You are on page 1of 8

THE 7th INTERNATIONAL CONFERENCE

"CIVIL ENGINEERING - SCIENCE AND PRACTICE"


GNP 2020 – Kolašin, Montenegro, 10-14 March 2020

Nina Serdar1, Jelena Pejović2

INFLUENCE OF ABUTMENT MODEL ON RESULTS OF BRIDGE


PUSHOVER ANALYSIS

Abstract
Just as response of bridge structure depends on bridge abutment behaviour, so does the
accuracy of abutment numerical modelling affects the results of seismic analysis. In case of
shorter bridge structures as well as bridges with irregularities (curved alignment, skewness,
irregular distribution of pier height, etc.) influences of abutment are even more pronounced.
Embankment infill mobilisation, soil-structure interaction under shear deformations as well as
shear key bearing capacity dominates bridge response and force distribution for column bents.
Since pushover analysis is powerful tool to quickly and relatively accurate estimate bridge
seismic response, it is important to evaluate all possible factors that influence analysis results.
In this paper, two possible approaches for bridge abutment modelling are investigated:
simplified model and more accurate model. In simplified model of an abutment, it is assumed
that soil resistant longitudinal pressure (represented only by two springs) is evenly distributed
behind the abutment and failure of shear keys are not taken into account. A more accurate
model is generated with assumption that skewness of abutments influence the distribution of
longitudinal pressures (represented with spring distributed along the width of abutment). Also,
shear keys are modelled and taken into account in this more accurate model. Comparison of
pushover analysis results are presented in this paper for the simplified and for the more accurate
method of abutment modelling. Results show differences in ultimate base shear strengths.
Totally, six RC bridges are selected for the analysis with variation in skew angle and pier type.
Two skew angles (20° and 30°) and non-skewed bridge are considered. Two types of piers:
single rectangular pier and double circular bent. It was noticed that skewness influence seismic
response especially for single column bent bridges. It is concluded that when the shear key and
distributed pressure behind the backwall of an abutment is taken into account higher values of
base shear can be expected.
Key words
RC bridge, seismic analysis, pushover analysis, abutment, soil pressure.

1
Teaching assistant, University of Montenegro, Podgorica, Montenegro, ninas@ucg.ac.me
2
Assistant Professor, University of Montenegro, Podgorica, Montenegro, jelenapej@ucg.ac.me

447
Civil En g in eerin g – S cien ce a n d Pra ctice

1. INTRODUCTION

Abutment behaviour under earthquake loading determines the response of bridge structures.
Embankment infill mobilisation, soil-structure interaction under shear deformations as well as
shear key bearing capacity dominates bridge response and force distribution for column bents.
Effects of the abutment –soil interaction is more pronounced in case of irregular bridge structures
or shorter bridges with stiff superstructures [1]. Also some bridge complex layouts, typical for
highway overpasses or bridges on access roads and ramps (skewed or curved bridges), should be
considered as irregular and soil-structure interaction should be considered in seismic analysis.
In bridge design abutments are considered as free standing retaining walls designed to resist
active and passive earthquake pressures. During the earthquake excitation, earth pressures induced
by internal forces from superstructure, generally can exceed the values of passive pressures
anticipated in design [2]. In seismic assessment pushover analysis can be considered as powerful
tool for quick and accurate estimation of bridge response. Vulnerability and loss analysis can be
based on the pushover analysis outcomes. It is important that the modelling of all bridge
components realistically represent bridge resisting mechanism. Abutments should be modelled with
accurate mass, stiffness and hysteresis behaviour in order to obtain relevant results. Also, level of
modelling complexity should be considered as the parameter in choosing the adequate approach.
Continuum models of embankment should be avoided, if possible, in order to generate less time
consuming calculations. Generally, behaviour of the abutment can be divided and considered in
three directions: vertical, longitudinal and transversal direction. Longitudinal response of the
abutment is governed by on the response of elastomeric bearing pads, gap, abutment back wall, and
backfill material and abutment foundations. Transverse abutment response is dominated by
elastomeric bearing pads, shear keys, wing walls, backfill material and foundations too. Vertical
response of the abutment should include vertical response of the steam wall, bearing pads, and soil.
In modelling, for the purpose of problem simplifying, some of the above components of response
may be omitted. Several studies were conducted [3], [4] on the effects of abutment modelling on
bridge seismic response.
In this paper, the evaluation of pushover analysis results is presented for two considered
abutment modelling techniques: simplified (S model) and more accurate model (MA model).
Results are presented for skew bridges. The effects of modelling assumptions on displacements of
pier top and ultimate base shear capacities are commented.

2. POSSIBLE CHOICE FOR BRIDGE ABUTMENT MODEL

In this chapter simplified and more accurate abutment models will be described in detail.
Simplified spring model (S model) is consisted of rigid element with the length equal to the
superstructure width and connected to the centreline of the deck through the rigid joint. At the end
joints of the rigid element tree nonlinear springs are assigned representing longitudinal, transversal
and vertical soil-structure interactions. A zero length element is used to model the abutment
backwall and embankment fill in the longitudinal direction. A compression gap of 0.12 m is taken
into account and after the gap closure, the abutments’ resistance is provided by the passive pressure
of the soil fill. The longitudinal stiffness of the abutment-embankment system is modelled

448
GNP 2 0 2 0

following the Caltrans (2006) [5] provisions. An elastic perfectly plastic force-deformation
relationship is adopted. The longitudinal abutment stiffness Ka depends on the initial embankment
stiffness, assumed to be 11.5 kN/mm/m, and the width (w) and height (h) of the backwall. Effective
stiffness is calculated by taking into account the gap width. The yielding force of the embankment
fill is calculated by multiplying the passive pressure of 240 kPa, suggested by Caltrans, by the
backwall area and height proportionality factor. The shear stiffness of the bearing pads is ignored
as it is considerably smaller in comparison to the embankment fill stiffness.
Transverse resistance in simplified model is modelled through zero length elements
assigned with the EPP force-deformation curve to represent a backfill-wing wall system. The
transversal abutment stiffness (Kt) is obtained by reducing stiffness in the longitudinal direction by
applying the reduction factors. The stiffness of the shear keys and bearing pads is ignored.
Vertical stiffness is represented with force-displacement response considering different
stiffness. Also zero length elements are used. Stiffness of a bearing pad is assumed to be 230
kN/mm. This stiffness is effective as long as the displacement is less than 19 mm. After the limit
displacement significantly larger stiffness is adopted to represent the rigid behaviour of a stem
wall, foundation and rock soil. Schematic representation of abutment simplified model is
presented in Figure 1.

Figure 1. Bridge finite element model and simplified abutment model

449
Civil En g in eerin g – S cien ce a n d Pra ctice

Figure. 2 Bridge finite element model and skew abutment model

More accurate model (MA model), applied here, was proposed in [14] is spine line model
that accounts for increase of backfill soil pressure from the obtuse toward the acute corner behind
the skewed abutment. In longitudinal direction expansion joint and backfill soil material are taken
into account in abutment modelling. It is assumed that due to low friction of bearing pads, their
shearing capacity can be ignored in both longitudinal and transversal directions. Also, backwall
shearing capacities negligibly contribute to longitudinal abutment bearing capacity. Five nonlinear
springs, uniformly distributed along with abutment width, are used to model passive resistance of
backfill soil. The initial stiffness and strength of nonlinear springs in longitudinal direction are
calculated in same way as for S model. For skewed bridges, stiffness and strength of nonlinear
springs increase linearly from obtuse to acute corner along skew length of backwall. All
longitudinal springs attached to rigid bar have uniform force-deformation response, in case of non-
skewed alignment. Passive pressure is taken to be perpendicular to backwall. A gap of 0.12 m
between backwall and deck is incorporated in force-deformation relationship by placing gap
element in series with nonlinear spring element. Shear key are taken into account in transverse

450
GNP 2 0 2 0

direction. Brittle shear key failure is assumed for force-deformation curve. Soil-structure
interaction springs (ssi) are used to approximate transversal response. Passive resistance of backfill
on wingwalls was assumed not to be mobilized. Vertical response is modelled in same way as for S
model. See Figure 2 for schematic representation of MA model.

3. BRIDGE DESCRIPTION AND OTHER NONLINEAR BRIDGE


MODELLING ASPECTS

All analysed RC bridges are 3 span bridges with span lengths: 32-40-32 m and concrete box
girder deck. Pier heights are 10 m. Bridges are designed according to European standards. Samples
with double column bents have 2 circular piers 1.4 m diameter. Single bent bridges have pier-wall
with cross-section 1 x 4.4 m.
Other common assumptions are made when nonlinear bridge model was generated.
Superstructure is modelled by the elastic frame elements (elfrm) placed at the centre of gravity of
the deck with geometric properties of un-cracked cross sections. Masses are lumped at adjacent
nodes. Asphalt cover and parapets are added to self-weight mass and they are calculated according
to the tributary length of the deck for every node. Concrete class C 35/45 is used for the deck, so
the modulus of elasticity and shear modulus were calculated accordingly. Elastic rigid elements
were placed between the deck and the pier elements. Whenever a rigid element is implemented its
stiffness is 100 times the stiffness of the adjacent elements.
Piers were modelled using inelastic force-based elements (infrmFB). Distributed plasticity
along the pier length is taken into account. For double pier bents, the beam integrated into the deck
at the pier’s top is modelled using the elastic element with increased torsional stiffness. Also it is
assumed that the top beam is rigid in plane of the deck. The piers are assumed to be fixed at the
base. The pier cross-section is divided into the material of concrete core - confined concrete,
concrete of the protective layer - unconfined concrete, and reinforcing steel to account spread of
nonlinearity across the depth and along the height of column. At every integration section the
stress-strain state is obtained by integrating non-linear uniaxial stress-strain relations of fibres.
Concrete class C 30/37 is assumed for piers. Tensile strength, both for confined and unconfined
concrete was neglected. Confined concrete material properties are calculated. Cyclic behaviour of
concrete is modelled Reinforcing steel used is class B 500 B for deck and B 500 C for columns
with yielding strength/strain of 575 MPa/0.2875%. Ultimate steel strain and strain hardening
parameter for column reinforcement are adopted as 7.5% and 0.0059 respectively. Mean values of
ultimate tensile strength is taken equal to 660 MPa, according to provisions given in EN 1998-2.
In order to examine seismic behaviour adaptive pushover analysis is used. At the end of
each load step, the forms of vibration and corresponding participation factor of the considered tone
are computed. Effects of the shifted vibration periods, caused by reduction of stiffness when the
structure reaches the point of large plastic deformations, and new distribution of stiffness in each
step of the load was taken into account. Applied load scheme, proportional to the mod, is updated
in each step for each significant vibration. These modal schemes are combined based on the rule of
the square root of the sum of squares (SRSS).

451
Civil En g in eerin g – S cien ce a n d Pra ctice

4. RESULTS AND DISCUSSION

Pushover curves in transversal and longitudinal direction, for both modelling approaches
are are presented on Figure 3. In transversal direction, for both abutments modelling techniques,
similar shape of curves are noticed. Slightly increase in stiffness and higher values of maximal
shear force due to the presence of the shear keys. It is important to state that shear key failure was
not observed. In longitudinal due to accounting and modelling of gap pushover curves have
different shapes. After the gap closure, passive earth pressures are developed behind the wing so
pushover curves for MA model do not follow smooth transition into nonlinear behaviour as it is
case for S model.

452
GNP 2 0 2 0

Figure. 3. Results of pushover analysis for different abutment models

Figure. 4. Comparison of base shear/weight ratio in transverse and longitudinal direction for
various abutment model, column bent and skew angle

In both modelling techniques same conclusion arises, skewness strongly affects response of
bridge structure with single bent pier, where weak bending axis is in transversal bridge direction.
The effect of an increase of skewness, for single bent piers, leads to a significant decrease of the
stiffness and maximum shear capacity in the transverse direction. An opposite trend is noticed in
the longitudinal direction of the straight bridges.
Comparison of base shear/weight ratio for different modelling techniques, skewness and
column bent is presented in Figure 4. Increase of total base shear is up to 15% when MA model is
used. Also, this increase is noticed for all skewness considered. In transversal direction it can be
contributed to shear key stiffness, in longitudinal direction to increase in resistance of distributed

453
Civil En g in eerin g – S cien ce a n d Pra ctice

pressure behind the backwall after the gap closure, both taken into account in MA model. Failure
of shear keys is not observed so the conclusions are limited only for such cases. Otherwise,
decrease of shear capacity could be expected after shear key failure. Considering the fact that
results do not vary greatly, simplified model may be a good solution for bridge structures whose
behaviour is not strongly controlled by the abutment response.

5. CONCLUSIONS

Abutment behaviour can significantly control seismic response of bridge structures. It is important
to to ensure that abutment model accurately represent the stiffness, mass and hysteresis behaviour
of all of its components: backwall, wingwall, shear keys if present, bearing pads and foundations.
In this paper six RC bridges are investigated. It is concluded that seismic response depend both on
skewness and bent type. More sensitive to increasing skew angle are bridges with single column
bent. Also, results of pushover analysis depend on approach used for abutment modelling even for
un-skewed bridges. Certain increase in base shear can be expected when more accurate model is
used. This can be contributed to the fact that more accurate model applied here takes into account
of shear key strength as well as it assumes distributed passive pressure behind the wall after the gap
closure.

LITERATURE

[1] Aviram A, Mackie KR, Stojadinovic B. (2008): Effect of abutment modeling on the seismic response
of bridge structures, Earthquake engineering and engineering vibration. 7(4):395-402
[2] Lam I and Martin GR (1986), Seismic Designof Highway Bridge Foundations, U.S. Dept. of
Transportation, Federal Highway Administration,
[3] Kaviani P. Zareian F. Taciroglu E. Performance-based seismic assessment of skewed bridges. Pacific
Earthquake Engineering Research Center; 2014.
[4] Aviram A. Mackie KR. Stojadinovic B. Effect of abutment modeling on the seismic response of
bridge structures. Earthquake engineering and engineering vibration. 2008;7(4):395-402.
[5] Caltrans. Seismic design criteria. California Department of Transportation. Sacramento (CA);
2006.Research, Development, and Technology.J. C. McCormac: "Structural Steel Design",
HarperCollins College Publishers, New York, 1995, 253p
[6] Kaviani P. Zareian F. Taciroglu E. Performance-based seismic assessment of skewed bridges. Pacific
Earthquake Engineering Research Center; 2014.

454

You might also like