Transportation Geotechnics 3 (2015) 3647

Contents lists available at ScienceDirect

Transportation Geotechnics
journal homepage: www.elsevier.com/locate/trgeo

Seismic analysis of highway skew bridges with nonlinear

soilpile interaction
Monirul Mallick, Prishati Raychowdhury
Dept. of Civil Engineering, Indian Institute of Technology Kanpur, India

a r t i c l e i n f o a b s t r a c t

Article history: Performance of highway bridges during an intense seismic event is an issue of utmost
Received 20 August 2014 importance. Of particular interest is the response of skewed highway bridges, where the
Revised 29 January 2015 skew angle and other related factors make the problem more complex. Although a number
Accepted 13 March 2015
of studies had been carried out to address these issues in the past, most of them have
Available online 20 March 2015
neglected or over-simplied the soilstructure interaction effects, primarily relying upon
the assumption that soilstructure interaction generally leads to a conservative estimation
Keywords:
of seismic demands. The present study focuses on investigating the effect of skew angle on
Skew bridge
Seismic response
seismic response of a bridge-foundation system including nonlinear soilpile interaction
Nonlinear modeling subjected to bi-directional ground motions. It has been observed that the rotational
Soilstructure interaction demand of the bridge deck is greatly affected by the skewness, indicating an increased
vulnerability of skewed bridges due to rotational movement of the deck leading to deck
unseating. It is also observed that the shear and moment demands of the piers increase sig-
nicantly with increasing skew angle, as much as 54% and 37%, respectively. The maximum
bending moment of the pile shaft is also found to increase upto 55%, indicating higher
design requirements for the foundation components of the skew bridges compared to a
similar normal bridge.
2015 Published by Elsevier Ltd.

Introduction of the stiffness for the abutment do not coincide, the

Skewe bridges are increasingly used in the highway
intersections and interchanges mainly to overcome the
space constraint. A bridge can be designated as a skew
bridge when the centerline of the bridge and the centerline
of the abutment and/or pier cap are non-perpendicular to
each other (Fig. 1). The skewed geometry of the bridge
affects the static and dynamic load transfer mechanism
of the system, and subsequently may lead to an altered
force and displacement demand. Moreover, the skew
bridges are strongly inuenced by the abutments, as the
center of the mass of the superstructure and the center
Corresponding author.
E-mail addresses: monirul@iitk.ac.in (M. Mallick), prishati@iitk.ac.in
(P. Raychowdhury).
inertial loading on the bridge tends to cause bridge
rotation about its vertical axis, leading to an excessive
transverse moment and unseating of the superstructure
and pounding to the abutment walls. For example, during
1971 San Fernando earthquake, the Foothill Boulevard
Undercrossing in California suffered a rotation in the
horizontal plane resulting in a permanent offset of about
10 cm (4 inch) in the direction of increasing skewness at
the abutment (Meng and Lui, 2000). Further, during 1994
Northridge earthquake, the abutments of the Pico-Lyons
skewed bridge located near Newhall, California, had
experienced a movement of about 51 mm (2 inch) in the
transverse direction (Apirakvorapinit et al., 2012).
A number of research efforts have been made in the last
few decades to understand the load transfer mechanism of

http://dx.doi.org/10.1016/j.trgeo.2015.03.002
2214-3912/ 2015 Published by Elsevier Ltd.
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647
Fig. 1. Schematic description of a skew bridge (Menassa et al. 2007).
skewed bridges under static and dynamic loading
conditions. Some of these studies have adopted numerical
modeling, whereas others relied upon experiments on
bridge component models, or eld observations. Followed
is a brief discussion on studies carried out to understand
the behavior of skewed bridges.
Literature review
In a pioneering study, Ghobarah and Tso (1973) adopted
a beam model to take into account the exural and torsional
modes of the deck of a skew bridge. Bakht (1988) critically
reviewed the design practice of skew bridges through ana-
lyzing skewed slab-on-girder bridges and concluded that if
the effect of skew angle is ignored, the maximum moment
demand will be over-estimated, but the maximum longitu-
dinal shear demand may be under-estimated leading to an
un-conservative design. Meng and Lui (2000) focused on
the superstructure exibility, substructure boundary con-
ditions, structural skewness and stiffness eccentricity in
their model. Menassa et al. (2007) investigated the effect
of the span length, slab width, and skew angle on the
response of a simple-span reinforced concrete bridge using
nite element method. It was observed that the AASHTO
LRFD design specications overestimated the maximum
longitudinal bending moment, and this overestimation
increases with increasing skew angle. Shamsabadi and
Yan (2008) developed a global three-dimensional nite-
element model for the seismically instrumented Painter
Street Overpass incorporating nonlinear foundationsoil
interaction. The results of the analyses showed that the
37
modeling of the pile foundations at the bent contributed
less to the overall bridge response as compared to the abut-
ments. Huo and Zhang (2008) studied the skewness effect
on live load reactions at the piers of continuous bridges.
Kalantari and Amjadian (2010) proposed an approximate
hand-method for dynamic analysis of a skew highway
bridge with continuous rigid deck. Apirakvorapinit et al.
(2012) conducted a series of pushover and dynamic analy-
ses using a nonlinear nite element model of Pico-Lyons
Bridge showing that at the end girders the percentage
increase in stress due to skewness can be 5060% for skew
angle of 40. In a recent study, Deepu et al. (2014) carried
out a dynamic analysis of a number of 3D nite element
models of various bridge congurations incorporating 0,
15, 30, 45 and 60 skew angles using SAP 2000 software,
with an assumption that the bridge piers are xed at their
bases. It was found that the skewed-bridge decks undergo
signicant rotations about the vertical axis during seismic
ground events and are permanently displaced from the
original location at the end of the shaking.
Provisions in the design codes
In the past, the skewed bridges were analyzed,
designed, and constructed in the same way as straight
bridges regardless of the magnitude of the skew angle.
For example, AASHTO (2003) standard specications used
to provide distribution factors for the interior girders of
simply supported bridges as a function of girder spacing.
It did not specify the consideration of the effect of the skew
angle and bridge continuity. In its more recent version,
38 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647

Deck slab Nodes

Skew angle
20 m 20 m
Springs for abutment-
backll connecon

Pier-pile
7m connecon
Piers node Pile head
length = 1 m

p-y springs

4 x 5 pile pile
group t-z springs
21 m elements Pile
embedded
length
Idealized
= 20 m
pile block
q-z springs

Fig. 2. Schematic of the 3D nite element modeling of the bridge in OpenSees (2008).

Fig. 3. Schematic sketch of the cross section of T-beam deck slab (Deepu et al., 2014).

AASHTO (2007) recommends a reduction factor for the
bending moments for skewed bridges, however, no magni-
cation factor for the shear is specied. Eurocode (2004)
species that the design methodology of normal bridge
may be applied for the skew bridges, however, adequately
conservative estimations based on capacity principles
must be adopted in order to avoid brittle failure modes.
It also species that highly skewed bridges should be
avoided in high seismicity regions. If that is not possible,
adequate modeling of the actual horizontal stiffness of
the bearings must be done, taking into account the concen-
tration of the vertical reactions near the acute angles.
Alternatively, an increased accidental eccentricity may be
used. Indian Standard Code (IS: 1893, Part-3, 2004) recom-
mends that the skew bridges with an angle of 30 and
above with span more than 60 m should be categorized
under irregular type and detailed dynamic studies should
be undertaken. However, any specic guideline for design
of these bridges are not provided in the code.
behavior of skew bridges, most of the studies have
neglected or over-simplied the soilstructure interaction
(SSI) issue. However, neglecting SSI may lead to an inaccu-
rate estimation of the seismic responses, especially for
high intensity earthquake motions, as the exibility of
the foundations may alter the period and damping charac-
teristics of the bridge-foundation system. Moreover, the
material and geometric nonlinearity of the soilfoundation
interface can inuence the force and displacement demand
of the overall bridge system. This may have more complex
effect in case of a skew bridge. This study focuses on 3D
nite element modeling of a two-span bridge with varying
skew angle incorporating nonlinear soilpile interaction. It
is intended to investigate the effect of skew angle on the
bridge-foundation-soil system through static pushover
and dynamic time history analysis using bi-directional
(two horizontal components) ground motions.
Numerical modeling

Scope of the present study A three-dimensional nite element model of the bridge-
pile-soil system has been developed using the open-source
It is evident from the above discussion that although a software platform OpenSees (OpenSees, 2008). In addition
number of studies have focused on understanding the to the normal bridge model, four skewed models with
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647 39

Table 1
Details of component modeling.

Component Material model Details

Deck 3D nonlinear beam-column element Thickness = 250 mm
Concrete compressive strength, fck = 40 MPa
Mass density, q = 2,400 kg/m3
Youngs modulus, E = 3  107 kPa
Poissons ratio, l = 0.2
Longitudinal 3D nonlinear beam-column element Cross-sectional area, A = 0.675 m2
girder Moment of inertia of the girder, I = 0.1266 m4
Young modulus, E = 3  107 kPa
Pier 3D ber-section force-based beam-column element with nonlinear Yield strength of steel, fy = 4.6  105 kPa
ber materials Youngs modulus, Es = 2  108 kPa
Strain hardening ratio, b = 0.01
Piles 3D nonlinear beam-column elements Modulus of elasticity, Ep = 2.5  107 kPa
Cross-sectional area, A = 11.70 m2
Shear modulus, G = 9.6  106 kPa
Torsional moment of inertia, J = 21.78 m4
Pilesoil Nonlinear spring Uniaxial PySimple1, TzSimple1 and QzSimple1 materials;
interface details provided in Table 3

Table 2
Idealized soil parameters for painter street bridge site (Shamsabadi and Yan 2008).

Layer no. Soil type (USCS symbol) Depth (m) Soil properties
c0 (kN/m3) /0 () c (kPa) vs (m/s)
Layer 1 Compacted sandy ll (SP, GP) 1.98 20.44 38 2.39 204
Layer 2 Stiff silt and clay (ML/CL) 9.60 20.12 11 158 305
Layer 3 Medium dense sand (SP) 4.57 8.96 34 0 NA
Layer 4 Dense sand with gravel (SP) 3.65 9.90 36 0 NA

Note: c0 = effective unit weight, /0 = friction angle, c = cohesion, vs = shear wave velocity.

skew angles 15, 30, 45 and 60 are developed. The
superstructure geometry and structural properties are
adopted from Shamsabadi et al. (2007) and Deepu et al.
(2014), which is very similar to the conguration of
Painter Street Bridge in California. The foundation details
are taken from in Shamsabadi and Yan (2008) for the pain-
ter street bridge site. A schematic of the 3D nite element
model with different components of the skew bridge is
shown in Fig. 2. Note that the nodes and elements shown
in Fig. 2 are just an idealization, and does not correspond
to the exact number of nodes and elements used in the
study. A detailed description of the superstructure and
substructure congurations, soil properties, modeling and
analysis methodologies are provided herein.
Bridge deck slab
Bridge piers
There are two piers of height 7 m from the center line of
the cap beam to the bottom. The piers have circular cross-
sections with a diameter of 1.7 m for normal, 15 and 30
skew models, and 1.8 m and 1.9 m diameters are used for
45 and 60 skewed models, respectively. Variable number
reinforcement bars are used for different skewed models,
such as 30, 31, 32, 35 and 41 bars for normal, 15, 30,
45 and 60 skewed models, respectively with 57.33 mm
diameter longitudinal bars and 35.81 mm diameter stir-
rups (after Deepu et al., 2014). The cap beam is assumed
to be of dimension 1 m  2 m and is modeled as a rein-
forced concrete beam with 40 MPa concrete. The bridge
piers are modeled using ve 3D ber-section force-based
beam column elements with nonlinear ber materials

The bridge superstructure consists of reinforced con- Table 3

Idealized strength and stiffness parameters for different soil springs.
crete continuous deck slab, with two equal spans of 20 m
length and 14.18 m width. The deck is rested on 6 rein- Layer P-y spring T-z spring Q-z spring
forced concrete T-beam girders having a cross sectional No. parameters parameters parameters

area of 0.675 m2 (1.5 m  0.45 m). The center to center pult y50 tult z50 tult z50
spacing between any two T-beams is 2.44 m. The cross- (kN) (m) (kN) (m) (kN) (m)
sectional area of T-girder deck slab is 6.9353 m2 and Layer 1 586 0.016 29 0.0009
moments of inertia along different axes are J = 0.3237 m4, Layer 2 1114 0.319 51 0.0091
Layer 3 12,966 0.126 316 0.0120
I33 = 1.4313 m4, I22 = 118.7519 m4. The cross section of
Layer 4 17,926 0.108 391 0.0136 142,634 0.024
the T-beam deck slab is provided in the Fig. 3.
40 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647

(a) P-y Spring Response (b) T-z Spring Response

6000 250

Vertical Force (kN/m)

Lateral Force (kN/m)

200

4000
150
Layer-1
Layer-1 Layer-2
100
2000 Layer-2 Layer-3
Layer-3 Layer-4
Layer-4 50

0 0
0 0.04 0.08 0.12 0 0.04 0.08 0.12
Displacement (m) Displacement (m)

Fig. 4. Spring behavior in different layers of the deposit for (a) P-y springs and (b) T-z springs.

1 elements. The bridge abutments are constrained in the

vertical direction, while free to move in the horizontal
Normalized Axial Force

longitudinal and transverse direction. However, once the

0.9 relative motion of the bridge deck exhausts the abutment
gap the bridge deck starts pounding on the abutment
backll. A summary of the component modeling including
0.8
values of the relevant parameters are provided in Table 1.

0.7 Piles and pilesoil interface

Each bridge pier is supported on a 5  4 pile group. The

0.6
100
Fig. 5. Convergence analysis result for axial force response of pier.
(OpenSees, 2008). The pier sections are further discretized
into steel and concrete section, where the reinforcing steel
is represented by a uniaxial bilinear inelastic model with
yield strength fy = 4.6  105 kPa, Youngs modulus,
Es = 2  108 kPa, and a strain hardening ratio, b = 0.01.
The top of the pile cap is restrained by a control node,
which connects the bottom nodes of the pier with rigid link
200 300 400 500 600 diameter of each pile is 30 cm, and the spacing between
Number of Nodes any two piles is equal to three times the diameter of the
pile. The depth of pile is 20 m below ground, whereas the
free head length of each pile is considered as 1 m. Since
the modeling of soilpile interaction of a total number of
40 piles (20 piles under each pier) was becoming too
cumbersome, it has been assumed that the piles and the
adjacent soil moves together to produce a typical block
failure, with a block dimension of 3.9 m  3.0 m. This
rectangular block is assumed as an equivalent single pile
of diameter 3.86 m for the convenience of the modeling.
Elastic beam-column elements are used to model the
equivalent single pile.

(a) Fixed Base Model (b) SSI Model

16 16
Shear Force at Pier Base (MN)

Shear Force at Pier Base (MN)

12 12

8 8
Skew angle=0o
Skew angle=15o
4 Skew angle=30o 4
Skew angle=45o
Skew angle=60o
0 0
0 0.4 0.8 1.2 0 0.4 0.8 1.2
Deck Displacement (m) Deck Displacement (m)

Fig. 6. Pushover curves for (a) xed base model and (b) SSI model.
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647 41

(a) (b)
0 0
Skew angle = 0o

Skew angle = 15o

-4 Skew angle = 30o -4
Skew angle = 45o
Skew angle = 60o
-8
Depth (m)

-8

Depth (m)
-12 -12

-16 -16

-20 -20
-20 -10 0 10 20 30 40 0 40 80 120 160
Shear force of pile group (MN) Bending moment of pile group (MN-m)

(c) (d)
0 0

-4 -4

-8
Depth (m)

-8
Depth (m)

-12 -12

-16 -16

-20 -20
0 0.04 0.08 0.12 0.16 -4 -3 -2 -1 0
Lateral displacement of pile group (m) Lateral resistance of soil (MN/m)

Fig. 7. Response of the pile and soil from pushover analysis: (a) shear force prole, (b) bending moment prole, (c) lateral displacement prole, and (d)
lateral resistance of soil.

The soilpile interface behavior is modeled using non-
linear Winkler foundation (BNWF) approach after
Boulanger et al. (1999), utilizing displacement-based beam
elements to represent the pile and a series of nonlinear
springs to represent the soil (as shown in Fig. 2). The soil
springs are zero-length elements that are assigned differ-
ent uniaxial materials in the lateral and vertical directions.
The spring nodes are created with three dimensions and
three translational degrees-of-freedom. One set of spring
nodes, the xed-nodes, are initially xed in all three
degrees-of-freedom. The other set of nodes, the slave
nodes, are initially xed in only two degrees-of-freedom,
and are later given equal degrees-of-freedom with the pile
nodes. The pile nodes are created with three dimensions
and six degrees-of-freedom (3 translational, 3 rotational).
The lateral resistance of the soilpile interface is repre-
sented by laterally oriented p-y spring elements, whereas
the frictional resistance along the length of the pile and
the tip resistance at the base of the pile are represented
by t-z and q-z spring elements, respectively. The con-
stitutive behavior of the above-mentioned springs are
dened from PySimple1, TzSimple1, and QzSimple1 uniaxial
materials, respectively from OpenSees. These material
models are well-calibrated and widely accepted for design
and analysis of pile foundations. The details of these spring
elements including their backbone equations, validation
and applications can be found in Boulanger et al. (1999)
and Boulanger (2000). The parameters to dene the
backbone curves for each spring is derived based on the
properties of each layer of the deposit, as discussed in
the following subsection.
Selection of soil properties
The soil properties are taken from the painter street
bridge site, California, as reported in Shamsabadi and Yan
(2008). The deposit consists of four layers with an upper
most sandy ll layer overlaid by a stiff silty clay layer
and two layers of dense sandy soil. The details of the soil
properties are given in Table 2. These soil properties are
used to determine the spring backbone parameters for
p-y, t-z and q-z spring materials. The backbone parameters
42 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647

(a) Fixed Base Model (b) SSI Model

1 1

Fundamental Mode

Natural Period (sec)

Natural Period (sec)
0.8 Second Mode 0.8
Third Mode
0.6 0.6

0.4 0.4

0.2 0.2

0 0
-15 0 15 30 45 60 75 -15 0 15 30 45 60 75
Skew Angle (deg) Skew Angle (deg)

Fig. 8. Eigenvalue analysis results: (a) xed base model and (b) SSI model.

for the above-mentioned springs for each layer of the Results and discussion
deposit are shown in Table 3. Note that pult, tult and qult
represent the ultimate capacity of p-y, t-z and q-z springs, To understand the behavior of skew bridges including
respectively, whereas y50 and z50 represents the displace- soilpile interaction, a number of analyses have been
ment in lateral and vertical direction, corresponding to carried out, namely, pushover analysis, eigenvalue
the 50% of the load capacity in the respective directions. analysis, and dynamic time history analysis. The following
Fig. 4(a) and (b) illustrate the responses of P-y and T-z subsections discuss the different analyses methods along
springs representing the behavior of the soilpile interface with the obtained results and their implications.
in different layers. Note that these responses correspond to
the non-skewed bridge model. Pushover analysis

In this study, a nonlinear static nonlinear pushover

Finite element mesh properties analysis is performed according to FEMA-356 (2000)
guidelines. For the pushover analysis, the bridge is stati-
The 3D nite element mesh involves the representation cally analyzed under monotonically increasing lateral
of the essential structural and foundation elements of the loads in the transverse direction until the displacement
bridge (as shown in Fig. 2). A convergence study has been at the deck slab exceeds a certain selected target displace-
carried out to determine the optimum number of nodes ment. The target displacement is intended to represent the
and elements. For the convergence study, the non-skewed maximum possible displacement that is likely to be experi-
xed base model is used and the responses such as axial enced by the bridge during a design earthquake. In this
force, shear force and bending moment at the pier base study, a target deck displacement is set as 15% of the pier
are recorded for increasing number of nodes and elements. height, and applied at the pier-deck connection node with
The convergence study indicates that the responses tend to an increment of 0.1% pier drift ratio.
converge when at least 377 number of nodes are used Fig. 6 presents the pushover curves showing shear force
(Fig. 5). However, to achieve a better convergence, a model at the base of the piers versus the deck displacement for
with 547 nodes is used in the present study. This number different skew angle models considering xed base as well
indicates the nodes for the superstructure for the xed as SSI. It can be observed from Fig. 6 that the shear capacity
base model. However, for the SSI model, an additional of the piers is signicantly inuenced by the skew angle for
274 nodes are used for piles and soilpile interface model- both xed base and SSI cases. The shear capacity of the
ing, making a model of total 821 nodes with 1110 nonlin- piers increases up to 54% from non-skewed to 60 skewed
ear beam-column elements, 94 zero-length elements, 82 model for the xed base case. For the SSI case, this increase
equal DOF constraints and 2 rigid link elements for the glo- is about 60%. However, the absolute value of shear capacity
bal bridge-foundation model. for the SSI case is lower than the respective xed base case,

Table 4
Details of selected ground motions with 10% in 50 years hazard level (adopted from Somerville et al. 1997).

Ground motion Direction Record Earthquake Distance (km) DT (s) PGA (g)
magnitude Mw
El Centro01 Longitudinal (along global X axis) Imperial Valley, 1940, El Centro 6.9 10 0.02 0.46
El Centro02 Lateral (along global Y axis) Imperial Valley, 1940, El Centro 6.9 10 0.02 0.68
Gilroy01 Longitudinal (along global X axis) Loma Prieta, 1989, Gilroy 7 12 0.02 0.67
Gilroy02 Lateral (along global Y axis) Loma Prieta, 1989, Gilroy 7 12 0.02 0.97
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647 43

(a)
Acceleration (g) 0.8 (b) 1

Acceleration (g)
0 0

El Centro01 Gilroy01
El Centro02 Gilroy02

-0.8 -1
0 10 20 30 40 0 4 8 12 16 20
Time (s) Time (s)

Fig. 9. Input acceleration time history: (a) El Centro motions and (b) Gilroy motions.

5 100
(a) (b)

Spectral displacement (cm)

El Centro01
Spectral acceleration (g)

4 El Centro02 80
Gilroy01
Gilroy02
3 60

2 40

1 20

0 0
0.01 0.1 1 10 0.1 1 10
Period (s) Period (s)

Fig. 10. A 5% damped elastic response spectra for (a) spectral acceleration and (b) spectral displacement.

(a) El Centro 01-02 (b) Gilroy 01-02

In-plane Deck Rotation (deg)
In-plane Deck Rotation (deg)

0.004 0.008
Skew angle = 30 deg Skew angle = 30 deg
Skew angle = 60 deg Skew angle = 60 deg
0.002 0.004

0 0

-0.002 -0.004

-0.004 -0.008
0 4 8 12 16 20 0 4 8 12 16 20
Time (sec) Time (sec)

Fig. 11. Rotation time history for (a) El Centro motions and (b) Gilroy motions.
Transverse Deck Displacement (mm)

(a) (b)
36 0.006
In-plane Deck Rotation (deg)

El Centro 01-02
Gilroy 01-02
32
0.004

28

0.002
24

20 0
0 15 30 45 60 0 15 30 45 60
Skew Angle (deg) Skew Angle (deg)

Fig. 12. Deck response: (a) peak transverse displacement and (b) peak in-plane rotation.
44 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647

with a reduction ranging from 1% to 15%. Nonetheless, the
displacement corresponding to the peak shear does not
change much with skew angle, ranging from 0.08 m to
0.1 m for the xed base models, and from 0.2 m to
0.26 m, for the SSI models. This indicates a softer behavior
of the SSI models due to increased exibility. Note that the
soil springs yielded prior to yielding of the piers in these
cases.
Figs. 7(a) and (b) show the shear force and bending
moment prole of the pile block along its depth. It can be
observed that the peak moment increases only 6% from
non-skewed to 30 model, whereas it increases about
30% for the 45 model and 65% for the 60 model.
Fig. 7(c) shows the lateral deection prole of the piles
for different models, indicating that lateral deection
increases signicantly for highly skewed models compared
to non-skewed and slighly skewed models. Fig. 7(d)
demonstrates the reaction of the soil adjacent to the pile.
It may be noticed that the soil reaction for each layer is
distinctly different than the other layer. However, for a
particular layer, the reactions vary linearly. Moreover, it
may be noted that the reaction of each layer is inuenced
by the skewness of the bridge. These observations indicate
that the superstructure skewness may induce higher foun-
dation demands, which is generally not accounted in the
conventional design practice.
Eigenvalue analysis
Although the pushover analysis gives a fair estimation
of the seismic demands of a system, however, the method
has some inherent limitations, and cannot be solely
dependable for seismic response prediction. Therefore, a
dynamic time history analysis is necessary for more
accurate estimation of seismic demands. In this study,
before conducting the dynamic time history analysis, an
eigenvalue analysis is carried out to understand the modal
properties of the bridge under free vibration.
Fig. 8(a) and (b) provides the rst three natural periods
of the bridge for xed base model and SSI model, respec-
tively, for varying skew angles. It can be observed that
for both xed base and SSI cases, the fundamental and
second mode period decrease as the skew angle increases.
This reduction is about 56% and 50% in the rst mode and
31% and 17% in the second mode, from non-skewed to 60
skewed model, for the xed base and SSI case, respectively.
This indicates that the global stiffness increases with
increasing skewness for these two modes. However, for
the third mode, the period slightly increases beyond 45
skew angle, implying that the transverse stiffness of the
bridge reduces for highly skewed models. Further, it may
also be noted that the periods for the SSI cases are always
greater than that of the xed base cases, which is a result of
induced exibility due to soilpile springs. The period
elongation due to SSI effect is observed to range between
1.86 and 2.12 in the fundamental mode, 1.27 to 1.53 in
the second mode and 1.04 to 1.12 in the third mode. This
indicates that SSI affects the modal properties of the rst
mode signicantly compared to the higher modes.
Dynamic time history analysis
Nonlinear time history analysis has been carried out on
different bridge models with gravity loads applied rst
followed by dynamic earthquake excitation. Since the
skewness of the bridge is the focus of the study, it is impor-
tant consider the bi-directional effect of the ground
motions. For this purpose, four ground motions are
selected from Somerville et al. (1997). These motions cor-
responds to Imperial Valley record of 1940 El Centro
Earthquake and Gilroy record of 1989 Loma Prieta
Earthquake, California. These ground motions are scaled
by Somerville et al. (1997) to represent a 10% in 50 years
hazard level with a return period of 475 years for a soil
type D (stiff soil) according to NEHRP (2003) classication
system. The selection is relevant for the chosen bridge site,
as the soil prole in the present study consists of a stiff
silty clay followed by medium dense sand as shown in
Table 2. The details of the selected ground motions are pro-
vided in Table 4. Two bi-directional events are considered
in this study. In the rst event, the El Centro01 motion is
applied in the global longitudinal direction, whereas El
Centro02 is applied in the global lateral direction.
Similarly, in the second event, Gilroy01 and Gilroy02

(a) (b)
8 16
El Centro 01-02
Gilroy 01-02
Pier Moment (MNm)

14
Pier Shear (MN)

6 12

5 10

4 8
0 15 30 45 60 0 15 30 45 60
Skew Angle (deg) Skew Angle (deg)

Fig. 13. Peak responses at pier base: (a) shear force and (b) bending moment.
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647 45

(a) Pile Moment: El Centro 01-02 (b) Pile Moment: Gilroy 01-02
0 0

-4 -4

-8 -8

Depth (m)
Depth (m)

-12 -12

Skew angle = 0o
Skew angle = 15 o
-16 Skew angle = 30 o -16
Skew angle = 45 o
Skew angle = 60 o

-20 -20
0 10 20 30 40 0 20 40 60
Pile Moment (MN-m) Pile Moment (MN-m)

(c) Pile Shear: El Centro 01-02 (d) Pile Shear: Gilroy 01-02
0 0

-4 -4

-8 -8
Depth (m)

Depth (m)

-12 -12

-16 -16

-20 -20
0 2 4 6 0 2 4 6 8
Pile Shear (MN) Pile Shear (MN)

Fig. 14. Peak pile response prole along depth: (a) bending moment for El Centro, (b) bending moment for Gilroy, (c) shear force for El Centro, and (d) shear
force for Gilroy.

motions are applied in the global longitudinal and lateral
direction, respectively. It may be noted that the two
motions of each event corresponds to the mutually per-
pendicular horizontal components of a motion recorded
in the mentioned station. The acceleration time history of
the input motions are shown in Fig. 9, whereas Fig. 10 pre-
sents the 5% damped acceleration and displacement
response spectra of the chosen motions.
For performing nonlinear dynamic analysis, Newmarks
average acceleration integration method has been used. A
5% Rayleigh damping is assumed for the rst and second
modes. Newtons line search algorithm has been used to
solve the nonlinear residual equation. The system band gen-
eral is used to construct the LinearSOE and Linear Solver
objects to store and solve the system of equations in the
analysis. The reverse Cuthill-McKee numberer command
is used to construct the RCM degree-of-freedom numbering
objects to provide the mapping between the degrees-of-
freedom at the nodes and the equation numbers. The RCM
numberer uses the reverse Cuthill-McKee scheme to order
the matrix equations. The transformation method is used
to construct the constraint handler object. The maximum
number of iterations and tolerance used to check the
returning failure condition are 100 and 108, respectively.
46 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647

1.6 Since the pile foundation component is one of the cru-

Moment: El Centro01-02 cial components of the bridge system, the response of the
Normalized pile force (F/F=0)

Shear: El Centro01-02 pile foundation is also emphasized. Fig. 14 presents the

Moment: Gilroy01-02 proles of the peak shear and peak moment along the
1.4 Shear: Gilroy01-02
depth of embedment of the pile. It can be observed that
the maximum moment occures at about 4.5 m below the
ground for all models, however, the values are different
for different models. There is a clear trend of increasing
1.2
demands in pile moment and shear with increasing skew
angle. For example, peak moment increases from 29 MN-
m to 36 MN-m from non-skewed to 60 skewed model
for the El Centro event, and 34.8 MN-m to 54 MN-m for
1
0 15 30 45 60 the Gilroy motion. Fig. 15 presents the normalized pile
Skew Angle (deg) shear and moment demands for different models as a ratio
of non-skewed model. It can be observed that pile force
Fig. 15. Peak normalized pile responses with skew angle variation. demands increase about 25% for the El Centro motion,
and about 55% for the Gilroy motion from non-skewed to
60 skewed model. These observations indicate that the
foundation components should be designed considering
The following responses of the bridge models are
the effect of skewness, as skewness may induce signicant
emphasized: deck displacement, in-plane deck rotation,
alteration in the force and displacement demands of these
force demands at the pier base, and peak force demands
components.
along the pile length. Figs. 11(a) and (b) show the time his-
tory response of the in-plane deck rotation for 30 and 60
skewed models. It can be observed that the rotational
Conclusions
demand is much higher for the 60 model compared to
the 30 model throughout the time series. At some
This study focuses on 3D nite element modeling of a
instances, the demand in 60 case is more than twice than
bridge-foundation system incorporating nonlinear soil
the 30 case, indicating that rigid body rotation may
pile interaction and evaluating the effect of skew angle
increase signicantly with increasing skew angle. Since
on the seismic response of the bridge-foundation-soil
the seismic design mainly depends on the peak demands,
system. Nonlinear static pushover analysis, followed by
the absolute maximum values of each response parameter
eigenvalue analysis and dynamic time history analysis
are summarized in the subsequent plots for different skew
using bi-directional (two horizontal components) have
angles.
been carried out. The following key observations are made
Figs. 12(a) and (b) show the peak transverse displace-
from the study:
ment and peak in-plane rotation of the deck, respectively.
It may be observed that transverse displacement increases
(1) Shear force capacity of the piers are signicantly
from non-skewed to 15 skewed model, but remains con-
affected by the skewness of the bridge. The effect
stant afterward. On the other hand, in-plane deck rotation
is more prominent for the structures considering
seems to be very sensitive to the skew angle. This response
soilpile interaction compared to the xed base
is negligible for the non-skewed model, and starts increas-
structures.
ing abruptly with increasing skew angle. A peak rotation of
(2) Fundamental and second mode periods of the struc-
0.004 and 0.006 is observed for El Centro and Gilroy
ture decrease with increasing skew angle as much as
motion, respectively. For the bridge considered, a rigid
56% from non-skewed to 60 skewed bridge.
body rotation of 0.004 corresponds to approximately
(3) Rotational demand of the bridge deck is greatly
1.4 mm movement of the deck, whereas 0.006 rotation
affected by the skewness, indicating an increased
corresponds to approximately 2.1 mm movement.
vulnerability of skew bridges due to rotational
The seismic force demand of the bridge piers are also
movement of the deck leading to deck unseating.
affected by the skew angle as shown in Fig. 13(a) and (b).
(4) Shear and moment demands of the piers increase
It can be noticed that the shear force and bending moment
signicantly with increasing skew angle, as much
has an increasing trend with increasing skew angle, indi-
as 54% and 37%, respectively. The maximum bending
cating higher force demand for skewed models. The shear
moment of the pile shaft is also observed to increase
force increases from 4.34 MN in non-skewed bridge to
up to 55%, indicating higher design requirements for
5.45 MN in 60 bridge for the El Centro motion, and 5.1
the foundation components of skew bridges com-
MN to 7.8 MN for the Gilroy motion. Similarly, the bending
pared to a similar normal bridge.
moment increases from 9.4 MN-m to 10.3 MN-m for the El
Centro motion, and 11.2 MN-m to 15.3 MN-m for the
Note that the ndings of this study are limited to the
Gilroy motion. The maximum deviation for pier shear is
parameter space considered herein (i.e., bridge
25% for the El Centro event, and 54% for the Gilroy event,
conguration, soil type, water table, ground motion
whereas the same of pier bending moment is 17% and
parameters, etc.) and require further investigation for the
37% for El Centro and Gilroy motion, respectively.
purpose of generalization.
 M. Mallick, P. Raychowdhury / Transportation Geotechnics 3 (2015) 3647 47

References Huo XS, Zhang Q. Effect of skewness on the distribution of live load
reaction at piers of skewed continuous bridges. ASCE J Bridge Eng
2008;13(1):1104.
AASHTO. Standard specications for highway bridges. Washington, D.C;
Indian Standard Code. IS 1893 (Part-3): Criteria for earthquake resistant
2003.
design of structures: bridges and retaining walls. 2004.
AASHTO-LRFD. Bridge design specications. American Association of
Kalantari A, Amjadian M. An approximate method for dynamic analysis of
State Highway and Transportation Ofcials; 2007.
skewed highway bridges with continuous rigid deck. Eng Struct
Apirakvorapinit P, Mohammadi J, Shen J. Analytical investigation of
2010;32(9):285060.
potential seismic damage to a skewed bridge. ASCE Pract Periodical
Menassa C, Mabsout M, Tarhini K, Frederick G. Inuence of skew angle on
Struct Des Constr 2012;17(1):512.
reinforced concrete slab bridges. ASCE J Bridge Eng 2007;12(2):20514.
Bakht B. Analysis of some skew bridges as right bridges. ASCE J Struct Eng
Meng JY, Lui EM. Seismic analysis and assessment of a skew highway
1988;114(10):230722.
bridge. Eng Struct 2000;22(11):143352.
Boulanger RW, Curras CJ, Kutter BL, Wilson DW, Abghari A. Seismic soil
NEHRP. Recommended provisions for seismic regulations for new
pilestructure interaction experiments and analyses. ASCE J Geotech
buildings. Washington, D.C.: Building Seismic Safety Council; 2003.
Geoenviron Eng 1999;125(9):7509.
OpenSees. OpenSees (Open System for Earthquake Engineering
Boulanger RW. The PySimple1, TzSimple1, and QzSimple1 material
Simulation). Berkeley: Pacic Earthquake Engineering Research
models. Documentation for the OpenSees platform. <http://
Center (PEER), University of California; 2008.
opensees.berkeley.edu>; 2000.
Shamsabadi A, Rollins KM, Kapuskar M. Nonlinear soilabutmentbridge
Deepu SP, Prajapat K, Ray-Chaudhuri S. Seismic vulnerability of skew
structure interaction for seismic performance-based design. ASCE J
bridges under bi-directional ground motions. Eng Struct
Geotech Geoenviron Eng 2007;133(6):70720.
2014;71:15060.
Shamsabadi A, Yan L. Dynamic soilabutmentfoundationstructure
Eurocode. Design of structures for earthquake resistance, part-1: general
interaction of an instrumented skewed bridge. In: The 12th
rules, seismic actions and rules for buildings. Brussels: European
international conference of international association for computer
Committee for Standardization (CEN); 2004.
methods and advances in geomechanics (IACMAG), 16 October, Goa,
FEMA 356. Prestandard and commentary for the seismic rehabilitation of
India; 2008.
buildings. Virginia: American Society of Engineers; 2000.
Somerville P, Smith N, Punyamurthula S, Sun J. Development of ground
Ghobarah AA, Tso WK. Seismic analysis of skewed highway bridges with
motion time histories for phase 2 of the FEMA/SAC steel project.
intermediate supports. Earthquake Eng Struct Dynam
<http://www.sacsteel.org/project/>; 1997.
1973;2(3):23548.