2_Dynamic Stability Simulation of a Counter Balanced Lift Truck

Dynamic Stability Simulation of a Counter Balanced Lift Truck Truck Joshua Hogg, PE, M.Sc., Project Engineer at Hyster-Y Hyster-Yale ale Group Unrestricted © Siemens AG 2016

•

Realize innovation.

Dynamic Stability Simulation of a Counter Balanced Lift Truck Present Pre sented ed by: Josh Joshua ua Hogg

About Hyster-Yale Group

Hyster-Yale Group is a global manufacturer of lift trucks. Produces a full range of electric and internal combustion engine lift trucks under the Hyster and Yale name brands. Product ranges from small warehouse pallet movers to large shipyard container trucks.

EN16203:2014 EN16203:2014 [1] introduces new dynamic lateral stability test requirements for trucks with capacities of 5000kg or less being sold into the European market. Test consists of navigating a 90 turn within a specified lane at 90% of maximum travel speed. °

Failure is defined by lift-off of an inside rear tire (for conventional 4 wheel truck) New standard allows use of computer modeling given physical test validation is performed.

A: B: C: D: E: •

•

Accelerating Area Entrance Corridor Maneuvering Area Exit Corridor Deceleration Area Cross line 1 at 90% maximum speed Exit line 7 without lift off or crossing outline lanes.

w 2

≡ Exit corridor lane size

Anatomy of a Lift Truck

Anatomy of a Lift Truck Rigidly mounted axles; zero suspension. Rear steering with an Ackerman type steering linkage Large variation in front end equipment based on customer specifications. Lift Height Attachments Capacity

Developing the Dynamic Model

Developing The Dynamic Model

Model topology consists of a TRUCK mass body, rigidly attached to a MAST mass body. Primary function of the initial model has been to physically test a base truck, and use simulation to evaluate multiple mast options.

Articulating steering axle, and tires represented as discrete bodies Tires were modeled using the native “Complex Tire” element in LMS VL Motion.

An open differential was modeled explicitly using a kinematic assembly of revolute joints and relative constraints. Vehicle was propelled by a constant RPM applied to the input shaft of the differential.


Model topology is driven by parameters defined in a data table. Linked Spreadsheet tool houses inhouse tire stiffness and weight data for all tire options available. Lane size definition is controlled by parametric axis systems. Vehicle position is monitored via sensor axis systems defining the lane as well as at the vehicle drive and steer wheels.

Developing The Dynamic Model Expressions:

Model topology is driven by parameters defined in a data table. Linked Spreadsheet tool houses inhouse tire stiffness and weight data for all tire options available. Lane size definition is controlled by parametric axis systems. Vehicle position is monitored via sensor axis systems defining the lane as well as at the vehicle drive and steer wheels.

Lane_Boundary_Distance_DT_X=DIST(SAS2,SAS5,SAS2,”X”) Exit_Boundary_Distance_ST_Z=DIST(SAS3,SAS4,SAS3,”Z”) Lane_Boundary_Flag_DT_X=TEST(Lane_Boundary_Boolean_1,1,0)

Parameters: Lane_Boundary_Boolean_1=EVAL_EXPR(Lane_Boundary_Distance_DT_X)<0 AND EVAL_EXPR(Lane_Boundary_Distance_ST_Z)<0

Z

X

SAS1

X

Z

SAS2 SAS4 SAS5

Z

X Z

X SAS3


A Visual Basic script copies results plot images and model parameters to a scratch directory and kicks off a Python Program to Generate an MS Word Report


Quantifying Model Uncertainty

Quantifying Model Uncertainty Using simulation in conjunction with physical testing significantly reduces the total testing required. Acceptable?

However, relying on the simulation entirely is a logical progression. This requires a more detailed validation. For example ASME V&V.

Experiment

Simulation

F D P

Assessing model correlation requires understanding model uncertainty due to input variation. Measurement

Quantifying Model Uncertainty We have a parametric MBS Model

Model Input

Solve time is ~5 seconds Sweep simulation inputs, and evaluate response variance. 1.

Determine Standard Deviation (σ) for each simulation input quantity.

2.

Normally vary inputs about their mean value within their respective variance.

Standard Uncertainty [%]

Front track

0.52 %

Rear track

0.50 %

Drive tire width

2.6 %

Steer tire width

2.9 %

Truck mass (no mast)

0.14 %

Truck HCG (no mast)

0.30 %

3.

Vary exit corridor within a range large enough to capture a failure.

Truck VGG (no mast)

2.9 %

4.

Generate design table with 5000 configurations for a given truck.

Steer axle articulation stop gap

7.4 %

5.

Using solution manager, run configurations and output Steer Tire Normal Force, and Exit Corridor.

Speed

0.10 %

6.

Determine variance in Exit Corridor at zero Steer Tire Normal Force

Inertia roll

5.0 %

7.

Determine each input’s contribution to total input variance.

Inertia yaw

5.0 %

Inertia pitch

7.5 %

Quantifying Model Uncertainty Take exit corridor and minimum steer tire force output for each trial. Develop a linear model of response. Extract mean prediction and standard deviation at Steer Tire Force = 0. Repeat for multiple truck configurations. Develop conservative estimate for model variance based on input variation.

+1σ μ

-1σ

Preliminary Comparisons to Physical Testing


Measures for Comparison Lateral acceleration was chosen as a parameter for comparison, as it is theoretically proportional to overturning moment causing tip-over. Test failure occurs when the inside rear tire lifts off the ground This occurs when the reaction force on the inside rear tire goes to zero. Simulation confirms proportionality between lateral acceleration and steer tire normal force. Additionally, a test witness would observe whether lift-off occurred during each test trial.

 =

 =

 

 −  ∙  ∙  2 

Overturning Moment

Assumption:

 ∝ 


Experimental Setup Vehicle instrumented with motion capture sensor and accelerometer Test track is laid out with multiple target exit corridors Array of 8 motion captures cameras distributed around perimeter lanes Vehicle top speed set in software


w 2=7.368 m Physical Test: Simulation:

w 2=6.368 m No Lift-Off (3/3) No Lift-Off

w 2=5.368 m Physical Test: Simulation:

Physical Test: Simulation:

No Lift-Off (3/3) No Lift-Off

w 2=4.368 m No Lift-Off (3/3) No Lift-Off

Physical Test: Simulation:

Lift-Off (1/3) Lift-Off

Implementation

Implementation

_ _

_

Implementation

1 Physical Test

103 Simulated Variants

Next Steps

Next Steps Accurate Parametric Model can be used to extract critical statistical relationships from an otherwise stochastic event. Using LMS simulation to develop statistical predictive model to eliminate physical testing completely. Use simulation based statistical analysis as a development tool.

Thank you! Contact Information: Joshua.Hogg@hyster-yale.com Hyster-Yale Group

2_Dynamic Stability Simulation of a Counter Balanced Lift Truck

Recommend Documents