August 1, 2007: The I-3spanned the Mississip

5W Bridge that pi River in Minneapolis

145.

tured under the weight of rush ious bridge renovations and

concentrated construction materials.

Underlying Issues: • Inadequate gusset plate design and insufficient review

process. • Added weight from renovations, traffic, and

construction materials. • Lack of attention to gusset plates in inspections and

load ratings. • Communication issues between the construction

contractor and the State.

collapsed, killing 13 and injuring

Proximate Cause: • Weak gusset plates frac

hour traffic, prev

SYSTEM FAILURE CASE STUDIES

SEPTEMBER 2009 VOLUME 3 ISSUE 06

A Half-Inch to Failure

At 6:05 pm, on Wednesday, AugustWest (I-35W) bridge over theMinneapolis collapsed. On the dathe bridge’s eight lanes were closed for planned construction. Four weak connector plates fractured under the combined burden of rush hour traffic, concentrated construction equipment, and previous heavy renovations. The bridge fell 108 feet into the Mississippi River. The police, fire department, and U.S. Coast Guard immediately initiated rescue operations. Of the 190 people on or near the bridge, thirteen died and 145 were injured.

BACKGROUND he I-35W bridge supported a 1,907 foot long, 8-lane

poln

bytorced

bu br

urally deficient proximately 12% of U.S. bridg

uef th

rais

(Fis artes

nnected three beams at each node: two diagonal beams and one vertical beam (Figure 2).

Over the course of forty years, the state of Minnesota conducted significant renovations on the bridge three times. In 1977, the State increased the bridge deck thickness about two inches. This renovation increased the dead load (weight of the structure itself) by 13.4%. In 1998, the State increased the dead load another 6.1% when it installed a median barrier. Together, these two renovations increased the weight of the bridge 19.5% over the original design.

renovations began in June 2007, two months ment of

ired a construction contractor, . (PCI), to resurface the bridge. moving two inches of concrete it with fresh concrete.

on equipment to mix In the weeks leading up to the collapse,

had poured concrete on seven he contractor placed heavy

dge ramps for five concrete some equipment on the bridge

1, 2007, the Interstate-35 Mississippi River in y of the collapse, four of

Figure 1: The I-35W Bridge

wide roadway that served Minneayears. The state inspected the bridge a

the National Bridge Inspection Standards set Highway Administration (FHWA). Inspeclabeling the bridge “structurally deficient” sinlabel indicated that the bridge requiremaintenance and repair to remain in service,was unsafe (inspectors would have closed thebelieved it was unsafe). A struct

T is for forty nually using the Federal s had been 1991. This

significant t not that it

before the collapse. The Minnesota DepartTransportation (MnDOT) hProgressive Contractors, IncThe renovations involved reon the roadway and replacing

Resurfacing required heavy constructiand pour concrete.

A third set of

idge if they rating is not es are rated

ntly labeled e 465 steel

the construction contractor bridge deck sections. Tconstruction equipment on bripours; one pour stationed

uncommon; apstructurally deficient.

Steel truss bridges, like I-35W, were more freqstructurally deficient; approximately 31% otruss bridges in the U.S. were listed as structuat the time of the collapse. Such bridges consbeams of steel formed into triangular units large steel truss bridges, the ends of the beamwith riveted metal plates called gusset plagusset plates co

lly deficient t of straight gure 1). In e connected . I-35W’s

2 | P a g e September 2009 System Failure Case Studies – A Half-Inch to Failure

deck and some on the ramps, and another positioned the equipment and materials on the bridge deck itself.

About a week before the

rin

about post-task cleanupinterpreted the inspectorpermission. However, thenot an engineer, and he permission. He was prescrew used the correct ma

and bfinished ing the

su

ntrn

e bridit for trucks (80,000 lbs.), so the contra

ridge. One cement tanker wughil

andveh. B con

ssetset of the nodes on the south side of the bridge

idge ct fe

oximy plunles. Policents.

pm, the Sherriff’s Department officially switched from rescue to recovery operations. Recovery continAugust 6, when the last victim was found. Of the 190 people on or near the bridge at collapse, 111 had minor injuries, 34 suffered serious injuries, and 13 did not survive.

PROXIMATE CAUSE According to the National Transportation Safety Board, the gusset plates at the nodes that failed were only half as thick as they should have been. In addition to being too thin, some of the gusset plates were bowed, a distortion that further decreased the bridge’s weight bearing capacity. The weak

als, and rush hour traffic finally stressed the bridge ses.

e cause of the collapse to the everal of the gusset plates, at fractured, were only half an

hick. The plates were not designed to support the bridge’s original weight, let alone the additional weight of the 1977 and 1998

ves that the original designers quality control to ensure the

set plates. Although igner’s original calculations before

t, it probably did not have the resources to fully verify the consultant’s work. This was not an unusual

collapse, the contractor was prepathe concrethe foremstate bridginspector and materialplaced on t e The forem

g for one of pours, and asked the

Observers immediately notified 9-1-1. Apprpeople nearby began rescue efforts; over thirtthe river to help those in submerged vehicminutes after the collapse, the Minneapolis followed by area Fire and Sherriff’s Departm

gusset plates fractured under the concentrated weight of the construction materials, the buincreased load from previodiagonal steel beam

te an e construction if equipment

an (worried turf

in toe conn

bo

s could be h

the rbridge.

al loading) concern as spector was grant such onstruction tract terms. tractor had th sides of

UNDERLYING ISSUESNo single element caused theseveral issues led to the failuroriginal gusset plate design was incorrdesigned, the plates are stronconnect. Inspectors, unawarthe gusset plates were unliketo

rather than struc’s expressed lack o bridge constructionlacked the authorityent to ensure that thterials and fulfilled cy August 1, the cotwo outside lanes on

the bridge, leaving the four inner lanes to be re

WHAT HAPPENED? On August 1, the day of the collapse, the coequipment to mix and pour the eighth sectioTwo cement tankers weighed more than th

Work proceeded resurfac

rfaced. ymateri

rden of rush hour traffic and the us bridge modifications. The

s within the failed nodes shifted to the west and fractured the gusset plates around the ends of the diagonals (Figure 3). Once the diagonal beams separated,

est of the truss fell.

I-35W bridge collapse. Rather,e of the weak gusset plates. The

ect. When correctly ger than the steel beams they

e of the design error, assumedly to fail and paid little attention

them during inspections. On the day of the collapse, the heav renovations, concentrated placement of construction

Figure 2: I-35WGusset Plate

actor set up of concrete. ge’s posted ctor did not eighed less t it on the

enough to reveal its weaknes

FAULTY DESIGN The Safety Board traces thoriginal bridge design. Sincluding the gusset plates thinch thick when they should have been a full inch t

legal limbring them onto the bthan the legal limit, and the contractor brobridge along with four piles of sand and four pa water tanker truck, a small excavator, propelled buggies. The material and concentrated in a section of two closed lanesall the equipment was set up, ready for thewhich was scheduled to begin at 7 pm.

The bridge collapsed at 6:05 pm, when the gu

es of gravel, four self-icles were y 2:30 pm,

renovations. The Board belieprobably did not exercise

crete pour, calculations were correct for these gusthe State reviewed the desthe bridge was buil

plates at a fractured.

ould not Once the gusset plates broke, the rest of the brsupport the extra weight; within seconds, iMississippi river.

ll into the

ately 100 ged into

Just five e arrived, At 7:27

ued until

Figure 3: Artist’s Rendering of a Fracturing Gusset Plate

3 | P a g e September 2009 System Failure Case Studies – A Half-Inch to Failure

Figure 4: I-35 him of any W Bridge on August 2, 2007, the day after the collapse

situation; states commonly lacked sufficient resources for thorough design review.

GUSSET PLATE BOWING AND INSPECTIONS The Safety Board could not determine when the gusset plates bowed. Photographs taken in 1999 show that they were bowed at least eight years prior to the collapse. One safety

bobutruc

ofete dodd

cause he had learned ed

er time. ied the

previous materials, Inspector

o zed the

lates, or s hazard. f these

e bridge’s , gusset ents that

ith little material

he foreman asked the r permission to

aterials on the bridge deck, er should r day-to-eer relied struction

supervisor to informproblems, but the project construction supervisor was not onsite on August 1. Afterwards, he said he was not sure if he would have voiced concern about putting heavy materials on the bridge. “My best guess is it would have been a 50-50 chance that I might have done something,” he said.

tractor had formally requested materials on the bridge deck, the state

model used to determine bridge load ratings. Model calculations, however, did not include gusset plates because engineers expected beams to fail prior to gusset plates. According to the state’s model, the bridge was capable of bearing the August 1 loads. The

ed the collapse.

Transportation Safety Board’s commends that bridge owners ver planned modifications may

significantly increase stresses.”

also increased design review nts for new bridge construction isting bridge inspections. a now requires that all bridges

by consultants undergo an independent design review.

pections in Minnesota were The State examined all 25

s bridges in the state and found ate problems in four bridges.

paired the gusset plates on bridges, closed the fourth

d accelerated its replacement ther states also re-inspected

l truss bridges, paying special gusset plates.

ABILITY TO NASA a number of parallels between NASA facilities. The I-35W dured forty years of use and

modifications. Many NASA are several decades old and

have been modified to house different ts.

king design error of thin gusset aped notice during its design he consequences of a design t be considered when allocating

resources and time to expert internal and ent design reviewers. If a had conducted a more review of the I-35W designs in s, they might have caught the

gusset plate design flaw and prevented the 2007 tragedy. The design process at NASA is subject to oversights in the

review process as well. Like state departments of transportation, NASA often hires design consultants and verifies their work internally or through third party review.

a If the construction conpermission to place might have referred to a technical

inspection engineer remembered noticing theplates during his inspection in the late 1990s, bowing occurred during the original constbridge. He did not report his observations plates because “our inspections are to find dfindings of deterioration on maintenance. Wedescribe construction or design problems.” Adid not think gusset plates were critical to bridge safety be

wed gusset thought the tion of the the gusset rioration or not note or itionally, he

unaware that the project enginehave been asked instead. Foday concerns, the project enginheavily on the project con

model would not have predict

AFTERMATH As a result of the National findings, the FHWA now reinspect gusset plates “whene

in college that gusset plates are designto support 2 to 3 times the expected bridge loads.

The gusset plates did not worry other bridge inspectors either. Inspectors’ reports since 1994 noted rust, corrosion, and section loss in the gusset plates, but did not measure changes ovOften, inspectors simply copgusset plate description from areport. Inspector training such as the FHWA’s Bridge Reference Manual, did not address gusset plates in steel truss bridges. Ntraining materials emphasiimportance of gusset pidentified distortion as a seriouThe cumulative effect ooversights was that, despite thstructurally deficient ratingplates were not one of the elemworried inspectors.

COMMUNICATION AND LOADRATINGS The state provided contractors wguidance on construction placement. Tconstruction inspector foplace m

MnDOT requiremeand exMinnesotdesigned

Bridge insrevamped. steel trusgusset plThe State rethree of thebridge anschedule. Otheir steeattention to

APPLICThere are I-35 and bridge enthree major facilities

projec

I-35’s lurplates escreview. Terror mus

independMinnesotthorough the 1960

4 | P a g e September 2009 System Failure Case Studies – A Half-Inch to Failure

reviewand access they nee

alwdedy bt

ere sound, butly s

rmined hazard idadt et hans a

tors’ cursory d not quantify

nd they often n, providing an

d and incomplete review. “You get whmto

ou the checklist.

anho structure. d to measure the weighty modified to for hazardous y. Conduct an agement best

practices to mitigate unexpected problems from modified infrastructure.

A final lesson from the bridge collapse comes from the communication issues between the State and the construction contractor. The State did not clearly communicate who could authorize the placement of equipment and materials on

the bridge. While it is not clear whether the authorities would have refused the contractor’s request to place heavy equipment on the bridge, the confusion contributed to the contractor’s belief that they had the state’s permission when, in fact, they did not. We should use clear lines of authority to communicate safety-critical information both internally and externally. Verify who is empowered to accept a risk

is needed.

REFERENCES ridge Types - Truss. In The Basic

2009 from http://www.matsuo-ics/truss.shtm.

portation. (2007). Aerial Photos ne Image. 2 Aug. 2007. Accessed ot.state.mn.us/i35wbridge/photos/

ortation Safety Board. (2008, November). Highway Accident Report: Collapse of I-35W Highway Bridge,

st 1, 2007 (NTSB/HAR-08/03). May 2009 from

.gov/publictn/2008/HAR0803.pdf

E CASE STUDIES

NASA project managers must ensure thatknowledge, experience, time

ers have the d to conduct

ays feasible sign. There

Matsuo Bridge Co., Ltd. (n.d). BBridge Types. Accessed 27 Maybridge.co.jp/english/bridges/bas

Minnesota Department of Transof I-35W Bridge Collapse. Onli27 May 2009 from http://www.daerial/aug-2/index.htm

National Transp

a thorough review.

When a structure is already in place, it is not to conduct a complete review of the original are other ways, however, that the I-35 tragebeen prevented. Inspectors did not recognizeplates as a serious safety concern. Hindsighfalse assumption that these gusset plates wprior to the collapse, inspectors were probabany such failure mode. Previous successebiased inspections and unde

could have owed gusset exposes the

Minneapolis, Minnesota, AuguWashington, D.C. Accessed 27http://www.ntsb

unaware of and failures entification

SYSTEM FAILUR

skills. The model used to determine bridge loalso biased by the assumption that gussestronger than beams; it did not include gusscalculations. At NASA, we must recognize cbaseline configuration and question assumptiosystems.

An object lesson is the bridge inspecobservation of the gusset plates. Inspectors digusset plate changes from year-to-year, amerely recycled the previous year’s descriptiooutdate

ratings was plates were plates in its nges to any bout aging

at you inspect” and a repetitive, narrow checklist lihazard discoveries. At NASA, we need encourage system and process knowledge in and reward active hazard identification beyond

Each I-35 bridge renovation presented opportunity to assess the current state of the w

its potential train and r inspectors,

excellent le

Safety inspectors should have been expectenew (and explain old) changes caused byrenovations. At NASA, where facilities areaccommodate new projects, we must check effects a change might have on an aging facilitimpact analysis and follow change man

before a time-critical decision

QUESTIONS FOR DISCUSSION • How do you ensure design reviewers have adequate

resources (funding, expertise, experience) to be thorough?

• Can you think of other examples of when something was incorrectly assumed to be safe?

How do you chang ctions as infrastructoo much on checkl

you employ odifying a facility?

w the c?

• e inspeDo your inspections rely t

ure ages? ists?

• What change management practices do when m

• How do personnel and contractors kno orrect authority to go to if they have a question

Executive Editor: Steve Lilley steve.k.lilley@nasa.gov Developed by ARES Corporation

This is an internal NASA safety awareness training document based on information available in the public domain. The findings, proximate causes, and contributing factors identified in this case study do not necessarily represent those of the Agency. Sections of this case study were derived from multiple sources listed under References. An proper use of source material is unintentional.

se studies online or to essage.

y misrepresentation or im Visit http://pbma.nasa.gov to read this and other casubscribe to the Monthly Safety e-M