Fourth US-Taiwan Bridge Engineering Workshop

DEVELOPMENT OF A BRIDGE FAILURE

DATABASE
George C. Lee, J. O'Connor, J.C. Qi and Z.Q. Wang
ABSTRACT
Bridge failure may be defined as loss of a structural component, loss of a bridge's basic
functionality, a catastrophic bridge collapse, or any damage condition in between. A bridge can
fail due to a variety of single or combination of reasons including material imperfection or aging,
overload, insufficient capacity, construction error or improper maintenance. Lessons can be
learned through proper studies of bridge failures. Similar to reconnaissance studies of damaged
or collapsed structures after a natural disaster, design guidelines can be improved through better
understanding of the cause and mechanism of failure.
In order to carry out useful post event studies and to have useful information for future
bridge engineering practice, adequate documentation of the damage or failure condition is
needed. This requires a database primarily consisting of two types of information: the original
design and construction information (e.g. the NBI) and the failure information (i.e. the NYDOT
bridge failure database).
In the current research project at University at Buffalo (UB) sponsored by Federal
Highway Administration (FHWA) on multiple hazard design of highway bridges, a subtask is
concerned with the development of a database for the following two objectives:
1. Short term objective: development of a bridge failure database that can conveniently
supply the necessary information for forensic studies. The relationship between the
extreme hazard load effect and the damage/collapse conditions can help to develop a
methodology for evaluation and comparison of extreme hazards and calibrate the
analytical models.
2. Long term objective: development of a comprehensive bridge damage/collapse
database (a National Repository) over time to benefit future bridge engineering
practice in design against extreme hazard events
This paper is a progress report on the planning and development of the intended bridge
failure database to satisfy the above two objectives.

_______________
George C. Lee, SUNY Distinguished Professor, University at Buffalo, 429 Bell Hall, Buffalo, NY 14260
Jerome OConnor, Sr.Program Officer, Transportation Research, MCEER, University at Buffalo, Buffalo,NY 14261
Jincheng Qi, Research Scientist, CSEE, University at Buffalo, 332 Bell Hall, Buffalo, NY 14260
Zhiqiang Wang, Visiting Research Scholar, University at Buffalo (On leave from Tongji University, Shanghai,
China)

Paper 15

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

INTRODUCTION
Extreme Hazards include natural hazards such as wind, earthquake, snow, flood, and
wave. Despite of their relatively low rate of occurrence, their effects to society can be quite
catastrophic. Understanding hazards and their destructive impacts requires careful analyses and
scrutiny of past events and failure cases (Lee and OConnor, 2008). If detailed information about
each catastrophic event is made available in a useable form, in-depth analysis of the capacity,
hazard load effect and the possible failure modes of the structure under various circumstances
may be reconstructed. Information of such forensic studies can help to develop a methodology
for evaluation and comparison of extreme hazards quantitatively and calibration of extreme
hazard loadings for bridge design (Lee, Cho et al, 2008; Ghosn, et al, 2003; Lee and Sternberg,
2008). Currently, much reconnaissance information are scattered in various reports and journals
that are difficulty to gather. The development of a database of failed bridges in the past and those
failed in the future will provide valuable information for bridge engineers.
This paper presents the research task to establish a comprehensive bridge failure database.
It will be linked together National Bridge Inventory (NBI) and the bridge failure data collected
by New York State Department of Transportation (NYDOT) and other available sources
(NYSDOT, 2006; FHWA, 2007).

OBJECTIVE
Short term objective: Develop a database for the study of bridge failures and help
establish the relationship between the extreme hazard load effect and the damage/collapse
conditions of failed bridges.
Long term objective: Establish the framework of a comprehensive bridge
damage/collapse database (a National Repository) with the data to be gradually established over
time to benefit future bridge engineering practice.
DATABASE CONTENT COMPONENT
Extreme hazards can be divided into three general categories:
(1)
(2)
(3)

Natural Hazards, (e.g., flood, earthquake, hurricane)

Man-made Hazards Unintentional (e.g. condition related, accidents such as
vessel collisions, deficient designs, lack of maintenance)
Man-made Hazards Intentional (e.g. fanatical acts of terrorism)

In order to develop a bridge failure database, it is necessary to first define the scope of
failure and the entities of the intended database. These are challenging issues. Today we do
not have a commonly accepted definition of the bridge failure.
For the above mentioned short term objective, we intend to work on a limited number of
past bridge failures (serious damaged or collapsed) for the purpose of forensic study and
calibrating the multiple extreme hazard load effects on highway bridges. Experience of such
studies will enable us to pursue the development of a more comprehensive bridge failure
database to benefit future bridge engineering practice.

Paper 15

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

For our current research project, the following hazards will be included: earthquake,
flood (scour), wind and vessel collision. Other hazards will be included in the future.
There are several existing information sources on bridges in US such as NBI (National
Bridge Inventory) and NYDOT etc. NBI includes the basic information for the US bridge, which
include Bridge Identification, Structure type and material Age and service, Classification,
Bridge condition, Bridge geometry, Inspection, Proposed improvement, Appraisal,
Navigation, load rating and posting etc.. It offers the basic information about the existing
bridges and the maintenance and inspection records, but not failure information. NYDOT bridge
failure database includes the date, type, cause, numbers of death and injured. It also includes
certain identity and location information. Both of the above provide useful information, but
insufficient to accomplish the objectives of forensic study. We propose to establish a bridge
damage/failure database, which integrates current existing bridge information sources, to include
the following:
(1)
(2)
(3)
(4)

Bridge design information (design document, construction, drawing etc).

Environmental (Geographical and weather) condition.
Hazards information and traffic condition.
Bridge damage and loss information (damage positions and damage modes, etc) and
economic and environmental impacts.

An example of earthquake damage to bridges is given in Table 1.

Table 1. Earthquake damage mode and location of bridge
Name of component
Superstructures
Beam or girder
Steel truss
Substructures
Pier or column

Abutment

Bearing

Other
components

Paper 15

Foundation
(footing)
Expansion joints
Shear key
Bumper block
Restrainers cable

Failure models
Unseating (Loss of span)
Local collision damage

Annotation
Collapse
Local damage
Local damage, may
be collapse

Local buckling or yielding

Moment damage, shear damage,
moment and shear damage, joint
damage, Inadequate ductility
capacity
Wing wall damage, back wall
damage, inclination of abutments
Failure of shear keys at abutments
etc.
Bearings damage, anchor bolt
damage etc.
Pile damage, footing damage
Expansion joints collision damage
Shear key damage
Bumper damage
Restrainers cable fracture

Local damage, may

be collapse

Local damage, may

be collapse
Local damage, may
be collapse
Local damage, may
be collapse
Local damage
Local damage
Local damage
Local damage

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

DATABASE STRUCTURE AND SOFTWARE INTERFACE

Database Structure
All the information necessary to carry out case study of bridge failure under hazards is
contained in a single database. The database structure, see fig. 1, consists of three components,
the NBI: data file of bridges, the NYDOT bridge failure database and the bridge failure database
(to be developed) under multiple hazards.
The failure bridge
database

The NBI data file

Hazards information

The bridge failure database

under multiple hazards

As-built design
information

The NYDOT bridge

failure database

Damage information

Economic loss

Figure 1 Database structure of bridge failure

Database Software Interface
An user friendly interface software to handle bridge failure database is the core effort
currently under development. This software will facilitate the search and statistical analysis of
bridge failure under hazards that are necessary for case studies. A comprehensive set of menus,
dialogues, tool bars, buttons and forms are supported. On-line help is available. Figure 2 and
Figure3 show typical interface of this database.

NBI information statistics of US bridge

Selected bridge and associated information

Paper 15

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

Obtained position of the bridge through the Google Map

Information statistics of the failure bridge database

Figure 2 Database software interface components-information statistics.

Damage information obtained from database

Design information obtained from database

Report generated from database

Hazards information obtained from database

Figure 3 Database software interface components-design and damage information

Example: A case study of a collapsed bridge
The objective of this case study is to illustrate how to use this database to obtain the
necessary information associated with damaged bridge and to carry out forensic studies. We
chose the Bull Creek Canyon Channel Bridge damaged by the Northridge earthquake for this

Paper 15

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

case study. This bridge was completed in 1976. The bridge carries the SR118 over a channel
approximately 9 km northeast to the epicenter (EERI, 1995; Priestsley et al, 1994;
Reconnaissance Team, 1994; Buckle, 1994).
For this case study, detailed analyses were carried out to simulate the earthquake
performance of the bridge. Results are compared with the observed damages. The case study
consists of the following steps:
First, searches for bridge information menu and obtain as-built design information of the bridge,
such as general layout, dimension of superstructure etc. As shown in Figures 4 and 5, this is a 3span bridge, approximately 30 m in length and is supported on two multi-column bents. The
superstructure is a continuous prestressed 13-cell box girder. The deck also features a
longitudinal expansion joint which separates the bridge in two parts: North and South. The skew
of the supports ranges from 36 at the West abutment to 47 at the East abutment.

Figure 4 As-built design drawing

Figure 5 General plan and dimension of girder

Second, general hazards information and the acceleration data of the bridge are directly retrieved
from the hazards and hazard loads menu, see Figure 6. Due to a lack of the ground motion record
at the bridge site, the ground accelerations recorded at the Sylmar county hospital are selected
and modified for nonlinear time history analysis.

Figure 6 Ground acceleration of nearby bridge site

Paper 15

Figure 7 Dynamic analytical model

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

Next, an analytical model is constructed for nonlinear simulation earthquake damage analyses.
Figure 7 is the dynamic analytical model of Bull Creek Canyon Channel Bridge. The effect of
unseating prevention devices, expansion joints and soil-structure interaction are carried out. The
hysteresis behavior caused by the bearing friction and nonlinear characteristic of structure
components under the large ground excitation is considered

Figure 8 Analysis result compared with the actual performance of the bridge
The earthquake response of the bridge is calculated through detailed nonlinear timehistory analyses and compared with the actual performance of the bridge under this earthquake,
see Figure 8. Based on the above capacity and response assessment, most of the columns at bent
3 began to yield simultaneously at the top and bottom, column of bent 2 finally yielded at the top.
Such simulation analyses can be used to explore the damages to the bridge, and to understanding
of the cause and mechanism of failure.
The soft clay around piers is usually vulnerable to significant scour in the flood. We
further carried out earthquake response analyses of this bridge by considering the effects of scour.
Nonlinear simulated seismic analyses under scouring underneath of bent 3 is postulated (the
currently database does not contain scour information of this bridge, the scouring depth and
position are assumed). Dynamic analytical model is shown in Figure 9. Figure 10 shows the
analytical results considering the effect of scour compared to that without scour, where the
seismic responses of the bridge are quantitatively given.
0.04

Displacement (m)

0.02

Scour depth

0.00

-0.02

-0.04

-0.06

without scour
scour

-0.08

10

Tim e (s)

Fig 9 Dynamic analysis model with scour effect

Paper 15

Fig 10 Analytical result considered scour effect

Princeton, New Jersey | August 4-5, 2008

Fourth US-Taiwan Bridge Engineering Workshop

SUMMARY
This paper briefly describes the objective and progress of developing a bridge failure
database due to various multiple hazards. For a short term objective, bridge failure database can
conveniently provide information to facilitate forensic studies of failure of the highway bridges,
and for the calibration of the analytical model for development of extreme hazard load effect in
LRFD. For a long term objective, a comprehensive bridge damage/collapse database (a National
Repository) may be gradually established over time to benefit future bridge engineering practice.

ACKNOWLEDGEMENT
The study presented in this paper is supported by the Federal Highway Administration
(Contract No. DTFH61-98-C-00094).

REFERENCES
Lee, G. C. and OConnor, J. S. (2008). Improvement of Bridge Safety through Forensic
Studies, MCEER Technical monograph on July 2008.
Lee, G. C., Cho, S. Y., Tong, M. and Yen, W. P. (2008). Developing A Comparison
Methodology of Extreme Hazards for Highway Bridge Design, Proc 6NSC, Charleston, SC
Ghosn, M., Moses, F. and Wang, J. (2003). Design of Highway Bridges for Extreme Events,
NCHRP 489.
Lee, G.. C. and Sternberg, E. (2008). A New System for Preventing Bridge Collapse, J Issues
in Science and Technology XXIV: 3 Spring 2008, 31-36.
New York State Department of Transportation (2006). Bridge Collapse Database Albany, NY,
October 2006
Federal Highway Administration (2007). National Bridge Inspection Standards (NBIS), 23
CFR Part 650, http://www/fhwa.dot.gov/bridge/nbis.htm.
EERI, (1995), Northridge Earthquake of January 17, 1994 Reconnaissance Report, Earthquake
Spectra, Earthquake Engineering Research Institute, Oakland, California.
Priestley, M.J.N., Seible, F. and Wang, C. M. (1994). The Northridge Earthquake of January 17,
1994--Damage Analysis of Selected Freeway Bridges, Report No. SSRP-94/06, University
of California-San Diego, February 1994.
Reconnaissance Team. (1994). Preliminary report on the Northridge, California, Earthquake of
January 17, 1994. Canadian Association for Earthquake Engineering Association,
Vancouver, B.C. May 1994
Buckle, I. G. (1994). The Northridge, California Earthquake of January 17, 1994: Performance
of Highway Bridges, Technical Report NCEER-94-0008.

Paper 15

Princeton, New Jersey | August 4-5, 2008