo freeways are dependent (that i.5• are related to each other). Numerous other nonpararneuic rests are available for use when uansponacion data are nor suitable for the cbssic~ Statistical tests and only one has been presented here. Complete explanations and illustrative examples of the proc~ dures may be found in Taylor and Young (1998) and Washington ec al. (2003).

Appendix c

• 53~

5.0 CALCULATION AIDS Data analysis and evaluation may be greatly &.cilitated by the usc of calculators and computers. Simple sciencific calculators have built-in programs for calculating descriptive statistics such as means, standard deviations and variances, while more advanced calculators can be programmed to perform more complex statis~ical testS. Notebook-sized, battery-powered computers may be used to collect and analyze data at field locations and/or store data for later office analysis. Desktop personal microcomputers are available to most engineers and numerous statistical packages are available for use on these. Many of these packages have been developed specifically for engineering usc and include graphical capabilities. Specialized computer programs for analyzing the results of transportation studies are available from commercial sources, government agencies and user groups. Similarly, most commercial data collection equipment is typically shipped with a software program allowing basic statistical analysis. Exhibit C-15 shows a simple frequency dUgram and descriptive statistics for the spot-speed study data ofExhibit C-1 that was produced by a typical statistical analysis program and desktop microcomputer/printer.

Histogram of SPEED Mldpolat

Count

33.~

0 2 2 4 9 8 15

35.5 ~7.5

39.5 41.5

43.5 45.5 47.5 49.5 51.5 53.5 55.5 ~1.5

SPEED SPEED

10

7 ~

3 1 0 N 66 MIN 35

N=a"

••

•• ...... •••••••••

.......... ••••••••••••••• ........... ••••••• ••••• •••

•

MEAN 45.5 MAX

MEDIAN ~.s

STDEV 4.43

~s

Measures of central tendency and variability are also readily calculated in spreadsheet software using built-in functions in Microsoft Excel such as MEAN(a-z), MEDIAN(a-1.), MODE(a-1.), MIN(a-z), MAX(a-1.) and STDEV(a-z), where (a..z) corresponds co the range of cells containing the sorted or unsorted data. Microsoft Excel also allows the analyst to conduct basic Hests and other tests of signifi=:tce as pare of its analysis tool pack. The reader should consult the sofrware documentation to become familiar with these procedures.

6.0 REPORTING RESULTS The results ofstatistical analysis are presented in the form of tables or graphs as discussed earlier in this chapter. Depending on the application, specialiud display methods are available co show measures of both central tendency and variability in an dlicienr form. Exhibit C 16 shows a table with mean observations, as wdl as the standard deviation of multiple (hypothetical) samples of~peed observations coUeaed before and after the installation of the same tcaHic calming treatment at multiple locations. The standard deviations may be shown in a smaller font as appropriate. The sample sizes are shown in the column headings, providing all necessary input to perform statistical testS of significance. The exhibit highlights the effect oflow sample siz.c. While the speed impact of this (hypothetical) aeatmenc was reduced by 1.3 mph (2.1 km/h) for both sites 2 and 3, that d..ifference is not significant at site 2 due to a low sample siz.c. "",; a MANIIAI nF TRAN<;P()RTATI()N FNiiiNFFRINii ~TIIIIIF~. 2ND EDITION

In anorher application, Exhibit C-17 shows resultS of multiple simulation scenarios. For each (hyporhetical) scenario, borh rhe mean and standard error of travel time are shown, calculated from multiple iterations or runs. This form of display allows rhe user to visually compare rhe performance of two scenarios, while acknowledging the variability in rhe estimate. The c:xh.ibit shoWs ~ both scenarios 3 and 4 result in a low uavd time along the tested corridor. However, rhe error bars show the variabiliiy across runs is less for scenario 4. So, while scenario 3 results in the lowest ove~ mean, soenario 4 may be preferred since it also enhances uavd time reliability.

Travel Time Comparison

~

I

1

!

Error Bars at one StandardErrcrfrom lOiteratiCll"lS

45

~====--~:_~

======~--

40+---------------------------------------~

3P 30 25 20 15 10

5 0 Scenario1

Scenario2

Scenario3

Scenario4

ScenarioS

Appendix C • 537

7.0 REFERENCES Greenshields, B. D. and F. M. Weida, revised by D. L Gerlough and M. J. Huber. Statiuits with Applicariomto Highway Tra.ffit Annf];es, 2nd ed. \X'estpon, CT: £no Foundation for Transportation, Inc., 1978. Meier, K. J. and J. L Brudncy. Appli~d St4Jittia for Puh/ic AdministT1JrWn, 5th ed. Belmont, CA: Thomson ~Wadsworth, 2002. Moore, D. S. and G. 0 . McCabe. Introduction to the Practiu ofSratistia, 5ch ed. New York: W.H. Freeman and Company, 2006.

Taylor, M. A. P. and W. Young. Tra.ffit Analy1is: Ntw Tt:chnol

Appen d ix D ·:

..

.

.. ..

.

.

..

... . . . . . . . .

.. ..

.

.

.

... .

......

.

.

.

... 0

.

.

...

.

. . . . . . . ..

.

.

.

.

.. .

.

..

.

.. ..

c .... .

. . . .. .

.. .

. ...

0

••

Supplemental Material on Communicating Data Origimd By: Dtnma C Nelson, Ph.D., P.E. H. Douglas Robertson, Ph.D., P.E.

Edited By: Bastian]. Schroeder, Ph.D. RobertS. Foyle, P.E. 1.0 INTRODUOION

540

2.0

DESIGN OF GRAPHICS

540

2. 1 Section Overview

540

2.2 Graphics Design

540

2.3 Tables

542

2.4 Types of Chart

543

2.5 Design Considerations

549

2.6 Use of Color

549

2. 7 Graphics Checklist

550

3.0 WRITTEN REPORTS


550

3.2 Organization of the Report

550

3.3 Body of the Report 4.0 PRESENTATIONS

5.0

550

552 554

4. 1 Section Overview

554

4.2 Oral Presentations

555

4.3 Purpose and Scope

555

4.4 Organizatl9n of the Presentation

557

4.5 Answering Questions

557

4.6 Preparation and Planning Checklist

558

REFERENCES

559

Appendix D • 53!JIIII"

1.0 INTRODUCTION his a.ppendix presents supplementary material to Chapter 3, ~Communicating Data to the Public." While Chapter 3 contains key concepts and principles of data display and report writing. this appendix provides more detail on additional graphic types, sections of a written reporc and presentation techniques. The three ways co ddivet content co an audience are presented sequentially, consistent with Chapter 3.

T

2.0 DESIGN OF GRAPHICS 2.1. Section Overview Data presentation is an essential pan of most engineering reports and oral presentations. Graphics (sometimes called graphic illustrations or illuscrations) include tables, charts, figures, drawings and phocograpru. Graphics are efficient, powerful tools for the presentation of data and results. Howevtt, graphics must be prepared carefully to ensure the intended message is conveyed. Poorly designed graphics can confuse, distort and f.ill co communicate the rel~t informacion. Clip an (a type of graphic not discussed in detail in this manual) can enhance a presentation slide by complementing the words on a slide or helping depict a situation where other graphics are not readily available. In recent years, a wide variety of computer-based cools have come into common use. Profes.sional-looking graphics can be created using spreadsheets, graphics managers and presentation software, paint programs, computer-aided design (CAD) and programs chat allow the integration of scanned video images with computer-generated graphics. Simulation techniques, discussed in Chapter 11, are also appealing in showing the audience what the future might look like. These new tools have made the creation of report graphics fast and easy, and shifted the bwden of producing final graphics and t~les from the graphics staff to the traffic engineeriQ.g mff. Speed and ca.se of we havt abo encouraged the tendency to illustrate everything. a practice that can be ineffective and confusing. As discussed in Chapter 3, both chart and table design should primarily be content-driven, rather than sofrware-driven. Specifically, the analyse should have an idea wlu.c the final chan should look like prior .to starting a sofrware "wizard" co aid in its production. The foUowing discussion highlights additional design considerations of graphs and charts.

2.2 Graphics Design Graphics are wed to convey or to clarify information. Page layout, type of graphic, content and skill of execution all conuibuce to the success or failure of a graphic as a communication tool. Graphics m~ be designed for the audience chat will view them and for the citcumscances under which they will be used. This section includes some general guidelines for the design of graphics.

2.2.1 Fonu tmJA.ttmliMJ Eye movement and memory recall studies suggest some positions on a page are more important than others. For example, the eye usually focuses first at the upper left comer of the page. then moves around the page in seeps toWO!I
~AAMII.At

r---•--+--~ I 1

r-- - - .... I -'--·-I I

I

-+ -+

.J._ 1/3 ~ 1/3 --,k- 1/3 --)-

l"''C TAA.MC:PflQTAT1()M ' tJr.tf\IS:I:gJt..lt: (Tttntt<: .,P\ln tf"'llTirlM

1/3

1/3

_.J..

a} This figure has a strong focal point.

b) This figure lacks a focal point.

[Q][Q][Q][Q][Q)[Q][Q] IQ] 131EJIQ] [Q] [QJ IQ] [QJ IKI [Q] [Q] [QI [Q]

[Q][QJ[QJ(Q][Q][Q][Q]

IQ][Q][Q]IQ][Q][QJIQ] [Q][Q]IQ]JQI[Q]{Q)[Q}

IQ]IQ][Q](Q)[Q)[Q)[Q] [QJ[Q](Q]IQ][Q][Q)[Q]

c) The shape of objects directs the eye away from the illustration

Q

[Q][Q)[Q][Q][Q)[Q}IQ] [Q] [Q] IQl [Q] [Q] [QJ [Q) .

d) Visual weight and objects can be used to balance an illustration.

·~ --I>

---

~

most likely to be remembered. The lower right intersection is the second strongest visual position and, as the last item viewed, is also likdy co be remembered. Items placed away from these locations tend to serve as background informacion. They suppon irems in areas ofstrength but are less likdy to be remembered. These points apply to photographs as well. For example, a photo of a landscape with one-third sky an4 two-thirds landscape is more visually appealing than one with one-half sky and one-half landscape. ' The visual weight of items or areas on a graphic is also imponant. V'uual weight is an elusive combination of position, shape. color, contrast and meaning. If there is more than one object on the page, the eye tends to focus on the visually heaviest object first. The eye either stays focused on this point or continually rerunJS to it from other items {Exhibit D-2a). If all the items on a page have the same visual weight, there is little for the eye to focus on. With no point to focus on, the viewer may become bored and the eye may move off the graphic (Exhibit D-2b). Items placed within the focus areas (as shown in Exhibit D-1} draw the viewer's attention. The shape of the items can also direct the viewer's focus around the page (Exhibit D-2c). Items shaped like arrows that point to the lower right corner tend to direct the interest of the observer off the page. Placing a shape that either stops the flow or points back onto the page helps the viewer focus on the page and on the message (Exhibit D-2d}. 2.2.2 Onnposilion The strucnue of a graphic can be compared to that of a written paragraph. In a paragraph, the first sentence informs the reader what the paragraph is about: It td.ls the reader what to expect. This is analogous to the first itern of focus on the graphi~ The main body of the paragraph develops the idea presented in the first sentence. Similarly, the main body of the graphic serves as the backgtound, supponing the main message of the visual. Points of inserest are estab~ lished by the objects of heavier visual weight. The final sentence reinforces, concludes and tells the reader what the

paragraph was about. For a graphic, this usually corresponds to the item at the lower right area of focus. This is the last thing the viewer sees before leaving the graphic.

2.3 Tables Formal tables serve rwo basic purposes: to summarize information or data discussed in the rext and ro compile reference data. The purpose of the table guides how it is designed and where it is placed in a report. Summary tables save space and enhance comprehension by providing condensed information in meaningful form. A short, weU-designed tabulation can replace a lengthy section of explanatory text loaded with statistics. Tables designed for oral presentations are usually most effective when they are simple. The design of tables for oral presentations is discussed in section 4. Tables included in written reports may be more complex, as the reader has the time to study and undemand them. Reference tables generally provide material to support the text but are nor needed in the text for clarity. These tables are generally placed in the appendices. 2.3.1 Structural Elements ofa Table Some general guidelines for the devdopment of effective tables are described in this section. For reference, Exhibit D-3 shows the strucrural dements of a table. Formal tables are titled and numbered for reference. Tables are normally numbered in consecutive order, starting from the beginning ofeach article, chapter, or book. Arabic numerals are used to identify tables occurring in the main body of a work. Double numbering (1-1, 1-2, l-3) can be used ro locate tables in their respective sections of a report. Double numbering is appropriate when sections contain numerous tables or graphics. Position tables as close to the texrural reference as possible but after the initial reference to them. The ririe should briefly explain the content of the table and should be pertinent to the text. If the data are not original, ieference the source. Data items can also be referenced in the body of the table, as footnotes to specific data items, in columns, or in the title of the table. A subtitle contains any additional explanatory or descriptive information required to make the main title clear. Use the same type face as the title, but use either a smaller size or lighter weight.

Column and row headings should be brief. Vertical column headings and subheadings usually contain the dependent variables. Horizontal row headings (shown as the srub in Exhibit D-3) generally contain the independent variables. Define any abbreviations used in the headings as footnotes to the table. The field (or body) contains the tabulation of numerical or verbal data referred to by the stub and headings. Each unit of the fidd where a horizontal row intersectS a vertical column is called a cell Whenever possible, data items in columns should have a uniform degree of acc:;uracy. Fractions ace usually converted to decimals and numbers are aligned by decimal point. Plus or minus signs ace placed immediately to the left of numbers; missing entries are identified by a dash. Shore answer entries that require multiple lines may be right-justified unless the spacing berween words is noticeably large. Lines (rules) or grids are used to separate certain columns or lines, or to frame the table in a box. This may help to distinguish the headings, separating heading area from the data in the fidd, and.improve the readability of the table. Rules should be thin lines; heavy lines may draw attention away from the table and make it difficult to

read. Rule.5 work best for very small or very large tables.

/Col~~ ~

~

Road ~..ene ~~ !2: Pavement Serviceability StUb head

'---.. Yeac

12

Otber Arterials 6

11

8

. 1981

13 10

.lL

~978

10

1915

10

10 11

fM'

1987

Stub

Needs immediate Rei!!!!:

1983

Interstates

1

Needs Repair Soon Other

Interstates 27

24 26 23 24 18

542 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDillON

.--

~Decked Heads

+-Column heads

Acterials

38 39 44 43 49 45

~Cells in

body of table

! Each

table should stand alone (be understandable without referenct to text). All references and notes are handled !separately, without reference to the text or ocher tables. Reference numbering stares at the upper left:, with a new serit:s Jor each table (for example, a, 1, or •). Number tabular elemenrs from left co right across the table. Foornoresshould be the same width as the table. If possible, place each note on a separate line. Footnotes titled "Note" qualify, explain, or provide information relating to the table as a whole. They are alwa~ placed first in the sequence offoomotcs. Footnotes titled "Source" cite the sou ret of the information and are always last in the sequence of foomotes.

2.3.2 Organizing DaUz

Readers are accustomed to reading information from lefr to right. However, by definicion, rabies are struct ured ~round venical columns. Data in vertical columns appear co be easier to read than horizontal comparisons. The c ricial problem is to make the left-to-right relationship clear with respect to dara in venical columns (Whice, 1988) _ Organize the data to make imerrelacionships as visible as possible. Presem data so comparisons are possibk both within a table and among cables. Organize columns of data to suppon the purpose of the table (for instan ce, to show similarities or differences, statistical trends, or interactions). For example, a table presenting data on high accident locations may be otdered from high to low based on the accident experience. Exhibit D-3 shows the repair perctncages grouped by the need for repair for "interstates; and "ocher arterials." If the designer wanted to com pate repair needs by faciliry rype, the second levd decked headings would be *In cerstate" llfld "Ocher Arterials." This would pur the data for interstates in the rwo left columns and for ocher arterials in the r;WO right columns. Thus, a different purpose for the cable corresponds to a different ordering of rhe data. If users must c:xtract data from a table based on a name, place the enuies in alphabetical order based on the mllj or sueet. When designing reference cables, consider the ways in which readers may want to use these rabies. Avoid large gaps berween cohunns. ·Gaps. tend to confuse the reader because the eye must move across empry artaS t:O locate the next item in a row. Using generous spacing berween the lines can help improve horizontal tracking, whlch makes it easier to read across a cable. If a cable is large or complc:x, place it on a separate sheet. Small, simple tables should be placed directly in rhe tc:xt. Make sure the cable supplementS and complementS, but does not duplicllce, informacion in the text. All tables should be referred to in the tc:xt.

2.4 Types of Chart Chapter 3 identified the most commonly used chart types, including bar chartS, line graphs and scatter-plots. This section presents some additional, more specialized display options.

2.4.1. Higb-L01D Graphs . High-~w graphs we twO data series to create a set of vertical lines, ojle for each pair of values. They can be used to illustrate the high, low and mean values of any series of data collecced. The horizontal tick marks within the high/lo-w vertical bars in Exhibit D-4 indicate the observed value within the range for that bar.

Appendix 0 • ~

--· -- -·-·- .

. ---·--+---·--·~~====--~-----

280 280

~

~

~- -·--····-·-·

240

····-··--··-- -···--·--·----- ··.,--·-------···----·

220

Ttd<. Maries Indicate Observed Values

200~ • 180 ------b·-----~-----·----~-.,-------·--HIO 140

· - - -- - - -·- - ·.., _ _ _ _ _ _ _ _ _ _ .. _ _ _.._ _ _ _ _ , __

120

2

3

s

4

6

I

...:___

1

.'---

__

B

11me period .Not!': HI· Lo graphs can be used to display expected end observed values.

2.4.2 Ana G?oaph An tma pph presena each series as a filled-in area. The X-axi.s represents the number of dat2 points; the Y-axi.s indicates the number of series or accumulated series points being plotted. It is similar in concept to a stacked bar graph; dat2 series are stacked rather than overlaid. An example is shown in Exhibit D-5.

45

40

ic:

.

35

!

30

I

25

)

15

0

,E z

20

10

s 0 1980

i:

CAA

•

l.tA IIolltl'l l '"',..

1982

..

,

1988

Veer

Tnlllt.lrnru'""' ' ""''"''' ,.. •• ...,.,., ....... . . - - ·• •

1~

\ 2.4.3 Pictogr11nu . ·!A pictogram integrates graphics, symbols, or icons into various types of graphs, including bar graphs, line graphs and ! pie charts. Exhibit D-6 is an example of the use of pictorial symbols to illustrate sign legibility distances. Pictograrns \hat are too complex may be difficult to interpret.

Di s tances shown in

~e e t

r- 800

800-

,.. ~ ~ ~ $ ···~ <$ · ~
.....f

~

w~0~

~~- ~~~

300.....f~

.

~~

~

... ~ <8> ~ 100 -

""" 700

1-

600

t-

500

t-

400

I-

300

t-

200

' - 100

Source: J.F. Paniaci. Legibility and Comprtthmsion ofTraffic Sign Symbols, Research Report. Federal Highway Mminisuation, McLean, VA.

2.4.4 Statistical Maps Statistical maps are designed to show the locations of a variable collected over a geographical area. The map shown in Exhibit D-7 illusrraces chis type of graphical presemarion. Mapping wols are included ·with many graphics packages. ' ,~q;.,_~~~~~~-~~

~~17.:

Accidental Death r--=--

Rates

by States, 1971

---,

,o

....... ==BELOW 50.0 • 13 STATES lllll 50.0 TO 58.9 • 11 STATES 8,_ 80.0 TO 6$.9 • 13 STATES 70.0 & OVER· 13 STATES • to..tu: filM".,,~-- o,. ~o;sc-.- 0t1 • • ~~

..-e... . ..,.

~v.a, c

*"'._, "'"""' o.o.,,,..,l,


546 • MANUAL OF TRANSPORTAnON ENGINEERING STUDIES, 2ND EDITlON

2.4.5 Organization Charts An organization cbart is shown in Exhibit D-8. Chart elemenrs are connected by lines thac trace the flow of aLt(horicy wichin che organization. The divisions of any department are placed below the department; all elem ents on the ..,$arne level have comparable levels of author icy. Elemenc:s in an organization chace need not be enclosed by recl'llngles or other geometric shapes. Enclosed icems work well on simple charts. However, on large complex charts they wake individual text items more difficult to focus on and may require smaller text to make che chan a reasonable si7.C·

Transportation O~ratlons and Services Organizational Chart

:::=====.

------·---------------,: !·-·Coordination Mth od\ttr OfPJlfzati:lns !

--------~Departments,

l~~~~~!~~~~!:~~~-j

Appendix D • 54::;;;o?

2.4. 6 Flow Diagrams Flow @grams arc useful for representing procx:s.ses or procedures. Geometric wpcs (boxes, circles, triangles) dcnore spccilic types of activities. Arrowed lines (veaors) trace possible actions. For example, a flow diagram might be used to illusuatc the required steps in gaining approval for a new project, or, as shown in Exhibit D-9, a decision-making process.

llWENTOitY OF CURAIENT COM>nlOHS

NO

•

~ ~ 8IQHT

DISTANCE

ADEQUATE P'OR SAFE GAP ACCEPTANCE 1

YES

f'OMCAST T8AI'FIC VOLUME ~· 1:0 HTEAM&DIATE l'liJ1..IAE (Le. a YRa.>

·'\'I!

YES

EXJa'TNi

DOES OR l'liJ1..IAE T'I'IAtPIC SA'nSFV ANY OF THE 8IQNAL ~AU.AT10111 WARRANTS?

F ftEASON OTHER THAN STANDARD WARRANTS JUSTFED ..STALLATION DO THEM REASON 8TLL PREVAIL 1

ACCC)ERT RECORD WORSE WrTH

~

NO IJIC»FFCANT CHANG£ 1'1 ACCIDENT AECORO OR' NODATA

YES

8TAGE I

Source: J. L Kay, L. G: Neudorff, and F. Wagner. Critnia for Removing TMjfic Signals aru/ Usm' Guide. (FHWA-IP-80-12). Alexandria, VA; JHK Associac~ Lnc. and Wagner-McGee AssociaEeS, 1980.

I

\

••••"'" " ' TIIMI~onRTomnN ~Ni,INFFRINI.

<;TIIniF'i. 2ND EDITION

12.4.7 Project Progress Charts \Projtct progms charts illustrate the time schedule for a study or project. The X-axis is the rime span of the project in ap! propriate incr.emenrs. Each cask or set of tasks is repr-esented as a line or bar spanning the time period over which that 'task should be accomplished. Progress charts are used as pbnning rools and to check progress as a project proceeds.

•IltiDJ;iiLI'' ••

..

..

~ .r~

Field and Sar.ty EveluaUon of Speclallnteraedlon Trealment

lint line

v.. t TASK

I* I

Proje
&I

J

..

~ole.;,qs.

B OMiopMm~l.tat.rill

c

Fllld &oluDin ol Woo\ Ploll

0

FealiUII 1111 Selection ol SUo

:

~ IIIIP..,...RMeo~Malbllng\1\doo

G

jFn..... en.. ~· ~Report

I

!

i

12

Year2 11 21 21 l i

t5

I

I

ij_~

~~

i .i

: !

·l

~

l

!

!I

I

i I I

I I

I I

I I I I I

I

·I

!!

i

'

I' _l

'

CGocilct Bolin/.... OltJ Colodiolln E PNpON F_, O.llion SirMary Rlpart

f

iI

"

fkl

~

i I I

i

I

i

!

i

I

'

!

l

Oral Rlpoltl VIdeo/ Brocbwe

i

I I

i

I

I

J

:

:

I

iI

l ·t

:I

i i !I !

t .

: I I

!

I I

ln i

I

I

l

I

i

:i .

i;

I

I

'

I I

YearS 54 57

51

'

I

i I

.

I ~

!

iI

:

l

J.

i

.l

t

! !

I

i

~

~

l

I !

J

i

i

!

i

I I

!

I

' ! I

• FNI Repolt/ VIdeo/ Broclwre

I

I

i

!; I

*

I

~

I

i

I

at

*!

i

_!r : l !

!

••

~

I

I

II

v....

Yur3

zn_ :10 J3 lS 3S .421

~ i

i

II I

, .~

Ptoject Mlllilg wM Sponsor

2.5 Design Considerations Graphs must be d-esigned for the intended viewer and with the purpose of the graph in mind. If the graph will be used to make a slide, it should be simplified. If possible, limit the number (or groups) of curves or bars shown to four for visual aids. Graphs showing trends need nor be as large as graphs provided with dau points to be wed in calculations.

-·

The scale (or ~pect ratio) of the graph also affects the way data are perceived. A ratio of height to width of 3:4 is common. Designers may need to experiment, changing the scale of :1. graph ro determine the most effective aspect ra.rio for its usc. Too great a range of values causes the graph to appear compressed. Some spreadsheet packag-es will allow the usc of more than one vertical axis scale. Label them clearly co :~ovoid confusion. Type fonts (such as Helvetica and Ttmes Roman) should be consistent with those used on tables and in the ten ifpossible. The we of different sizes and weights of the same font type works weU.

2.6 Use of Colo'r C
• develop associations; • build rerenrion; and • create an aesthetically pleasing product. There arc some general rules for using color in pleasing ways. A color whed (which shows the 12 basic hues or colors) is a very useful roo! in choosing colors. h can be purchased in any art supply store. The primary colors are red, yellow and blue. The other (secondary) colors are produced from these three. Cool colors (blue and predominately blue) are relaxing and appear to recede on the page. Warm colors (red, or red and yellow) are stimulating and advance to the foreground. Green, brown and red-purple lie berwcen rhe warms and cools and arc therefore relatively neurral. An important use of color is to aruact attention, primarily through conrrasr. Color should be applied to the elements of greatest significance. A bright color used with black is the most effective. Complementary warm and cool colors can also be used sparingly. Blue appears to be the most popular color. The colors used in a graphic should fir the overall mood. Too much color can detract from rather than improve the readers understanding of rhc graphic.

2.7 Graphics Checklist It takes time and experience to produce the most effective graphics. Slides (or overheads) oli:en make or break a presentation. Before finalizing graphics, go over the checklist below. I. Are the graphics suitable for the intended audience? 2. Are the graphics legible? 3. Is the message clear? 4. Is the layout neat and properly aligned?

3.0 WRITIEN REPORTS 3.1 Section Overview This ·section expands on the discussion in Chapter 3 on considerations for a written report summarizing results from transportation studies. The language, style and presentation of a report should match the intended audience, whether it is directed to a technical audience of engineers or the general public. All reports should begin with·a brief summary of the purpose, important findings, conclusions and recommendations to serve readers who do nor have time to read the entire report. The report should be of adequate length to present all pertinent information, while remaining clear and concise. The formatting and layout of should be professional and attractive to the read~r. Tables and £gures should be used as necessary to illustrate key concepts. ·

3.2 Organization of the Report The components of a transportation engineering report are determined by the report purpose, length and complexity. In general, shorter reports have fewer components. The following is a list of components, any or all of which may be included in a given report. • Letter of transmittal

• Tide page • Copyright notice • Disclaimers


• Forward or preface • Acknowledgmenu • Table of comenrs • List of tables • List of figures • Summary or executive summary • Body of the reporr • List of references or bibliography • Appendix • Glossary • Index The sequence presented follows normal conventions but may be altered to fir a given situation. Each component is discussed briefly below.

3.2.1 Letur of1hmsmitttd A letter addressed ro the person or agency for whom the report is prepared is often attached to the fron t of the report. The representative of the organization that conducted the study and prepared the report signs the letter. This lmer is usually a short sratemenr char the signer is pleased to submit the report in the number of copies required. lr nuy aJso briefly explain how and why the study was conducted and irs single most important outcome. The principal purpose of the letter is to document the submission of the report. Copies of the transmittal letter may be included in copies of the report. 3.2.~ Tule Page The title page includes the name of the report, the date the report was prepared, the authors' names and the sponsors' names.

3.2.3 Copyright Notice If the report is copyrighted, this notice appears on the title page or the page that immediately follows. Authors ~ obtain information and an official copyright application fonn from the Register of Copyrights, Copyright Office;, Library of Congress, Washington, DC 20025. Copyright informatiqn is also available online at www.copyright.go'Y'3.2.4 Disclaimers lo general, a disclaimer alerts the reader that the results, findings, conclusions andlor recommendations presented i.J:1. the report are those of the author(s) and do not necessarily reflect the opinions, views, or policies of the spon.sori.nS agency or organization. Disclaimers most frequently appear in research reportS. Disclaimers may also include explici. 1: information about the report that might otherwise be assumed incortectly. 3.2.5 Foreword or Preface . A foreword or preface; provides the history leading to the report and, if applicable, its relationship to other reporu. ~ foreword may be written by the author(s) of the report but is commonly written by a representative of the sponsorin~ agency or organization. A preface is generally written by the author(s). 3.2.6Achwwktlgmenu This section recognizes the persons and agencies who assisted or contributed to the study. Acknowledgments mayalso appear in the lerrer of transmittal. 3.2. 7 Tllhk of Cmtenu Major divisions, sections, or chapters of the report are listed in order of appearance along with the page num~r ~- . ning each division. Subdivisions of a section or chapter may also be listed in the table of contents. Ap~ndix D • 55f

3.2.8 List ofTables The number and title of e:a.ch table and the page number where it is locued :a.re presented in order of appeannce in :a. list of tables. The inclusion of this list depends on the length of the report and the significance of the tables as stand· alone presentations.

3.2.9 List ofFigures Figures are listed in :a. similar fashion to the tables. If tables :a.re listed, figures should also be listed, and vice versa. The list of tables precedes the list of figures. 3.2.10 S~nn.-ry or benaiw S~nn.-ry This section briefly summarizes the purpose of the srudy, major findings, conclusions and recommend:a.tions. It is designed for those who do not have time to read the entire report and as :a. means of refreshing :a. reader's mind at a later time. Bec:a.use it is almost certain to be read, it is the most imporeant section of the report. The SIUilllW')' must not overstate nor understate the findings or cause the reader to take conclusions or rccornmend:a.tions out of context. Sometimes the executive Sll!Illii.:U}' is written more for the decisiorunaker, someone with very limited time who needs to know the bot· tom line very quickly. A rypical SIUilllW')' ranges from one to five pages. However, an c:xccutive summary for a complex, multivolume report may be 25 to 30 pages and published in a separate volume. Once put together, reread the executive SumtiW')' to make sure it Bows together so it is not just :a. cut-and-paste dfort out of the main report. 3.2.11 B.J.y of the /Uport The body is the heart of the report and is supported by all the other components in this lise. The orgmization and content of the body are discussed in :a. later section.

3.2.12 List of~ rw Bii11Hvapby When material from :a.nother report or book is used, it should be properly acknowltdged. Autho£8 should cite the dowment in a lise of references and refer to the list at the places in the text where the material is used. Place the list either :a.t the end of the report or at the end of each chapter. Another option is to cite each reference in a foomote on the page where the materia.! is first used. Examples of refuence ~ can be found at the ends of the chapters in this llWlual. A bibliography is :a. list of books and reports that contain materials that arc useful to readers who wane to pursue the

subject m:mer further. Rq>orts describing cransporution engineering studies do noc usually contain bibliographies unless the author wants to encourage further study or aplanation. 3.2.13 Appnulix Detailed material that supports but is not essential to the body of the report should be placed in an appendix. Appropri:a.te appendix materials include supporting d:a.ta, detailed aplanations of methodologies or procedures, derivations of formul:a.s, conversion f.u:tors, lists of symbols, d:a.ta collection formaa, d:a.t.l. collection P.rotocols and checklists. Appendices are effective means of fully documenting :a.nd supporting the results of a study without cluttering the body of the report.

3.2. U Glosury If the report is written for nontechnical re:a.ders or if new and unf.uniliu terms are introduced, a gloss:a.ry .or list of defin.itio~;~s should be included in the report. If few new terms are wed, footnote definitions may suffice. 3.2.15lrulet Transport:a.tion engineering reportS seldom contain indices unless the report is voluminous or intended for frequent reference. Th~ inda is more aact and detailed than the table of contents. The index liru all major subjects alphabetically along with the .,ages where each subject is addressed in the report.

3.3 Body of the Report The body of the report is composed of a series ofsections or chapters.In a cransport:a.tion engineering report, the body contains answers to the following questions. 1. What was the objective or purpose of the study?

2. Why was the srudy necess:a.ry? .. . ............ ·~ •. ""'""""'"'' ,.,"''' '"'"r".r·,-nu.tr

C"Tlll"''~c:c

-,,.,u"\ s:nrTtf'\M

3. When, where and how was the srudy conducted? 4. What were the findings? S. What conclusions were drawn? 6. What recommendations were made?

3.3.1 OrftmiUng the &Jy The first step in effeCtively axranging the body of the repon is to construct an outline. The outline should list the chap· ter, S(C[ion and subsection headings and may conl:a.in notes or copies to be covered under each heading. The ou tline helps the writer organize the body in a logical, comprehensive and complete manner, and makes the task of writing easier. There arc a number of ways to organize and present the material in the body of the repon. Some of these uc; described bridiy in the following paragraphs. Sequeotial Lopcal Statcmeau Ptoceed.ios &om Problem to Solutioa. The problem is stated in the opening chap· ter or seCtion. The recommended solution in logical ste~ is developed in succeeding chapters. Cause ud EfRc:t. The report begins with a description of the causal factors surrounding a problem. Then the report addresses the resulting effects, as indicated by the appropriate measureS of effectiveness (MOEs) . For example, the report of a congested route study mighr begin with a description of traffic volumes, vehicle classification, occupancy and roadway geometries. The effects of these factors would then be described in terms of measured speed, delay, levels of service (LOS) and accidents. 1-...me SeqllCDce Previous conditions are addressed, followed by present conditions, followed by projected future conditions. Pro~eBU in Order of Importance. Each ·problem and proposed solution (if known)

is presented in order of priority begiru:!.ing with the most severe, or anticipated time of implemencation, or financial feasibility, as e:nmples. l..ocation. For data coUection reports, each intersection or roadway section (link) is described in rum. The order of prescncation may be based on type of road, siu of facility, uea of the city. magnirude of the problem, or o«her logical scileme.

Order of.Andie:nce lntuest or Familiarity. Treating the most populu subject first, rcgardiess of its imponanc;e, may be a way to gain the readers' interest and lead them into less familiu subjects. Grouping Similar Subjecu T~ Similar subjects should be grouped together and presented in the same section/chapter to ease reading and understanding. For e:nmple, planning chapters may be followed by design chapters, which in turn ue followed by operations and safety chapters.

3.3.2 Ttp tm Botl] Ctmtmt. Format. The body of a typical transportation engineering srudy report generally includes: • purpose or objectives of the srudy; • background (that is, what led to the study or why was the srudy needed); • scope of the srudy (that is, what limits were placed on the srudy); :

• methods used; • data collected (for example, type, amount, when, where, etc.); • analyses performed;

• findings; • conclusions; and • rccomiiiendations.

Appendix 0 • 553

Examples of subjects chat would be appropriate for a transportation engineering study report include: • alternatives developed or examined: • selection of alternatives, traffic control devices (TCDs), or routes; and • evaluation results, such as: • cost analysis or financial impact; • environmental impact; • traffic impact; and • implementation plans or recommendations (for example, preferred alternative). Use Simple, Clear and Concise Language. Remember, a transportation engineering study report is a presentation of technical faccs and their implicarions, not Literary art. On me other hand, write the report to hold the reader's attention. Do not use complicated wording in an effort to ·sound important." Most readers prefer simple, common language.

s

- Use a Good Style Manual. The elements of style in wriring rechnical reports vary &om one manual to the next. Pick a good manual, adopt a set of conventions and maintain uniformity in spelling, grammar, punctuarion and formar throughout the report. Sevetal good technical wriring style manuals are included in the references at the end of this chapter. Agencies such as U.S. DOT and the Transportarion R=rch Board (TRB) also maintain technical wriring style manuals. Page Numbering. Number every page except the tide page and the letter of transmittal. The summary or the ~st page of text can be "Page 1." Advance pages are nwnbered with lowercase Roman numerals. Appendices may continue the numbering of the main body or can be numbered A-1, A-2 and so on. Place page numbers at the top or bottom center of each page. Break Up Lengthy Narratives of Pure Text. Headings and subheadings lead the reader through the author's train of thought and presentation of &as. Make ample use of summary tables and figures. Photographs may also be useful in describing certain situations or conditions. &plain the Methodology Used. The credibility of the results of a uansportacion engineering study often hinges on the type and amount of data collected and the methods used to analyze those data. Therefore, the report should clearly, yet concisely, provide the reader with sufficient infon:nation about the data and methodologies employed t.o establish an acceptable level of confidence in the findings. Explain the application of standard methods by referencing a text or manual on the subject. Special methods require more explanation that may be placed in an appendix to avoid cluttering the body of the report. The sources of data and informacion used in the study should be clearly identified. Authors should also specify the amount, place, time and conditions under which field data were collected. Rdationsb.ip of Findings to Conclusions and Recommendations. This relationship is similar to the construction of a ho~ The findings are the foundation on which the conclusions rest. The conclusions, in rurn, support the· recommendations. Reviews and Editing. A good technical writer, a panel of reviewers, or an e:xperienc.ed editor should review the report, if possible. This simple practice often turns a dull, Lisdess recitation of facts and figures into an interesting, understandable and perhaps even enterta.ining piece of writing. The goal for any report is that it be read and understood correctly. A report that readers ill?ore or misunderstand has no value and the work and dfort to produce it will have been wasted.

4.0 PRESENTATIONS 4. 1 Section Overview This section discusses some additional aspects on making oral presentations that adds to the discussion in Chapter 3. Sometimes the only information on a transportation project the public sees or hears about is from a presentation. Effective presentations help explain important results of a srudy, can inform and educate the public about the issues 554 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND EDmON

. involved in a project and can provide feedback on which of several alternative designs are preferred and why. There~ fore, the presemer needs to understand these addidonal concepts in preparing for and presenting information.

4.2 Oral Presentations Oral presentations arc an imporcant means of communication for transportation professionals. Presentations range from simple progress statements 'to extensive rtports on traffic studies to comprehensive tranSportation plans. All effective communications must fit the requirements and experience of the audience. This is especially im portam in oral presentations. Define all unfamiliar terms. Arrange and present the nuin points of the presentation so they are easily understood and allow a logical conclusion to be made. If a listener cannot follow the presentation wit h at least a reasonable degree of understanding, it makes lirde difference how thorough the investigation was. There is little charice for a listener, once lost, to catch up. 4.2.1 Presenuztions for Public Access There are many opportunities within communities for making presentations that are recorded or broadaut 1M o-ver public access channels or the Internet. The same principles apply in these settings as in others. The main differences are in positioning of the speaker and use of decuonic presentations, displays, etc. Should there be an opportunity for communicati.ng in this manner, make an appointment with the studio supervisor for a discussion on how your presentation can be most effective for their recording setup.

There may be facsimile or call-in questions. These should be filtered by someone in the studio and then shared with the presenter{s) by the session moderator. See Section 4.5 on how to answer questions &om the audience.

4,3 Purpose and Scope Effective oral presentations are not condensed versions of written reportS, nor au they speeches read aloud. Ea.ch presentation should be prtpared for the specific conditions under which it is to be given and adapted to the prc:scnccr, the audience and the presentation environment. The following five basic characteristics distinguish effective oral pre· senracions from written reports: • Specific audience Limited scope • Personal presentation

Need for instant understanding • Limited time for presentation 4.3.1 A.rulknce A number ofpeople with widely diffi:rent technical backgrounds may read a written report. Therefore, the writer ml.l.st tty to adapt the writing to a broad levd of reader understanding and interests. The oral report, however, is uruallY prepared for a specific audience. A speaker must orient the report to the particular interests and l~ls of undCI'SQncling of her audience. ~egin the presentation talk with a founda.tion the listener will understand and is interested in. 0 o not talk down to the audience; be careful to avoid the trap of condescension. Remember: The listeners have given u..J? their time to come co hear your presentation. Do not waste their time. 4.3.2Scope Readers may read the entire wriro:n repott or only certain portions. However, an oral presencacion is usually prqnre~ with a specified time limit in mind. The scope of the material must be appropriate to this limit, which often indude:::.S a question-and-answer period. The presentation should include a clear explanation of the subject and the gea~ conclusions. Use brief summaries of subscantiaring data when they are critical for undemanding the repon. Abri~! description of the procedure is usually all that is required. If a written report is available, the talk is probably bes -t: limited to the· most important parts of the written repon, concentrating on those items that are of greatest in teres "1:

Appendix D • 55~

to the audience. It is better to present less and cover it thoroughly than-to discuss too much and have the listener feel you did a superficial job.

4.3.3 Presentation The medium of the oral report is the speaker. Therefore, the speaker's adivery style is critically important. Posrure, gestures, eye contact and &.cia! expression, voice projection, enunciation, pronunciation and the degree co which the speaking style seems relaxed, conversational and expressive all affe~ the attitude and receptivity of the audience. Four styles of presentation delivery include extemporaneous;-impromptu, manuscript and memorized techniques (Michaelson, 1987). The o:tnnporaMOus talk is usually preferred for. the oral presentation of a· technical report. In this method, the oudine of the speech is carefully prepared but not committed to memory. The impromptu (offthe-cuff) method is obviously inappropriate for the presentation cif.lil\e report; however, it is an appropriate style for the question-and-answer session. Prepared and read manuscripts dr'Jm.f"'orized methods tend to be inflexible and sound artificial Practice, however, will improve the quality of an~tation.

'""'-'

Oral presentations frequendy include questions from the audi~ec. ~uestions may be asked during the presenta· cion or be deferred to a question-and-answer period that follo'IIJ t}tt formal presentation. While each method has its advan cages, the mOS[ satisfactory arrangement is to defer q~t0tfonittg until the formal presentation has been completed. Potential questions are often answered during the remainae'r of the talk.

.

-

.

The presentation should be made in an appropriate location. ldeallr.,the space needs to be comfortable and sized for the number attending. It should contain all the necessary equipmerX for the presentation, usually a lectern with a light and computer projection equipment. Sometimes the hos~ presenters to bring their own laptop for use with the projection equipment, but that is quickly changin~tlttrs are presenting during one session, put your presentation on a memory stick for use in the host computer. '

.

4.3.4lnstant Urukrstaruling

: .. ~

~ · ~,..

·

If some portion of a written report is not immediately clear, th~ay reread it, look back or ahead, or even consult other sources. However, the spoken word only lases during , ment of presentation. The speaker must be exceedingly careful co be as clear as possible at all times, payin . . . ar attention to audience reaction. Instant understanding is &.cilitated through voice, language and the usc:., s~tions and summary statements. .

~:...,~

Proper use of voice includes adequate projection and distinct enli:nt:t.rtion to permit even those persons Sealed farthest away to hear and understand the material presented. The speaker should use pauses freely to break up the flow of ideas into meaningful thought unirs. Also, speaking wilh sufficient forcefulness and using a variety of inflections avoids monotony and gives life and meaning to words. A speaker must be sensitive to the rate of delivery and must pace the various remarks for understanding, variety 1-nd emphasis. Practice is essential to master all of these techniques. ...The audience must understand the vocabulary used. If the readers~a. written report encounter an unfamiliar word, they can refer to a dictionary. In an oral repon, the sp~ mmnrezine unfamiliar terms foe the listeners. Obviously, the problem varies from one situation to another, but the speaker should be sure the listeners understand the terminology. A broad vocabulary permits an expression of thoughts with clarity and precision in a vigorous ~d colorful whion. Transitions and summaries help guide the listener through the developmen·c of the presentation topic. After a section of the repon is developed, summarize the main points briefly before moving on to the nat section. Also, show the relationships among various sections.

4.3.5TOrganize and 'rehearse the presentation for the established time limit. The average rate of speaking is 100 to 150 worqs per min. (with listener comprehension at up to 600 words per minute). Tuning can be checked by practicing with audiotape, videotape, or with a live audience. If available, videotape is the most effective. Ic gives the presenter the opportunity to see. how he or she is coming across and how 10 impc~ his or her presentation skill.s. Simply speaking aloud hdps check the organization of the calk; however, whc:P-Practicing in an •empty room• there is a ten~ency to speak much faster than on tape or to an audience. Rehearsal will help the speaker achieve a relaxed, at-case posture, and a smooth, confident delivery. ·

14.4

Organization of the Presentation

)A well-conceived talk, carefully tailored to fit ilie audience, will fail if it is difficult to follow. A confused reader of a

report can regress as needed to wade through an obscure passage; a confused listener is likely to be lost forever. P resentations are organized around three major divisions: the introduction, the body and the conclusion.

4.4.Ilntroduc1Um The introduction prepares the audience for th'e body of the presentation. The introduction should motivate the listener, catch the interest and inform about the'concent of the report. Motivation is accomplished by dwelling briefly on the importance of the subject and by establishing a distinct impression that something worthwhile is to be offered. The speaker can inform listeners about the tof'•ic of the presentation by:

• identifying and defining the general sl1bject; • providing the necessary background; and • giving a pr~vieY{ of the main divisions to bt covert
·2. general conclusions that are drawn flom the subconclusions I

3. recommendations that arise from the general conclusions The 'conclusion section summariz.es the material land may provide a transition to the question-and-answer period. It should flow logically and clearly from the material presented in the body. Surprise endinp may bt dramatic, but are rarely useful.

4.5 Answering Questions Often the most difficult part of a presentatiolt is fielding questions.,If the audience is hostile or upset, this time can be particularly challenging. It is enremdy important co anticipate questions and objections that may arise during and after a presentation. Rl:sponses should be re~earsed so the speaker appears to know the topic well and can remain cool and thoughtfUl even in confrontational sjruations. The following guidelines suggested by Berthouc:x and Hindle (1981) provide a few thoughts on handling a fl.Uesti.on-and-answer session. 1. Be brief. do not say more than is re
2. Do not use an opening line of nonsense patter to gain time to think of your answer. For c:xample, do not start with, "Let me just think ouc loud for a minute" or "That is a very difficult question. • Pause briefly before starting to speak if you need a moment to organize your ideas. Repeating the question clearly will allow you to gain some time and confirm that you undeiStand the question. Do not assume the questioner will deduce the answer correctly froin a long discussion of pros and cons. Be explicit. If your answer must be long or compia, make a brief SUflllllary statement.

3. Do not bluff. If you do nor know rhe answer to a quelition, the following choices are available: "I am sorry, I cannot answer your question;" or "If you please, I would like to consult my notes (references, partners) before I answer that question;" or "! can nor answer char now. T wUI have the answer chis afternoon (tomorrow, next week) ."

4.

Don't be evasive. Answer the question you arc asked, or do nor answer. Do not answer a different question. The questioner will not be impressed or fooled. Avoid hedging. Sometimes the answer is not simple; ir must be qualified or have limits set. In these cases, tty to make che limirs quantitative and make the qualifications precise.

5. Be specific. Use precise words. Favor citing a number over using a descriptive phrase. Never assume you and the listeners have the same scale of reference for imprecise words (words such as dup, high, wann,

smam. Cite dara or numerical facts whenever possible.

6. Do not try co answer a question you do not understand or one chat is imprecise. If the question is unclear, your answer will be unclear. Even worse, it may be wrong. Do not guess at what the quelitioner has in mind, ask for clarification. You may rephrase the question slighdy, but do not change it. Ask for confirmation that your interpretation is correct. If you are answering questions from a group and receive a com· ment far off the point, or far wrong,_decide if it is best to assume others also recogni-ze this and move on as politely as possible. Be polite but firm in your posicion; however, never be argumentative or insulting. Above all, remain calm. If you arc: asked a hostile question, restate it using positive phrasing. Avoid single or double negativeli. It may be helpful to restate a negatively phrased question or one that uses angry words or emotionally loaded terminology into one that is clear, precise and contains less inflammatory language.

7. Never use slang and avoid jargon. Slang is imprecise and it can give the impression the speaker is poorly educated. Jargon is vocabulary known to a group of specialistS but generally unfamiliar to outsiders. Often, jargon can be replaced by a few more common words.

4.6 Preparation and Planning Checkl ist Preparation and planning will help ensure a quality prelienration. During the process of planning and giving a presentation, the following items will help assure success.

4.6.1 Organization Well qeforc: the prelicntacion date, consider the following: • Does the introduction aplain why the subject is significant? • Arc: the major points covered withour excessive detail?

• Will the audience understand all the terminology? Is the talk simply arranged in a logical sequence? Are the visl12ls ~mple and visible? 4.6.2 Rehetn'SII! • Rehearse the presentation until you feel comfortable with the material. • Rehearse with the visuals you intend to usc:. • Identify and practice fielding possible questions.

4. 6.3 At Presentation Tzme On the day of the pres.entation, arrive early enough to check the following: • Does the sound equipment function; do you know how co usc: it?

• Is the room equipped with the necessary projection equipment; is it in working order? • Is the temperature of the room neither too warm nor coo cold?


• How is the lighting adjusted?

• Is rhe searing arrangement right for your needs? • Wu(everyonc in the audience be able ro see your visual aids? Iris roo lace ro correct any deficiencies or learn how to use rhe equipment when walking ro rhe podium.

5.0 REFERENCES Berthouex, P.M. and D. Hindle. "Abour rhe Presentation of Your Paper: Oral Communication: So me Guidelines on AnsWering Questions." Engineering Educarion (December 1989}: 243-244. Bel(, P. C. Manual ofTraffic Enginuring StuditJ, 4rh ed. Washington, DC: Institute ofTransporralion Engineers, 1976. Editorial Scaff of the University of Chicago Press. Tht Chicago Manual ofStyk. Chicago, IL: Uni~ersity of Chicago Press, 1982. Hodges, J. C., M. E. Wh.inen, W. B. Homer, S. S. Webb and R. K. Miller. Harbrau Colkgt HaMboolt., lith ed.: San DiegO• CA: Harcourt Brace Jovanovich, 1990. Hodges, J. C, M. E. Wh.inen, W. B. Horner, S. S. Webb and R. K. Miller. Harbr«t Colkgt Handbook, 13th ed. Chicago: University of Chicago Press, 1998. . Insurance lnstirute for Highway Safety. Fan:s. Washington, DC: Insurance Institute for Highway Safety, 1991. Kay, J. L. L. E. Neudorff and F. Wagner. Crittri4 for &moving TraffitSignaiJ and U1m Guitk, FHWA-IP-80 12. Alexandria, VA:. JHK and Associates Inc. and Wagner McGee Associates, 1980. Ksanier, M.G., J. W. Presley and D. C. Rigg. Prmtiu-HaU WOrkbook for Writen, 4th ed. Englewood Cliffi, NJ: Prentice Hall• 1985.

Lala!li, N. and D. G. Gerard. "Communicating Eff=ivelywith the Public." ITE]D-urnai(Occober 1995): 51}.-52. Lalani, N. and S. B. Colman. "Making Effective Technical Presentations." ITEjournal (January 1995}: 29-38. Mason, J. "Communicating with Elected Officials.• ITE]ournai (December 2005): 18-22. Michaelson, H. H()IQ UJ wn~ and Publish Enginming Papm and Repom, 2nd ed. Philadelphia, PA:. lSI Press, 1987. National Safety CouncU. kcidmt Fan:s. Chicago, IL: National Safety Cou~cil. 1986. Paniati, j. F. Ugibility and Comp"hmsion ofTTYJjfic Sign Symbou. Research Report. McLean, VA: Federal Highway Administration, 1988. TurnbuU, A. and R.N. Baird. Rindurt and WU!Ston, 1975.

Th~ GraphicJ ofCommuniciJJion: Typography.

Layout and Design. San Francisco, CA: Holt,

White, J. Graphic De1ign for tht Ekctronic Agt. New York: Watson-Gupcill f?ublicarions, 1988.

Appendix 0 • 55~

.......

Append ix E !, •••• • • •• •• • •••• • •••••••• ••• •••• •• ••••••••••• •• • •• ••• • • •• ••••• • • ••• • • •••••••• •

•••••

Useful Resources for Transportation Studies By: Bastian]. Scbroetler, Ph.D. Chrisropher M. C#nningham. P.E. DanielJ Fitulley, P.E. Robert S. F07k, P.E. 1.0 DEVELOPING A TIME-STAMP MACRO

io

561

1.1 Overview

561

1.2 Coding

562

USEFUL DATA COLLECTION FORMS

564

This appendix conu.ins useful ~esources for transporracion studies. The first part discusses how analysts can create a time-stamp macro in Microsoft Excel that can be adapted to various data collection applications. The second pan presents commonly used data collection forms. Each form may be removed or photocopied !Or use. The fOrms are ident:i.fied as to source in the manual by chapter number. The details of use are described in the text near the page on which the form appears or is first mentioned.

1.0 DEVELOPING A TIME-STAMP MACRO 1.1 Overview his section describes how Microsoft Excel and Visual Basic (VB) can be used to create a macro that records time stamps for observed events. The macro can then be copied o,nto a laptop computer to perform a wide range of customized. field studies. For example, the macro can be psd to collect raw volume counts that are later aggregated into bins. Similarly, the macro can be used to perform a speed study by collecting time stamps of vehicles (front bumper) arriving at two points a known distance apart. The macro can also be customized to a range of other applications, such as pedestrian crossings by signal phase, or vehicle compliance srudies.

T

The method can be applied in real-time field studies or from video observations. In all cases, the analysts predefines ce.ru.in keys to correspond to specific events of interest. It is important to practice the actual data collection befOre •going live,~ especially if the macro was previously used for orher sru.dies. Coding a large number of events can create a high cognitive codirtg load, and human factors' principles should be considered when laying out the key structure. Foe example, the aforementioned signal compliance srudy might use keys "A," •s• and •o• to denote vehicle arrivals in three lanes. These keys would be operat~ with the lefr hand. The right hand may then be used for example to and "K.• The c:nct configuration depends on record signa! phases "green; "yellow" and •red; using keys •H; the specific study and usee preference.

·r

If the coding load becomes too high, the analyst may decide to supplement additional data at a later time (from video) using a common uro-time reference point. Events ~ then be ordered by sorting by time stamp. Alternatively, two coders can be used co collect different data, either on the same or separate keyboards. Note that MS Excel will not recognize an external number pad for this macro, but will support a full external USB keyboard.

1.2 Coding The time-stamp macro relies on a builtrin MS Excel function, NOWQ, that r~pons the instantaneous computer system time. The function is automatically updated every rime the Excel worksheet is updated in any form (for example, by entering a value). During data coUecrion, every rime the analyst presses one of the preconfigured burrons, the macro (very quickly) performs three taSks: 1. It enters the key code in the accive column (A) and row (l) and moves the input cursor inco the next column (B).

2. It copies and pastes the cell containing the NOW() function into the active cell. 3. It moves the input cursor back into column A, buc down one row (2), where it waits for the next key input. The actual macro code is wriucn in VB, a programming language included in Microsoft Office produces. The macro is activated by hitting a uSrarc Recording~ burton that is in turn associated with a function in.VB. The macro continues to respond to keys until the analyst hits "Stop Recording," at which cime Excel is returned co its normal srate. Exhibit E-l shows a possible layout of che input screen. The Excel worksheet contains a title bar, the "scan:" and "stop" buuoru, the system time stamp with the NOWO function and a data table.

··- ····· ' LLO !

23

24... i

.........

• ••• •

"l Hf,•• ,..

-

_. ..,..

rrrrt ¥ '": . . , . , . .

251 2§'

_,._ lfltlfll ............ .........

V I 2S

IT L IIITJIJNfllf)ll .......... lfllffiiUJIIIIII Ill ........

-

r

--- --------- --- --------

.! '

- .. ..... ..

~

· i---·-------·i·:~:·~~~~~=-~:~~t~~·: _::.~::_ ~- ·.·...... -· ·- --· ~

562 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES. 2ND ED1110N

·\The macro only records data in the first two columns. Column C contains equations that automatically deduct the :active row time sump (see cdl Bll in Exhibit E-1 ) from the first time sramp in the study (cell $B$10) using the ·.function [=Bll-SB$10]. Since the NOWQ funCtion works with the system time in che format DD:HH:,MM:S$.00, column D multiplies the values of column C by the factor 86,400, which is the number ofseconds in a day. The values in column D are easily added or subtracted in the analysis. The analyst may use the remaining columns in the worksheet to code other functions that can be used as real-time coding aids. For example, in a co.mpliance study, a function could indicate the active signal phase recorded in the macro through the function {=lF(Allr"H","RED","")]. This function will return the statement "RED" if cell All cont:~.ins the letter "H" and returns a blank cell otherwise. During a busy li~ld study, this em help the malyst double-check visually whether a key was pressed correctly.

r·)

The "scan recording" and "stap recording" buttons are associated with subroutines in the VB Macro. Exhibit E-2 shom the components of the VB.

'!'~"...

~--

:::r

::np,:~

Sub StartReco«ll"'l () .q .Lppl1cae1on.OnXey "a", •PrlntTiJDe.l" .lp£1l1cacion.onJCey "b", "Ptio.tTiiDeR"

..,. pE.:::!ld::...::Sc=ub"----------- -- -- - --

::~

- - - --

- - --

-

- - -'.

Sub StopReeorcliDq ()

l9P l1cat ion. OnJte r "a" .lpplication.'.OnKey "h"' End Sub Sub PdntTimd() lppl ieacion.lc't iveCell. Valutt • •1• .lpp11cat1on• .&c~1veCeU. Offset (0, 11 •Value • lpplication.lta"ll'! (" J7"t . Value lppUcation • .&ctiveCell.ottset(l, Ot • .lctiV1lte E!ld Sub Sub

Prln~Timei!O

l ppliattioD. • c:t1veC'e11. Value: • •a• lppllcation. ActiV!!Cell. Offset (0, 11 . Value • lpplication. Ranqe ("J7"). Value lpplicaeion • .lctiveCell.ottsec(l, Ot .le
... : ··

I·

The VB macro contains three principle types of subroutines:

• Sub Start &cording, which is associated with the "St:lrt Recording~ control button in the MS Excel worksheet. It assigns the •print1imeA" subroutines to a specific button pressed by the user (in this case the:: letter":~."

or "A").

• Sub SUJp &cording, which is associ:~.ted with the "Stop Recotding" control button in the MS Excel worksheer. It disassociates the "PrintTime" assignmentS from the button press.

• Sub Print1imd, which is the copy and paste procedure discussed above (in this case for the letter "a" or "A"). It prints the selected letter, moves into the next column to the right, copies and pastes the time stamp and moves back to the initial column and down one row. All three subroutines need to be uscr-cusromiud to reflect wharever keys the analyst wishes to use in the study. In rhc oeample shown in Exhibit E-2 only keys "A" and "H" are defined and all remaining keys will simply rerum the key vall.l(. The macro oould easily be enended to include mOSt keys on an ASQI keyboml, including all letters, numbezs and symbols.

Appendix E • S631J

2.0 USEFUL DATA COLLECTION FORMS

Sou=: Rd'er to Exhibit 1-1.

Symbol

In

tin ,ft yd ml

....

"'ml' yd'

... ..

_

..,.._

-

"'

011

"'

yd'

211.4 "2.114 ' G.3

~

0-$ 1.1.

--mbo

NIEA

.....,..._

tqt,l. . wrd.

--

..

I GilloN

""'"" ....

oobloywdo

...., om m

'" .....

1m' h•

...

211.4

..... ...... ....

VOLUME

JU

..... 0.~

u

o.oa ._, .

ldlo-

... 1100.-'1

·--1......

·--....

rniiU..

...... .............. .--.

I

Symbol~

T.oFtnd

1Symbol!

""

mna-

0.00

.. ....

G.4.

lnoh..

"'m

.......

u

foot

!em

.... ""

""" hi

............. m...,.

1.1

110

to

H

N

NIEA

0.18 1U 1.2 0.4 . .

f\lfUit' I'MM'*

I _ . ..._._

-(100-..,

~-

o.m

N-

-

I I I

....... ...... ................

1-·-..1:=1--- j••c

"C

vd

--

........

,;·

"'' ftll

d'

mil

..........

...

poundo

lb lb

~d-

ftoz

...........

VOLUME

mlllllilon

ml

I

h

....... mR•

MASS'"""'......... 2.2

ml

I

,,.. "'

2.1

1.1

I

In-

-.. --·-'"' ·- - -.... -· -· ·--o.e

1......

to

...""

I

"''" om

ltNI'EM'MIE 1 -

''f

I"-:ply I lENGTH

2.1

-

~

MAss~h11

abort ClOne t:2000 lbt

fq-

-

Symbol

....

Q.t

I poundo

RuWounoM

........ 1-...

""

To Find

....... .... ~--u

~~-

-- .... --- ·

I

1~1 LENGTH

"""' .... "

.. ,._

ft01

ApptoxlrMM Convwtion. from Metric MMiurM

..,pn>...,.,.C<>-..........10MolrioMoliour"

I

_..,.. _

METRIC CONVERSION FACTORS

0.03 2.1

.-

PI

.., ....... ............ ue o.a

q.....

• . . . .et

,~

qt

lll• l ftll yd' .

TEMPBIA'IUM 1-a

Jc.o..-

..--1 ..~

1~1-

.b.nn,mrfiY J: •

1;1:,~

Passenger Vehicles, Trucks and Miscellaneous Vehicles Tca.flic

Bound on

Strtet

Weather

From

To

Numbers of

I

1/4 Hour Starting

I

Passenger Vehicles Including Taxicabs

Other Vehicle Trucks

Excl. Bus & Str. Cars

- - I - -Dace

Recorder

Pedestrians Traffic

Side of

Scrcct

Weather

Hours From 1/4 Hour Starting

To

IDbound

Outbound

---- -Due

Recorder

Ccntinud on next page Source: Box and Oppenlander, 1976, p. 40. Refer to Exhibit 4-4.


, 'jJID-'"· :,- g)iii:it·'''!o·,~ ·"'{f.,''f~.(''. '(C· ''jffi;'":li}T-'"7-J';<;;;---y;; •· ~ f_,;,•'>·•••, . ;~ ~W!)l t£% 'ij_:W~Ijf · ' .fu\t '·, w!iJX\lf ~~ . .•.•~ . •- -~•• -~ @JYJiit.l.(l:?"!li . •lYIP£.tj~;.;i'~~H?~...t·'C;r.
i ,, i." ' " ' - ' . · . :0 "I

Pedc.mians Summary Sheet Ped~mians

Dac~

on

Str~CI

Weather

·Street

Compiled by

Street

Screec

Sc l/4 .Hour Stutins

Side Ia

O.c

Side In

Our

Sc Total lo

Ouc

Side In

Out

Side lo.

Ow

St

Sc Total

"'

O.c

Side

"'

Out

Side lo.

O.t

Total lo.

Ouc

Side Ia

Out

Side Ia

Ow

Toed ho

--

ou<

-To
Pealt Hocu:

--Continud on nat page

Source: Box and Oppenlmder, 1976, p. 40. Refer ro Exhibit 4-4.

-

Appendix E • 56-::;;ii'

Pas.enp Vehides, Truda and Miacd.l.aneoas Vehldes Summary Sheet

Tttfficon --------------~-------

Date-------------- - - - - - Weather ______________

114 Hoou

Compil~bY------~--------------

I ..._,..,. Vcbldar

I

Tracb

I

TtoAiit

Srreet

I

Nioc

Teool

I

=• I

Total Tracb

I

Tn..uit

I

MiK

I

Teal

lloo'::"

Cmtimud rm next pag~ Sow~

Box and Op!>fnlande.r, l976. p. 40. Ihler ro Exhibit 4-4.

568 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, lND EOITlON

~liB~·~All Person Summary Sheet Date

Traffic on

Weather

Compiled by

U4

s..ru.., """' .......... Vdides

P
TNdoo

1iuoit Mloc: VcWda v.b;da

'WoiDoc

Toal w.o-1

.._

Tooal

A--ac r..k

"""' Source: Box and Oppcnlandcr, 1976. ~- 40. Refer to Exhibit 4-4.

VdUdes

S trtet

PcooaaOuobowociV~a

T....D

Tn.uit Vdoideo

Mite Vd.ida

loal

Wall
Toea! Ia

o...t.o...cl PluO.t

1:~ ' fjj)'r,;;r,c;,~, ··•··"")'z.,.J~""~•n!' •·' ···~ ,,, .. ., ·®.~Jii.!J.Jk" t¥;rJ!,i.f?u.;{,l\lu·(.!:~jij;~·~;,~-o/rofo'-!AJ.lr'~~ 1.1Sll1>'9$,r~.:; •;..~>~·. ,.,._ ,·;.~Th~,l- ·

, , ... ,..

~~ -~

,. -. - .

VEHICLE TURNING MOVEMENT COUNT FOUR-APPROACH AELD SHEET Time NI S Street - - - -- - - - -- - EI W Street P •

tO------

Oay·- - - - - -

Date

Weathe r - - - - - - -- - - - - - : - - pas:Cilger cars, ttationwtQOI'I. motorcyd.s, pidc.-\lp lrudu.

Ob~rvt<------------------

T .. othe-r ltucM. (Recotd •nr sQ'Iool b'.l$ as sa: olher b\lsu •• 8).

J

p

:T IP

T

I I I I I

I J

I I I

IP

I

~

I I I

II

I I

I

IT

I

I I

I

p

IT

II

p

T

I

I I

'II I

I I

I I

p

I I

L

p

JT

II

I I

I

I I

I

I I

p

,

IT I

I

I I

I

I

T

.I

I I

I I I

p

I

~

I

I

Source: Box and Oppenb.nder, 1976, p. 21. Refer to Exhibit 4-10.


r

-I

-

~

'0 t1>

"'a.

)(.

. m

~

'

DATA COLLEC110N FORM Site 10 I

Route

MilepOsts

County

Weather--------------

Day

Date Start Time

£xrractlon Date - -- - - - - - - - -

End Time

TSR Unit

Diskette I

Data Collected: Raw Vehicles

Length State Speed Blns Mat configuration:

_

TSR State

Director(-- - - -- - - - - -

Speed Stats - - - - - - -

Stream Stata - - - - - - - -

Counts

Length Bin~ ---------------

I_ I I_

Lane _ _ _ Direction _ _

State ;..·- - - - - -

Expanded M i l e p o s t s - - - - - - - - - - - - -

_

.Lane--- Direction - - - -

,_

lane---Olredlon - - - -

lane--- Diredlon - - -

Noted Problems - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Site Charecteristlca beginning It (Milepost or loc:atlonl - - -- -- - -- - - - - - Lane Width SllouldelType &Width _ _ __ Median Type lntenecdona (4-legl: Signalized lntenectlng Roadways: Right Side left Side Otiwwlys: Right Side

It Width

Bridges - -- - Unslgnalized - - - - - -

Slg.

Unsig. - -- -- - - -

Sig.

Unslg. - - - - - - - -

Left Side - - - -

HorlzontaiCUrvwture - - - - - - -- - - -- - -- - - - - - - - - - -

·Terrain - - - - - - - - - - - - - - - - - - - - - - - - - _ . : . - - - - , OevelopmentType - - - - - - - - . . , . . . -- - - -- - -- -- -- --

om-Cha~ ---------~---------------

'

.

.-.

.- . .

• '

. Data Collection Log Removal

Deployment

i I

Site ID

Route

Coonty

State

Milepost

Date Collected

Driving Time ----

Setup

Driving Time

-

Retrl~a1

-

.:r~~~~~~~~:~::~:~~~~~!~:.:;:·~~~~~~~~;~.;Ar;~~;~~~r Darn CoUection Summary Site ID

lAtta Typ•l

Speed Limit

I I I

No. of No. of Miltpo.t lues Ow.ys


Horitonlal Tma.in

Cuswtun:

I Other

..

"d~ifil~·~an.m

..

~.!~ :.,,_-

~-

-- ;.,·lf_~'C'#-~!it.u. - ~--· ~-~.c:, ., - ' _.. - ,-,-- ··----. il'i •• -. ~1@'k :::'.~ . :.· •>;;·. ,:;,
.r..,l}il·~·

_@l)jG,@I.~~tK:~

<-r

-

INTERSECTION CONTROL DELAY WORKSHEET -···

- "~·-·

........

*-

· ·-

•••

.

•.. • •••.• - -

. ........ . . , ... .. .

·····-- ·· ... .. . Site Information

.

Genem//nformllllon

-

.Ii

......•

#Oo

Cic~o !\ ~~-:~L~ rr.

l n~
Anitr'it A;J!:ncy or t/Jmpaflj Oi!e P
•

a esc

Ae
~ C!h:r

Juri>dicti~

FM

_______

/npvlln/1/a/ Param111rs

193.9

Analy:isYm

_______________

-~

.,

Humberul h~E~ . Il fr«!.flow Sll!'d, FFS (milh) SIJn;sycount in~nial, ~ (s)

;crri.in3. v.,

2

T
40

Sl~pp~d vehi:les count. v,~-~

15

C~le

::'-0 22~

lerg:h. C (.;)

... Input FJeld Data Hu~r of V:hidesln

Court n:er.sl

Cbck

C-;ele

Tim:

tlumber

1

f

,:)

2

4:1)·~

4<42

+.47

.. .. .

Ou":u=

3

4

5

a

7

6

1l

1:3

:2

2

0

2

e

:2

6

16

6

0

0

2

,:)

7

It

...

14

2

0

0

-~

5

7

:0

:;,

:;,

0

i

5

6

:0

:2

,

2

..,.

0

0

:

G

5

7

9

13

4

0

0 0

2

7

,

6

8

12

i2

0

6

-~

7

r.

u;

a

0

37

Total

- Computations --Totah-.!tldES In queue, lV01 =

64

Ti~iii--QIEU2 JErVEhicle, d_q =(Is • ~) • 0.9 "' ll Ho. of -.rehicl~ stowing per lue each C)-tie =(i(i'Ui A:c~lf{)ec!l corr£di:m fattor.

Source: Refer to Exhibit 6-1.

Cf (fx. A1 S-2)

8C '"-- - --G: --l--..-

t4

·~

s

10

G

= v

Fr.~t'JJn of vehicles slcppng. F\'S = V,: Accet!Oe-:el corredinn dela·f. d,t~ =FVS • CF CCJ~trol delayl<~£hicle, d

9

0

0

lll.fllh.:roft;tlessur\'E)cd. N.:

3iY

9,5

4

8

=ci.q • ci.4

7.o 0.42 !.7

$

. :1.2

s

Appendix E • 57S

I I

.J

FIE1D SHEET· SA11JIIA fiOI! I'UJW ~DY

L~~--------------------~----~--------------------~-----------------

_______ BoundT~

~:

Unk--------------------------------~----------------------

G,.do·

~:

~------------~-------------

w..tl>v. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Oboe,._

,. UM Width: ______ft_ AtH:

I~

Oilier.--------------------Obo

l1me IMcondsl bolwN n 41hwlllcloand'lllo wh.

8thv.h.

8111 ..11.

21 '

2

22

a

22

4

24

I

21

•

21

7

·27

~

:za

•

21

10

30

11

31

12

22

13

sa

14

34

16

,.

17

t
10!1> wh.

I

I

37

38

1t

38

20

40

ColumnS

---lv!lhl· to Exhibit 6-4.

8111wh.

35

,.

~fer

411>-•""'lllo wh.

lOCI! - .

1

1e

Source:

Tomo t - c l l l -n

No.

-

·-

Cal

011

1!00 • TCI!I! Nymb!t o! Obo!l'f!!lont

If

+

'

+ \' + '

lcl

lAS)

. ... .

..

~-

·~

•

.....

'•

.

;o.·- .

.·:-·· ..

.f

,.,.~

'

.

..

-.... ·'

"" "

.

..

..

Survey Dare ' Location

Crosswalk across

End of Survey (to nearest minute)

Numb_er of lanes 'N'

Starr of Survey (to nearest minute)

Roadway Width 'W'

Total Survey time (minutes)

Adequate gap time 'G' Number of Gal s Tally Total

Gap Siu Seconds 8

Computations

Multiply by Gap Size Ga

9 10

R + [W/3.5] + 2(N- l)

G c _ _ _.sec

11

l2 13 14 15 16 17 18 19 20 21 22 23

T=

total survey rime x60

T=

_ _ _ sec

D=

t
Ds

_ _ _%

24 25 26

r=

. total time of an gaps equal ro or greater than G

27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43

'

Totals

Source: ITE. 1998. ~ferro Exhibit 7-8.

"'

~~.

-

:&OO'MJ1l.&>'~b rJ~:i".JW'l;tfu1:~~~f.'i[~;furiij)£~f'5 :Z," ~~">'::-\";,::.>;:;~~;;;."t.":..'t-?t:rf.;'l1J;: ..~'::.::?.•: .~~'J)• ; .. . ·-··· ,_ .-· - . . ·, '" _- __ ' •.,., " - --- -. - .·<>· :;...... . . .:J :· ~ .·

"'--~''

~-~

~-,-.

~····--

Survey d:tte Crosswalk surveyed

Crosswalk across Divided Ro3dway?

Yes

No

Curb ro curb distance Number of groups Group Size

5 or fewer

Number of rows

Tally

Total

Cumulative

Computations

1

6 [0 10

2

11 ro 15

3

16 to20

4

21 to25

5

26 tO 30

6

31 to 35 36 to 40

7 8

41 to 45

9

46 ro 50

10

lime period studied Number of adequate rraffic gaps Number of minutes in the same period ls the warrant satisfied? Source: ITE, A Program for School Cror1ing Protection: A &commmtkd Practict, 3rd Edition, 1972. Refer to Exhibit 7-9.


Driver Observance of Stop Signs Field Sheet

loca~on

__________________________________________________________

Time

to

Weather ---------------- - ---- - - - - - - - Non-Stopping

Practically Stopped- 0 to 3 mph

Stopped by Traffic

Voluntary Full Stop

c: 0

~

ul

vi z

Right__ _ .:....__

Straight

y

---

~

Straight Voluntary Full Stop

Left

4 Right

Stopped by Traffic

Practically Stopped - 0 to 3 mph

Non-Stopping

c:: 0

~

.... vi z

----- - - - - - - - -

Date

Recorder _ _ _ _ _ _ __ _ _ _ _ _ _ __

Source: Box and Oppenlander, 1976. Refer to Uhibir 8-4.

Appendix E • >1~

Direction ofT ravel 1rune Date

Area Type._ _ _ _ __ _ _ _ _ _ _ _ _ __ _ to _ _ _ VVeamer ____________ ______________ _

Observer

c A R

s

T R

u c

K

s

Source: Adapted from MokJrist Compliance with Srandltrr/ Traffic. Refer to Exhibit 8-5.

Driver Observance of Traffic Signals Field Sheet

l~tion

_______________________________________________________________________

ilme

to

Weather --------------------------------------------•• N.S.E. ,., ..... W ••.. un

il!! !" !"

~

1eui!S padwnr

0

"

;: &'

~

c

pa)j

3

l

~

"' "

~

~

."en

60 !!.

Uili~J9 J~ljt MOll~,\

.,..t ~

C'l

;;;

...

"

~ UllaJ!) ___.,._____ ,

··-- .

~ N. S.E.W

· - -· - -· - -· •.. Ufl

~'""''-

Grl!i!n

j ,..~ ;:

c

...:!.3

.."'

Q,

~!!.

,...f ~

C>

Yellow after Green Cl

a

%

["'

~

"

"' ~

,1 Date _______________

:oaie"

Red

~

=

Jumped Signal

I 5

;:

...

.,;

z

'

~~------------------------------

·:Recoril-e;: · · .......... - ... · ····---· ·-···· ·--· --·.. -

Sou=: Box md Oppcnlander, 1976. Refer 10 Exhibit 8-6.

AnnPnrlil< !' • '!i:R 1

,;~~1;1,~~iWiiJ~~·~t.~ift.~~@Wi~&nur11i1£~~::j;~~;;,'t.w!;-.~r~~i:"~;; ··· ... · ·n· • r ·c- · • · • . ,. • • . .. •·• ,~-

Locarion , Direction ofTravel Time Date I

--~·

t ·

.

· , · · . • ,.

·- -

-·

• • • . . ., , .. ,,_.

- .

•

·

.. · .... .

. . . . . . . .~

Area Type

ro

Indication

Wearher Ol»crver Acci.o n

Cars

Trucks

Total Cars

Total Trucks

Turned on GorY

Stopped on R, Waited for G, Turn~ on G On Green Behind a Waiter (above) Turned onG Accempt~

'

to rum on R, turned

onG Full Stop

'

Stopp~ by Cross Traffic

On~No Queue

Stopped by Pedestrian Crossing

No Stop

Full Smp On Red Queue Sropp~ by Cross Traffic

Smpp~ by Pedesrria.n Crossing

Toals Notes: _____________________________________________________________________ Souoce: Adapr~ from Mowrisr Ormp/Uma with Sumtlttrri Traffic Control Drnca. FHWA-RD-89-103. ~fer to Exhibit 8-7.


~~}·

•;· ,,· ....,~~ -,·l'•.

-.-.~::. !' ..f.···n'•";~;•·;::'·~· ~.;.;: , .··..;.:--.•v•... ,... ':'.l'

:[:· . . ·1•.,.

-:· · r

.

-e- 1 :,

· ....... • .. - . ••. · ,.

.::r..">-'~·•.. ~-........\

,."'

,._.

~

·~:II'~J!'.ll!'J~·'flt~J.fr!'..q..\11@'ftl,.~~ l!W%,i!i~W.~v.U~Jif("'' ·;'1/ ·:~ ·~ ______,_

Pedestrian Observance of Traffic Signals F'..ld$1><01

LOC
T~ Pedo•tll•••cr=in&

to

of

wea~' --------------------------SLonlhe{N.U .W.) _ __ S!de_ _ _ __

St. in

Oir~cdon

_ _ _ _ _ _ _ __ _ _ _ _

~============~~~----------~~~~ M;~s~~·~hl,-----------lii____TI;o;,,.~,------

1

-

lcrosswalk}

Stepped from Curti on

" ~ I~ ~

-

~ 1:§p .

o!!

,S!

J ~~

"'

~jiiJ

~

I.

,!(

t;

-z l~

I

Croued

O~con•lly

I

::::::::

"' 3:

-

hl~ j~t Q

"

Total

O>ll _ _ _ _ _ _ __ __

Reoorder _ __ _ _ _ __ _ _ _ _ _

Source: Box and Oppenlander, 1976. Refer to E:iliibir 8·8.

Appendix E • ~ 3

Tr.avd-Tune and Delay Study Test Vebide Method Dace

Weacher

Route

Direction

Trip Started at

At

Trip Ended at

Trip #

(l..ocauon)

(Mileag<)

(l..ocation)

(Mllea&
At

Location

Tunc

Location

Symbols or Delay ca= S-Traffic Signals, SS-Srop Sign, Ped-Ped.csuians, BP-Bus

Source: Refer co Exhibit 9-2.

\l 1 C RA a

U.cHJIIAI f\J; TRl\N<;I)f\Q T/\Tif'\1\1 t J.Ir:lf,I CC: Oit.lr.. <"TI I n ii:C

~t. IC"'\

Cr'IITinM

Type

Delay (accond.s)

l:r.l;

.

•

.... . .

.

.

.

.. -· ..

..

-

••

·~

.

lind-Time and Delay Study Test Vehide Mechod -Volume E:rte11.1ion Dare

Wearhec

Route Sw-rPoinr

E.nd Point

Ru.n

StartTtme

FuilibTame

Travel Tame

Vehides Met

Vehides Ovuuking

Vehicles hssed

V~des

Start Tame

FuilibTimc

TcavdT..me

VehidesMct

Ovutaldng

Vehicles Passed

Bound 1

2 3

4

5 6

7 8

9 10

Total

Average Ru.o _ _ Bound 1

2

..

3

4 5 6 7

8 9 10

'

Total

Avenge Comments O~cvec


Location ______________________________________________________________________

Time

Weather _ _ _ _ _ _ __ _ _ _ _ _ _

tQ

Date _ _ _ _ __ _ _ _ _ _ _ __ _ _ Adults

Observer _ __ __ _ _ __ _ _ _ __ _ Children

tN 1:

c:

::2

::2 :E

~

~

6

u

11:I

~

:;

~

~

Street Name

....

E z"' ~

~

VI

Adults

Children

Source: Refer To Exhibit 12-1.

586 • MANUAL OF TRANSPORTATION ENGINEERING STUDIES, 2ND EDm ON

~

~~~~~~~:~~~~~~-fl~~... l-~ti:~:·~~*;~ti~:,;~~~:,;:~.fi~~ :·~;~(7~.~-~~Y:~::·~!ff~r\!~i~:o·7~~-~1 Loc3tion

1ime

Date

Walking Speed

Crossing Dimnce

Critical Gap

Gap Siu (sec)

Tally

10

13 14

--'

--

17

I

19 20 21

-~I

15 16 18

Total

--

12

I

---

-

II

.

-

:..--

-

22

-

23 24

--

25 26 27 28 29

-

30 . 31

-

32

..-

~

-

33 34

-

35 36 37 38

--

'

39 40 Tow Adequau Gaps

-

Source: Refer to Exhibit I 2-8.

Appendix E • 547'

Take a walk an d u se this checklist to rate your neighborhood 's walkability;

Loca tion of walk - - - - - -- -

Rating Sca le:

1

2

1. Did you have room to walk? 0 Y•

0 S\ile"WalSu oc pelbl tQft.t 1104 .opJ~4!'d $\d.......Ju, ...... ~

«ceck.&

O S".&~..,.~-""""*'pol'"~ .ap..

.....)l.ery. . . ., .... . c. [J No .~.s-m.. or ibo.Sdtu 0 Too JnOCh tnlic

'Ia;;;=====

c_ Loc:uior».ofptoblcmc . , . .... Rodac; (clodt on<) t2S4S6

Ov..

ONo

c v...

ONo

ov..

ONo

ov..

ONo

0 Yo

IHMdbe"'""bJ41n.tn:~

0~.!'-ot~ -------

=~-:.~=:~:O:o~ Cro. 'Witb ctM ape?

OScvy~

0Non...UIIJII>oo4 Dlny, kltt crnn.te or ltrllllh CJ Dlny olr 4., ., n Knabilt ftlUo!l 0 SoiiWthlna • • _ _ __ .

0

Loaodonootp-

6

How does your neighborhood stack up? Add up your r&tings and de cide.

0 Scnv ~ Dtlwn... c--·ot.-,....-~ C OWDOt",.W.to~~chl:~

0 Spe4 ~co....._. It ttuo11Qb ....c Uytau «

4...... ""-c-> .....•llal>nl

..... ------ -

U-l4 C4hbatd 'Jbt ha¥e • put

l . __

21 ...'15 Cclebc:u•a We. Y~r

s.__

1'-20 Obr.bat\t n...,d,1 WOf'l:.

··--

oo......-r..

c~,:..:-:-,

·--

:t. _ _

C Tu...t iDOl» pro,.•croc&lng:dt• slr'Ht

T otal _ _

....,...._. ... -cans.

..........hood lo ,....,. _...

u .. u , ,.. ,o

It oeeddon ot'WOitc. Yo. dottft'W blcl.rtbu tt.t. h''• • d.ialt.c , . ......,

~~·"""--> I 2 ) .. I 6

Now th~t you've identified the problems, go to the neJ!,'t page to f ind out how to fix tllem.

Source: FHWA. 2009. Rdcr to Exhibi< 12-13.

lAhf\111/\f ()C

- ----

1 2a •s'

3. Did drivers behave w ell?

a

-

0 Some~l..-.,rt~\ngl.: 0 NH4fd ,.oot SCU&. Oowtn. ot tNft

RAdacol
a...._ (clo<\t_,

CtlR

I ..

~bdi::ftc~•JMO!

OS
occpw~-.coei'Ot'

C N..-..,p.l c_.._« ............. 0 ~can Wodr.M4 ou "rit'W' ot....:lc 0~« ptan• ~ow ntw oCwalk 0 NHded ~rb runpt ot c:anp ..eded ~

OY..

C'ftlaa~wwbr ~ ~oo~~lj

5. Was your walk pleasant?

0 Somt pa>~~~.....

s

•

s.., .... _ld.,.. """ ..... t-adono ~pod>.._ --

O~WIJKIO WW.

2 ) ..

'

R.dll&l (
0 1'b6: s[. . . -.&4• u - - a:.o ~or did

s

4

.... ...,..,.... •c.•.-

.,)456

2. Was it easy to cross streets? 0 ......

·-·

_~

4. Was it easy to follow safety rules? Could you and your child ...

0 Sonw pfObl.ttb: 0

J

........ ...

---

TO/\M('Qf'\QTATint.l

rl~lrl•trrnl•tr

I"Tl ,,.., ... ,..

""'•"' ,.....,,_,_. ,

""

~·

!

:

0.

,

-. :

.

... . . . ·'

..

'.•. ••

.....~~""#.'« ·''.t"': ...

~~

,

••

~Xt~~Jii~~~gn?fjj

Boarding Count Field Sheet

\ Route

Block Number

Day

Date

Weather

Observer Route Segment

Boardiog Passengers Reduced

From

Souzce: Refer to Erltibit 13-6.

To

Full Fare

Reduced Fare

Full -+ Transfer

Transfer

..

Transfer

Alll'assq

.

~~~~-~~.:r~~r~z,~~_: ~.J~:~tf¥:· ~~~:~~~~-~J~j,t~::·:.?~~~~!·~:~ -ttJ£~~-¥1~~~:~~!t~~·::·-~·$!I Point Check Fidd Sheet Bus Stop Number - - - - - - -- - - -- -

Rome (s) - -- - -- - - - - - -- - Dar -~--0 Depu ting U>ad

Dar~ - - -- - - - - -

a Arriving U>ad Route Number

I Direction

Wearher - - - -- - - - - -- - - - Observer

Block Number

Vehicle Capacity

Arriving Tune

Scheduled

I

Source: Refer To Exhibit 13-7.


Actual

Passengers

, ·~~~t.~~(>·~iji.~~;/ft:_',;;:·;:i~J¥,ir,t),..- ,,p, .. •v

- · ' ' .,,

.~··,•-;_,,•'1"

.;1

.~,

'•'#o•'·

RIOE tHEtK fiElD SHEET

8l0 tK HUMBER

ROUTE NUMBER OATE

DAY

WEATHER 08 SERV Eft

DIH CTI ON OF TRIP SCHEOULECI START TIME

Lou:lo11

I

Oo

Pnsan4ers Oil lu4

Ti..-• Chtc11:

A :marlr.s

!

I

Sou~ ~fer To Exhibit

13-10.

Appendix E • 591

-

,UBUC TRANSPORTATION vtHia.E OELAY FIEI..O SHEET Oq.

o-:

M elhod:

TllpNumber. Tllp lkan llrM: TrlpEnd11""':

Weather:

"oute: Olrection:

VehldoT~:

111

121

locellon

llme•t

131 Time

corwol

point

-d

elo than

.•.

141

(51

(Ill

-d

O.loy

,nmo,

Stop

lime

than

welldng .

COUM

welldng

TOTAL Dt!i.AY TIME, SECONDS

Symbolo lor-Y ca.-: P • po._loedlng. S •

•.w~c

olgnol,

SS • otop algt>. PIC • _....., caro. DP • double patDd. PEO • pectMtrian~ RT • right tuma, LT • a.tt tuma, T • een«ll

congestion, KT •lntenllonally ldlledllrne, 0 • other IOICPialnl•.

Renwrb:

Sou=: Refer to Exhibit 13-11.

m 151-131 Deloy

lime (MeJ

;~:5'1i11 ~ .

.

..

D•

..

.•

..

~

~.

.•

~-

...

.,

..... ..

"

~~j.i:w. o:,.. ?",· .· ~'~~~ ~. ,_ ....

r

Truck Daily Log Sheet

\Dare

Ciry/Area

Beginning Odometer Reading (Starr of First Trip) Ending Odometer Reading (End of Last Trip) Commodity Handled

Stop

Number

Scop Location

Arrival

Departute

!we

ll.IDC

Type of Basiness or Activity

Load Factor for Truck or

Trailer Quantity

Type

After Scop

Notes

Jsr 2nd

~ 3rd

~

: 4th 5th 6th 7th 8th 9th lOth ll"th 12th 13th 14th 15th 16th 17th 18th 19th 20th 21st 22nd 23td 24th 25th Comments Driver

Soutu: Refer to Exhibit 14-7.

:

""''

..

~~~ ;·...-~!110 -~~~w~t-~~~r~ruful&?~~~:v~;-.;;x¥J:~~~lffit~~JJ21J;f~~~~ Suee! Parking Tlltnover Su.mmazy Sheet

Dace S~t

Weather_ __ _ __ _ _ _ __ _ _ _ _ _ __ _ _ _ _ _ 1imc

_ _____________________________ End1imc__ _____ _ _ _ ____ _ _ __

Location

Facility

Block

Facility

Type

Number of Stalls

Total Tumover Vehicles

I

Ho11tly

Turnover Rate (vehid es/stallf bour)

~mmcn~--------------------------------------------------------------------~----Observcr Source: RdU 10 Exhibit 16-1I.


~~~~~a~

_' .!.,.•

'iif~~~a1i/'¥ ~~~~.r.:-t; · . .,._.,. .... :·~.+::::;"':;·.v.O::.·!t··.;~:·G·-~~¢.-~;r~..,.,-;,4'~~~- ~ ..,-;* -.~.:,_ ~ .. : ·~ , . . .-J~ y~~ ...~ • ::;t;·lf1!~ ~: .o"t-.,..,~9'""';a~-~~~;"i~~-:::4·· :?(·~'_;~~-:.·.}•.i;.:-:~t,:-.'i,.·.~-':'1;~~.:-·!!;.~-':~~k~.)~·:Jo:,:__·:.-- · ·.,,......:~~.-\ ~-~\

!

Parking Survey

The \ Department ofTr:mspomuion is conducting an analysis of the patlcing conditions in chis area. We are asking each driver to complete the following questionnaire. Your answers will be kept anonymous and data will be summarized so that the analysis and report will in no way disclose data on individual!. We thank you in advonce for filling out this brief survey. Your time is valuable, and the answers you give will help us improve parking in the area. I. What is the primary purpose of your trip? (check only one)

OWork

0

Tourism/~creation

0 Personal Business (Medial, Banking, Socia.!, etc.)

O School

0 Shopping

I

0 Other (please specify):

I

2. What was the specific location you visited immediately after parking your car here? Street address or bwinw narne: 3. How long were you parked at chis locaLion? hours

minules

4. Where did your trip start (where did you come from just prior to parking here)? Nearest street intersection: 5. Please share any additional comments you have about parking in chis area; Source:: ~er to Exhibit 16-12.

Appendix E • 601

~iB;~~-.~ .:~~~~~~~~r Parking lotervicw Sheet Weather

Date

EndTtme

Start lime Location Trip Purpose

Tune

(Work, School, Shopping. Start (Parlred)

End {Left Stall)

Trip Destination

Office Use Only DW"atioo

Personal,

Street adress or Bo.siness

Other)

name

Hours

Miou~

Walking DisWICC

'j I

( I

Comments O~rvet


i'~~·!)ibit E;_39. Traffic Conflict F,orm Wlth Or~ line Per,Tun~ · P~rl~~ :;~t:~: ;.:. : ·..

.. '·!

.:

I

:i

;~,,-:.~} · ~~~

. .. . -r•O

~-

•( .'·:: ~···.' ~ ~'~.\:.:t}:~\~:;~ ~~~;~~

INTERSECTION TRAFFIC CONFLICTS SUMMARY

Location

Leg Number(s) Date

..

..

~---

i .i 1 "".,

Cl3 :::

'§e

8

~

~ ~

1 <

Observer(s)

C=Conflict I;el\~1\111:

Same Directio~

Right-

RightSlow Lane Opposing Thm Same Vehicle Change Left-Tum From· Right Direction Thm

RightThrough Tum FromFromFrom· R!ght Right Left LeftThm

c sc c sc c sc c sc c sc c sc c sc

Total C + SC \

Length of Recording Period

SC - Secondary Conflict

Daily Count • Rate Per 1,000 Veh

Source: Parlcer and Zegeer, 1988. Refer to Exhibit 18-22.

0

LeftTum

il

·A ll Through RightAll Sam~ Through Thm On From- From·· Direction CrossLeft Red Left Traffic

sc c sc c sc c sc c sc c sc c sc

~

.'i';:;.::,titW~~ ~~tr.v....;...r~~·~;..-~'.i;!'!la~'#"'"'''-· !'!!~~,,,,, "~'''"''"""*~;,!~•<;,.~":r"':>-,;P,"-:-:.,~;,~1".f''·

~~~~-u,~~~· ··.~ ~~·~.JI.1,~1Wr~-~W!-~i~l~.~Jf~!:t·~~~ ~1.3f=t~~~.~~;;. :::sz~~..::.~~~- ~~r ~f~:?:o:~~l?:r.-~~:~.~J:t~:t~,~~t~i~

ACTOR CODES

ACTION CODES

Name: Date: Time Period:

Intersection:

Direction (leg with actor I ): Weather:

Time

Actor I

Action I Actor 2

Action

Source: Hummeretal.• 1989. Refer To Exhibit 18-23.


Comments

rnrn rnrn rnrn rnrn IIJQOO'

G.oQII'liPI'I0c1oc•lioon

Ftoo'

FJ~

Ft~r

"•-'CS.nc• tvP•

,.,....,..

0 ~: :: ;;~

O•M•tp(Oc)f\ or .....,..,

ll•.tldonc• type

"'u' fl

0~: :: ~-- ~':,

""H

R•~t'fiM

o~:: ~~:::r Retklonc• tviM

"'"'"

0!::::

~~:;.:, ·

us.

1TrnTrr·n. rrcr1 O.aetipllo" o t

""d V t-$

lr iTT"itnTr TiTJ IT.iTrrrnlrriTl •-""-'*""···-... ,.,.. •. I I I I I I I I I I I I Ill O..CtfpciOn ol U nd UM

tt~·i>

o . .c ripUon ofLaMU c•

~ Olhtt

Source: Refer to Exhibit 20:2. [~~~-~'2!;71\{Q~-;~;~ +

,..

'

•

...

" ..~~~ - .··· ~

_;:·.~:' · ~~~~ :.:iiii:t~~':~~~~~fl!f~J-~

Origin-Destination· S tudy Field Sheet Location

Station Number

Time: Begin:

\nbound

End:

1

Origin

Outbound

J

2

3

4

5

Oestinatfon

Route Used

Parking

Othe r

IMUftW ~ ..._oOCI:- Wftt. 20M, er OUI.w C'*1

Date~

S"'""'-ZIOnltt,.«li.laJI,..y

Loc:uiOOMIIII1)'Jie

Observez

Source: Box. and Oppenlander, 1976. Refer to Exhibit 20-6.

-

Appendix E • 6CP 5

8 ~

i d

:I (I ,. : !

.,'

..... .,.....

..----..------, ...------

!· ~-

i

SoutCC:

ftaak t • u Y e"J' M\IIA

Box and Oppcn.lan
:!,

• • • •" • • •

-

• -~ • • · ~--u·,nT
f"t

t..-u ,,.,..,"' lr' rTI U"'\lrf"

""~"'"

f"r"IITM"\fd

~or

'f-

Coo,.rau ...

License Plate Study Field She~t Oi,...ctlon or Trafllc

Location nme: Segln: End:

Ucens• Number

Statlon Number - - - - - - - - - - - - - - - Weather

Time

Truc::l< or

8WJ

I

Out•

ot

State

?

Dater._ _ _ _ _ _ _ _ _ __

ueense

Number

Time

Truck or Bus

I

OUt·

·or State ?

~n~'---------------------

Source: Box and Oppenlander, 1976. Rtfer 10 Exhibit 20·1 0.

Aoot>ndix E • 607

...

0

"'•

. , ..t;

•:

~

A

)>

z c

)>

r-

~

..... ::0 )>

z

"'""'0 ;D

E 5 z

5 6

Mor AcctpCtd Gaps:

7

Nor R• Hltd Cops: Attc_p11nc:e• ,.. Re.Jcctions:

9 10

0

22

23 24 25 26 27 28 29

0 m 0

=l

5z

IS ~SUM(G6:07)

•06/08

•ll6/E8

•F6/F8

I 0

1 l

3

4

4

6

.

.

'·,. ·. •,

~

1'01'.\L

9

~sUM(l>t.:li~J

s ..

•06108

-SUM(07:117)

0 • SUM(H6:H7)

-sv~·

•H6/H8

Tobit A

13

-"' N z

;D

Y'ES

12

..... "' c

zm m zGl

4S •SUM('E6:E1}

IS 30 •SUM(F6:F7)

. ,_

11

;;;

Gl

s

0 60 -sUM(06:07)

-

..

7 2S

'·' lncr<8llng Proponions?

'-'~ ~

G·

0

Auumc lhat 1ho followi~_i_l*_E_etCCC1 tancc dam set wilh 1 2as:ecc>nd interval was eol1eeted: , I Gop/Loa Slu ( soc) J s

4

14 15 16 17 18 19 20 21

"' z

c

B

30

31 32 33 34

Colnmn N6.: Criricol Cop (scc): At«plcd Gop (se<)

0 0 -DI71(100-SDSI6)'100 0 • 018/(IOI)..SD$16)"100 ..0 I !!I(100-SD$17-SOS I6}" 100 -DI91{100-$DSI6)"100 -1>19/(1 OO-SDS17-SOSI6)"1 00 -G81S1S8' 100 •HS/SI$8' 100 •D2Difl 00-SD$1 6)0 I00 cOlO/( I OO-S0$17-$0SI6)'1 00 -sUM(D 16:0 20) •SUM(E16:1l10) ~SUM(fi6:F20)

I

• 08/SISI'IIJO c£8/SI$8'1 00 •F8/S1S8"100

3 5 7 9 TOTAL(''!.)

0 0 0

..

s 8

·.,

.. 0 0 0

• DI9/(IOO-SUM(SDSI6:SDSI8))"100 • 0201(100-SUM(SDS 16:SDSISll'l 00 •Sll~ I(G 16:020)

0 •0201(1 00-SUM(SDS16:$DS 19))'100 • StJM(HI6:H?O)

s

Table U Col¥mh Number Crlrlul Gop (Itt) Atuprod Gop (sro\ 3 5 7

'TOTAl..

.

Cri
1 2

]

l

4

4

6

R

• E6 •EIS/100'0532 •B19/IOO"OSJ2 •E2lYIOO"DS32 •028" 100/£17 • Dll/SRSJ2

0 •Hl9-029

0 0 •f00-SUM(D30:.EJO) -(Ol«VV 00)' F532

0

•E6

0 0

•F6 -
-fi31-SUM(D31 :Fll)

•H6

-GJI'IOO/HlO -G321SHSJ2

• SUI'-t(lll8:HJ Q

"'(FI9/IOP)'E$32 i"'(FlD/lOO)'E$32 •l!1t"100/F18 • IJ.l/SI!$32

-(Dll'P26+E32'

F30'100/GI9

Total

•lf.\ l/(11$32

.

'

. '\

'" '"

In dEx .. . . .

~

~

••••

~

••••

•

••••••• "

......................

A MDT. S,·c average annual daily

0

•••

•

0

.................

ADA. See Americws wirh Disabilities Act

traffic

ADT. Su average daily traffic

MSHTO. See Amcri~n Association of State Hiehway and Transpon:a tion Official; .

ad,•anced traffic management (ATM), 9, 179

MSHTO Highway Safecy Design and Operations Guide, 385 accderacion/deceleradon dday, 100 on fr~eways, 191 accelerometers, 407-408, 407f advisory speed and, 409r, 412 for bicycles, 242 centerline markings and, 139 definition of, 9 for pedestrians, 242 acccpred gap, 110- 112 definition of, 9 Ramsey and Roudedge method and, lllc accepted lag. 9, 110 access, 463-483 classification for, 435r accessibility for bicycles, 252 definition of, 9 for pedemians, 238, 252 for public rraruport:aion, 270 Acccssibilicy Communicy Transportation in our Nac:on .(ACTION), 264 accident modification fac[Qr (AMP), 9, 375-378 accumulation srudies, of parking, 330- 332,332f acknowledgments, on written reports, 551 ACTION. Se-e Accessibility Community Transportation in our Nation

advisory speed, 406-4 12 accelerometers and, 409t, 412 ball-bank indicator and, 409c, 4ll-412,4lh data reduction and analysis for, 409-412 definition of, 9 design equation method for, 410, 4!0t · design speed and, 406 samples for, 408 aerial swveys for densitj•, 187 for iTD, l72 air cargo, 293 air quality, 457-460 nonaruinmcnt area and, 457 reportS on, 460, 460f standards for, 458c algorithms, 205 for CF. 10, 202 delin iri<>n of, 9 DTA, 211 for gap acceptance, 13, 222 for lane-changing, 15, 225 for micro~copic model, 202 NGSIM, 204, 204f for operating speed, 227 optimization, 16 for simulation, 19, 205,218 alleys, 419 ill-vehicle sampling for spor speed, 86-88 spot speed and, 136-137 American Association of Stare Highway and Transpon:ation Officials (~HTO), 9, 385 collisions and, 348

0

0

•

•

••

0

•

•

•

•

0

.....

~

••

freeways ~nd, 178 JSD 3nd, 1I 2, 115 TCDand, 144 f..meri~n

National Standard Practice for Roadway Lighting, 41 5

Americans wirh Disabilities Acr (ADA), 128. 385

AM F. See accident modification fuctor · analogy, 475

.~ .I

AJ'J.OVA Su one-way analysis of variance APC. See automatic passenger counting

.,i ~;

appendix, in wri tten reports, 552 approaches collisions ac, 369 conuol delay at, 222 definition of, 9 delay ac, 210 to interseaions, 28, 210, 488 lefc-rurn signals at, 355 of minor screec, I 25 roundaboulS and, 46 speed at, 79. 98, 115,401 video ar, 65 volume :u, 44, 47, 146t

..

;

.. ; :;

approach sight triangle, 113 appropriate knowledge, principle of. 26 area chart, 544, 544f classifications, 419-420 counts, 49-57 cransponation plan, 477-478 arithmetic mean, 523-524 Arizona Freight Network Analysis Decision Support System, 295,

296f urival volumes, inrersecrion counts and,45. 45f Index • 609

d

·. '

ATM. See advanced traffic management ATR Su automatic traffic recorder attainment area Clean Air Act and, 9 definicion of, 9 EPA and, 9

!ij.! j! ~i

u '·

it ·. I IJ

~!

i

i

r: '

ii .. i

1·

i. !

''

automatic coums for bicycles, 242-245 data reduction for, 68 for Aow, 183 freeways and, I ~Sf for pedesuians, 242-245 periods for, 69 for public transportation, 281-283 fortesr vehicles, 165 for TID, 161, 165 for volume data collection, 62-66 automatic passenger counting (APC), 272, 281 ITS and, 281 automatic traffic recorder (ATR), 192 automatic vehicle identification (AVI), 173 automatic vehicle location (AVL), 281-282 GPSand, 281 ITS and, 281 uavel time variability from, 282f visualization with, 282f pavement markings. s~~ also cemerline markings compliance with,_144 at crosswalks, 388 reaordlecrivity of, 139-140, 139t reuorellecrometer for, 139f average, 522. mobility, 197 for spot speeds, 523f average annual daily .traffic (AAD1},49,57, 192 .collisions and, 355 ddinition of, 9

Avcrage-CarTcchnique, 161

average daily traffic (ADT) conrrol counrs and, 55- 56 cordon count and, 50 definition of. 9 volume and, 119 average day, 474 definition of. 9 eight-hour vehicular volume warrant and, 124 four-hour vehicular volume warrant and, 125 peak hour warrant and, 127

AVI. s~~ automatic vehicle identification AVL. s~e automatic vehicle location axle counts, 68

8 ball-bank indicator, 407-408, 407f advisory speed and, 409t, 411-412, 411! centerline markings and, 139 definition of, 9 bar graphs, 27t grouped, 28, 28f stacked, 29, 29f for volume counts, 71f beacon, 10 before-and-after test, 92, 93t analysis for, 492 for bicycles, 255 comparison in, 494 with conuols, 492-494, 493f drawbacks to, 489-490 in experiments, 489-494 hismry in, 489, 491 for lighting, 415 maturation in, 489,491 overcoming drawbacks, 490-492 for pedesuians, 255 forTCD, I36 units in, 489, 493 warm up period in, 491 benefit-cost ratio for councermeasures, 374t, 377379, 378t definition of, 10

for lighting. 425-426, 42)t biased responses, in surveys, 513 bibliography, in written reporrs, 552 bicycles, 237-259 accelerometer:l for, 242 accessibility for, 252 automatic counts for, 242-245 before-and-after test for, 255 classification for, 4!9 compliance and, 252 ,,.. #-·confiict anp, 251- 252 data collection for, 253- 256 definition of. I0 GPS for, 242 handheld count boards for, 239 HCM and, 240 . intersection counts and, 46 laptop computers for, 239 .LOS and, 240 manual counrs for, 238-2·h map for, 259f MOE for, 253-254 networks and, 252-253 'fCD and, 252 time-lapse photography for, 242 uavel paths and, 252 visualization for, 258 volume and, 238-245 bicycle lane definition of. 10 QOS and, 253 bi.keabilicy checklists, 252 bikeways, 419 bins, for gaps, 110 blanket method, 139 block and curb face numbering system, for parking, 326f block designs, 496

A Blu~rintfor NEPA Docummt Contmt (NCHRP), 452 bottlenecks, graphs for, 32

BRT. s~e bus rapid transit

buffer index, 197 build-up, 477

i: Bureau ofU:nsus, 450, 508t

central tendency, 522-526

bus rapid transit (BRT) ddinicion of, 10 public transponat.ion and, 264

CF. &(car-following CFCs. &( chlorolluorocarbons CFS. S(( Commodity Flow Survey

c CAD. s~e computer-aided design calibration, 205 ddiniuon of, 10 of inputs, 223-227 for lasers, 85 for radar, 85 from screen-line ~unts, 54 forSLM, 455 for spot speed, 88 with video-base counts, 65 capacity, 479-480 definicion of, 10

charts, 27c area, 544, 544f Row diagrams, 548, 548f high-low graphs, 543, 544f organization, 547, 547f picrograms, 545, 545f project progress, 549, 549f scatistical maps, 546, 546f cypcs of. 543-549 checkers data collection and, 264, 272 definition of, 10 lice~-placc marching by, 334 for public uanspomtion, 280 surveys by, 284

capacity limitations, principle of, 27 i· I

carbon dioxide (C02), 457 car-following (CF), 225-226 algorithm for, 10, 202 definition of. 10 sensicivity analysis and, 206

chcckiiscs bikeability, 252 for graphics, 550 for prescntacions, 558-559 quescions, 506 for transportation studies, 6t walkability, 252

Carload Rail Waybill Sample, 294t chlorolluorooubons (CFCs), 457 causal chaio for'coUisions, 356, 357, 372 definition of. 10 CBD. &( ccncnl business district cellular phone observ:~.cion method probe vehicle and, 174 for TID, 172 centerline markings accderomctcrt and, 139 ball-bank indicator ~d, 139 definicion of, I0 middle ordimte and, 15 rai.scd reflective pavement marker and, 139 sensors and, 66 cencnl business district (CBD) definition of, 10 parking for,-325 trucking and, 300

TTDand, 161

chord for curve radii, 408-410 definition of, I 0 CL S~t conlidence interval classificaciorts a.rea. 419-420 for areas, 419-420 for bicycles, 419 counts, 54 ddinicion of, 10 of inventories, 316 for mobility and aecess, 43 5t for pedesuians, 419 for road sumces, 41St of streets, 312f. 418£,419,435, 436f of vehicles, 185-186, 185f

Clean /Ur Act attainmem ue~ and, 9 construction impacts from, 459 clearance.interval compliance wirh, 151 ddinirion of, I 0

~

r.

;. ,;

clearance lose time, I OS, 1 09

(

clear sight triangles, 113

..

clipart, 27t

{

closed questions, 506 cluster sampling, 502-504 COr Set carbon dioxide

t/ (.

coding and reduction errors, 513 codliciem of variation (CV), 400, 401 collectors, 419 collisions, 119, 347-379 MDT and, 355 AASHTO and, 348 at approaches, 369 causal chain for, 356,357, 372 causes of, 357 countermeasures for, 369-379 data analysi.s for, 357-379 dara collection for, 348-357 data reduction for, 354-355 EB and, 368 engineering judgment and, 355, 365,426 erroncow data for, 356-357 frequency of, 359 involvtmena and, 358 lightingand, 420-423 maps and, 354 . MLand, 191 N/D for, 424-425 n.etwork screening for, 359-369 night percentage for, 424 numbcts and trends with, 358 person injury in, 351-352 random narure of, 356 rates of, 359--360 RQCand,364 spot maps for, 359 SWP and, 365-368, 366t,.367t types of, 351, 35h .

' I

·' ~

i

.

;

,.

unreported, 356 volume and, 354

.

~

':~ .

'

;:-,

:;I -.

collision diagrams, 369-371, 370f dclinit.ion of. I 0 TCD and, 1 ~ 7 collision rare for section (RSEC), 359-360

·.

·...I

collision rare for spoc (RSP), 360

- .. i

collision races, 11 9

·~

.

~I

· ·: i ·I

;I

~~I

. :[

. '

:i

!

:l

:_..:

!· ? • ••

!

~i

-..!i ·:.~

·• !l

:i

collision reporcs, 348- 352, 349f- 350f definition of, 10 lighting :>nd, 42 I RSA wd, 384 TCD wd; 144 collisions road surfaces and, 370 SPF and, 368 collision severicy definition of. I 5 EPDO wd, 360-362 I
;. j ;

··'

' ..L j,

~

i! ·' .;:

comparison in before-and-after rest, 494 in experiments; 487-438 history and, 494 maturation and, 494 regression to the mean and, 494 units and, 494 compatibilicy, principle of, 27

compliance. Su al.!o TCD comp!iance bicycles and, 252 with clear;mce interval, 151 with crossw-•lk1, 252 definicion of, I 0 freeways and, I 86- 187 with HOV lane, 186 with no·left-curn, 150, 150f with pavement marking, 144 pedestrians and, 144, 154- 155. 154(, 252 with IITOR, 152-1 53, 152f with school crossings, 144 with STOP signs, 149, 149£, 155-156 with 'fCD, 143-156, I46t, 186-187 with traffic signals, 151, l 5Jf composition of graphics, S41-542 of vehicle cypes, 219 computer-aided design (CAD), 540 definition of. 10 for signs, 315f concurrent flow HOY lane, 10, 180 condition diagrams, 119,371- 372, 371f definition o(, II for intersections, 313f

plrking and, 325, 331 r
....

-·

controls, lx:fore-and-afte'r-test with, 492-494,493f oontrolcounu,54-57,243 MDT and, 57 ADT.and, 55-56 daily factors in, 55-56, 56r DDHVand,57 definition of, II growth factors and, 55 pe:U: hour and, 57 sample3 and, 69 seasonal factors in, SS-56, 56c control delay, 98 at approaches, 222 definition of. 12 6eld procedure3 for, 100-102 HCM wd, 98, 100 LOS and, 102 queue and, 100

confidence interval (CI), 147, 376, 529

control poinu, on freeways, 161

confidence level, 364, 399t, 486 definition of, II speed and, 83, 83t for spor speed, 92, 92t with TCD compliwce, 148t, ISS

convenience sampling, 504

conflict. Su also traffic conflicts bicycle3 and, 251- 252 definicion of, 11 at intersections, 391 f-394f multiple threat and, 252 pedestrians wd, 251- 252 vehicles and, 251- 252 congestion gra.phs for, 32 HOY lane wd, 179


control station, 69, 243

coordinated signal system, warrants

for, 131 copyright norice, on written reports, 551 cordon count, 50-53 accumulation computations for, 53r ADTand,50 definition of. 11 elW!lple of, 53f O·D and, 50 summary sh~t for, 51f-52f cordon line, 431,438,449

cosine error

crosswalks complil'l~e

wirh, 252 definition of. 12 pavemera marking at, 388 pedestrians and, 239f QOSand, 253 unsignalized inrersecrions and, 252 volume and, 238

·definitioH or, 11

for lasers, 80, Sit for radar. 80, 81 r, 81 t cost-effectiveness. See benefic-cost racio counrs. See specific count types count bo:l.l'ds. See handheld count boards

cumulative frequency diagrams, 519-520

countermeasures benefic-cost rario for, 374t, 377- 379, 37Bt for collisions, 369- 3 79 definition of, 11 e\•aluarion of, 379

cumulative frequency distribution, 518 of spot speeds, 520r

count expmsion, 70t, 238 -lefinicion of. II sampl~s and, 69 coverage counts AADTand, 57 definicion of, ll sarnpl~s and, 69

curb parking, 326-327 map for, 328f

Current Industrial Reportr, 297 curve radii, 190, 406 chord for, 408-410 middle ordinate for, 408-410 CV. See coefficient of variation

cycle length definition of, 12 for signali:z.ed intersections, 44 simulation and, 202 TID and, 99

crash, 119. See also collisions definition of, 11 warrancs for, 131 crash frequency definition of, 11 SPF and, 18, 368 crash race, 364 definition of, II elderly and, 348 nonmotoril:ed, 237 regional, 35) crash reduction factor (CRF), 11, 375- 376 crash severity, 11 , 360 CRF. See crash reduction facror crime data collection and, 8 lighcing and, 426 critical gap, 247-249 definition of, 11 estimation of, 111-112 gap acceptance and, 226 for pedestrians, 247, 249

D daily factors, in control councs, 55- 56, 56t daca collection for bicycles, 253-256 checkers and, 264, 272 for collisions, 348- 357 crime and, 8 forms, 564 for freeways, 191-195 individual vehicle selection method for, 79-84 for inventories, 318-320 for lighting, 415-424 for multi-use paths, 241 f for parking, 332-339 for pedestrians, 253-256 pitfalls o( 8 point data, 221-222 for public transportation, 272-285, 273f

for RSA, .185-388 s:lfec-f with, 7- 8 s~ga:ents, 222 for sirnulati.-.:>, 218-228 for sp<>c speed, 84-89 for surveys, 500-50 l forTCD compliance, 146-155 wich resr vehicles, 161-165 for traffic conAiccs, 394-405 training for, 5 for cransponarion planning, 438-448 for tra!lsponarion studies, 4 forTTD, 161-165 for volumes, 58-66 data display types and purposes, 27t data mining, 12, 40 data reduction, 516-522 for advisozy speed, 409-412 for automatic counts, 68 foe collisions, 354-35) for freeways, 195-198 fo r lighting, 424-426 from manual counts, 67 for parlting, 339-340 from peak hour, 67c, 68f for RSA, 388-389 for simulacion, 229-233 for spoc speed, 88-93, 89t for traffic conBiccs, 405-406 foe rransportacion planning, 448-450 for1TD, 165-166

i :

:l

.I

· '!

i

!

·;

dB. See decibels DDHY. See directional design hourly volume decibels (dB), 452-453 default parameters and distributions, 205 delay. See also specific delay types ac approaches, 210 definicion of. 12 intersections and, 98- 104 pedesrrians and, 249 public transportation and, 279f TCDand, 123

'

)

j.

I ' I

·~ · ·;!

ft

!! ·l

..,,

;~ :1 ·~

·u p l!

.f:r.

delay per person, 197 Index • 613

·~

ddinearor, 12, 370, 374r

detector occupancy, 184

definition of, 12

demand

diagrams, 27c

density. Su also spttd-flow-densicy

:rr.-~··:1>-;-

i:'Zi:.:~~

departure sight rriangle,l13 STOP signs and, 114.

departure volumes, intersection counts and, 45, 45f descriptive star~scics, 522-528 population and, 522 samples and, 522 for spot speed, 91- 92

design of experiments, 485-496 factorial, 494-496 of ~phics, 25-34, 540-550 for interchanges, 203 principles of. 26 of surveys, 499-514 of tables, 542-543

~i

:i

il,. ll ll

I~

I I

i.

design equation method advisory speed and, 410, 41 Or design speed and, 410

design spttd advisory speed and, 406 definicion of. 12 design equation method and, 410 on freeways, 178 stopping sight
Duktop Rrfirmu for Crash

&dut:tion Factors (FHWA), 375

ti !' I

I

:•

collision, 10, 137,369-371, 370f condition, II, 119, 313f. 371-372, 37lf C\J mulativc frequency, 519-520 Bow, 71, 72f. 548, 548f frequency, 519 for intersection Aow, 71, 72f site, 372, 372t

dilemma zone

··1·

.: Department of Mo\or Y~hicles (DMV), 353 .: . .

I.

definicion of, 12 laser and, 79-82 radar and, 79-82

dererminisric, 205

definition of, 12 rraffic, 219 transportation, 463, 481 volumes, 223-224 aerial surveys for, 187 definition of. 12 detector occupancy and, 187 on freeways, 187, I 88f LOS and, 187, 188f occupancy and, 187. . .,... UAVs for, 187 ! \ ~~~

Doppler effect

density and, 187

definition of. 1i lighting and, 370

directional design hourly volume {DDHV) control counts and, 57 definition of. I2

double-barrel questions, 507 double negatives, 507 driver behavior, 273 ITS and, 223 RSA and, 384 simulation and, 210-21 2, 225-2;2.7-·' · ;r·

dr:iveways, intersections and, 97-1 15 DTA Su dynamic traffic assignment dwelling-unit inrervic:w3, 447 dynamic traffic assignment {DTA) algorithm, 211 4~finition of, 13 simularion and, 211

direct measurements, of speed, 79 disclaimers, on written reports, 551 discrirninabiliry. principle of. 26

E

dis ranee measuring instrument

EADT. Su estimated average daily

(DMI), 12, 162, 165

traffic

disuibucion

Easter Seals Project AC110N, 264

cumulative frequency distribution, 518, 520t default, 205 definition of, 12 frequency disuibucion uble, 90t, 517-520, 517t Poisson-distribution, 491, 492, 492f relative frequency
diverging diamond interchanges, 49 DMI. &e distance measuring instrument DMV. &e Department of Motor Vehicles Domestic Waterborne Commerce, 294t

EB. See Empirical Bayes edge line markings, 13, 372t eight-hour vehicular volume warrant, 124-125, 125t ElR &t environmental impact report

EIS. Set environmental impact statement dderly, crash race and, 348 emergency scenario modeLs, simulation and, 212-2 14, 213f emissions, 459 from RR. 292 from truck.!, 292 from vehicles, 217-218, 217f

Empirical Bayes {EB), 35~ coUi.sions and.• 368 HSM and, 368

~ngineering judgment

\ collilion5 and, 355. 365, 426 1 definition of, 13 •. growth factors and, 55 STOP 5igns and, 132 warrants and, 21, I23 YIELD ligos and, 133 engineering plans, 32-34, 33f engineering study definicion of, 13 speed limit sign warn.na and, 134 sratutical analysis and, 515 TCDand, 135 written reports and; 513-554 environment, 451-460 simubtioo and, 217-218, 2!7f

Evaluating lntnucticn improvnnmt.t: An Enginuring Study Guidt(NCHRP), 208 Excel,536

expert sampling, 504-505 ,

environmental impact statement (EIS), 452, 464

expressways, 419

Environmental. Protection Agency (EPA), 9, 16, 459 atuinment area and, 9

figures, in written reports, 5 52

experiments before-and-after rest in, 489-494 comparison in, 487-488 design of, 485-496 inferential statistics wim, 486-487 paired comparison in, 488, 488t random assignment in, 487 unpllied comparisons in, 487

express toll lane (El), 177 definition of, 13 MLand, 179-180

~lation method, for lTD,

172

F

EPDO. &e equivalent property damage only

factorial design, 494-496 ANOVA and, 494-495, 495c means te5t for, 495t random assignment in, 496, 496t

s~~ evacuation

response curve

estimated average daily traffic (EAD1),119 estimation, in inferential statistics, 528-529 ET. &e express toU lane evacuation freeways and, I90 simulation and, 212-2 I4 evacuation demand zones, 213 evacuation response curve: (ERC), 13, 213, 214f .

Floating-urTc:chnique, 161 flow automatic couna for, 183 definition of. 13 freeways and, ~ 82-183 HCMand, 182 speed and, 183 vph and, 182

flow rate. Su also satur2Iion Bow rate ddinition of, 13 peak hour and, 57 FMCSA. s~~ Federal Motor Carrier Safety Administration

FABCO, 351-352

ERC.

flashing, 13, 492, 520

flow di.agrarns, 548, 548f for imecscecions, 71, 72f

EPA.. Su Environmental Protection Agency

equivalent property damage only (EPDO) eomsion severicy and, 3~362 ddinicion of. H

FHWA. Su Federal Highway Administration field reviews, 388

executive summary, in written reports, 552

environmental impact report (EIR); 464

FFS. Su fm~-flow speed

facton, 486

focal points, in graphics, 540-542, 541 f foorways, 419 forced stop, 149 Foreign Waterborne: Commera, 294t

40 visualization, 24, 39-40, 40f

FARS. s~~ Fatality Analysis Reporting System

four-hour vehicular volwne warrant, 125-127, 126t

Federo.l Highway Administration (FHWA), 13, 143, 192,204, 205t, 353,375,384,456 on freeway work zone5, 190 noise and, 453 shared-usc pam and, 253 Federal Motar Cartier Safety Administration (FMCSA), 353 Federal Transit Administration (FTA), 13

l ]

I

I

I I I

I

foceword, on written reports, 551

FAF. &~ Freiglu Analysis Framework

Fatality Analysis Reporting System (FARS), 353

I

;I

fractional factorials, 496, 497t frames, 501-502 free-flow speed (FFS), 99 ISO and, 112 lTDaod, 104 freeways, 177-198, 178f. 4 19 AASHTO and, 178 ac:celeracion/decdcration delay on, 191 automatic counts and, 19.5~ . compliance and, 186-:187

I .I

-

control poinrs on, 161 data collection for, 191- 195 data reduceion and analysis for, 195-198 definition of, I 3 deruiryon, 187, IS8f des;gn speed on, 178 evacuation and, 190 Aow and, 182-183 gaps and, 186 GP on, 178 HCM and, 178, 184£ headway and, 186 incidents and, J90 occupancy and, 184-1 8.5 performance measu res for, 198t queues and, 189 queue length on, I 89 safety and, I 90 segment srudies for, 187-190, 195- 196 SMSand,l83 speed-Row-densi ty on, 184f, 191 spot srudies for, 195- 196 system monitoring for, 197 TID on, 188-189 vehicles on, 185-186, 185£ volume on, 196£ weavilig segmenu and, 183 work zonc:5 on, 190

handheld count board< for, I 23 in- road seMors for. 186 measurement of, 249- 250 pedesrrians and, 246-250, 248f percentile and, I I 0 rejected, 17 school crossingsand, 128-129, l30r TCD and, l23 time, I 14-115 unsignaliud incersecrions and, 123

General Estimates System (GES), 353 general purpose lanes (GP) concurrent Row HOV lane and, 10 definition of. 13 on freew2ys, 178 HOV and, 179 HOV lane and, 180 genec:~.tion stud ies,

of parking.

330-332

frequency, of noise, 453 freq uency diagrams, 51 9

gcom~ tric dday, 98

definition of, 12 field procedures for, l 02- I 03

frequency distribution table, ) 17-520, 517t for spor speed, 90t

GES. Su General Estimates Sysr~m

FTA Su Federal Transit

GHG. Su gr~enhous~ gases

full scop, 149

GIS. See geographical information

gaps, 109- 112. Su also accepted gap; critical gap bins for, 110 definition of. 13 field pro~dures for, II 0 freewa)'$ and, 186

Graeco-Lari n square, 496

graphics -~ ch: ckliscs for, 550 communicalion with, 2S-34 com posilion of. 541-542 design of, 25- 34, 540-550 foc:t.l poinrs i.n. 540- 542, 541£ visual weight in, 541f

global positioning sysrem (GPS) AVL and, 281 for bicycles, 242 defin ition of, 14 for pedestrians, 242 probe vehicle wd, 174 for rest vehicles, 16 1- 162 for iTD, 161-162

616 o MANUAL OFTRANSPORTATION ENGINEERING STUDIES, 2ND EDITION

..--·

graphs, 27t, 3 l f, 32f, 516 bar, 27t, 28, 28f, 29, 29f, ? If line, 27t, 30- 31, 30f pie, 27t, 29, 29f seep, 27t gravity models, 475 Green Book. See Policy on Geometric Daign ofHighway1 and Struts greenhouse gases (GHG), 457 ground-based radio navigation, 173 grouped bar graph, 28, 28f

growth factors, 477-478 con trol counu and, 55 engineering judgmenc and, 55

Guitknu for lmpkmmtation of tlu AASHTO Srraugic Saftry Plan (NCHRP), 375

guide signs, 11 8, 134-135 definicion of. 14

sysr~ m

global calibration, 223

G

GPS. Su global position ing syscem

graphical display of daca, 24

(G IS), 40, 174 definicion of, 13 spuial distribution and, 522

Ad ministration

GP. s~c gene raJ pu rpose lanes

gap accepcance, 109- 112, 226227. Su also accepred gap algorithm for, 13, 222 erilical gap and, 226 daca analysis for, I I lt-1 Pt definition of, 13 field procedures for, 1 10-111

g~ogr:~.phical info rmacion system

Freighr Analysis Framework (FAF), 294t

gioss:~ rr. in written repo rts, 552

goods movcmenc, 29 1- 305 dm scrs, 294t- 295r urban problems wirh, 299t

H handheld counc boards, 59-60 for bicycles, 239 for gaps, 110, 123 for pedestri:ws, 239 for TCD compliance, I46

handheld speed measurement devices, 80f

hi~ wry

hardwa re-in-the-loop (H[L), 224

in before-and-after rcsc, q~~~. 49 I comparison and, 494 MOE md, 489

hau'rdous materials, 292, 304-305, 304t HCM. Sa Highway Cnpaciry Manual head ways, 109 definition of, 14 freewaysand, 186 in-road sensors for, 186 heavy vehicles, 185-186 HFCs. Su hydro fluorocarbons high-low graphs, 543, 544f h!gh occupancy and toll faciliry (HOT), 177 definition of, 14, 15 HOY and,180 ML and, 179- 180

HOT. See high occupancy and wli faciliry HOV: Su high occupancy vehicle HOY lane compliance with, I S6 concurrem Bow, 10, 180 congestion and, I 79 definition of. 14 GP ;md, 180 occupancy in, 179, 219 HPMS. Su Highway Performance and Moniroring System HS IS. See Highway Safery Informacion System

in-road sensors, 62, 63-64 for gaps, I 86 for headways, 186 sensors for. 64f for spot sp:c:d, 86-87, 87f Institute c.fTransportation Engineers (JTE), 13, 119,464 incelligem transportation systems (ITS) APCand, 281 AVLand, 281 definicion of. 14 driver behavior and, 223 simulation and, 210, 212,223

intermediate areas, 420

I

imerrarer reliabiliry, 13, 254

13, 32, 57. 385, 464 bicycles and, 240 control delay and, 98, 100 l!owand, 182 freeways and, 178, 184f "lmcn:eccion Control Delay Worksheet" from, 100, 101£ LOS and, 185, 240 PCEand, 185 service m=ure and, 227

impact score, 13, 270

intersections. Set also Michigan U-turn intersection; signalized intersections; unsignaliz.ed intersections approaches ro, 28, 210, 488 condirion diagram for, 313f conflict ar, 391 f.-394f conrinuous llow, 49 delay and, 98-104 driveways and, 97-115 flow diagrun for, 71, 72f sarurarion Aow rate for, 215 traffic conflica ar, 391, 39 1f- 394f. 398t

Highway Performance and Monitoring System (HP~S), 294r Highway Safery Information System (HSIS), 353

Highway Saftty Manual (HSM), 13 EB and,368 HIL. s~~ hardware-in-the-loop histogram, 518 for spot speed, 90f. 9lf. SI 8r

incidents definition of. 14 freeways and, 190 index, in wr;tten reports, 552 indirect measurements, of speed, 79 individual vehicle selection method, for d:ua collection, 79-84 inferential statistics, 92, 528-535 estimation in, 528-529 wirh experiments, 486-487 nonparameuic tests in, 533-535 proportions in, 531 run pies and, 528-53 I significance resting in, S32- 533 informative changes, principle of. 27 inputs, 205 calibration of. 223-227 simulation and, 228r

:r.

:'ji l; l

impact, 463-483

hypothetical questions, 508

"

;~

· ·I;

~ j~:

Highway Cnpaciry Mam1al (H CM),

hydrofiuorocarbons (HFCs), 457

~

f

interchanges definicion of, 14 design for, 203 diverging diamond, 49 expressways and, 419 overpasses and, 194 single-point urban, 49 weaving segments and, 192

high occupancy vehicle (HOY), 177. Su also HOV lane definition of. ! 0, 14, 15 excess capacil)" a'!d, I 80f GP and, 179 HOT and, 180 MLa.nd, 179-180 person occupancy of. 187

HSM. See Highway Safoty Manna/

~

-e'

j:r.,.

;~tl ' ,J.

:i~

,•:

.,, " !' '' 1

:

; ,~

:li :/r-1 -!r1 ,;,

internal studies, for transporrarion planning, 4-46-448

"Intersection Control Delay Worksheet," from HCM, 100, lOlf intersection counts, 44-49 arrival volumes and, 45, 45f bicycle and, 46 departure volumes a.nd, 45, 45f for Michigan U-rurns, 47at signalized inrersccrions, 44-45 Index • 617

· ~ .

~ H~i -! il j

i:J'j

lid :- ;f. ~

;,,i(iI ~-
.]f'i ~ ~. ·~

ifl I

-

:n:i

:~1,,.: H~~1 · ;I~

i iI ~

'f . -: fi ··~

'

!ri

; i t-1 ,ii.J : ,,1

;fj li I

I

?'-;,i

.1, ' l I

.j

~

t!

;: '0:

i> ~·.

i':

:· i

L 1-

f: f:.

i'

•'

~r~

~

h_:

::-3 r·.

:l :

-1 for superscreeu, 47 at unsignaliz.c:d intenections, 44 intersection sight distance (ISO), 112- 11 5 MSHTO and, 115 measurementS for, 113f STOP signsand, 114-115 TCD and, 135 time gaps and, 114-115 YIELD signs and, 114 interviews dwelling-unit, 447 for parking. 336--339, 338f su~and, SIO

training for, 510 for uanspon:acion planning. 438-440 forTTD,l71-172 at workplaces, 447--448 in-vehicle counting technology, 62 inventories, 309- 321 access ro, 317 classification of. 31 6 data collection for, 318--320 for land use, 432-434 oflighting, 415 location systems for, 316 maintenance of. 32~321 parking. 325-332 retrieval of. 317 ston.ge of. 3 17 ror uuupon2tionpbnrung. 432-437 updates for, 321 inverse sampling, 400

K KABCO, 351-352 collision severity and, 15

key counts, 55 L lag. 15, 110 land use, inventories for, 432-434 la.ne-changing, 225-226 algorithms for, 15, 225 definicion of, 15 traffic c.onftica and, 394 laptop computers, 60 for bicycles, 239 for gaps, 110 intersections and, 98 for pedeStrians, 239 for saturation How, 105 forTCD compliance, 146 for test vehicles, 162 la.sers calibration for, 85 cosine error for, 80, 8 II definition of, 15 Doppler effect and, 79-82 lSD and, 112 round off error for, 80 for spot speed, 84-86 Luin square, 496 learning elfect, 210

involvements, collisions and, 358

letter of transmittal, on written reportS, 551

lSD. Stt intersection sight distance

level, 15

ITE. Stt Institute ofTransportarion

lcvd of service (LOS), 32 bicycles and, 240 conrrol delay and, I 02 de6n.i cion of. 15 density and, 187, 188f HCM and, 185, 240 MLand, 198

Engineers iteration, 13, 229 ITS. Stt intelligent transportation systems

J jay-W21ki.ng, 13, 155, 253 judgfnentsampung, 504-505

license-place matching by cbcckers, 334 for parking. 332-336, 333f for path-based councs, 48

for traruporution plan'ning, 44 1-444, 442f, 445f forTTD, 171-172 lighting, 413-426 before-and-after tm for, 415 bencfit-
lights-on studies, 446 Likert scale, 507 line graphs, 27c, 30-31. 30f

Livabk StrtttJ, 482 load, 266t, 267t definition of, 15 public ~n.nsportarion and, 278 loading, trucking and, 299-302 local roadways, 419 location ~ems, for invemories, 316 long-range ua.nsporta.cion plans (LRJ"P), 431 LOS. Stt level of service lose time, 105-109 6dd procedures for, 108--109 LRTP. Su long-rwge transportation pla.ns

M macroscopic models, 202-203 definicion of, 15 for queue length, 105 magnitude, of noise, 452-453

Maintaining Traffic Sign 137

R~trort!foctivir:y,

major street, 419 definition of. 15 parking on, 331 peak hour wur:tnr and, 127 w.arnncs and, 132 managed lanes (ML), 177-198 coUisions and, 191 definition of, 15 LOS and, 198 mea.sures for, 198 types of. 179- 180 VMTand,179 Managed Mocorways, 179 manual counts for bicycles, 238·-242 dar:t reduction from, 67 for pedestrians, 238- 242 periods for, 69 . for public transportation, 273-281 for cest vehicles, 162-165 forTTD srudy, 161 for volume d2u coUcction, 58-62

ManWll on Uniform Traffic Control D(vices.(MUTCD), 16 pedc:suians and, 238 signs and, 132 TCD and, 117- 118, 144 wunnts and, 123, 124 manual speed tr.a~s, 85-86

Maximum-CarTcchniquc, 161

misery index, 197

mean, 91 arithmetic, 523--524 ccnrraltcndency and, 522 definition of. 15 MOE and, 492 public tr:tnsporcation and, 288 str:tci.fied r:tndom sampling and, 502

mitigation measures, 464

mems test, for &aorial design, 49Sc measures of dfectiveness (MOE), 486 for bicycles, 253-254 definition of, 15 history and, 489 maturation and, 489 mean and, 492 for pedestrians, 253--254 person occupancy and, 187 regression to the mean and, 490 TCQSM and, 266t- 277t

ML.

s~~ managed

lanes

MMIRE. s~e Model Minimu m Invcncory of Roadway Elements

medium-term planning. 43o' mesoscopic models, 15, 202 method of sampling. 528 metropolitan planning organizations (MPOs), 429, 457 Michigan U-turn intersection . counts for, 47 de.finirion of, 15 layout of, 48f

maps, 27c. &~also online mapping cools; spot maps for bicycles, 259f collisions and, 354 for curb parking. 328, 328f s~cisricd, 546, 546f with street classifications, 312f forTCD, 312f time-<:antour, 161, I 67f for ~raffic count, 71. 74f for traffic Bow, 71 , 73f

middle ordinate for radii, 408-410 definition of, 15

maritime cargo, 293

midlock, 50

marwacion in before-and-after test, 489, 491 comparison and, 494 MOE 211d, 489

minor sneet approaches of, 125 definicion of, 15 peak bour wunnt and., 127

:. •

: :~

modal splir, 476-477

'~~

mode, 91, 524

..

Model Minimum Inventory of Roadway Elements (MMIRE), 353

'

:,

models, 205. &(also spfcijic models ormodaty;m definicion of, 15 for simulation, 219-220 30,24 modified binomial test, for Poissondistribution, 492, 492f

1:

;• :. I

:i :i

.

:! ·' ';i "

:!

MOE. See measures of effectiveness

Motorin umplianu with Standard Traffic ControlDevim (FHWA), 143 MPOs. &t metropolitan planning organizations multiple panels, 27t multiple threat conBiet and, 252 definition of, I 5 mulriscage random sampling, 504

microscopic modd, 201, 107f algorithms for, 202 definition of. I 5

multi-use paths, data coUection for, 24lf

Microsoft Excel, 536

multiway stop control signs, warrants for, 133

cum

J

mobility averages, 197 classification for, 435r

median, 91,524 median U-curn. Su Michigan U-rum intersection

•

MOBILE6, 459

MUTCD. Su Manual on Uniform Traffic Control Devica

,,I

r: :

j:

L

.'

·!l

..; !~ I

'. !i

;! i

..

I

!

.

I I , I'

'i

N NAICS. See North American Industry Classification System

:i i

.i

~· .

·'

..,'•

.,

,.

.

National Highway Planning Network (NHPN), 294 r

~~

~~tiona!

:I

•I

~.

I! ~

~. ~

f

r

'~

§ •j

.

.•.

ji

l !Iil !i

~

'I

...Ji1, •·. r. ji

\

I- ,;nway Traffic Safery Administration (NHTSA), 353 National Transit Database (NTD}, 264 National Transportation Atlas Dacabase (NTAD), 294t National Transportation Statistics, 295t NCHRP. See National Cooperative High\vay Research Prografu ·

N!D. s~~ ratio of rates near collisions, 215 nearly stopped, 149

I!

nested designs, 496

n

networks, 205 bicycles wd, 252-253 definicion of. 16 mosuresof,223 pedestrians and, 252-253 $Creening of, 359-369 $imulacion and, 220f warnuus for, 132

ll

'

frequency of, 453 rn~gnirude of. 452-453 during pe:.k hour, 455 pollution from, 292 prcdicrions about, 456 samples for, 455r temporal distribution of, 453 drne variance of, 453

N3rional Cooperative Highway Research Program (NCHRP), 16, 2
. j.'I ,.J! ,:

., il

r

11

l,;

'I! !! ~

·1:

·!l t:. ti

:i

night percentage, for collisions, 424

'Ji ~

!

,,'i ·I

online mapping tOols, _(iJ,.J62 definition of. 16 / for parlcing, 326

noncoverage, 514

on-sire circulation, 480

nonparametric rests, in inferential statistics, 533- 535

open questions, 506

nonrandom sampling, 504-505 convenience, 504 expert, 504-505 judgment, 504-505 quo~:a, 505 snowbaU, 505 nonresponse, in surveys, 513-514 nonsite traffic forecasts, 477-478, 477t North American Industry Classification System (NAlCS), 297

not-at-home, 514

NHTSA. s~~ National Highway Traffic Safety Administration

node-aggregared data, 223 noise, 452-457 abatement criteria for, 454r COntOUrs, 456, 457f determination of existing levels, 454-455 FHWA and, 4.53

one-time solution, 210, 212

nonattainment area air quality and, 457 definicion of. 16

NGSIM. Su Next Generation Microsirnulacion

i•'

0 -D. Su origin-destination

one-way analysis of variance (AN OVA), 487 faccorial design and, 494-495, 495r

North American Transportation Statistics, 29 5

,, 'I'

parking and, 330, 332, 340f

no-left-turn, compliance with, 150, 150f

Next Generati-J n Microsimulation (NGSJM), 204, 204f algoridun,204,204f

NHPN. See National Highway Planning Network

density and, 187 frcewoys and, 184- !85 in HOV lane, 179, 219

no stop, 149

NTAD. s~e National Transportation Atlas Database NTD. Su National Transit Database Nu. name seleaion technique, 504 null hypothesis, 486 number of approaches, 316

0 occupancy

definicion of, 16


operating speed algorithm for, 227 definicion of. 16 $Unulation and, 227 validation and, 227 opcimiz.ation algorithm, 16 delinition of, 16 simuh.cion and, 208 steady-stare, 210 oral presentations, 555 ordinal scale, 507 organiz.ation charts, 547, 547f origin-destination (0-D) cordon count and, 50 da.t2 c:xpa.ruion for, 443t definicion of. 16 field $heet for, 440f parking and, 326 path-based counts and, 47-48 peak hour and, 47 screen-line councr and, 54, 449f simuh.cion and, 212 survcys,437-438 TAZs and, 432f transportation planning and, 437, 448-449 ortho-recrilication dcfirtition of. 16 with video-base counts, 65

outputs, 205 simuLa!ion and, 228t

validation for, 227- 228 overpasses, interchanges and, 194

Jicense-pl•tc macching for, 48 O·D and. 47--48 samplc.s for, 47-48, 49c pavement markings, edge line, 13,

372t

p pace speed, 16, 91 paired cowparison, in experiments,

488, 488c paratransit, 16, 264 parking, 323-343 accumulation ~cudic:s of, 330-332, 332f block and curb face numbering system for, 326f forCBD, 325 congestion and. 325, 331 curb, 326-327 dara collection for, 332- 339 data reduction and analysis for, . 339-340 data tabulation form for, 341f-343f · duration summary shecc for, 335f generation srudies of, 330-332 interviews for, 336-339, 338f invenrories, 325-332 license-plate matching for, 332336,333£ on major street, 331 occupang•and, 330, 332,340f 0-D and, 326 online m
PCE. Su passenger-car equivalent

PDO. Su property damage only peak hour, 472t control counts and, 57 data rcduaion fcom, 67r, 68f ddinition of. 16 Row rare and, 57 noise duri'ng, 455 · 0 -D and, 47 TID and, 161 volume for, 473f warrants, 127, 127t

peak hour faeroe (PHF), 57 peak 15-minute Row rare, 57

Pedestrian Road Safity Audit Guieklin~.r and Prompt Lists (FHWA), 384 pc:dc:srrians, 237-259 aocderomecers for, 242 accc.ssibi!ity for, 238, 252 automatic countS for, 242-245 before-and-after test for, 25 5 behavior of. 250:..2.513 classification for, 419 compliance and, 144, 154-155, 154f, 252 conflia and, 251-252 i critical gap for, 247, 249 crosswalk and,. 239f data collection for, 253-256 definicion of. 17 delay and, 249 gapsand,246-250,248f GPS for, 242 handheld count boards for, 239 intersection counts and, 46 laptop compiuc:rs for, 239 lighting and, 4!7t, 426 manual counts for, 238-242 MOE for, 253-254 MUTCD and, 238 nerworks and, 252-253 platoon and, 228

RSA for, 387 r school crossingund, I 2S- 129, l3i t TCD and, 120-1 2 1. 252 time-lapse photog raphy for, 242 rraffic signals and, 154- 155. 154{ unsignaliu:d intersections and, 247- 249 vehicles and, 251-25 2 visuali1.acion for, 2 58, 258f volum~ and, 128, 238-245 walking spero by, 246 wall..-w2ysfor. 252, 419 warrants and, 128

pedometers, 242 percentile definicion of, 17 gaps and, 110 of speed, 520 for spot sp«ds, 523f percentile speed, 85 th, 84, 84t,

91-92, 520

.

definition of, 17 peak hour warrant and, 127 YIELD signs and, 133 percent of congested travel, 197 percent variation, 197 Perception, Identification, Emotion, and Volition (PIEV), 134 perceptual organization, principle of, 26 perflurocarbons (PFCs), 457 perfocmance measure, 17. Se~ also specific pcrformanu mtas11m permissive turn definicion of, 17 at signalized intersections, 45 permicted error, 83 pcrson-houcs per year, 197 person injury, in collisions, 351352 person occupancy, 184 ofHOV,187 MOEand, 187 PFCs. Se~ pcrflurocarbons

Index • 621

!

i ~

i

trip distribution, 475--476, 475f. 476f

I

I

trip generation, 473-474, 474t

1

Trip Gm"ation (ITE), 119, 464

i;

I

i !! I

i j

1 l ; g

~

'~ ,. ~

~

i

~ ~

I)

i

trucking, 292 CDD and, 300 loading and, 299-302 routes for, 298 surveys for, 448 unloading and, 299-302 weight and dimension !imics for, 297t, 302-304 WlM and, 303-304, 303t

croSswalks and, 252 gap and, 123 pedestrians and, 247- 249 right of way at, 398

vehicle miles mvded (VMT), 26, 192 MLand, 179

unwilling to answer, 514

vehicle observation srudy, 123 forTID. 171-174

Urban Transportation Planning Package, 450

vehicle owner mail questionnaires, 447

U.S-Canada border crossings, 294t

vehicle registrations, 446

U.S. Census Councy Business Patterns, 295r

vehicle signarure matching method •.. forTID, 172 ,.,; •••~

U.S. Department ofTransporruion (US DOT), 353

vehicle.~

truck only toll (TOT), 177 definition of, 15, 20 ML and, 179-180

US DOT. Su U.S. Dcpartmem of Transportation

TSM. Su transportation systems managemenc

user perception definition of, 21

U.S. Economic Census, 295t

<{05,252,253,25~256

TSP. S(t transit signal priority

per hour (vph), 119

video, 60-61, 65 at approachc, 65 for imenections, 98-99 for saturation Bow, I 05 signal phase and, 255 for speed decemiliutioo, 82 for speed craps, 85-86 with test vchiclc, 161

~

·;~ 'I

TIC. &t time· t
s,, travel-time dday

~

TID.

'! ).

r-eese, 488

'I

] • !

type I error, 486 type II errcir, 486

u UAVs. S" unmanned aerial vehicles i

l l

i 1

1 }

unable to answer, 514 units in before-and·ahcr test, 489, 493 comparison and, 494 dcfin.ition of. 21 replication and, 486 samp!ing,501 universal design, 252 .

I ~ •r,

l.Ii

!i· !I

I

Ul).)oading, trucking and, 299-302 unmanned aetial vehicles (UAVs), 172 for density, 187 unpaired comparisons, in experiments, 487 unsignaliz.ed intersections, 43

U.S.-Mexico border crossings, 294t U.S. Ports and Warerways Facilities Database, 294t

YIN. &t vehicle idcntific:uion number virrual detectors, 65 virrual earth, definicion of, 21

v validation, 205 definition of, 21 operating speed and, 217 for outputs, 217-228 variabilicy, 526-528 v/c. s(( volwne-to-capacicy ratio vehicles. Su also probe vehicle; test vehicles classification of. 18~186, 185f conflict and, 251-252 definition of, 21 emissions from, 217-218, 217f on freeways, 185-186, 185f heavy, 18~186 pedestrians and, 251-252 routes by, 219 types of. 219 vehicle identification number (VlN), 352 vehicle intercept method, 446

visual aids, in podium presencacions, 38 visual arl2lyrics, 21, 40 visualization withAVL, 282f for bicycles, 258 definition of. 21 4[),24,39-40,40f for pedestrians, 258, 258f posters and, 40 forpubllc~portation,280f

for simulation, 232-233, 233f 3[), 39-40 visual table.~, 27t visual weight, in graphics, 541 f VMT. Su vehicle miles traveled volume, 43-74 ADTand, 119 at apptoachc, 44, 47, 146t arrival,45,45f bicycles and, 238-245

col!isions and, 354 count expansion for, 69~ 70t count periods for, 69 crosswalks and, 238 data collection for, 58-66 data presentation for, 70-71 DDHV. 12 ddinition of, 21 demand, 223--224 eight-hour vehicular volume warrant, 124-125, 125t four-hour vehicular volume warrant, 125- 127, 126t on freeways, 196f lighting and, 423--424 for peak hour, 473f pedestrians and, 128, 238--245 s:~mples and, 69 simulation and, 227 TCD and, i 19-121 test vehicles and, 166-171, 169f, 170t time series distribution of, 521 t TTD and, l(i6-171, 169f. 170r volumes DDHV,57 deparrure, 45, 45f volume-to-capacity ratio (v/c), 102 vph. See vehicles per hour

four-hour vehicular volume, 125- 127, !26t major meet and, 132 for mulciway stop control signs, 133 MliTCD and, 123, 124 for ne[Work, 132 peak hour, 127, 127r pedestrian volume and, 128 for regulatory sign, 132-134 for school crossings, 128-129, 130t, 13lt for speed limit, 134 for STOP signs, 132- 133 forTCD, 120, 120f, 123 for traffic control signals, 124, 128 for YIELD signs, 133-134 waterborne commeroe, 294t weaving segments frcewayund, 183 interchanges and, 192 Web site design, 41 weigh-in-motion (WIM), 43, 173, 192, 193f definition of. 2 L scale technologies comparison, 303t trucking and, 303-304, 303t

WIM. See weigh-in-motion wireless technology method, for

TID, 172

w walkability checklists, 252, 257f walking speed, by pedestrians, 246 walkways, for pedestrians, 252, 419 warm up pedod, in before-andafter test, 491 warning signs, 118 definition of. 21 placemenr of, 134 warrants . for collision severity, 131 for coordinated signal system, 131 for crash, 131 definition of. 21 eight-hour vehicular volume, 124-125, 125t engineering judgment and, 21, 123

WOrk Zone Operations Best Practices Guideline; (FHWA), 190 . work z.ones, on freeways, 190 written reports, S50-S54. See also collision reporu on air quality, 460, 460f appendices in, 37 body of. 36-37, 552-554 communication with, 34-37 engineering srudy and, 513-554 exhibits for, 37 organization of, 550-552 sections of. 35 for s~:~cistical analysis, 536-537 target audience for, 35-36 writing style for, 35-36

y {

Yellow Book See MSHTO

Highway Saftt:y Design a11d Operatwns Guide YIELD signs, 44 design speed and, 114 engineering judgment and, 133 lSD and, 1i4 . percentile speed, 85th and, 133 right of way and, 133 TTD and,I03 war~tsfo~ 133-134

<

Manual of Transportation Engineering Studies

Recommend Documents