Вы находитесь на странице: 1из 15

Structure and Infrastructure Engineering

Maintenance, Management, Life-Cycle Design and Performance

ISSN: 1573-2479 (Print) 1744-8980 (Online) Journal homepage: https://www.tandfonline.com/loi/nsie20

Fragility analysis of tall pier bridges subjected to


near-fault pulse-like ground motions

Xu Chen, Jianzhong Li & Zhongguo Guan

To cite this article: Xu Chen, Jianzhong Li & Zhongguo Guan (2019): Fragility analysis of tall pier
bridges subjected to near-fault pulse-like ground motions, Structure and Infrastructure Engineering,
DOI: 10.1080/15732479.2019.1683589

To link to this article: https://doi.org/10.1080/15732479.2019.1683589

Published online: 04 Nov 2019.

Submit your article to this journal

Article views: 47

View related articles

View Crossmark data

Citing articles: 1 View citing articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=nsie20
STRUCTURE AND INFRASTRUCTURE ENGINEERING
https://doi.org/10.1080/15732479.2019.1683589

Fragility analysis of tall pier bridges subjected to near-fault pulse-like


ground motions
Xu Chena,b, Jianzhong Lic and Zhongguo Guanc
a
Shanghai Institute of Applied Mathematics and Mechanics, Shanghai University, Shanghai, China; bDepartment of Civil Engineering,
Shanghai University, Shanghai, China; cState Key Laboratory of Disaster Reduction in Civil Engineering, Tongji University, Shanghai, China

ABSTRACT ARTICLE HISTORY


More than 40% of bridges in Southwest China have piers over 40 m height. The seismic behaviour of Received 1 February 2019
these bridges is critical, especially exposed to near-fault pulse-like ground motions. This paper con- Revised 27 June 2019
ducts fragility analyses to assess the seismic vulnerability of tall pier bridges under near-fault motions. Accepted 25 July 2019
Fragility curves are developed for a typical 50-m tall pier and a 10 m conventional pier with probabilis-
KEYWORDS
tic seismic demand models (PSDMs). When developing the PSDMs, curvature ductility at pier base is Tall pier bridges; near-fault
used as the demand parameter; the 5% damped spectral acceleration of the fundamental period motions; seismic
(Sa(T1, 5%)) is employed as an intensity measure of the motions. Results show that the 50 m pier is performance; fragility
less vulnerable than the 10 m pier under both the near-fault and far-field motions and is unlikely to analysis; vulnerabil-
collapse in real earthquake events. Furthermore, the seismic fragility of the 50 m pier is found highly ity assessment
related to the pre-defined damage state capacities, indicating that the damage states should be
carefully specified to properly estimate the vulnerability of tall piers. Finally, analyses are conducted
for piers with heights of 40, 60, 70, and 80 m, extending the conclusions obtained from 50 m pier to a
class of tall piers.

1. Introduction analysis procedure. Besides the higher mode effects observed


in previous studies, the authors also pointed out that the
Numerous highway bridges have been built or are under
displacement at the pier top did not respond in phase with
construction in the mountainous areas of Southwest China
the section curvature at the pier base for tall pier bridges.
in recent years. Because of the rugged topography of the
This fact implies that the displacement at the pier top is an
region, more than 40% of these bridges contain piers with a
invalid damage index for tall piers and that the section
height over 40 m, with some exceeding 100 m. Southwest curvature should be used instead.
China is known as a region of high seismicity, experiencing The aforementioned conclusions from numerical analyses
many strong earthquakes (e.g., the magnitude 8.0 were then validated by the shake table tests for tall pier sys-
Wenchuan earthquake in 2008, and the magnitude 7.0 tems conducted in Tongji University (Chen, Guan, Li, &
Lushan earthquake in 2013). As these bridges with tall piers Spencer, 2018). Using a typical near-fault ground motion
are key links in the local highway network, their seismic (Rinaldi motion from 1994 Northridge earthquake) as input,
performance and safety is of great concern. the results showed the formation of additional plastic hinge
For tall pier bridges, the large distributed pier mass can at mid-height, and the appropriacy of utilising section
lead to considerable seismic inertia forces; furthermore, curvature as damage index. The results of these studies also
these tall piers are flexible, resulting in long fundamental demonstrated that the seismic intensity for ultimate or col-
periods which are out of the range of the dominant energy lapse state of tall piers was quite large and unlikely to be
content of many seismic events. However, higher modes reached during an earthquake event. Therefore, only focus-
may still make a significant contribution to the response ing on ultimate state was insufficient in engineering practice
(Chopra, 2001). Some numerical investigations (Guan, Li, for tall pier bridges; more damage states should be incorpo-
Xu, & Lu, 2011; Tubaldi, Tassotti, Dall’Asta, & Dezi, 2014) rated to better characterise their seismic behaviour.
have proved that due to the higher-order modes of pier col- Note that all these aforementioned studies in last para-
umns, the distribution patterns of seismic forces differ sig- graph on tall piers mainly concentrated on the seismic per-
nificantly from those for conventional piers; and an formance corresponding to far-field ground motions, on
additional plastic hinge might form at the mid-height of tall which the current seismic design are based. However, the
piers as well. Moreover, Chen, Li, and Guan (2016) dis- influence of near-fault pulse-like motions was not consid-
cussed the seismic performance of tall piers from the initial ered, which has been recognised as the most severe and dev-
yielding to the ultimate state using an incremental dynamic astating seismic loadings that structures undergo. To date,

CONTACT Zhongguo Guan guanzhongguo@tongji.edu.cn


ß 2019 Informa UK Limited, trading as Taylor & Francis Group
2 X. CHEN ET AL.

investigations have shown that due to the directivity the seismic vulnerability of tall pier bridges, especially sub-
(Somerville, 2000) and fling effect (Wang, Zhou, Zhang, & jected to near-fault motions, is not found in the literature.
Igel, 2002), near-fault motions are mainly characterised by This paper focuses on developing fragility-based seismic
significant velocity pulses (Mavroeidis & Papageorgiou, vulnerability assessment for tall pier bridges when subjected
2002). These features cause a moderate descending portion to near-fault motions. Four damage states are first defined
of acceleration and velocity spectra of these pulse-like for tall piers, and the capacity for each damage state is rep-
motions, leading to greater seismic demands for structures resented in the form of maximum section curvature ductility
with long period such as isolated bridges (Cao & Li, 2019; along the pier height. Then, using nonlinear time history
Li & Cao, 2015) and tall pier bridges (J onsson, Bessason, & analysis (NTHA) procedure, the PSDMs are established for
Haflidason, 2010). both a typical bridge with piers of 50 m height located in
Park et al. (2004) investigated the seismic performance of Southwest China and a conventional bridge with 10 m pier
the Bolu Viaduct in the Duzce, Turkey, earthquake of when subjected to near-fault ground motions; while the
November 1999. The authors pointed out that due to the same procedure is conducted for a suite of far-field input
near-fault motions, excessive relative displacement between motions for comparison purpose. The results of both piers
the superstructure and the pier was observed at an early are presented in a probabilistic format of fragility curves
stage of the earthquake, causing damage to the bearings and based on the PSDMs, focusing on the seismic vulnerability
energy dissipation units. Furthermore, with larger modal of the 50 m pier for near-fault motions. Similar analyses are
acceleration values of higher modes, more significant conducted for tall piers with height of 40, 60, 70, and 80 m,
higher-mode effects might be observed in the case of tall respectively, extending the conclusions obtained from 50 m
piers subjected to these near-fault motions. Thus, studies are pier to a class of tall piers. Finally, conclusions of this study
required investigating the seismic performance of tall pier are presented.
bridges located in near-fault regions of South-west China.
Seismic fragility analysis is an essential tool for investigat- 2. Bridge prototype and analysis model
ing the vulnerability of bridge structures, offering a means
of communicating the probability of damage over a range of 2.1. Bridge prototype
potential earthquake ground motion intensities (Padgett & The prototype bridge considered herein is identical to a pre-
DesRoches, 2008). To develop fragility functions, the rela- vious study (Chen et al., 2018), as shown in Figure 1. In
tionship between peak seismic responses and ground motion this four-span continuous bridge, the joints are socket con-
intensities is required, which is provided through probabilis- nected to the pier top and thus provide hinge-like connec-
tic seismic demand models (PSDMs). To obtain enough tions between the girders and the piers. Each bridge pier is
data for PSDMs, the structural demands and/or capacities composed of variable hollow sections with external dimen-
can be estimated through various analytical methods such as sions from 3.35 m  5.0 m to 2.1 m  5.0 m and wall thick-
elastic spectral (Jernigan & Hwang, 2002; Shinozuka, Feng, ness of 0.6 m; detailed information of the cross section is
Kim, & Kim, 2000), nonlinear static (Dutta & Mander, presented in Figure 2. More information about this typical
1998), and linear/nonlinear time-history (Kunnath, Larson, tall pier bridge in Southwest China could be referred to
& Miranda, 2006; Mackie & Stojadinovic, 2001). Chen et al. (2018).
An extensive amount of fragility analysis has been con- To better understand the fragility of the tall piers, a typ-
ducted to assess the vulnerability of bridges structures to ical conventional bridge constructed in Southwest China
date. For example, Kim and Feng (2003) conducted fragility with piers of short-to-medium height was used as a refer-
analysis for highway bridges subjected to ground motions ence case for comparison as well. The superstructure of this
with spatial variation. The authors found that ductility conventional bridge was the same as that of the tall pier
demands of bridge columns could be underestimated if the bridge; however, the height of the pier was 10 m with uni-
bridge was analysed using identical support ground motions form rectangle section as shown in Figure 2, where the unit
rather than differential support motions. More recently, of steel diameter is millimetre. The axial compression ratio
Muntasir Billah and Alam (2015) conducted fragility ana- at the base section of the 10 m pier was about 0.10, which is
lysis for concrete bridge piers whose plastic hinge region similar that of the 50 m pier (0.094); and the longitudinal
was reinforced with super-elastic shape memory alloy steel ratio of the base section is 1.37%, a value that is com-
(SMA) considering various demand parameters. The results monly used in high seismicity regions of Southwest China
showed that the vulnerability of bridge piers was signifi- and the same as that of the 50 m pier.
cantly affected by the selection of the seismic demand
parameters (EDP); when displacement ductility was
2.2. Analysis model
employed as EDP, the pier reinforced with SMA was more
vulnerable while the opposite phenomenon was observed To focus on the fragility of piers, the prototype bridge was
when residue drift was used. However, all these fragility simplified and represented as a single mass-column system
analyses are applied to conventional bridges with short- as shown in Figure 3. In this single column model, the
to-medium piers subjected to far-filed motions on which tributary mass (M ¼ 723 tons) from the two-adjacent half
design specifications are based. To date, fragility analyses for spans of the superstructure (see Node 1 in Figure 3(a)) was
STRUCTURE AND INFRASTRUCTURE ENGINEERING 3

30 30 30 30
Cast-in-situ joint

2.10 Precast Precast


girder girder

Rubber pads
47 45 49 Investigated 48 46 Socket of
portion shear key
Pier column

(reinforcement and tendons are not shown)


3.35
Details for girder-pier connection

Figure 1. Elevation of the prototype bridge (Unit: m).

500
500
150
Longitudinal rebars
Φ32 @ 15
60
Longitudinal rebars
210
350

150
60 Φ32 @ 15 Longitudinal rebars
Φ24 @ 8
Rectangular ties Φ12
60
Rectangular ties Φ10
60

Rectangular stirrups Φ10


Rectangular ties Φ12 @ 10 cm inside intensified region
@ 20 cm outside intensified region
Rectangular stirrups Φ12
@ 10 cm inside intensified region
@ 20 cm outside intensified region
(c) Cross section of
(a) Base section of tall pier (b) Top section of tall pier conventional pier
Figure 2. Cross sections of the 50 and 10 m pier (Unit: cm).

Stress
Node 1
(e)
element 1 εu ε0
Node 2 Cross Strain
element 2 Section
fpcu

Beam-column
element (c) fpc
E0=2fpc/ε0
Node i Longitudinal
Stress

steel fibers
Node i+1
Unconfined cover Fy b×E0
concrete fibers E0

Confined core Strain


Node N concrete fibers E0
element N (b)
(a) (d) b×E0 Fy
(f)
Figure 3. Numerical model: (a) cantilever beam-column; (b) scheme of fibre-element model; (c) fiber section; (d) Discretisation of fibres; (e) constitutive relationship
of concrete; (f) constitutive relationship of steel.

concentrated at the pier top. The distributed pier masses The finite element models for both the tall pier and the
were lumped at the nodes of column elements, ranging conventional pier were developed based on the OpenSees
from 55 to 94 tons. Soil–foundation interaction effects were platform. To consider the nonlinear behaviour of the
neglected in current model. Similarly, the conventional reinforced concrete piers, force-based fibre beam-column
bridge was also simplified and modelled as a single column elements were adopted (Taucer, Spacone, & Filippou, 1991).
system for comparison. The cross sections were divided into concrete and steel
4 X. CHEN ET AL.

Table 1. Characteristics of the near-fault ground motions.


Sa(T1, 5%) (g)
No. Earthquake Year Magnitude Epicentral distance (Rjb, km) PGA (g) 50 m pier 10 m pier
NF1 Northridge-01 1994 6.69 0 0.874 0.473 1.952
NF2 San Fernando 1971 6.61 0 1.238 0.408 1.154
NF3 Tabas_ Iran 1978 7.35 1.79 0.871 0.447 0.875
NF4 Imperial Valley-06 1979 6.53 0 0.287 0.184 0.359
NF5 Imperial Valley-06 1979 6.53 8.54 0.220 0.110 0.295
NF6 Imperial Valley-06 1979 6.53 7.31 0.235 0.244 0.383
NF7 Imperial Valley-06 1979 6.53 0.07 0.317 0.287 0.513
NF8 Imperial Valley-06 1979 6.53 8.6 0.232 0.234 0.233
NF9 Imperial Valley-06 1979 6.53 10.79 0.267 0.276 0.357
NF10 Imperial Valley-06 1979 6.53 4.9 0.370 0.332 0.426
NF11 Imperial Valley-06 1979 6.53 1.76 0.529 0.283 0.528
NF12 Imperial Valley-06 1979 6.53 0 0.447 0.424 0.499
NF13 Imperial Valley-06 1979 6.53 0.56 0.469 0.443 0.938
NF14 Imperial Valley-06 1979 6.53 5.09 0.481 0.185 0.541
NF15 Imperial Valley-06 1979 6.53 5.35 0.258 0.154 0.355
NF16 Irpinia_ Italy-01 1980 6.9 8.14 0.190 0.187 0.305
NF17 Irpinia_ Italy-01 1980 6.9 6.78 0.321 0.313 0.423
NF18 Superstition Hills-02 1987 6.54 0.95 0.432 0.689 0.945
NF19 Loma Prieta 1989 6.93 10.27 0.285 0.173 0.418
NF20 Loma Prieta 1989 6.93 10.38 0.370 0.110 0.296

fibres, as shown in Figure 3(c,d), and the concrete fibres near-fault motions as reported (Xu & Gardoni, 2016).
were further discretised as confined (core fibres) and uncon- Moreover, the Sa(T1, 5%) values of the 10 m pier are larger
fined (cover fibres) ones. The constitutive relationship for than those of the 50 m pier and the PGAs for most motions
the concrete material was calculated according to the as well.
Kent–Scott–Park material law (Figure 3(e)) (Scott, Park, & Note that amounts of investigations have demonstrated
Priestley, 1982). The unconfined compressive strength of the the significance of avoiding bias in structural response
concrete was set as 26.8 MPa, whereas the confined com- during seismic analysis to date, selecting ground motions
pressive strength was calculated as 28.5 MPa. A uniaxial based on epsilon (e) has been shown a promising method
bilinear relationship with kinematic hardening and optional achieving this aim (Baker & Cornell, 2006; Kazemi,
isotropic hardening was adopted for the steel fibers Ghafory-Ashtiany, & Azarbakht, 2013, 2017; Mousavi,
(Guirguis & Mehanny, 2013); the steel’s behaviour (Figure Ghafory-Ashtiany, & Azarbakht, 2011). To validate the
3(f)) is characterised by an initial linearly elastic portion of selected ground motions, the CMS-e (conditioned mean
the stress–strain relationship, with a modulus of elasticity of spectrum considering the effects of e) proposed by Baker
E0 ¼ 200 GPa, up to the yield stress Fy ¼ 400 MPa, followed (2011) is employed developing the target spectra for near-
by a subsequent region of strain hardening (b ¼ 0.02). fault and far-field motions, respectively. The conditional
mean spectra are established based on seismic hazard disag-
gregation results and ground motions models of Southwest
3. Selection of ground motions China (Lei, Gao, & Yv, 2007), and the period of interest
Two suites of ground motion records from the Pacific (T) is defined as the first mode period (T1 ¼ 2.1 s).
Earthquake Engineering Research Centre (PEER) database Detailed steps of developing the conditional mean spectra
were selected and used in this study. One suite is composed can be referred to in Baker’s work (2011).
of 20 near-fault motions, while the other contains 20 far- Figure 4 displays the spectra of the selected motions and
field motions; the detailed information of these two record the conditional mean spectra, for 50 m pier in the case of
sets is listed in Tables 1 and 2, respectively. both near-fault and far-field motions; the situation of 10 m
As shown in Table 1, the near-fault motions are selected pier is similar. In this figure, the near-fault and far-field
with epicentral distances (Rjb) less than 15 km, and magni- motions are scaled with Sa(T) ¼ 0.3 and 0.026 g, respect-
tude between M6.5 and M7.5. Table 1 shows that the PGA ively, which are the target values utilised in CMS-e method.
levels range from 0.19 to 1.24 g, and the Sa(T1, 5%) corre- Figure 4 shows that the selected ground motions match well
sponding to the 50 m pier and 10 m pier model range from with the conditional mean spectra obtained from CMS-e
0.11 to 0.69 g and 0.23 to 1.95 g, respectively. Note that the method for both near-fault and far-field motions. This phe-
Sa(T1, 5%) of the tall pier are observed much lower than the nomenon indicates that no significant bias is observed with
PGAs for most motions, due to the inherent long funda- these selected motions, and they could be utilised for fragil-
mental period of the tall pier. On the other hand, Sa(T1, ity analysis.
5%) values of the 10 m-conventional pier are similar to or
larger than the PGA levels, and significantly greater com-
4. Demand models for investigated bridges
pared with those corresponding to the 50 m pier.
For the far-field motions (Table 2), the epicentral distan- In current seismic fragility assessment for bridges, different
ces are over 50 km, and the PGA levels range from parameters (Alam, Bhuiyan, & Billah, 2012; Hwang, Liu, &
0.6  10–2 to 0.136 g, which are generally lower than the Chiu, 2001; Kowalsky, 2000; Muntasir Billah & Alam, 2015;
STRUCTURE AND INFRASTRUCTURE ENGINEERING 5

Table 2. Characteristics of the far-field ground motions.


Sa(T1, 5%) (g)
No. Earthquake Year Magnitude Epicentral distance (Rjb, km) PGA (g) 50 m pier 10 m pier
FF1 San Fernando 1971 6.61 61.75 0.052 0.007 0.008
FF2 San Fernando 1971 6.61 61.75 0.026 0.004 0.007
FF3 Taiwan SMART1 1983 6.5 91.54 0.006 0.003 0.010
FF4 Taiwan SMART2 1983 6.5 91.54 0.020 0.003 0.018
FF5 Taiwan SMART3 1986 7.3 51.35 0.052 0.035 0.045
FF6 Taiwan SMART4 1986 7.3 51.35 0.136 0.032 0.141
FF7 Loma Prieta 1989 6.93 79.16 0.049 0.019 0.190
FF8 Loma Prieta 1989 6.93 79.16 0.039 0.019 0.084
FF9 Loma Prieta 1989 6.93 79.71 0.125 0.038 0.324
FF10 Loma Prieta 1989 6.93 54.97 0.049 0.023 0.056
FF11 Loma Prieta 1989 6.93 54.97 0.050 0.019 0.061
FF12 Loma Prieta 1989 6.93 71.23 0.098 0.042 0.110
FF13 Loma Prieta 1989 6.93 71.23 0.113 0.059 0.155
FF14 Loma Prieta 1989 6.93 76.4 0.036 0.012 0.050
FF15 Loma Prieta 1989 6.93 76.4 0.077 0.030 0.098
FF16 Loma Prieta 1989 6.93 75.07 0.029 0.018 0.054
FF17 Loma Prieta 1989 6.93 75.07 0.068 0.059 0.076
FF18 Landers 1992 7.28 144.13 0.026 0.018 0.047
FF19 Landers 1992 7.28 144.13 0.028 0.023 0.035
FF20 Landers 1992 7.28 50.85 0.050 0.017 0.033

4 0.5
Sepctral Acceleration (g)

Imperial Valley Sepctral Acceleration (g) Loma Prieta


0.4
3 Other events Other events
Conditional mean spectrum Conditional mean spectrum
0.3
2
0.2
Sa(T*) = 0.3 g Sa(T*) = 0.026 g
1
(a) 0.1 (b)
0 0.0
0 1 2 3 4 0 1 2 3 4
Periods (s) Periods (s)
Figure 4. Conditional mean spectrum for Southwest China (with T ¼ 2.1 s, the fundamental period of 50 m pier) and response spectra of selected motions: (a)
near-fault motions; (b) far-field motions.

Padgett & DesRoches, 2008; Zhang & Huo, 2009) have been cloud approach, a suite of ground motions with various inten-
used to measure the seismic demands of piers, such as drift, sities are selected and used as input for bridge structures.
residual drift and displacement ductility, etc. Note that all While in the scaling approach, each of selected motions is
these damage measures are related to the displacement at scaled up and down to generate an ensemble of motions with
the pier top, which has already been shown to be an invalid various intensities. Series of NTHA procedures are then con-
damage index for tall piers (Chen et al., 2016; Guan et al., ducted to compute the structural demands at each intensity
2011). On the other hand, the maximum response of section level, followed by regression analyses constructing the PSDMs.
curvature is most widely accepted and used damage measure To incorporate the seismic responses of pier columns from
for these tall pier bridges (Ceravolo, Demarie, Giordano, linear-elastic to nonlinear-plastic range, scaling factors are
Mancini, & Sabia, 2009; Chen et al., 2018). Thus, seismic selected from 0.5 to 5.0, covering a wide range of ground
demand of the 50-m tall pier is represented in the form of motion intensities. Furthermore, to provide sufficient data
curvature ductility (l/ ), which is defined as sets for analysis, 10 scaling factors equally distributed between
0.5 and 5.0 (i.e., 0.5, 1.0, 1.5 … 4.5, 5.0) are generated and
/ applied to each of the selected ground motions listed in Tables
l/ ¼ (1)
/y 1 and 2. Thus, for both the cases of far-field and near-fault
motions, 200 data sets were eventually generated.
where / is the curvature demand, and /y is the curvature
In a PSDM, the median of seismic demand (Sd) and the
for the first yielding of longitudinal steel bars, which can be
ground intensity measures (IM) are assumed to follow a
obtained from the moment–curvature analysis.
logarithmic correlation according to the power-law function
In current study, two methods, namely cloud approach and
(Cornell, Jalayer, Hamburger, & Foutch, 2002), which can
scaling approach (Billah, Alam, & Bhuiyan, 2013), are used to
be expressed as
provide sufficient data for the development of PSDMs, In the
6 X. CHEN ET AL.

6 6.0
10 m pier (a) 10 m pier (b)
y = 1.4645x + 2.968 4.5 y = 1.482x + 2.596
R2 = 0.737 R2 = 0.806
3 3.0
ln( )
1.5

0 0.0
50 m pier 50 m pier
y = 1.336x + 1.397 -1.5 y = 1.385x + 1.911
R2 = 0.763 R2 = 0.815

-3 -3.0
-3 -2 -1 0 1 2 -4 -3 -2 -1 0 1 2 3 4
ln (PGA) ln (Sa(T1, 5%))
Figure 5. Comparison of the PSDMs of 50 and 10 m piers subjected to near-fault motions for (a) PGA-based; (b) Sa(T1, 5%)-based.

Sd ¼ aIMb orlnðSd Þ ¼ lnðaÞ þ blnðIMÞ (2) leads to inherently long fundamental period of the 50 m
pier (T1 ¼ 2.15 s), which is out of the range of the dominate
where a and b are unknown regression coefficients, which
energy content of many seismic events (as shown Figure 4).
can be obtained from a regression analysis. With the
Thus, the demand of curvature ductility at the base of tall
assumption that Sd follows the lognormal distribution
piers, which is mainly dominated by the first vibration
(Padgett & DesRoches, 2008), the dispersion of the seismic
mode, is generally smaller than that of the 10 m pier, whose
demand conditioned on IM, bDjIM , can be computed as (
fundamental period is 0.92 s. Furthermore, the 10 m pier
Muntasir Billah & Alam, 2015)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi also shows a larger dispersion than the 50 m pier, which is
PN presented by both a smaller R2 in Figure 5(a) and a larger
ð Þ
i ðln D lnðaIM ÞÞ
b 2
bDjIM ¼ (3) bDjIM in Table 3.
N2 From Figure 5(b), where Sa(T1, 5%) is employed as IM,
where D is the seismic demands obtained from NTHA pro- similar conclusions can be obtained when comparing the
cedures for each input motion; N is the number of total results for the 50 and the 10 m pier. Note that the range of
simulation cases. horizontal axis values, i.e. lnðSa ðT1 , 5%ÞÞ, for two pier
For the establishment of PSDMs, various parameters can models are different in this figure. The reason is that, even
be used as IMs. While the spectral acceleration at funda- subjected to the ground motion with the same PGA level,
mental period for a damping ratio of 5% (Sa(T1, 5%)) is the Sa(T1, 5%) of two piers will be different due to different
widely recognised as an efficient IM (Shome, Cornell, fundamental period, as listed in Table 3. On the other hand,
Bazzurro, & Carballo, 1998; Tothong & Luco, 2007), Mackie for the PSDM corresponding to far-field motions (Figure 6),
and Stojadinovi¢ (2001) and Padgett and DesRoches (2008) the 10 m pier is observed with higher median values and
employed PGA in their investigations as well. Thus, this larger dispersions compared with the 50 m pier as well,
study employs both PGA and Sa(T1, 5%) as IMs for the con- using either PGA or Sa(T1, 5%) employed as IM.
struction of PSMDs, and the one that meets the criteria of Comparisons between the two PSDMs developed for
efficiency and sufficiency (Luco & Cornell, 2007; Padgett, 50 m pier subjected to near-fault motions (see in Figure
Nielson, & DesRoches, 2008) is then selected for the follow- 5(a,b) and Table 3) show that when Sa(T1, 5%) is employed
ing analysis developing fragility curves. as IM, the regression results yields less dispersion; i.e.,
Figure 5(a) depicts and compares the PSDMs developed Sa(T1, 5%) is more efficient in current study compared with
on the basis of section curvature ductility (l/ ) for the 50 PGA. For example, the R2 of the PGA-based regression line
and 10 m pier subjected to near-fault ground motions, with is 0.763, while that of the Sa(T1, 5%)-based one is 0.815, and
PGA employed as intensity measure; Figure 5(b) presents the bDjIM for PGA-based and Sa(T1, 5%)-based results are
the corresponding results for the two pier models using 0.636 and 0.562, respectively. Similarly, in the case of 10 m
their Sa(T1, 5%) values as intensity measure. Similarly, the pier, when PGA and Sa(T1, 5%) are used as IMs, the R2 val-
PSDMs developed for two pier models subjected to far-field ues are 0.737 and 0.806, respectively, and the bDjIM are
motions are presented and compared in Figure 6(a,b), using 0.747 and 0.642, respectively. Furthermore, when Sa(T1, 5%)
the PGA and Sa(T1, 5%) as IMs, respectively. The parame- is used as IM, the values of b are 1.385 and 1.482 for the 50
ters for Equation (2), i.e. ln(a) and b, are computed through and 10 m pier, respectively, larger than those corresponding
regression analyses and listed in Table 3 for each case, as to PGA (1.336 and 1.465). This result shows that Sa(T1, 5%)
well as the dispersion (bDjIM ) obtained with Equation (3). is a more sufficient IM than PGA as well.
When subjected to near-fault ground motions with PGA This fact indicates that for both the two pier mod-
as IM (Figure 5(a)), the 10 m pier yields higher median val- els, lnðl/ Þ possesses a better linear relation with
ues of curvature ductility than the 50 m pier. This phenom- lnðSa ðT1 , 5%ÞÞ, rather than lnðPGAÞ: The reason might be
enon is reasonable that the significant structural flexibility that Sa(T1, 5%) directly presents the spectrum acceleration
STRUCTURE AND INFRASTRUCTURE ENGINEERING 7

Figure 6. Comparison of the PSDMs of 50 and 10 m piers subjected to far-field motions for (a) PGA-based; (b) Sa(T1, 5%)-based.

Table 3. Parameters of PSDMs for various cases.


PGA as IM Sa(T1, 5%) as IM
Input motions Pier height ln(a) b bDjIM ln(a) b bDjIM
Near-fault 50 m 1.397 1.336 0.636 1.911 1.385 0.562
10 m 2.968 1.465 0.747 2.596 1.482 0.642
Far-field 50 m 0.817 1.120 0.699 1.852 1.118 0.485
10 m 3.202 0.992 0.797 3.259 0.868 0.695

Table 4. Damage states for both 50 and 10 m piers.


can be utilised for the tall pier as well. From previous inves-
Damage state Performance level Curvature ductility (l/ )
tigations about conventional piers, l/ at pier base is corre-
Slight Minor Cracking and spalling 1.70 > l/ > 1
lated to the displacement ductility (ld ) at the pier top,
Moderate Moderate cracking and spalling 3.67 > l/ > 1.70 which can be computed by the following equation:
Extensive Degradation 14.19 > l/ > 3.67    
3 /y H þ H 2 lp lp l/ 1 /y
1 2 1
Collapse Collapse l/ > 14.19 D
ld ¼ ¼ (5)
Dy 3 /y H
1 2

corresponding to the first vibration mode, which dominates where D and Dy denote displacement demand and yielding
the seismic response of section curvature at the pier base, displacement at the pier top, respectively; H represents the
where this parameter reached the maximum value. pier height; lp is the equivalent length of plastic hinge at
Consequently, the fragility analysis of both the 50 and 10 m pier base region. Considering lp approximately as 0.1 H
pier, subjected to either near-fault or far-field ground (Berry, Lehman, & Lowes, 2008), the curvature ductility can
motions, are conducted with the PSDMs using Sa(T1, 5%) as be expressed as
intensity measure in the following parts. ld 1
l/ ¼ þ1 (6)
0:285
As presented in previous research (Hwang et al., 2001),
5. Estimates of capacity for investigated bridges
the range of displacement ductility of conventional piers
Definition of different damage limit states and their quanti- corresponding to the four damage states are 1 < ld < 1.20,
tative measures is critical to assess the seismic fragility of 1.20 < ld <1.76, 1.76 < ld < 4.76, and 4.76 < ld , respect-
bridges. Damage states of a bridge should be defined in ively. Consequently, the damage states of tall piers defined
such a way that they represent the bridge functional per- in terms of l/ can be obtained using Equation (6), as
formance during an earthquake. In this study, four perform- shown in Table 4. Note that the damage states of 10 m pier
ance levels as defined in HAZUS-MH (FEM, 2008) in terms are also represented in the form of l/ for comparison pur-
of observed experimental responses, namely slight, moder- pose in this paper, although it can be characterised by dis-
ate, extensive, and collapse, are adopted for the investigated placement-based indexes.
piers as presented in Table 4. Table 5 presents the capacities of each damage state for
Note that the curvature ductility (l/ ) is a performance the two bridge piers, where the median values (Sc) are com-
index in section level, which ought to be independent of the puted using Equation (6) and thresholds of ld recom-
structure type, i.e. the structural capacity expressed in terms mended in previous work of Hwang et al. (2001).
of this parameter will not be affected by such factors as pier Furthermore, the lognormal standard deviations (bc ) for
height, format of girders, etc. Therefore, l/ corresponding each damage state are shown in this table as well, which
to each damage state of conventional short-to-medium piers arises from the inherent randomness of system, i.e. the
8 X. CHEN ET AL.

Table 5. Limit capacities of two piers (Lee & Billington, 2011; Muntasir Billah of exceeding slight damage (see in Figure 7(a)) is 0.25 g,
& Alam, 2015; Nielson, 2005; Saiidi & Ardakani, 2012).
38.9% higher than that of the 10 m pier (0.18 g); while con-
Damage state Slight Moderate Extensive Collapse
sidering the extensive damage state as shown in Figure 7(c),
Sc 1 1.70 3.67 14.19
bc 0.25 0.25 0.46 0.46
the Sa(T1, 5%) of the 50 m pier for 50% probability (0.64 g)
is 56.1% higher than that of the 10 m pier (0.41 g). And
when input PGA levels are considered, the values corre-
randomness of the material properties that affect strength. sponding to 50% probability of damage of 50 m pier can be
The capacities of the two piers are assumed to follow a log- even more than twice those of the 10 m pier for all damage
normal distribution as proposed by Nielson (2005).The val- states. These phenomena match the results plotted in Figure
ues of bc recommended by previous investigations (Lee & 5, where the 10 m pier yield higher seismic response than
Billington, 2011; Muntasir Billah & Alam, 2015; Nielson, the 50 m pier, and the discrepancies between the expected
2005; Saiidi & Ardakani, 2012) are used in this study and responses increase with the input Sa(T1, 5%) values.
listed in Table 5. Note that larger uncertainty is adopted for Similarly, Figure 8 compares the fragility curves of the
the extensive damage state (bc ¼ 0.46), while a smaller one 50 m pier when subjected to near-fault and far-field
is assumed for slight and moderate damage state (bc ¼ motions; the Sa(T1, 5%) values corresponding to 50% prob-
0.25). The reason for this assumption is that the higher ability of exceeding each damage state are presented as well.
damage states are associated with more aleatoric uncertainty This figure shows that at lower damage levels, namely slight
than the lower ones (Ramanathan, DesRoches, & and moderate damage, the 50 m pier is less vulnerable under
Padgett, 2012). near-field motions. For the slight damage, the ratio between
the Sa(T1, 5%) for 50% probability of damage corresponding
to the near-fault motions (0.25 g) and that of the far-field
6. Fragility assessment
motions (0.19 g) is about 1.32, while this ratio is 1.19
Fragility curves represent the ability of a structure to with- (0.37 g/0.31 g) in the case of moderate damage state. This
stand a specified event (Padgett & DesRoches, 2008; Tekie fact means that with the development of damage states, the
& Ellingwood, 2003), defining the possibility of a structure seismic vulnerability caused by near-fault motions increases
being damaged when subjected to a given level of IM; that more quickly than that corresponding to the far-
is, the conditional probability of the seismic demand (D) field motions.
exceeds corresponding structural capacity (C), as shown in When considering the extensive damage state (see in
the following equation: Figure 8(c)), the fragility curve of the 50 m pier under near-
fault motions almost coincide with that corresponding to
Fragility ¼ P½D  CjIM (7)
the far-field motions. Furthermore, for the collapse damage
Therefore, with the PSDMs and estimated capacities cor- shown in Figure 8(d), the 50 m pier is observed more vul-
responding to different damage states, the fragilities can be nerable under near-fault motions, on the opposite side to
generated by the following equation (Padgett & DesRoches, the results observed for slight and moderate damage states.
2008): The Sa(T1, 5%) for 50% probability of damage is 1.72 g in
0 1 this case when near-fault input motions are considered,
lnðSd Þ  lnðSc Þ
Fragility ¼ U@ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi A (8) while that for the far-field motions is over 2.0 g.
b2DjIM þ b2C These phenomena can be explained by referring to the
PSDMs plotted in Figure 9(a), where l/NF and l/FF present
where UðÞ denotes the standard normal cumulative distri- the curvature ductility demand due to the near-fault and
bution function (CDF). far-field motions, respectively. From this figure, the PSDMs
Utilizing the parameters from the capacity of each dam- corresponding to the two sets of ground motions are
age state as well as those from the PSDMs, the fragility observed with an intersection at Sa(T1, 5%) ¼ 0.801 g and
curves are generated and presented in Figure 7 for slight l/Inter ¼ 4.97, indicating that the near-fault and far-field
through collapse damage for both the 50 and 10 m pier ground motions will lead to identical seismic demands at
under near-fault ground motions. Sa(T1, 5%) values for 50% this point for the 50 m pier. For intensities Sa(T1, 5%) less
probability of exceeding each damage state are presented for than 0.801 g, the seismic demands from the far-field
each case, and the PGA levels corresponding to each Sa(T1, motions are observed greater than those from the near-fault
5%) value are also estimated by considering the averaged motions; otherwise, relatively larger seismic demands are
ratio between PGA and Sa(T1, 5%), which can be computed induced by the near-fault ground motions than the far-field
from Table 1. ground motions.
Figure 7 illustrates that the 50 m pier is seismically less This phenomenon can be attributed to the different char-
vulnerable compared to the 10 m pier for all the damage acteristics of near-fault and far field motions. Velocity pulses
states, especially the more severe ones, namely extensive and in near-fault ground motions tend to result in larger spectral
collapse, are considered; that is, higher input intensity (rep- accelerations in long-period range, while far-field ground
resented by Sa(T1, 5%) herein) is required for 50 m pier motions generally have larger spectral accelerations in short-
exceeding a certain damage state. For example, the Sa(T1, period range (Bray & Rodriguez-Marek, 2004; Dicleli, 2006).
5%) of the 50 m pier corresponding to the 50% probability This tendency is also shown in the near-fault and far-field
STRUCTURE AND INFRASTRUCTURE ENGINEERING 9

Figure 7. Fragility curves for 50 and 10 m piers subjected to near-fault motions for various damage states.

Figure 8. Fragility curves for 50 m pier when subjected to near-fault and far-field motions.
10 X. CHEN ET AL.

Figure 9. (a) Demand model and (b) fragility curves for 50 m pier subjected to two sets of motions.

ground motion groups used in this study (Figure 4). Thus, near-fault motions; while the comparisons for 50 m pier under
the structures could behave elastically at low intensities, and different ground motion sets are presented in Figure 10(b).
the far-field ground motions would likely result in larger The comparisons in these figures confirm previous results
seismic demands than the near-fault ground motions. When that: (a) the 50 m pier is less vulnerable than the 10 m pier
the structures undergo significant damage with increased when subjected to near-fault for all the four damage states; (b)
intensities, the mode periods are elongated, and the near- for both the slight and moderate damage states, the 50 m pier
fault ground motions would likely lead to larger seismic is more vulnerable under far-field motions; while these two
demands than the far-field ground motions. The change sets of motions lead to similar seismic fragility for extensive
consequently led to the fragility curves presented in Figure damage, and the tall pier is more vulnerable under near-fault
8(c) crossing each other. motions when the collapse state is considered.
Particularly, if the capacity of a damage state is exactly
set as l/Inter , the fragility curves of the 50 m pier will be
nearly the same when subjected to both ground motion sets,
7. Brief description of the results for tall piers with
various heights
which is shown in Figure 9(b). In this figure, the term
‘Inter’ denotes the damage state with capacity of l/Inter , and To extend the conclusions obtained from the 50 m pier
bC is assumed as 0.46. This fact shows that the assessment model to a class of tall piers, similar fragility analyses are
for the vulnerability of tall piers significantly depends on also conducted for tall pier models with heights of 40, 60,
the damage state capacity, which should be carefully speci- 70, and 80 m, respectively. These tall piers are subjected to
fied by the designers. ground motions listed in Tables 1 and 2 as well. Due to
Note that the 50 m pier model possesses extreme low vul- space limitations, the corresponding results are briefly sum-
nerability for the collapse damage state, subjected to either marised in this section and will not be discussed in detail.
near-fault or far-field motions. As shown in Figure 8(d), the Table 6 shows the Sa(T1, 5%) values of 50% probability
Sa(T1, 5%) for the 50% probability of collapse damage are of damage for each tall pier model when subjected to near-
1.72 g and over 2.0 g for near-fault and far-field motions, fault motions; the PGA levels corresponding to each Sa(T1,
respectively. By referring to Tables 1 and 2, the correspond- 5%), estimated with the average PGA/Sa(T1, 5%) ratios for
ing PGA levels are estimated over 4.0 g for some motions, each pier model, are presented for illustrating as well. Note
which are unlikely to occur in real earthquakes. that PGA directly depends on the input ground motions,
On the other hand, however, the extensive damage state while Sa(T1, 5%) is related to the structural dynamics (fun-
is possible to be exceeded, as the 50% probability Sa(T1, 5%) damental period, T1) in addition to motion characteristics.
are 0.64 and 0.61 g for near-fault and far-field motions, Thus, when comparing the results of tall pier systems with
respectively. Although the corresponding PGAs might be different heights, PGA serves a better index presenting the
over 1.5 g for some earthquake excitations, exceeding the input intensity of ground motions; i.e. a higher level of
values considered in current design specifications, quite a PGA indicates a stronger earthquake excitation, and vice
few earthquakes with particularly larger intensity were versa. From Table 6, all the tall piers with heights from 40
observed in recent years. For example, the maximum PGA to 80 m are observed with greater input PGA levels com-
levels were 1.82 and 0.96 g for the 1994 Northridge earth- pared with the 10 m pier for each state, when 50% probabil-
quake (Yegian, Gazetas, Dakoulas, Makris, & Ghahraman, ity of damage is considered. This fact indicates that when
1995) and 2008 Wenchuan earthquake (Krausmann, Cruz, subjected to the same near-fault ground motions, tall piers
& Affeltranger, 2010), respectively. Consequently, future are expected to be seismically less vulnerable.
research for tall piers might focus on the issues about the Table 6 also shows that the Sa(T1, 5%) and PGA values
seismic performance before final collapse. corresponding to 50% probability of damage for each pier
Figure 10(a) plots the median values (Sa(T1, 5%)) of model increase with pier height up to 50 m and then
the fragility curves for the 50 and 10 m pier subjected to decrease with higher pier heights. The reason for this
STRUCTURE AND INFRASTRUCTURE ENGINEERING 11

2.0 2.0
(a) 50 m pier (b) Near-fault
1.5 10 m pier 1.5 Far-field

Sa(T1, 5%) (g)


1.0 1.0

0.5 0.5

0.0 0.0
Slight Moderate Extensive Collapse Slight Moderate Extensive Collapse
Figure 10. Comparisons of median values of Sa(T1, 5%): (a) two piers subjected to near-fault motions; (b) 50 m pier subjected to both sets of motions.

Table 6. Sa(T1, 5%) and PGA values for 50 % probability of damage for each pier model under near-fault motions (Unit: g).
Damage state
Slight Moderate Extensive Collapse
Pier height (m) Sa(T1, 5%) PGA Sa(T1, 5%) PGA Sa(T1, 5%) PGA Sa(T1, 5%) PGA
10 0.18 0.14 0.25 0.20 0.50 0.40 1.03 0.82
40 0.19 0.24 0.29 0.37 0.56 0.71 1.79 2.28
50 0.25 0.34 0.37 0.50 0.64 0.86 1.72 2.32
60 0.072 0.32 0.11 0.49 0.18 0.84 0.47 2.09
70 0.039 0.26 0.061 0.41 0.12 0.81 0.38 2.56
80 0.033 0.27 0.052 0.44 0.096 0.81 0.28 2.33

1.2
from significant distributed mass of pier columns becomes
T1-10m = 0.92 s
more important and results in greater seismic demands,
Sepctral Acceleration (g)

Sa(T1, 5%) = 0.58 g

0.9 T1-40m = 1.50 s Thus, lower input intensities are required leading to a cer-
Sa(T1, 5%) = 0.43 g
tain level of damage, represented by a decrease of Sa(T1,
T1-50m = 2.16 s
Sa(T1, 5%) = 0.29 g 5%) and PGA for 50% probability of damage. Note that the
0.6 T1-60m = 2.91 s trend of ‘decrease’ in IMs herein is compared with the case
Sa(T1, 5%) = 0.23 g
T1-70m = 3.74 s of other tall piers (with heights of 40 and 50 m); while com-
0.3
Sa(T1, 5%) = 0.16 g T1-70m = 4.68 s
pared with the 10 m-height pier, the PGA levels of tall piers
Sa(T1, 5%) = 0.12 g
corresponding to 50% probability of exceeding each damage
state are still significantly greater.
0.0 Furthermore, despite the Sa(T1, 5%) values in Table 6 differ
0 1 2 3 4 5 6
significantly, the estimated PGAs are comparable for each tall
Periods (s) pier for all the damage states considered. For example, the
Figure 11. Mean acceleration response spectrum for near fault motions with T1 greatest differences between the PGAs for 50% probability of
of various piers. damage for tall piers from 40 to 80 m are 23.5%, 18.0%, 5.8%,
and 18.4%, respectively, for slight, moderate, extensive, and
phenomenon is that the factors dominating the seismic collapse damage state. This fact means that under near-fault
demands of piers change with the increase of pier height. motions with similar input intensity (PGA), dynamic
To better understand this issue, Figure 11 shows the mean responses of piers with heights ranging from 40 to 80 m are
acceleration response spectrum of the selected near-fault similar and can be represented by the results of 50 m pier. On
motions, where the fundamental periods for each considered the other hand, the estimated PGA values corresponding to
pier system and the corresponding Sa(T1, 5%) values are collapse damage state exceed 2.0 g for all tall pier models,
marked for comparison. For piers with heights increasing which is unlikely to be reached in an earthquake event, as pre-
from 10 to 50 m, the variation of seismic performance is viously stated for the 50 m pier. While for the extensive dam-
mainly influenced by the elongation of the fundamental age state, the corresponding PGA levels are around 0.8 g, a
periods as shown in Figure 11, which leads to smaller spec- commonly recorded intensity level in catastrophic earth-
tral accelerations and consequently lower seismic demands. quakes as Tohoku earthquake (Takewaki, Murakami, Fujita,
This is represented in the results of fragility analysis by Yoshitomi, & Tsuji, 2011), Wenchuan earthquake
increased values for both Sa(T1, 5%) and PGA correspond- (Krausmann et al., 2010), and Northridge earthquake (Yegian
ing to 50% probability of damage presented in Table 6. et al., 1995). Therefore, damage states other than collapse are
When pier heights increasing over 50 m, although the indeed to be incorporated when investigating the seismic per-
fundamental periods continue to increase, the discrepancies formance of tall pier bridges.
between the corresponding spectral accelerations become For illustrating purpose, Figure 12 presents and compares
smaller (see Figure 11). On the other hand, the inertia force the medium values of Sa(T1, 5%) of piers with heights of 40,
12 X. CHEN ET AL.

2.0 2. For both 50 and 10 m pier models, the PSDMs developed


40 m (near-fault) employing Sa(T1, 5%) as intensity measure (IM) show less
40 m (far-field) dispersions than those with PGA, for both near-fault and
1.6
60 m (near-fault)
60 m (far-field)
far-field ground motions. Consequently, the fragility anal-
Sa(T1, 5%) (g)

1.2 80 m (near-fault) yses are conducted based on the PSDMs with Sa(T1, 5%)
80 m (far-field) as IM for each investigated case.
3. When subjected to the same near-fault ground motions,
0.8
the tall piers are demonstrated seismically less vulnerable
for all the four damage states compared with the 10 m pier.
0.4 And the final collapse damage is unlikely to be reached for
tall piers for both near-fault and far-field ground motions,
0.0 implying the necessity of incorporating more damage
Slight Moderate Extensive Collapse
states for assessing seismic performance of tall piers.
Damage state 4. Compared with far-field motions, near-fault motions
Figure 12. Median values of Sa(T1, 5%) for piers with heights of 40, 60, and 80 m. generally lead to higher seismic vulnerability for tall
piers with height ranging from 40 to 80 m.
60, and 80 m, respectively, subjected to both near-fault and
Furthermore, the seismic fragility of these tall piers is
far-field ground motions. This figure shows that the 40 m
comparable, and thus the results of the 50 m pier dis-
pier is more vulnerable under far-field motions when slight
cussed in detail can be used as representative of a class
and moderate damage states are considered, while near-fault
of tall piers with heights in this range.
motions are more harmful for the extensive and collapse 5. The assessment of vulnerability for the tall piers sub-
state. This phenomenon is similar to that of the 50 m pier, jected to near-fault and far-field motions changes for 40
implying that the fragility assessment is dependent on the and 50 m piers with different damage states considered.
predefined damage state capacities and may change with the That is, when considering the slight and moderate dam-
damage states considered. age states, the 40 and 50 m piers are more vulnerable to
However, on the other hand, for the 60 and 80 m pier, far-field motions than near-fault motions; in contrast,
the near-fault motions are observed to induce higher seismic when the collapse damage state is considered, near-fault
fragility for all the four damage states, as the medium values motions are more harmful.
of near-fault motions are always smaller than those of the 6. The other three tall pier models, i.e. those with heights
far-filed motions. This fact indicates that the results of fra- of 60, 70, and 80 m, are more vulnerable under near-
gility assessment for tall piers also depend on the structural fault motions for all four damage states. These phenom-
features in addition to the damage state capacities. The ena indicate that the seismic vulnerability of tall pier
results presented in Table 6 and Figure 12 illustrate that the bridges depends on both the damage state capacities
seismic vulnerability of tall piers with heights ranging from and the structural characteristics.
40 to 80 m exhibits similar tendency with the 50 m pier.
Thus, the conclusions obtained from the 50 m pier case can Note that tall piers are unlikely to collapse during an
be generalised to this class of tall pier bridges with standar- earthquake event, future research might focus on the post-
dised superstructure and hollow pier section. earthquake serviceability and repair, which is challenging
due to the rugged topography of Southwest China.

8. Conclusions
Disclosure statement
This study focuses on the seismic fragility assessment for
No potential conflict of interest was reported by the authors.
bridges with tall piers subjected to near-fault ground motion
and compared to results for far-field ground motions. The
procedure of developing the probability seismic demand Funding
models (PSDMs) is described. The analysis results corre-
The authors gratefully acknowledge the support by the China
sponding to tall piers are presented in the format of fragility Postdoctoral Science Foundation (No. 2019M651468), the National
curves and compared with those of a conventional bridge Natural Science Foundation (No. 51678434, No. 51908348), and the
with piers of 10 m height. From the presented results, the Project of State Key Laboratory for Disaster Reduction in Civil
main conclusions are as follows: Engineering of China (No. SLDRCE15-A-01).

1. Four damage states, namely slight, moderate, extensive, References


and collapse, are developed for the tall piers. Instead of
Alam, M. S., Bhuiyan, M. R., & Billah, A. M. (2012). Seismic fragility
displacement-based indices used in previous investiga-
assessment of SMA-bar restrained multi-span continuous highway
tions, these damage states are presented in terms of sec- bridge isolated by different laminated rubber bearings in medium to
tion curvature ductility in this study, which has been strong seismic risk zones. Bulletin of Earthquake Engineering, 10(6),
shown as an appropriate damage index for tall piers. 1885–1909. doi:10.1007/s10518-012-9381-8
STRUCTURE AND INFRASTRUCTURE ENGINEERING 13

Baker, J. W. (2011). Conditional mean spectrum: Tool for ground- International Journal of Advanced Structural Engineering, 5(1), 23.
motion selection. Journal of Structural Engineering, 137(3), 322–331. doi:10.1186/2008-6695-5-23
doi:10.1061/(ASCE)ST.1943-541X.0000215 Kazemi, H., Ghafory-Ashtiany, M., & Azarbakht, A. (2017). Development
Baker, J. W., & Cornell, A. C. (2006). Spectral shape, epsilon and of fragility curves by incorporating new spectral shape indicators and a
record selection. Earthquake Engineering & Structural Dynamics, weighted damage index: case study of steel braced frames in the city of
35(9), 1077–1095. doi:10.1002/eqe.571 Mashhad, Iran. Earthquake Engineering and Engineering Vibration,
Berry, M. P., Lehman, D. E., & Lowes, L. N. (2008). Lumped-plasticity 16(2), 383–395. doi:10.1007/s11803-017-0388-7
models for performance simulation of bridge columns. ACI Kim, S. H., & Feng, M. Q. (2003). Fragility analysis of bridges under
Structural Journal, 105(3), 270. ground motion with spatial variation. International Journal of Non-
Billah, A. M., Alam, M. S., & Bhuiyan, M. R. (2013). Fragility analysis Linear Mechanics, 38(5), 705–721. doi:10.1016/S0020-7462(01)00128-7
of retrofitted multicolumn bridge bent subjected to near-fault and Kowalsky, M. J. (2000). Deformation limit states for circular reinforced
far-field ground motion. Journal of Bridge Engineering, 18(10), concrete bridge columns. Journal of Structural Engineering, 126(8),
992–1004. doi:10.1061/(ASCE)BE.1943-5592.0000452 869–878. doi:10.1061/(ASCE)0733-9445(2000)126:8(869)
Bray, J. D., & Rodriguez-Marek, A. (2004). Characterization of for- Krausmann, E., Cruz, A. M., & Affeltranger, B. (2010). The impact of
ward-directivity ground motions in the near-fault region. Soil the 12 May 2008 Wenchuan Earthquake on industrial facilities.
Dynamics and Earthquake Engineering, 24(11), 815–828. doi:10. Journal of Loss Prevention in the Process Industries, 23(2), 242–248.
1016/j.soildyn.2004.05.001 doi:10.1016/j.jlp.2009.10.004
Cao, L., & Li, C. (2019). Tuned tandem mass dampers-inerters with Kunnath, S. K., Larson, L., & Miranda, E. (2006). Modelling considera-
broadband high effectiveness for structures under white noise base tions in probabilistic performance-based seismic evaluation: Case
excitations. Structural Control and Health Monitoring, 26(4), e2319. study of the I-880 viaduct. Earthquake Engineering & Structural
doi:10.1002/stc.2319 Dynamics, 35(1), 57–75. doi:10.1002/eqe.531
Ceravolo, R., Demarie, G. V., Giordano, L., Mancini, G., & Sabia, D. Lee, W. K., & Billington, S. L. (2011). Performance-based earthquake
(2009). Problems in applying code-specified capacity design proce- engineering assessment of a self-centering, post-tensioned concrete
dures to seismic design of tall piers. Engineering Structures, 31(8), bridge system. Earthquake Engineering & Structural Dynamics,
1811–1821. doi:10.1016/j.engstruct.2009.02.042 40(8), 887–902. doi:10.1002/eqe.1065
Chen, X., Guan, Z., Li, J., & Spencer, B. F. Jr. (2018). Shake table tests Lei, J., Gao, M., & Yv, Y. (2007). Seismic motion attenuation relations
of tall-pier bridges to evaluate seismic performance. Journal of in Sichuan and adjacent areas. Acta Seismologica Sinica, 29(5),
Bridge Engineering, 23(9), 04018058. doi:10.1061/(ASCE)BE.1943- 500–511. ( in Chinese) doi:10.1007/s11589-007-0532-y
5592.0001264 Li, C., & Cao, B. (2015). Hybrid active tuned mass dampers for struc-
Chen, X., Li, J., & Guan, Z. (2016). Effects of higher modes on tall
tures under the ground acceleration. Structural Control and Health
piers. IABSE Symposium Report, 106(12), 136–143. doi:10.2749/
Monitoring, 22(4), 757–773. doi:10.1002/stc.1716
222137816819258348
Luco, N., & Cornell, C. A. (2007). Structure-specific scalar intensity
Chopra A. K. (2001). Dynamics of Structures: Theory and Applications
measures for near-source and ordinary earthquake ground motions.
to Earthquake Engineering. Upper Saddle River, NJ: Prentice Hall Inc.
Earthquake Spectra, 23(2), 357–392. doi:10.1193/1.2723158
Cornell, C. A., Jalayer, F., Hamburger, R. O., & Foutch, D. A. (2002).
Mackie, K., & Stojadinovic, B. (2001). Probabilistic seismic demand
Probabilistic basis for 2000 SAC Federal Emergency Management
model for California highway bridges. Journal of Bridge Engineering,
Agency steel moment frame guidelines. Journal of Structural
6(6), 468–481. doi:10.1061/(ASCE)1084-0702(2001)6:6(468)
Engineering, 128(4), 526–533. doi:10.1061/(ASCE)0733-
Mavroeidis, G. P., & Papageorgiou, A. S. (2002). Near-source strong
9445(2002)128:4(526)
ground motion: Characteristics and design issues. Proceedings of the
Dicleli, M. (2006). Performance of seismic-isolated bridges in relation
Seventh US National Conference on Earthquake Engineering (pp.
to near-fault ground-motion and isolator characteristics. Earthquake
Spectra, 22(4), 887–907. doi:10.1193/1.2359715 25). Boston, MA: 7NCEE committee.
Dutta, A., & Mander, J. (1998). Seismic fragility analysis of highway Mousavi, M., Ghafory-Ashtiany, M., & Azarbakht, A. (2011). A new
bridges. Proceedings of the INCEDE-MCEER center-to-center pro- indicator of elastic spectral shape for the reliable selection of ground
ject workshop on earthquake engineering – Frontiers in transporta- motion records. Earthquake Engineering & Structural Dynamics,
tion systems, pp. 22–23. 40(12), 1403–1416. doi:10.1002/eqe.1096
FEM. (2008). Hazus-MH MR3 Edition [Computer software]. Earthquake Muntasir Billah, A., & Alam, M. S. (2015). Seismic fragility assessment of
loss estimation methodology, Technical Manual. Washington, DC: concrete bridge pier reinforced with superelastic shape memory alloy.
National Institute of Building Sciences for Federal Emergency Earthquake Spectra, 31(3), 1515–1541. doi:10.1193/112512EQS337M
Management Agency. Nielson, B. G. (2005). Analytical fragility curves for highway bridges in
Guan, Z., Li, J., Xu, Y., & Lu, H. (2011). Higher-order mode effects on the moderate seismic zones (PhD thesis). Georgia Institute of Technology,
seismic performance of tall piers. Frontiers of Architecture and Civil Atlanta.
Engineering in China, 5(4), 496–502. doi:10.1007/s11709-011-0131-9 Park, S. W., Ghasemi, H., Shen, J., Somerville, P. G., Yen, W. P., &
Guirguis, J., & Mehanny, S. (2013). Evaluating code criteria for regular Yashinsky, M. (2004). Simulation of the seismic performance of the Bolu
seismic behavior of continuous concrete box girder bridges with viaduct subjected to near-fault ground motions. Earthquake Engineering
unequal height piers. Journal of Bridge Engineering, 18(6), 486–498. & Structural Dynamics, 33(13), 1249–1270. doi:10.1002/eqe.395
doi:10.1061/(ASCE)BE.1943-5592.0000514 Padgett, J. E., & DesRoches, R. (2008). Methodology for the development of
Hwang, H., Liu, J. B., & Chiu, Y. H. (2001). Seismic fragility analysis of analytical fragility curves for retrofitted bridges. Earthquake Engineering
highway bridges, Project MAEC RR-4 Report. Urbana-Champaign, & Structural Dynamics, 37(8), 1157–1174. doi:10.1002/eqe.801
IL: Mid-America Earthquake Center. Padgett, J. E., Nielson, B. G., & DesRoches, R. (2008). Selection of opti-
Jernigan, J., & Hwang, H. (2002). Development of bridge fragility mal intensity measures in probabilistic seismic demand models of
curves. 7th US National Conference on Earthquake Engineering, highway bridge portfolios. Earthquake Engineering & Structural
Boston, MA. Dynamics, 37(5), 711–725. doi:10.1002/eqe.782
J
onsson, M. H., Bessason, B., & Haflidason, E. (2010). Earthquake Ramanathan, K., DesRoches, R., & Padgett, J. E. (2012). A comparison of
response of a base-isolated bridge subjected to strong near-fault pre- and post-seismic design considerations in moderate seismic zones
ground motion. Soil Dynamics and Earthquake Engineering, 30(6), through the fragility assessment of multispan bridge classes. Engineering
447–455. doi:10.1016/j.soildyn.2010.01.001 Structures, 45, 559–573. doi:10.1016/j.engstruct.2012.07.004
Kazemi, H., Ghafory-Ashtiany, M., & Azarbakht, A. (2013). Effect of Saiidi, M. S., & Ardakani, S. (2012). An analytical study of residual dis-
epsilon-based record selection on fragility curves of typical irregular placements in RC bridge columns subjected to near-fault earth-
steel frames with concrete shear walls in Mashhad city. quakes. Bridge Structures, 8(1), 35–45.
14 X. CHEN ET AL.

Scott, B., Park, R., & Priestley, M. (1982). Stress-strain behavior of con- Tothong, P., & Luco, N. (2007). Probabilistic seismic demand analysis using
crete confined by overlapping hoops at low and high strain rates. advanced ground motion intensity measures. Earthquake Engineering &
ACI Journal Proceedings, 79(1), 13–27. Structural Dynamics, 36(13), 1837–1860. doi:10.1002/eqe.696
Shinozuka, M., Feng, M. Q., Kim, H. K., & Kim, S. H. (2000). Tubaldi, E., Tassotti, L., Dall’Asta, A., & Dezi, L. (2014). Seismic
Nonlinear static procedure for fragility curve development. Journal response analysis of slender bridge piers. Earthquake Engineering &
of Engineering Mechanics, 126(12), 1287–1295. doi:10.1061/ Structural Dynamics, 43(10), 1503–1519. doi:10.1002/eqe.2408
(ASCE)0733-9399(2000)126:12(1287) Wang, G. Q., Zhou, X. Y., Zhang, P. Z., & Igel, H. (2002).
Shome, N., Cornell, C. A., Bazzurro, P., & Carballo, J. E. (1998). Characteristics of amplitude and duration for near fault strong
Earthquakes, records, and nonlinear responses. Earthquake Spectra, ground motion from the 1999 Chi-Chi, Taiwan Earthquake. Soil
14(3), 469–500. doi:10.1193/1.1586011 Dynamics and Earthquake Engineering, 22(1), 73–96. doi:10.1016/
Somerville, P. (2000). Seismic hazard evaluation. Bulletin of New S0267-7261(01)00047-1
Zealand Society for Earthquake Engineering, 33(3), 371–386. Xu, H., & Gardoni, P. (2016). Probabilistic capacity and seismic
Takewaki, I., Murakami, S., Fujita, K., Yoshitomi, S., & Tsuji, M. demand models and fragility estimates for reinforced concrete
(2011). The 2011 off the Pacific coast of Tohoku earthquake and buildings based on three-dimensional analyses. Engineering
response of high-rise buildings under long-period ground motions. Structures, 112, 200–214. doi:10.1016/j.engstruct.2016.01.005
Soil Dynamics and Earthquake Engineering, 31(11), 1511–1528. doi: Yegian, M., Gazetas, G., Dakoulas, P., Makris, N., & Ghahraman, G.
10.1016/j.soildyn.2011.06.001 (1995, April 2–7). The Northridge Earthquake of 1994: Ground
Taucer, F., Spacone, E., & Filippou, F. C. (1991). A fiber beam-column motions and geotechnical aspects. Paper presented at Third
element for seismic response analysis of reinforced concrete structures. International Conference on Geotechnical Earthquake Engineering
California: Earthquake Engineering Research Center, College of and Soil Dynamics, St. Louis, Missouri, Vol. 3, pp. 151–157.
Engineering, University of California Berkekey. Zhang, J., & Huo, Y. (2009). Evaluating effectiveness and optimum
Tekie, P. B., & Ellingwood, B. R. (2003). Seismic fragility assessment of design of isolation devices for highway bridges using the fragility
concrete gravity dams. Earthquake Engineering & Structural function method. Engineering Structures, 31(8), 1648–1660. doi:10.
Dynamics, 32(14), 2221–2240. doi:10.1002/eqe.325 1016/j.engstruct.2009.02.017

Вам также может понравиться