Norbert Delatte, P.E., Ph.D., F.ACI, F.ASCE Professor an and d Chair, Department of Civil an and d Environmental Engineering
“An unacceptable difference between expected and observed performance” – Leonards (1982) “Nonconformity with with design expectations” – Feld (1964) – if so, if so, there are many many failures failures
1
Engineers design Simplified two part definition of of design design
Figure out everything that can possibly possibly go go wrong wrong Make sure it doesn’t happen
Knowing how systems perform and fail – failure literacy
“The art and science of of molding molding Materials Materials we we do not fully understand, fully understand, into Shapes Shapes we we cannot precisely analyze, to resist Forces Forces we we cannot accurately accurately predict predict – in such a way way that that the society society at at large is given no reason to suspect the extent of of our our ignorance.” James Amrhein, James Amrhein, on structural engineering, cited by Carper (1989)
2
Forensic engineering is the application of of engineering engineering principles to the investigation of of failures failures or other performance problems. Forensic engineering also involves testimony testimony on on the findings of of these these investigations before a court of of law law or other judicial other judicial forum, forum, when when required. Failures are not all catastrophic, such as as when when a building or bridge collapses, but include facilities or parts of of facilities facilities that do not perform as intended by the owner, design professional, or constructor.
Develop failure literacy literacy –– identify key identify key failure failure mechanisms Develop theory, refine and calibrate models Theory should Theory should explain three cases
Factor of of safety safety (FS) (FS) > 1, stable configuration FS = 1, impending failure FS < 1, failure
3
Pisa tower Lower San Fernando Dam failure Carsington Dam failure Kettleman Hills W Hills Waste aste Landfill slope failure Desert View Desert View Drive Embankment failure La Conc Conchita hita slide Landslide Dam on the Saddle River (Alberta) Transcona Grain Elevator failure Buffalo Creek Disaster, Disaster, W West est Virginia, Virginia, and Vaiont Dam Landslide, Italy
4
The Tran ransco scona na and Fargo Grain Elevators, October 1913 and June and June 1955 The V The Vaiont aiont Dam Landslide, October 9, 1963 Hurricane Katrina Levee Failures, Failures, August August 29, 2005
October 1913 an and d Ju June ne 1955 Great Plains of North Am Amer eric icaa
5
Near Winnipeg, Manitoba Near Winnipeg, Construction started in 1911 Structure finished September 1913 Filling with Filling with grain began Starting October 18, 1913, about 88% full, settlement increased to about 1 foot Structure tilted about 27 degrees from from vertical vertical The elevator stayed intact and and was was jacked jacked back upright, although now 14 feet below grade
6
Top layer tan and gray gray slickenslided slickenslided clay clay,, inorganic high plasticity plasticity clay clay (CH) (CH) – unconfined compressive strength 1.13 tons per square foot (tsf) Lower layer layer weaker weaker gray gray sil silty ty cla clay y Plate bearing tests before construction indicated capacity of capacity of 4 45 tsf Total estimated pressure due to structure 3.3 tsf Wash borings taken immediately immediately after after failure determined layer thicknesses
7
qn
1 2
qu N e sN e
Investigated four decades after the fact by by Ralph Ralph B. Peck – landmark paper by by Peck Peck and Bryant, 1953 Used to to verify verify bearing bearing capacity capacity theory theory –– predicted capacity agreed capacity agreed with with presumed pressure on soil qn
1 2
N e 51
qu N e sN e N e 51
B
D 1 5 L 5 B
3.6 77 12 1 51 1 5.56 5 59.5 5 23.5 5 195 5 77 23.5
Calculated bearing capacity capacity 2.57 2.57 tsf Estimated stress at time of of failure failure 2.24 tsf Calculated factor of of safety safety of of 1.09 1.09 “The development of of soil soil mechanics after the Transcona failure eventually eventually provided provided a basis for computing the ultimate bearing capacity capacity of of soils. soils. It was subsequently subsequently realized, realized, therefore, that the Transcona failure served as a ‘fullscale’ check of of the the validity of validity of such such computations.” (Shepherd and Frost 1995, p. 5).
8
42 years after Trascona, but soil part of 42 years of the the same ancient Lake Lake Aggassiz Aggassiz deposits Fargo elevator elevator was was narrower and longer than Transcona structure Filling began began April April 1955 Readings taken starting May May 10, 10, 1955, showed considerable settlement Collapsed and broke apart apart June June 12, 1955
Date of Observ ation May 10 May 10 May 18 May 18 May 25 May 25 June 1 June 8
Settlement Readings, mm (in.) BM 1 0 (0.00) 15 (0.60) 49 (1.92) 125 (4.92) 241 (9.48)
BM 2
BM 3
0 0 (0.00) (0.00) 3 (0.12) 18 (0.72) 27 52 (1.08) (2.04) 119 137 (4.68) (5.40) 222 265 (8.75) (10.42)
BM 4
BM 5
BM 6
BM 7
0 (0.00) 24 (0.96) 58 (2.28) 140 (5.52) 277 (10.92)
0 (0.00) 30 (1.20) 73 (2.88) 150 (5.89) 293 (11.52)
0 (0.00) 40 (1.56) 89 (3.49) 152 (6.00) 308 (12.13)
0 (0.00) 37 (1.44) 76 (3.00) 150 (5.89) 287 (11.30)
9
10
View of of the the collapsed Fargo Grain Elevator after subsurface soil failure in 1955. (Photo from the Institute for Regional Studies, NDSU, Fargo).
Stratum A –– 3 5 ft/ 0.91.5 m black to gray Stratum A gray mottled mottled silty clay Stratum B – 8 ½11 ft/ 2.63.4 m gray gray and and tan sil silty ty cla clay y Stratum C – 2 6 ft/0.61.8 m layer of of interbedded interbedded sand, silt, and clay, clay, weak weakest est shear strength Stratum D – thick deposit of of dark dark gray gray clay clay with with occasional pebbles
11
Assumptions in bearing capacity capacity equations equations
The ratio D/B should be less than 2.5, The shear strength should be averaged over a depth of 2/3 B below the raft – in this case, a depth of of 10.6 10.6 m (34.6 ft), which ft), which includes all four strata, The foundation should be loaded concentrically, The soil considered should be entirely entirely cohesive, cohesive, and The shear strength strength within within the layers considered should not vary not vary by by more more than 50 % from average
Cohesive soil condition not satisfied in layer C
Calculated average shear strength 0.77 – 1.22 kips per square foot (ksf) Multiplying by by N Ne = 5.35, bearing capacity capacity 4.11 4.11 – 6.52 ksf Estimated FS = 0.80 to 1.37
12
“…classic example of of aa fullscale bearing capacity capacity failure. failure. Even the most unsophisticated testing program and computation would computation would have revealed that a net net working working pressure of of 55 ksf was courting failure. A A soil soil investigation limited to unconfined compression tests on untrimmed samples would samples would have been adequate. Using these test results, a net failure pressure of of 4.11 4.11 ksf would have been calculated. The minimum factor of of safety safety is is 1.5 so that a maximum working maximum working pressure of of 2.74 2.74 ksf would have been allowed… a simple plot of of load load versus versus settlement for any one of of the the bench marks marks would would have shown the elevator to be in imminent danger of of collapse. collapse. Prompt unloading would have saved it. Why Why this this data data was was not analyzed is a mystery.” (Nordlun (Nordlund d and Deere, 1970, p. 605).
These impromptu full scale field tests tests verified verified bearing capacity theory capacity theory Need to account for complexity complexity of of layered layered soil systems and actual foundation geometry Prompt unloading unloading would would have saved the Fargo elevator after the settlement settlement was was observed
13
Peck, R. B., and Bryant, F. G. (1953). “The Bearing Capacity Failure Capacity Failure of of the the Transcona Elevator,” Geotechnique , 3, 201–208. Nordlund, R. L, and Deere, D. U. (1970). “Collapse of Fargo Grain Elevator,” J. Soil Mech. Mech. and and Found. Found. Div. 96( 2), 585–607.
October 9, 1963 Vaiont Vaio nt Go Gorge rge,, Italy
14
Extensive system of of dams, dams, reservoirs, and hydroelectric powerhouses in Piave River V River Valley alley,, high in Italian Italian Alps Alps Thin concrete arch dam Dam held back 169 million cubic meters of water water June 1957, owner increased capacity capacity by by 30 30 %
15
March 22, 1959 landslide at nearby nearby Pontesei Pontesei Reservoir of 33 million cubic meters of of of rock rock killed one person Concern about stability stability of of sides sides of of reservoir reservoir Young geologist Edoardo Semenza found evidence of ancient landslide, uncement uncemented ed mylonitic zone Designers thought landslide unlikely
16
The 1.5 km (1 mile) zone of of unc uncement emented ed catac cataclasist lasistes es along the base of of the the left left wall wall of of the the valley valley,, along along with with solution cavities, sinkholes, and springs, Ancient landslide masses had filled the the valley valley,, and then had been cut into two by by the the new new V Vaiont aiont stream, The southern slope of of Mt. Mt. Toc had a “chair like” structure of of bedding bedding planes, dipping steeply steeply at at the top and more shallowly shallowly near near the base, and A fault A fault separating the in situ rock mass from the ancient landslide (Gen (Genevo evois is and Ghirotti 2005).
In 1960, owner began filling dam and monitoring earth movements Movements increased as dam filled Landslide of of 750,000 750,000 cubic meters into reservoir, on November 4, 1960 Owner thought movement could be controlled by lowering water lowering water Boreholes drilled to reduce reduce water water pressure
17
October 9, 1963 at about 10:41 pm, 270 million cubic meters of of rock rock fell into reservoir Landslide speed up to 25 m/s Wall of water went water went over the dam Destroyed town of of Longarone, Longarone, destroyed other hamlets and and villages villages 2,043 people killed
18
From http://www.uwsp.edu/geo/projects/geoweb/participants/Dutch/VTrips/Vaiont.HTM
From http://www.cnsm.csulb.edu/departments/geology/people/bperry/Mass%20W http://www.cnsm.csul b.edu/departments/geology/people/bperry/Mass%20Wasting/V asting/VaiontDam.htm aiontDam.htm
19
Scope of of investigation investigation
Whether the hydrogeological examination of of the the dam area was given proper consideration in planning and construction, and whether and whether the previous landslides in the area area were were taken seriously, Whether the dam’s testing testing was was still continuing at the time, The level of of the the reservoir in the 10 days prior to the disaster, and whether and whether safety safety recommendations recommendations for the level level were were followed, and Whether a previous landslide in the area a few days before the disaster should have have warranted warranted an evacuation order downstream, and and whether whether officials acted properly properly (Ross (Ross 1984 p. 133).
20
Blamed “bureaucratic inefficiency, muddled withholding of of alarming alarming information, and buck passing among top officials” 11 men charged charged with with crimes ranging from manslaughter to negligence Some served short short jail jail sentences
The creation of of the the lake basin, as as well well as the the variations variations in the level of of the the reservoir, The clay clay seam seam along the failure surface, The ancient landslide, The geological structure, Seismic action, and A confined A confined aquifer behind and below the failure surface (Se (Semen menza za and Ghirotti 2000).
21
“Engineers and geologists are now generally generally obliged obliged to examine the slopes of of proposed proposed reservoirs for the owners. Where unstable slopes are identified, their impact on the project must be explained. When the unidentified slides are large and the effects on the project could be significant, there is an obligation to explain explain why why such such slopes are different from and safer than the the V Vaiont aiont slopes… If If the the engineers cannot give a reasonably reasonably complete complete and consistent explanation of of the the V Vaiont aiont Slide, in terms of of the the currently available currently available methods of of stability stability analyses, analyses, then it is difficult to see how they they can can feel confident about their evaluation of of other other reservoir slopes. The disturbing aspect of previous of previous reviews of of the the V Vaiont aiont Slide is that there are gross inconsistencies inconsistencies when when the field data, slide behavior, and the results of of the the analyses are compared” (Hendron and Patton, 1985, pp. 1 – 2).
Frank Patton, a consulting engineering geologist, and his colleagues began investigating the slide in 1975 and visited the failure plane. They They found found a layer of of plastic, plastic, low strength clay clay (also (also known as fat clay) at the base of the slide, between 13 and 100 mm (1/2 and 4 inches) thick. Considered
The three dimensional shape of of the the slide surface, Actual laboratory laboratory shear shear strengths of of material material from the site, and Piezometric levels taking into account rainfall and reservoir levels (Hendron and Patton 1985, p. 2)
22
It had been observed that movements increased as the reservoir level increased, but the reservoir level also increased when increased when it rained. Therefore, two possible causes for the increased movement were higher higher water water pressures due to higher higher water water levels within the reservoir, or increased pressures pressures within within the mountain from rainfall against the fat clay. “The erroneous assumption assumption which which led to the conclusions… was that all other factors factors were were remaining constant and the reservoir level level was was the main main variable variable controlling the stability of stability of the the slide. In fact, rainfall rainfall was was significant and was not remaining constant” (Hendron and Patton 1985, p. 54).
Since the failure had occurred, it it was was necessary necessary for for stability analyses stability analyses to demonstrate a factor of of safety safety near near 1.0 under the failure conditions. It was It was also necessary necessary to to demonstrate factors of of safety safety near 1.0 at the times times when when significant movement movement was was observed, as as well well as somewhat greater for the periods where movements movements were were insignificant. The periods of of time time where where the factor of of safety safety should should be near one are the prehistoric landslide, the major movement of of October October 1960 1960 when when the cracks formed, and October 9, 1963.
23
24
Thin layer of weak weak clay clay –– provided a lubricant Higher water Higher water pressures behind slide plane – correlated with higher rainfall and higher reservoir levels Sinkholes in kar karsti sticc pla plain in allowed allowed water water to infiltrate Boreholes did not go deep enough Ancient landslide confirmed
Mt. Toc in local dialect means crazy crazy –– locals were locals were aware of of ground ground movements Necessary to Necessary to carry carry out out reservoir slope stability stability analyses analyses Low permeability permeability clay clay layer layer provides lubricant and holds in in water water pressure
25
The dam remains mostly undamaged, but unusable since the reservoir is full of of rock rock
Hendron, A. J., Hendron, A. J., and Patton, F. D. (1985). “The Vaiont Slide: A Slide: A Geotechnical Analysis Analysis Based Based on on New New Geological Observations Geological Observations of of the the Failure Surface, Vol. I, Main Text, Technical report GL–85–5, Department of the of the Army Army,, U.S. U.S. Army Army Corps Corps of of Engineers, Engineers, U.S. Army Engineer Army Engineer W Waterwa aterways ys Experiment Station, Vicksburg,, Mississippi, Vicksburg Mississippi, June. June. Wearne, Phillip (2000), Collapse: Collapse: When When Buildings Fall Down, TV TV Books, Books, L.L.C. (www.tvbooks.com), New Y New York, ork, N.Y.
26
Ap rill 5, 1987 Apri Schoharie Creek, New York State
Schoharie Creek Bridge collapsed after three decades of service, of service, April April 5, 1987 Near record flood Five vehicles Five vehicles fell into river, ten occupants died Over time, scour protection for bridge piers had been compromised
27
Five simply simply supported supported spans across river Spans supported by by concrete concrete pier frames Columns on lightly lightly reinforced reinforced plinth, on shallow spread footing Spread footing to be protected by by dry dry riprap riprap
Contract awarded February February 11, 11, 1953, bridge opened summer on 1954 100 year 100 year flood occurred the next next year year Bridge survived, but damage may may have have had bearing on later collapse Asbuilt plans showed that sheet piling had been left around piers, but it had been removed
28
Plinths formed cracks shortly shortly after after completion, 1/8 to 3/16 inch inch wide wide High tensile stress in top of of plinth plinth – similar to uniformly loaded uniformly loaded beam upside down Tension reinforcement added to top of of pier, pier, but not extended into columns Other problems noted
112.5' 27.75'
57'
27.75'
Symmetrical about C L
Deck
Stringer at 8'-6" o.c.
Floor Beam at approx.20' o.c.
Knee Brace
Main Girder
Cantilever Floor Beam Ends
Bearing 7'-0" sq Column 5'-0" wide X10'-0" deep Tie Beam
Column
Plinth Reinforcement
Plinth
Footing Figure 1- Pier Section ( after "Collapse," 1987 )
29
Column Forces Plinth Tensile Reinforcement
Soil Pressure
Sloped Embankment
NORTH Riprap r ap Sloped Embankment East Abutment
West Abutment Flow PIER 1
100' SPAN1
PIER 2
110' SPAN2
PIER 4
PIER 3
120' SPAN3
110' SPAN4
100' SPAN5
Figure 2 - Schemati atic c plan of bridg bridge ( after after "Coll ollaps apse," 1987 )
30
April 5, 1987, 6 inch rainfall plus snow melt, estimated 50 50 year year flood Pier three toppled, and spans three and four fell into creek One car and semi on bridge, three drove off off into into gap – 9 bodies found, one never recovered Pier two and span two fell 90 minutes later
Photo by Sid Brown of the Schenectady Gazette
31
32
Two teams investigated and cooperated
Wiss, Janney,, Elstner (WJE) Wiss, Janney (WJE) Associates, Associates, for New New Y York ork State Thruway Authority Authority Thornton Tomasetti, P.C., for New New Y York ork State Disaster Preparedness Commission
Cofferdam constructed around failed piers, site dewatered and excavated Scour identified as cause
33
Shallow footings Erodible soil under footings – layers of of gravel, gravel, sand, silt, interbedded interbedded with with till Insufficient backfill Riprap protection, inspection, maintenance inadequate
34
After 1955 flood, berms constructed upstream – would increase velocity increase velocity Upstream end of of pier pier 3 fell into 9 foot deep hole, 25 to 30 feet undermined Sheet piles had not been left in place Riprap too light – was about 300 lb., should have been 1,000 to 1,500 lb.
Levy and Salvadori, 1992 Levy and “in the contract issued in 1980 for maintenance maintenance work, work, all reference to new stone riprap had been deleted by by aa nonengineer state employee employee who who decided, after viewing the site from shore, that it it was was unnecessary.”
35
Flood greater than anticipated by by designers, designers, riprap already disturbed already disturbed Curve upstream direct direct water water to pier 3 Debris directed directed water water to base of of pier pier 3 Berms directed directed water water under bridge Embankment increased increased velocities velocities Dam upstream set for for winter winter conditions
Bridge bearings allowed spans to lift or slide Simple spans not redundant Concrete piers did not have enough steel for ductile frame action Plinth reinforcement kept plinth from settling gently, broke suddenly
36
3 scour mechanisms
Long term aggredation/degradation – long term change in level Contraction scour – Q = Av Av,, if A smaller, A smaller, v v increases increases Local scour at pier or abutment
Countermeasures include riprap, piles under piers, and cofferdams
Bridge design must consider structural, hydraulic, and geotechnical factors Piers were Piers were shallow, not protected by by riprap riprap Bridge inspections critical, including underwater If riprap If riprap moves, replace replace with with bigger rocks
37
494 bridges failed between 1951 and 1988 due to scour Need to predict critical storm Need regular inspections of of superstructure, superstructure, substructure, and underwater features Erosion protection necessary necessary around around piers and abutments subject to scour
National Transportation Safety Safety Board Board (NTSB). (1988). “Collapse of of New New Y York ork Thruway Thruway (1 (190) Bridge over the Schoharie Creek, near near Amsterdam Amsterdam,, New Y New York, ork, April April 5, 1987.” Highway Accident Accident Report: NTSB/HAR 88/02, 88/02, W Washington, ashington, D.C.
38
ThorntonTom omas aset etti ti,, P. C. (1 (198 987) 7) “O “Ovverv ervie iew w Repo Re port rt In Inve vest stig igat atio ion n of th thee Ne New w Yor ork k St Stat atee Thruw Thr uway ay Sch Schoha oharie rie Cre Creek ek Bri Bridg dgee Co Colla llapse pse..” Prepared for: New York State Disaster Preparedness Commission, December. Wiss, Janney Wiss, Janney,, Elstner Elstner Associates Associates,, Inc., and Mueser Rutledge Consulting Engineers (1987) “Collapse of of Thruway Thruway Bridge Bridge at Schoharie Creek,” Final Report, Prepared for: New New Y York ork State Thruway Authority, Authority, November.
39
Aug us ustt 29 29,, 2005 New Orleans, Louisiana
Massive failure of of the the levee system Deaths and property property damage damage Near destruction of of aa major major American American city Series of of major major breaches
40
Failed at 6:30 am, 450 ft/137 m section of of II wall Failed when Failed when water water 5 ft/ 1.5 m below top of wall wall Flooded Lakeview neighborhood Wall built on organic marsh soil over a layer of of soft soft clay, shear strength low and and variable variable Waterfilled gap next to to wall wall
41
Water pushes pushes wall wall back Gap forms and fills fills with with water water,, presses presses wall wall Failure plane forms
Failed around 6 to 7 am, am, water water also below top of wall wall Soil below marsh here here was was sand, not clay Water seeping through highly highly permeable permeable sand lifted (floated) the levee
42
Failed around 7 or 8 am Levee rested on marsh over a thick sand layer, but looser and and weak weaker er than LA LA Canal Canal South levee Failure mode probably probably similar similar to LA LA Canal Canal South levee
Multiple breaches starting around 5 am Soil here here was was marsh over soft clay clay over over sand Probably sliding Probably sliding failures, exacerbated by water waterfilled gaps Some I walls overtopped – scour eroded soil from land side
43
Pumps were not part of Pumps were of hurricane hurricane protection system Few stations stations work worked, ed, many many operators operators had evacuated Electrical power failed Pump stations flooded Many pumps Many pumps discharged into canals and and waterwa waterways ys that had been breached – just recirculat recirculated ed the the water water
Lack of of appreciation appreciation of of risk risk Piecemeal construction of of hurricane hurricane protection system System under designed – standard project hurricane only 101 only 101110 mph/ 162177 kph Many levees Many levees not high enough No single entity entity in in charge No external peer review No stable funding
44
Understand risk and embrace safety Reevaluate and fix the hurricane protection system Revamp the management of of the the hurricane protection system Demand engineering quality
https://ipet.wes.army.mil/ Ignore the “security “security certificate” certificate” warning warning See also also American American Society Society of of Civil Civil Engineers. (ASCE). (2007). The The New New Orleans Hurricane Protection System: by Hurricane Hurricane What Went Wrong and and Why Why, Report by Katrina External Review Panel, Panel, ASCE, ASCE, Reston, Reston, V Va. a.
45
Thorough forensic investigations investigationsare are important to understand behavior of of engineered engineered systems Forensic observations are important to to verify verify theory theory and calibrate and adjust models Failures are often complex, so beware of of simple simple answers – “There is always an easy easy solution solution to every human problem — neat, plausible, and and wrong wrong..” H.L. Mencken
46
