Content uploaded by C. C. Fu
Author content
All content in this area was uploaded by C. C. Fu on Dec 24, 2014
Content may be subject to copyright.
Study of Damaged Wushi Bridge in Taiwan 921 Earthquake
By Yao T. Hsu1 and Chung C. Fu2, Members
Abstract: This paper reports on the damage of Wushi bridge in a recent Taiwan 921 earthquake. Damage
to Wushi bridge appeared in the superstructure, the substructure and the approaches. Typical types of
damage are discussed and illustrated in this paper. A review of the bridge design specifications in Taiwan
is also presented to give the background on the seismic design of Taiwan highway bridges.
Introduction
. At 1:47 AM (local time), Tuesday, September 21, 1999, a devastating earthquake with a
magnitude of 7.3 on the Richter scale struck central Taiwan. According to the seismic report published by
the Taiwan Central Weather Bureau, the epicenter of this earthquake, so called Chi Chi or 921 earthquake,
is located at 23.85° N and 120.81°E at a depth of 7.0 km (Fig. 1). The 921 earthquake was associated with
two closely spaced faults, Chelungpu and Shuangtung faults (Fault lines 18 and 20 in Fig.1). These two
faults are 10 km apart and almost in parallel. The hypocenter at the town of Chi Chi is at the intersection of
these two faults. It was caused by the reversive fault movement at the subduction zone boundary of
Euroasian and Philippino plates. The official estimates of the casualties and losses are 2,161 casualties,
8736 injuries and $3.7 billion property loss (NSF/ROC 1999). This is the strongest earthquake to hit
Taiwan within the past 100 years and the most costly natural disaster.
Most casualties were due to numerous failures of non-ductile concrete buildings. Since the
earthquake struck in the middle of the night, very few casualties were caused by failures of bridges.
However, million of dollars were lost due to damage or collapse of bridges. Many damaged bridges along
the key routes were repaired on a temporary basis. Others were put under investigation to find the strategy
of retrofitting. Based on the severity of the damage and also the use for strategy of retrofitting, bridge
failures due to this earthquake were divided into three groups:
(1) Severe case: Traffic was interrupted by failed bridge piers or falling beams;
(2) Moderate case: Controlled traffic was imposed due to settlement, damaged bearing, and cracking
of decks, beams or piers;
(3) Minor case: Normal traffic is maintained with slight settlement, minor cracking, or minor
horizontal movement.
1 Associate Professor, Department of Transportation Engineering and Management, Feng Chia University,
Taichung, Taiwan, R.O.C.
2 Director and Associate Professor, Bridge Engineering Software and Technology (BEST) Center,
Department of Civil Engineering, University of Maryland, U.S.A.
Highway damage was widespread throughout two central Taiwan counties, Taichung and Nantou.
Hundreds of bridges from expressways to county roads are located in these two counties. Except for about
10% of the bridge population experiencing moderate-to-major damage, most escaped serious damage.
Figure 2 shows measuring stations and their measured peak ground accelerations (PGAs). On the top of
the cross are vertical accelerations, on the left are E-W horizontal ground accelerations and on the right are
N-S horizontal ground accelerations. All those considered to be under 'near-field' action are subjected to
intense horizontal and vertical ground motion as well as surface fault movement. The worst displacement
caused by the 921 earthquake was about 7-8 meters along certain sections of the Chelungpu fault. It is
understandable that if faults pass under the bridge and the dislocations are large, catastrophic incidents and
even bridge collapses are bound to happen.
Site Investigation
The second day after the earthquake, even with the interrupted traffic, the first author led twelve
students, divided into six groups, who rode motorcycles to visit the damaged bridge sites and took hundreds
of pictures to build a large inventory for future reference. The second author also visited the sites a month
later to collect more information, assess the damage and evaluate the causes. Wushi Bridge is one of the
bridges inspected and is reviewed in this paper. Wushi bridge, located across the Chelungpu fault line (Fig.
2), shows multiple bridge failure modes and gives a representative case of bridge failure under earthquake.
In general, it may be noticed that most of the problems can be blamed on designs based on early codes and
the severity of the earth movement. Based on the measurement records along the fault line (Fig. 2), most of
the ground accelerations were over 300 gal, some even as high as 1G, which are much higher than the latest
design ground accelerations of 0.33G, 0.28G, 0.23G and 0.18G. If the measured acceleration records were
used in the design, unexpected seismic actions and soil effects caused damage and collapses.
This paper gives a brief overview of the investigation and possible causes for damage to Wushi
bridge. The bridge shows some complex failure modes caused by this earthquake. An overview of the
causes of damage and collapse in this earthquake may offer reliable guidelines on deficiencies in bridge
design and later considerations in improving code provisions.
Bridge Damage Investigation
Wushi is located on Provision Route 3 and is an essential link between Taichung and Nantou
counties. The total length of the Wushi Bridge is 624.5 meters and the total width is 26 meters. The bridge
consists of one northbound bridge and one southbound bridge, each with two lanes of 12.5 meters (Fig. 3).
The northbound bridge was finished in the 1960's and the southbound one was completed in 1973. The
superstructures used prestressed concrete I-beams with constant span length of 34.84 meters.
The cross section of the northbound and southbound super- and substructures are shown in Figure
4. It is noticed that the northbound substructures are wall type concrete piers and the southbound
substructures are hammerhead concrete piers. They are all supported by 6-meter diameter, 13-16 meter
shaft foundations.
Most of the design and construction information of the northbound bridge, constructed in the
1960's, is unavailable. The southbound bridge adopted Kh = 0.15 (equivalent to the peak ground
acceleration PGA = 150 gal) as the earthquake design coefficient. Boring records show that the river bed
was covered with pebbles and underlined by clay rock.
The Wushi bridge is a river crossing bridge and it incidentally crosses the Chelongpu Fault. The
northern end of the bridge was the more severely damaged. One side of the fault was lifted vertically about
2.1-2.3 meters. Movement along the bridge=s longitudinal direction was estimated to be about 2.2-2.3
meters and the movement along the bridge=s transverse direction was about 2.1-2.3 meters. The third span
of the northbound bridge crosses the subduction zone boundary with 45o reverse fault movement. The
movement caused the superstructure of the first several spans to fall to the ground. Inspection (Fig. 5)
showed clear evidence of fault uplifting on both sides of the fallen spans.
According to the published record from the Central Weather Bureau, the East-West PGA was 518
gal, North-South was 639 gal, and vertical was 416 gal. The surface permanent horizontal movement was
2.3 meters and 2.9 meters along the East-West and North-South, respectively.
The record closest to the site was made at the Freefield Strong Seismic Station TCUD71, which is
about 5.5 Km southeast of the nearest town of Chao-Fung. After the Chi-Chi earthquake, Wushi bridge
was severely damaged and the road was closed. Due to the slip movement, the first and second spans of
the northbound bridge fell to the ground and the third span clearly showed lateral movement (Fig. 6). The
northern end abutment of the northbound bridge was pushed by the span and the jigsaw type of expansion
joint and the backwall board were demolished. Under the pressure, the backfill was pushed by the
superstructure and moved upward (Fig. 7). Under horizontal vibration, substantial shear cracks showed on
the southbound hammerhead concrete pier (Fig. 8). The cap of the first pier of the northbound bridge
(P1N) cracked and the East side exterior PCI beams were flexured. The end diaphragms and shear blocks
designed to prevent concrete beams from moving laterally were crushed as the whole superstructure moved
westward (Fig. 9).
Review of Design Codes
Taiwan is located in an active seismic area. Bridge design based on earthquakes in Taiwan has a long
history. There have been three major stages, starting from 1960:
(1) November 1960, the Department of Transportation published the first edition of “Highway Bridge
Engineering Design Specifications” (DOT/ROC 1960) which divides the Taiwan area into two zones
with 0.1G and 0.15G, respectively. The second edition of the Specifications modified the coverage of
the high seismic zone into one large area but still with ground accelerations of 0.1G and 0.15G. In
1970, the first freeway connecting northern and southern Taiwan was in the planning stage. A special
project focusing on earthquakes was also underway and the coefficients, based on the geographic area,
soil condition and importance of the bridge, were modified to 0.2, 0.15 and 0.1.
(2) January 1987, the Department of Transportation published “Highway Bridge Design Specifications”
(MTC/ROC 1987) Based on the latest earthquake theory available at that time, the design horizontal
coefficient kh was determined by :
kh = ZSIC0 if the height of the bearing cap ≤ 15 m (4)
kh = βZSIC0 if the height of the bearing cap >15 m (5)
Where kh is the design horizontal coefficient (≥0.1); C0 is the baseline design earthquake coefficient
(=0.15); Z is the zoning coefficient (1.2 for Strong Seismic Zone A, 1.0 for Strong Seismic Zone B, 0.8
for Moderate Seismic Zone and 0.6 for Weak Seismic Zone); S is the soil coefficient associated with
ground period TG, divided to four categories (Category 1: TG <0.2, S=0.9; Category 2: 0.2≤TG <0.4,
S=1.0; Category 3: 0.4≤TG <0.6, S=1.1; Category 4: TG >0.6, S=1.2); I is the importance factor with
1.0 for important bridges and 0.8 for common bridges; β is the adjusted factor based on the bridge
fundamental period and soil strata.
(3) January 1995, Department of Transportation published “Highway Bridge Seismic Resistance Design
Specifications” (MTC/ROC 1995). Based on the latest codes of the United States and Japan, plus the
local conditions, the National Center for Research on Earthquake Engineering (NCREE) in Taiwan
conducted a study on highway bridge seismic resistance design and published new Specifications
with a new design horizontal seismic coefficient V:
V = ZIW(C/Fu)m/1.2αy (5)
where V is the least design horizontal seismic force; W is the total superstructure and substructure dead
load (V/W can be considered as the design coefficient); Z is the zoning acceleration coefficient (0.33G
for Seismic Zone 1A, 0.28G for Seismic Zone 1B, 0.23G for Seismic Zone B and 0.18G for Seismic
Zone C); Importance Factor with important bridge 1.2 and common bridge 1.0; αy is the initial
earthquake amplification factor; C is the normalized acceleration response spectrum coefficient; Fu is
the structural system seismic reduction factor and (C/Fu)m can be considered as the adjusted
acceleration response spectrum coefficient.
It can be seen from the above evolution that bridge design based on seismic force has gone from a less
governing force to a main governing design force, especially with substructures. It is also evident from
site observation that very few bridges designed by (or satisfied) current codes (MTC/ROC 1995,
AASHTO 1998, AASHTO/LRFD 1998) were damaged by the 1999 earthquake.
Study of the Failure Modes
921 earthquake severely damaged Wushi bridge. This bridge may give the impression of being
rather straightforward structure system, composed of superstructures, substructures and foundations.
However, the bridges damaged by this earthquake show some complex failure modes and they are
discussed as follows:
(1) Abutment and Wingwall Failure: Potential slumping at bridge abutments has traditionally been
ignored during design in Taiwan. As far as earthquakes are concerned, there are two types of bridge
abutments: integral abutment and seat-type abutment. The essential difference is that one (seat-type)
permits relative movement to occur between the superstructure and end support and the second type
(integral) does not. After filling the gap between the seat and the abutment back wall in Wushi bridge,
the backfill and road surface did not allow further movement. Impact of the superstructure generated
high passive pressure, resulting in rotation with damage to the top and pile supporting system. The
recommendation for minimizing the drop of the road elevation is to be able to install approach slabs.
Simple span reinforced concrete slabs can be designed with a minimum length of 3 meters for the
approach slab. Figure 7 shows clearly the back-fill settlement. It has also been learned that abutment
back-wall, back-fill and approach slabs should be carefully designed and constructed to prevent
collapse.
(2) Substructure damage: Wall piers and hammerhead single column piers are used in substructures on
north- and southbound, respectively. It is well known that it is economically feasible to design
columns to yield and dissipate significant amounts of energy and to perform in a ductile manner.
However, with relatively short columns and much stiffness in the transverse direction, shear cracks
appeared on piers P1S, P2S, P3S, P5S, P7S, P8S. Typical shear crack shows in Figure 8.
Shaft foundation was used in Wushi bridge to support piers. Since the fault line crosses between piers
2 and 3, parts of the foundation, especially ones close to the fault line, were severely damaged and
some settled (Fig. 10).
(3) Superstructure Failure: The main reason for the severe superstructure failure was the major fault
movement in the horizontal and vertical directions. Since the fault line intersects the bridge at 45
degrees, the superstructure pushed and pulled in the longitudinal direction and, in the mean time,
impacted upon the shear blocks in the transverse direction. Due to large movements, the bridge
collapsed with fallen beams on the first two spans. Span failure in this case was due to unseating,
which includes pier tilting, insufficient seat length and inadequate restraining force capacity.
Conclusion
Most of the observations briefly addressed in this paper were collected from the field survey
conducted by the authors. In general, it may be noticed that most of the damage and collapses arise from
design or large movement near the fault line rather than from construction or material inadequacy. As
stated earlier in this paper, an overview of the causes of damage and collapse from this earthquake may
offer reliable guidelines on the deficiencies in bridge design and considerations for improving code
provisions later.
Reference
AASHTO, "Standard Specifications for Highway Bridges," 16th Edition, 1996, with Interims up to 1998,
American Association of State Highway and Transportation Officials, Washington, DC
AASHTO, "AASHTO LRFD Bridge Design Specifications," 2nd Edition, 1998, American Association of
State Highway and Transportation Officials, Washington, DC
The Department of Transportation (DOT/ROC) “Highway Bridge Engineering Design Specifications,"
Taiwan, the Republic of China, 1960.
The Ministry of Transportation and Communication (MTC/ROC) "The Design Specifications for Highway
Bridges," Taiwan, the Republic of China, 1987.
The Ministry of Transportation and Communication (MTC/ROC) "The Seismic Design Specifications for
Highway Bridges," Taiwan, the Republic of China, 1995.
National Science Foundation (NSF/ROC) "Summarized Report on Damaged Bridges due to Chi-Chi
Earthquake, 1999/9/21," Taiwan, the Republic of China, December 1999.
Figure 1 Epicenter and Active Faults in Taiwan
Figure 2 - Wushi Bridge, Epicenter, Chelungpu Fault line and Surrounding Stations
(Ref: Central Weather Bureau, The Ministry of Transportation and Communication, Taiwan, ROC)
Figure 3 - Plan and Elevation of Wushi Bridge
Figure 4 - Cross Sections of Wushi Bridge
Figure 5 - Oblique Fault Movement between Piers P3N and P3S
Figure 6 - Fallen Beam due to Superstructure Movement in the Longitudinal Direction at Pier P2N
Figure 7 - Backfill Upward Movement at the North Abutment of the Northbound Bridge
Figure 8 - Shear Cracks on the Southbound Hammerhead Concrete Pier p2S
Figure 9 - Crushed End Diaphragms and Shear Blocks at Pier P1N
Figure 10 - Pier Settlement and Shaft Shear Cracks at Pier P3N
Figure 1 Epicenter and Active Faults in Taiwan
Figure 2 - Wushi Bridge, Epicenter, Chelungpu Fault line and Surrounding Stations
(Ref: Central Weather Bureau, The Ministry of Transportation and Communication, Taiwan, ROC)
Figure 3 - Plan and Elevation of Wushi Bridge
Figure 4 - Cross Sections of Wushi Bridge
Figure 5 - Oblique Fault Movement between Piers P3N and P3S
Figure 6 - Fallen Beam due to Superstructure Movement in the Longitudinal Direction at Pier P2N
Figure 7 - Backfill Upward Movement at the North Abutment of the Northbound Bridge
Figure 8 - Shear Cracks on the Southbound Hammerhead Concrete Pier p2S
Figure 9 - Crushed End Diaphragms and Shear Blocks at Pier P1N
Figure 10 - Pier Settlement and Shaft Shear Cracks at Pier P3N
