Analysis of Recent Bridge Failures in the United States

Kumalasari Wardhana1 and Fabian C. Hadipriono, P.E., F.ASCE2

Abstract: Over 500 failures of bridge structures in the United States between 1989 and 2000 were studied. The age of the failed bridges
ranged from 1 year 共during construction兲 to 157 years, with an average of 52.5 years. The most frequent causes of bridge failures were
attributed to floods and collisions. Flood and scour, with the major flood disaster in 1993, contributed to the frequency peak of bridge
failures 共almost 53% of all failures兲. Bridge overload and lateral impact forces from trucks, barges/ships, and trains constitute 20% of the
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

total bridge failures. Other frequent principal causes are design, detailing, construction, material, and maintenance. Comparison made
among three periods of similar studies 共1977–1981, 1982–1988, and 1989–2000兲 revealed almost similar trends, with most failures
occurring during the bridge’s service life. Also, human-induced external events occurred frequently in all three periods, but were most
dominant in the first and third periods. Technological advances in information systems have a great impact on data collection and analysis.
DOI: 10.1061/共ASCE兲0887-3828共2003兲17:3共144兲
CE Database subject headings: Bridge failure; Bridge inspection; Bridge maintenance; Collapse; United States.

Introduction database does not constitute a complete listing of all failures that
have occurred in the United States.
In its latest report, the National Bridge Inventory 共FHWA 2001兲 Despite a dearth of information regarding this important issue,
revealed that 691,060 bridges currently exist in the United States. studies have been conducted on the failures of constructed facili-
The Federal Highway Administration rated nearly 30% of these ties. For example, Eldukair and Ayyub 共1991兲 researched 604
bridges as substandard 共although as compared with the 1988 data, structural and construction failures in the United States from 1975
this figure was 12% lower兲. Despite this discouraging figure, de- to 1986. A more specific study about bridge failures in the United
tailed information on the number of U.S. bridges that have failed States was conducted by Harik et al. 共1990兲 for the period of
or were in a severe condition is not readily available elsewhere. 1951–1988. In addition, earlier failure analyses of constructed
The New York Department of Transportation 共NYDOT兲 is thus facilities were carried out by Hadipriono 共1985兲 and Hadipriono
far the only agency that is attempting to collect information and and Diaz 共1988兲 for the periods 1977–1981 and 1982–1988, re-
develop a database on bridge failure cases in the United States. spectively.
Following the tragic collapse of the Thruway Bridge over This paper continues the latter studies to investigate and ana-
Schoharie Creek in 1987, NYDOT took several steps to reduce lyze failures of bridges that have occurred in the past 12 years,
and prevent future bridge failures. One of those steps was the i.e., the period between 1989 and 2000. The information con-
creation of the Bridge Safety Assurance Unit, which began in tained in this paper is collected from the NYDOT database; from
1990. The initial undertaking in 1990 was to collect as much engineering journals and magazines 共Engineering News Record,
information as possible regarding bridge failures in the United Roads and Bridges, and Civil Engineering兲; from the home pages
States. This information was then used to create several bridge of the Federal Highway Administration 共FHWA兲 and the Depart-
vulnerabilities that are classified into hydraulics, steel details, ment of Transportation of New York, Ohio, Utah, Wisconsin,
concrete details, collision, seismic, and overload vulnerabilities Texas, and Illinois; from personal experience; and through E-mail
共Scott Lagace, personal communication, 2001兲. Each class has a contacts. Although numerous minor bridge failures may not have
procedure that helps arrive at a vulnerability rating. This then been reported in published sources, the writers believe that the
identifies bridges that require corrective actions. NYDOT also information assembled here is sufficient to draw useful conclu-
stated that the information has been obtained through the news sions.
media as well as through responses to a survey they send every 12
years to all 50 states. The unit has received much information
from some states and little to none from others; hence, despite the Failure Defined
valuable source of data the unit has compiled, at this stage, the
Earlier analyses that have become the basis for the study in this
1
Graduate Student, Ohio State Univ., Columbus, OH 43210. paper revealed 57 cases of published bridge failures that occurred
2
Professor, Civil Engineering, Ohio State Univ., Columbus, in the United States between 1977 and 1981 共Hadipriono 1985兲.
OH 43210. The second study of such failures between 1982 and 1988 dis-
Note. Discussion open until January 1, 2004. Separate discussions closed 24 cases of bridge failures 共Hadipriono and Diaz 1988兲.
must be submitted for individual papers. To extend the closing date by
These figures represent a number of much publicized and well-
one month, a written request must be filed with the ASCE Managing
Editor. The manuscript for this paper was submitted for review and pos- known failure cases that were collected rigorously yet manually,
sible publication on February 28, 2002; approved on July 9, 2002. This without the advantage of the information technology we currently
paper is part of the Journal of Performance of Constructed Facilities, have. In reality, there could have been unrecorded bridge failures
Vol. 17, No. 3, August 1, 2003. ©ASCE, ISSN 0887-3828/2003/3- 共depending on how we define the term failure兲 that were over-
144 –150/$18.00. looked during the analyses; hence, the above figures could have

144 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003

J. Perform. Constr. Facil. 2003.17:144-150.

 been higher. In this paper, the writers have collected 503 cases of the bridge components. Maintenance deficiency, such as corrosive
bridge failures that occurred from 1989 to 2000. The writers be- or damaged components, takes place during postconstruction or
lieve that defining the type of bridges and the term failure is the service life of the bridge.
important to present sufficiently accurate figures of the failed When construction components are precast or prefabricated,
bridges in the United States. material deficiency originated by the manufacturer may contribute
The Association of American State Highway and Transporta- to bridge failures. Examples of such deficiencies are the use of
tion Officials 共AASHTO 1999兲 defines a bridge as ‘‘a structure, defective and substandard materials. The first five deficiencies are
including supports, erected over a depression or an obstruction considered as those associated with problems having internal ef-
共such as water, highway, or railway兲, having a track or passage- fects on the bridge components. On the other hand, the bridge or
way for carrying traffic or other moving loads and having an its components may also suffer from external effects, such as
opening measured along the center of the roadway of more than vehicle impact or a corrosive environment. Such causes may be
20 ft between undercopings of abutments or spring lines of classified as external events. Note that these deficiencies may be
arches, or extreme ends of openings for multiple boxes; it may correlated but that such correlation may not be readily apparent;
also include multiple pipes where the clear distance between hence, in this study, only the most probable principal cause in
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

openings is less than half of the smaller contiguous opening.’’ The each case was considered.
types of bridges investigated in this study fall under this defini- In addition, with respect to the effects on the bridge or its
tion. components, these deficiencies may be categorized as enabling,
While there is no concerted opinion regarding the definition of triggering, and procedural causes. The enabling causes are related
failure, throughout this paper, the term failure refers to two con- to the internal condition or performance of the bridge or its com-
ditions, collapse and distress. Failure is defined as the incapacity ponents. Hence, the first five principal causes 共design, detailing,
of a constructed facility 共in this case, a bridge兲 or its components construction, maintenance, and material-related problems兲 dis-
to perform as specified in the design and construction require- cussed in the preceding fall into the category of enabling causes.
ments. Bridge collapse is the failure of all or a substantial part of The triggering causes are external events that could initiate failure
the bridge, where full or partial replacement may be needed. In of a structure. The procedural causes are related to management
terms of functionality, collapse occurs when the entire or a sub- problems and the interrelationship between parties involved in a
stantial part of a structure comes down, in which the structure project. The latter causes are difficult to prove because they are
loses the ability to perform its function. Collapse can be further usually hidden and unpublished; thus, their evaluation is beyond
classified into two categories, total collapse and partial collapse. the scope of this study.
Further, total collapse implies that several primary structural
members of a span have fallen down, such that no travel lane is
passable. Partial collapse suggests a condition where some of the Results of Study
primary structural members of a span have fallen down, where
Results of the study presented in this paper include discussions of
such a condition endangers the lives of those traveling on or
failure occurrences, principal causes, and specific causes of
under the structure. Distress is the unserviceability of a structure
bridge failures.
or its component共s兲 that may or may not result in a collapse.
Moreover, distress is a particular condition of the structure, which
has undergone some deformations without losing the whole struc- Failure Occurrences
tural integration. In sum, both collapse and distress are subsets of Investigative studies conducted for bridge failures revealed that
failure. 503 bridges of various types failed in the past 12 years 共1989–
2000兲 in the United States. Out of the total recorded failures, 456
cases of bridge collapses were found from the NYDOT database.
Causes of Failures Defined Publications such as engineering journals, magazines, and web
sites revealed 65 failure cases, 18 of which overlapped between
The principal causes of bridge failures were categorized as defi- the two sources. Hence, only 13% of the major failure cases were
ciencies in design, detailing, construction, maintenance, use of reported in the civil engineering news media. The age of the
materials, and inadequate consideration of external events. The failed bridges ranged from one year 共during construction兲 to 157
first four deficiencies represent integral roles in the building of a years, with a mean of 52.5 years, a median of 52 years, and a
bridge. Deficiency in design constitutes errors, mistakes, over- mode of 63 years.
sight, omission, or conceptual flaw that could have taken place Table 1 shows failure occurrences of over 17 bridge types that
during the design process of the bridge. Detailing is a process range from arch to tied-arch. In addition, the study also identifies
between design and construction periods, in which the details of floating and pedestrian bridge failures. About 12% of these
the structural design are prepared for their implementation bridges could not be identified; hence, they are classified as mis-
through shop drawings. Design detailing is commonly performed cellaneous. Components of these bridges are primarily made of
by the contractors and approved by the engineers. Changes are steel, concrete, and timber.
often made emphasizing workability and constructibility of the Table 1 shows that the dominant types of failed bridges are the
facility. Previous studies revealed that this process is vulnerable steel beam/girder and steel truss bridges, with 145 共29%兲 and 107
to discontinuity or loss of the original design concepts 共Hadipri- 共21%兲 occurrences, respectively. Note that these failed bridges
ono 1985兲. Therefore, deficiency in design detailing may be con- constitute over 50% of the total bridge failures. The next signifi-
sidered as a class by itself. It includes errors, mistakes, omissions, cant cases involve failures of concrete beam/girder and concrete
and discontinuity/loss of design concept. Construction deficiency slab bridges, representing 29 共6%兲 and 25 共5%兲 occurrences, re-
occurs as problems with workmanship and deviation of results spectively. While significant, failure occurrences of these concrete
from the specifications. Examples of such deficiencies are im- bridges are pale in comparison with the former two types of failed
proper installation and inadequate temporary structure to support bridges.

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003 / 145

J. Perform. Constr. Facil. 2003.17:144-150.

 Table 1. Type and Number of Bridge Failures Table 2. Number of Failures with Respect to Phase of Failure
Bridge type Material Number of failures Percentage Occurrences
Types of failures Construction Service Unknown
Arch — 17 3.38
Bailey Steel 1 0.20 Distresses 0 17 0
Bascule — 2 0.40 Partial collapses 3 80 13
Beam/girder Concrete 29 5.77 Total collapses 5 12 21
— Steel 145 28.83 Unknown 0 277 75
— Timber 13 2.58 Total 8 386 109
Box Concrete 2 0.40
— Timber 5 0.99
Box girder Concrete 9 1.79
— Steel 3 0.60 the continual growth of the bridge population. According to the
Cable Steel 1 0.20 U.S. Department of Transportation 关Federal Highway Administra-
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

Corrugated pipe Steel 4 0.80 tion 共FHWA兲兴, the number of new bridges being added each year
Covered Timber 6 1.19 ranges from 1,400 to 4,000 units, with an average of about 2,500
Culvert Steel 17 3.38 共FHWA 2000兲.
— Other 2 0.40 The types of failures and the phase in which these failures took
Slab Concrete 25 4.97 place are shown in Table 2. The number of failures that occurred
— Steel 1 0.20 during service life 共386 occurrences兲 is far greater than that dur-
Span Steel 7 1.39 ing construction 共eight occurrences兲. Such is expected for most
— Timber 8 1.59 structures, including bridges, because, at any point in time, the
Stringer Steel 12 2.39 number of existing bridges during service life is far greater than
— Timber 12 2.39 that during construction. Also, the duration of the service life is
Truss Steel 107 21.27 much longer than that of the construction of bridges. Furthermore,
— Timber 9 1.79 loads applied to bridges increase with time, while efforts to up-
Tied arch Concrete 1 0.20 grade and maintain bridges remain relatively the same throughout
Floating — 2 0.40 the years. Table 2 also shows that a large number 共109 occur-
Pedestrian — 2 0.40 rences兲 of unknown failure phases exists because of the lack of
Miscellaneous — 61 12.13 information on the time the bridges were built.
Total — 503 100.00 The types of failures classified as distress, partial collapse, and
total collapse are also presented in Table 2. Among these three
failure types, partial collapse 共80 occurrences兲 dominates, fol-
Distribution with respect to the year when failures occurred is lowed by distress. Here, too, an overwhelming number of un-
presented as a bar chart in Fig. 1. From a total of 503 bridges that known failure types 共277 occurrences兲 are associated with incom-
failed during the 1989–2000 period, 112 failures 共22%兲 occurred plete data.
in 1993, which seems to be an anomaly. Further investigation It would be interesting to know which states have experienced
reveals that most of these 1993 failures coincided with the occur- the most failures. The writers tabulated the 10 highest ranked
rence of a major flood in the Midwest. In 1993, the Mississippi states as shown in Table 3. In terms of the number of failed
and Missouri Rivers and their tributaries overflowed and flooded bridges, the highest ranked state is Iowa, with 85 failure cases,
several Midwest states, namely, Illinois, Iowa, Kansas, Minne- most of which were attributed to the flood disaster described ear-
sota, Missouri, North Dakota, South Dakota, and Wisconsin. This lier. Iowa happened to be the most effected state during the 1993
flood caused damage in many constructed and transportation fa- flood. The state of New York is ranked second, with 64 failures,
cilities, including the failures of numerous bridges, particularly in followed by a distant third 共Virginia兲.
Iowa, Minnesota, and Missouri. Table 3 also shows the percentage of these failures with re-
Fig. 1 also shows other peak occurrences of failures such as spect to the total number of bridges in the state. Note that Iowa
the years 1989 and 1996. The 1989 Loma Prieta earthquake and a and New York have about the same percentage of failures 共0.3%兲.
1996 flood contributed to these peaks. The chart also reveals that, Since New York was not effected by flood, yet ranked on the
in recent years, failure occurrences seem to have declined, despite same par with Iowa, it would be interesting to see the reasons
behind the high failure frequency in New York. Also, a compari-
son can be made with other states having about the same or a
greater number of bridges, i.e., Minnesota 共21,696 bridges兲 and
Missouri 共26,060 bridges兲. These states were effected by the great
flood of 1993, yet they have much lower percentages of failure
frequency 共0.08 and 0.05%, respectively兲 as compared with New
York. One possible reason for New York’s high failure frequency
may be attributed to the fact that the state’s Department of Trans-
portation has initiated the creation of the bridge failure database
and, hence, has the most complete information regarding bridge
failures in New York. In addition, New York is an older state as
compared with the aforementioned midwestern states; therefore,
there is bound to be a greater number of older bridges that were
Fig. 1. Number of failed bridges distributed by year 共1989–2000兲
rendered obsolete and vulnerable to failures. The highest percent-

146 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003

J. Perform. Constr. Facil. 2003.17:144-150.

 Table 3. States, Ranked by Bridge Failure Frequency
Rank Name of state Failures Percentage of total failures Total number of bridges Percentage of failures to number of bridges
1 Iowa 85 16.90 26,035 0.33
2 New York 64 12.72 22,121 0.29
3 Virginia 37 7.36 16,481 0.22
4 West Virginia 34 6.76 8,156 0.42
5 Arkansas 33 6.56 13,901 0.24
6 Maryland 29 5.77 5,895 0.49
7 California 22 4.37 16,752 0.13
8 Minnesota 18 3.58 21,696 0.08
9 Mississippi 14 2.78 17,979 0.08
Missouri 14 2.78 26,060 0.05
10 Georgia 13 2.58 16,752 0.08
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

age of failures with respect to the total number of bridges in a The most dominant figures are those related to flood and scour
state is held by Maryland, with 0.49% 共29 failures of 5,895 共165 and 78 cases兲. The Federal Highway Administration defines
bridges兲. scour as ‘‘erosion or removal of streambed or bank material from
bridge foundations due to flowing water, usually considered as
long-term bed degradation, contraction, and local scour.’’ There-
Principal Causes
fore, failures caused by floods and scour are often one and the
As alluded to before, causes of bridge failures are classified into same 共Scott Lagace, personal communication, 2001兲. While both
six principal causes, which include both enabling 共design, detail- causes may have produced similar failures, the figures in Table 5
ing, construction, maintenance, and material-related problems兲 were compiled from original sources available to the writers. One
and triggering 共external-related events兲 causes, as shown in Table possible explanation for the different classification is that those
4. who entered the information into the database might have used
Observation shows that only a small proportion of bridge fail- both terms interchangeably as the same source of failures 共Scott
ures experienced distresses, while the majority of bridges col- Lagace, personal communication, 2001兲. Another possibility is
lapsed. Table 4 also reveals that 54 共11%兲 out of 486 collapses that the flood-related cases are associated with the great flood of
were attributed to enabling causes, while the majority of collapses
共415 cases, or 85%兲 were due to triggering causes. In the distress
mode, the percentage of enabling causes is higher than the trig-
Table 5. Type and Number of Failure Causes
gering causes. This seems to suggest that triggering causes tend to
result in collapses and enabling causes incline toward promoting Failure causes and events Number of occurrences Percentage of total
distresses. While this may be true for the triggering causes, expe- Hydraulic 266 52.88
rience shows that enabling causes could also result in collapses Flood 165 32.80
when distresses are ignored over a certain period of time. Table 4 Scour 78 15.51
also shows that, aside from external events, maintenance and Debris 16 3.18
construction-related deficiencies predominantly caused the bridge Drift 2 0.40
failures. The table indicates that 43 共8%兲 collapses and distresses Others 5 0.99
are maintenance related. Several of the latter cases are associated
Collision 59 11.73
with obsolete and deteriorated bridges.
Auto/truck 14 2.78
Barge/ship/tanker 10 1.99
Specific Causes Train 3 0.60
Other 32 6.36
Detailed information on various deficiencies is presented in Table
Overload 44 8.75
5. The leading causes of bridge failures are flood/scour, collision,
and overload. These causes fall under the category of external or Deterioration 43 8.55
triggering causes. General 22 4.37
Steel deterioration 14 2.78
Steel-corrosion 6 1.19
Table 4. Number of Principal Causes of Failure Concrete-corrosion 1 0.20
Fire 16 3.18
Principal cause Collapse Distress
Construction 13 2.58
Design 2 1 Ice 10 1.99
Detailing 0 0 Earthquake 17 3.38
Construction 11 2 Fatigue-steel 5 0.99
Maintenance 37 6 Design 3 0.60
Material 4 2 Soil 3 0.60
External 415 5 Storm/hurricane/tsunami 2 0.40
Others 共NA兲 17 1 Miscellaneous/other 22 4.37
Total 486 17 Total 503 100.00

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003 / 147

J. Perform. Constr. Facil. 2003.17:144-150.

 their dump bodies raised, and garbage trucks with the forks up
共Scott Lagace, personal communication, 2001兲.
Deterioration of bridge components is also an essential cause
of several failure cases. Forty-three cases have been observed, 14
of which are attributed to the deterioration of the bridges’ steel
components and six cases related directly to steel corrosion.
In addition, earthquake-related bridge failures represent over
3% 共17 occurrences兲 of total failures. All occurrences are in Cali-
fornia, and most of these are attributable to the 1989 Loma Prieta
共six occurrences兲 and 1994 Northridge 共10 occurrences兲 共Cooper
et al. 1994兲 earthquakes. Note that, in this study, the collapse of
Fig. 2. Bridge failures caused by floods, distributed by year an overpass of a similar type to bridges is considered as a single
bridge collapse occurrence.
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

1993, while the scour-related cases may have occurred elsewhere

or are unrelated to the 1993 Midwest flood. Be that as it may, the Comparison of Three Periods Surveyed
combined figure of 266 flood/scour-related cases constitute 53%
of the total causes of failures. The total number of bridge failures in this study is far greater than
Further analyses of the failure data caused by flooding shows those of earlier studies. In order to present a compatible compari-
that 31% of the failures took place in 1993 共Fig. 2兲. This result son, failures due to natural disasters 共e.g., inevitable earthquakes,
also supports the role of Midwest flood of 1993 in causing nu- storms, fires, and floods兲 and deterioration/obsolescence were ex-
merous bridges to fail. cluded 共Tables 6 – 8兲. While, on one hand, these exclusions will
Table 5 also indicates debris as another specific cause listed in improve the compatibility of the comparative analysis, they also
the hydraulic category with flood and scour. There are 16 cases of isolate human-induced deficiencies from act-of-God natural
failure derived from debris flows, all of which occurred in the events. However, not all deterioration is due to old age; bridges
same year, 1995, and in the same state, Virginia. This coincidence that failed at a relatively young age 共less than 50 years of service兲
is attributable to a flash flood that took place in Madison County were considered as having their service life expired prematurely.
in June of 1995. A flash flood is a flood that occurs in a short Thus, 18 out of 43 deterioration cases 共40%兲 were included in this
period of time as a result of rain falls at such a high rate that water analysis as maintenance deficiencies. Table 6 is an abbreviated
does not have time to be absorbed into the ground 共Watson 2001兲. form of Table 4, where principal causes due to natural disasters
Heavy rains in the mountains of Virginia caused mudslides and and deterioration/obsolescence were excluded. From Table 6, a
washed debris up against the bridges 共National Weather Service total number of 157 collapses and nine cases of distresses were
2001兲. found. A dominant cause of 115 occurrences 共77% of all human-
The next dominant cause is overloading 共44 occurrences, 9%兲 induced failures兲 is associated with external causes such as over-
of various types of bridges. While only two pedestrian bridges load and lateral impact on bridges by land and marine vehicles.
共Table 1兲 were observed to fail due to this cause, such a failure By eliminating natural disasters and deterioration/
could produce a devastating effect on public safety. For example, obsolescence as causes of failure, the trend of bridge failure oc-
the collapse of the Speed Motor Lowe’s Walkway in Concord, currences over the years appears to fluctuate every 2 or 3 years
North Carolina, resulted in 107 people being injured 共GoCaroli- 共Fig. 3兲.
nas.com 2000兲. The number of bridge failures in Table 6 is relatively large as
One particular bridge—the Evergreen Point Floating Bridge compared with that of the previous studies. One possible expla-
on Lake Washington, WA—experienced a recurrence of distresses nation is the use of current technological advances in information
and collapses during its service life. The first failure took place dissemination that were nonexistent in the past. Most of the in-
when the pontoon bridge deck cracked open in 1989 due to an formation gathered in this study is originated from the NYDOT
unknown cause. Then, again, in 1991 another problem emerged as database and the Internet.
the pontoon cracked. The third setback occurred in 1999, when Table 7 shows the comparison of failure occurrences among
the bridge was split in two by a strong wind. In 2000, the bridge three periods of study, from 1977 to 2000. Failures are divided
was again damaged by collision. into collapses and distresses and, subsequently, classified into the
Of the 503 bridge failures observed, the total number of re- time of occurrences, during construction or service life. Table 7
corded human-induced fatalities and injuries 共excluding natural
disasters and deterioration/obsolescence-related failures兲, is 76
and 161, respectively. Hence, the injuries resulting from a single Table 6. Principal Causes of Failure Excluding Natural Disasters and
case such as the Speed Motor Lowe’s Walkway accident could Deterioration/Obsolescence
easily reach 66% of the total injuries or 45% of the total human- Principal causes Collapse Distress
induced fatalities and injuries observed in this study. Design 2 1
Another significant cause of bridge failures is collision by land
Detailing 2 0
and marine vehicles 共Table 5兲. Notable among these are impacts
Construction 10 2
from trucks 共14 occurrences兲, barges/ships 共10 occurrences兲,
Maintenance 17 1
trains 共three occurrences兲, and others/unknown 共32 occurrences兲.
Material 6 2
These cases 共59 occurrences兲 constitute 12% of the total bridge
External 115 2
failures. Most of these failures were attributed to lateral impact
Others 共NA兲 5 1
forces of vehicles on bridges. Examples are collisions caused by
Total 157 9
backhoes improperly loaded on flatbed trucks, dump trucks with

148 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003

J. Perform. Constr. Facil. 2003.17:144-150.

 Table 7. Comparison of Failure Distribution with Respect to Stage of Occurrence in Three Survey Periods
1977–1981 1982–1988 1989–2000
Type/stage of failures Number Percent Number Percent Number Percent
Collapses during construction 3 9 4 17 8 5
Collapses during service 11 33 5 21 149 90
Distresses during construction 3 9 5 21 0 0
Distresses during service 16 48 10 42 9 5
Total 33 100 24 100 166 100

concludes that collapses and distresses consistently took place trend of failure occurrences and causes. In all three observations,
during the service life more often than during the construction most failures took place during the service stage. This is expected,
process, for reasons described earlier. because the population of bridges during service is larger than
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

Table 8 compares the three periods of time, in which the fail- during construction, service life duration is generally longer than
ure events are separated into six principal types of causes. The that of construction, and there is a generally disproportionate in-
dominant principal cause of collapse from the first and third crease of loads working on bridges and enhancement of bridge
共present兲 studies is external causes, while the second study found resistance through time. External events in the latest 共third兲 study
construction-related deficiencies as significant. In all studies, con- period are overwhelming; they stand out in comparison to all
struction and material-related deficiencies are the leading factors principal causes in all three studies. Even if—to achieve a more
of distresses. compatible comparison—information obtained from the NYDOT
database is excluded, external events in the third study period are
still dominant. This suggests that attention should be given to
Summary and Conclusions minimize such events.
Another point of comparison is the relatively larger number of
The study of over 500 bridges that failed in the past 12 years in
failure occurrences in the latest study period as compared with the
the United States 共1989–2000兲 revealed an average age of 52.5
previous two studies. A reasonable explanation is that the latest
years, with a range from one to 167 years. About 50% of bridges
study was conducted by taking advantage of information technol-
that failed are typically steel beam/girder and steel truss bridges.
ogy that was not available in the past. For instance, if the latest
Other frequent occurrences are associated with failures of con-
study was performed in a fashion similar to the first two 共i.e., by
crete beam/girder and concrete slab bridges. The study shows that
excluding databases and the Internet兲, only 65 failure cases were
failures took place primarily during the service life of the bridges.
Records show that the states of Iowa and New York are ranked observed. Despite these advantages, there is an urgent need to
highest in terms of failure occurrences. Failures in Iowa are as- improve data collection and processing. The NYDOT database is
sociated with the 1993 major flood, while those in New York are clearly a winning start, but without a concerted effort from all
related to the bridges’ obsolescence 共of the total 64 cases in New U.S. states in maintaining a reliable repository of bridge failures,
York, 25 bridges were over 50 years old when they failed兲. In such information may be rendered nugatory. For example, multi-
addition, New York is the first to develop a major database of tudinous failure cases observed in the latest study are incomplete;
bridge failures. many of them are not furnished with bridge types and phase of
Overwhelming external events, both natural and man-made, failures, two variables that are paramount for conducting statisti-
representing 83% 共420 occurrences兲 of all principal causes, trig- cal or other quantitative risk analyses.
gered the bridges to fall. Nature-induced external events include Aside from cases familiar to the writers, the majority of the
floods, earthquakes, fires, ice, and hurricanes 共with floods repre- cases investigated in this study are provided with only limited
sents 53% of all failures兲, while the human-induced external information as to the causes of collapses and distresses. While
events that constitute 20% of all failures include bridge overloads interpretation can be readily made on the enabling and triggering
and lateral impact of land and marine vehicles on bridges 共vehicu- causes, the procedural causes are not apparent; hence, no attempt
lar impact represents 12% of all failures兲. was made to discern the latter cause. Procedural causes are often
Upon completion of this observation, a comparison among associated with inadequate responsibility delineation, communi-
three periods of similar studies was conducted to discern possible cation problems, legal and contractual issues, and other indirect

Table 8. Comparison of Principal Causes of Bridge Failures

Collapse Distress
Principal causes 1977–1981 1982–1988 1989–2000 1977–1981 1982–1988 1989–2000
Design — — 2 共1%兲 共11%兲
2 2 共13%兲 1 共11%兲
Detailing 2 共14%兲 2 共22%兲 2 共1%兲 共5%兲
1 1 共7%兲 —
Construction 2 共14%兲 4 共44%兲 10 共6%兲 共32%兲
6 3 共20%兲 2 共22%兲
Maintenance 1 共7%兲 1 共11%兲 17 共11%兲 共11%兲
2 2 共13%兲 1 共11%兲
Material — — 6 共4%兲 共26%兲
5 3 共20%兲 2 共22%兲
External 9 共64%兲 1 共11%兲 115 共73%兲 共16%兲
3 3 共20%兲 2 共22%兲
Others 共NA兲 — 1 共11%兲 5 共3%兲 — 1 共7%兲 1 共11%兲
Total 14 共100%兲 9 共100%兲 157 共100%兲 19 共100%兲 15 共100%兲 9 共100%兲

JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003 / 149

J. Perform. Constr. Facil. 2003.17:144-150.

 References
American Association of State Highway and Transportation Officials
共AASHTO兲. 共1999兲. Federal codes of regulations: revised as of April
1, 1999, Washington, D. C.
Cooper, J. D., Friedland, I. M., Buckle, I. G., Nimis, R. B., and Bobb, N.
M. 共1994兲. ‘‘The Northridge Earthquake: progress made, lessons
learned in seismic-resistant bridge design.’’ Public Roads Online,
具http://www.tfhrc.gov/pubrds/summer94/p94su26.htm典 共Jan. 4, 2002兲.
Eldukair, Z. A., and Ayyub, B. M. 共1991兲. ‘‘Analysis of recent U.S. struc-
tural and construction failures.’’ J. Perform. Constr. Facil., 5共1兲, 57–
Fig. 3. Number of bridge failures excluding natural disasters and 73.
Federal Highway Administration 共FHWA兲. 共2000兲. ‘‘National bridge in-
deterioration/obsolescence distributed by year
ventory record count 1998 –2000.’’ 具http://www.fhwa.dot.gov/bridge/
allbridg.htm典 共Dec. 5, 2002兲.
Downloaded from ascelibrary.org by PENN STATE UNIV on 08/12/14. Copyright ASCE. For personal use only; all rights reserved.

GoCarolinas.com 共2000兲. ‘‘NC DOT: all private bridges must be in-

spected.’’ 具http://www.gocarolinas.com/news/charlotte/2000/05/23/
bridge – collapse.html典 共Jan. 2, 2002兲.
Hadipriono, F. C. 共1985兲. ‘‘Analysis of events in recent structural fail-
problems from which enabling and triggering causes may stem. ures.’’ J. Struct. Eng., 111共7兲, 1468 –1481.
Understanding all causes including procedural causes can moti- Hadipriono, F. C., and Diaz, C. F. 共1988兲. ‘‘Trends in recent construction
vate design and construction professionals to proactively mini- and structural failures in the United States.’’ Foren. Eng., 1共4兲, 227–
mize the recurrence of bridge failures. These latter causes can be 232.
Harik, I. E., Shaaban, A. M., Gesund, H., Valli, G. Y. S., and Wang, S. T.
obtained from legal records and insurance claims; they should be
共1990兲. ‘‘United States bridge failures, 1951–1988.’’ J. Perform. Con-
collected and included into existing or new databases. str. Facil., 4共4兲, 272–277.
National Weather Service, Weather Forecast Office Baltimore/
Washington. 共2002兲. ‘‘The greatest storms of the century in the
Acknowledgment Greater Washington-Baltimore Region.’’ 具http://205.156.54.206/er/
lwx/Historic – Events/StormsOfCentury.html典 共Dec. 21, 2001兲.
Watson, B. M. 共2001兲. ‘‘Virginia floods.’’ National Weather Service,
The writers wish to thank Scott Lagace, P.E., from the New York Weather Forecast Office Baltimore/Washington, 具http://
State Department of Transportation for providing the writers with 205.156.54.206/er/lwx/Historic – Events/va-floods.html典 共Dec. 5,
the NYDOT database along with his comments about this paper. 2002兲.

150 / JOURNAL OF PERFORMANCE OF CONSTRUCTED FACILITIES © ASCE / AUGUST 2003

J. Perform. Constr. Facil. 2003.17:144-150.