Modelling of Bridges For Inelastic Analysis: Saiidi@unr - Edu

Chapter 2
Modelling of Bridges for Inelastic Analysis
M. Saiid Saiidi, Antonio Arêde, Donatello Cardone, Pedro Delgado,

Mauro Dolce, Matej Fischinger, Tatjana Isaković, Stavroula Pantazopoulou,
Gokhan Pekcan, Rui Pinho, and Anastasios Sextos
2.1 Introduction
Engineers rely on mathematical models to help analyze and design structures while
being aware that the actual structure might behave differently than that assumed in
the mathematical model. The ability of engineers to reduce the gap between the
actual and assumed behaviour of structures has increased substantially due to new
knowledge about material behaviour and the rapid development of computing tools.
Earthquake engineering of bridges is particularly challenging because of the variety
of components and parameters that affect the bridge seismic response. Inelastic
analysis of bridge response has the potential of being a powerful approach towards
realistic estimate of the actual behaviour of bridges. However, the power and
M. Saiid Saiidi (*)

Department of Civil and Environmental Engineering, University of Nevada,
Reno, NV 89557, USA
e-mail: [email protected]
A. Arêde
Departamento de Engenharia Civil, Faculdade de Engenharia, Universidade do Porto,
Rua Dr. Roberto Frias, s/n, 4200-465 Porto, Portugal
D. Cardone
Department of Structures, Geotechnics and Applied Geology, University of Basilicata,
Viale dell’Ateneo Lucano 10, 85100 Potenza, Italy
P. Delgado
Escola Superior de Tecnologia e Gestão, Instituto Politécnico de Viana do Castelo,
Avenida do Atlântico, 4900-348 Viana do Castelo, Portugal
M. Dolce
Department of Civil Protection, Via Ulpiano 11, 00193 Rome, Italy
A.J. Kappos et al. (eds.), Seismic Design and Assessment of Bridges: Inelastic Methods 5
of Analysis and Case Studies, Geotechnical, Geological and Earthquake Engineering 21,
DOI 10.1007/978-94-007-3943-7_2, # Springer Science+Business Media Dordrecht 2012
6 M. Saiid Saiidi et al.
versatility of inelastic analysis also makes the results particularly sensitive to

improper application. The material presented in this chapter is intended to inform
researchers and designers about available modelling methods and important
parameters to help ensure that inelastic analysis of bridge earthquake response is
conducted properly.
2.2 Superstructure (Deck)
This section addresses modelling issues of bridge decks occurring in dynamic

analysis of bridges. Generally, today’s finite element programs contain libraries
of elements that can support a range of options regarding the accuracy of represen-
tation of superstructure’s geometry and loading – however, practitioners rarely
resort to detailed 3D solid elements for that purpose, in order to minimize the
modelling effort and computational time. Rather, the degree of detail used in
idealization (modelling) of the superstructure in practical bridge analysis is often
associated with the type of loading applied.
In gravity load analysis the geometric complexity of the deck is usually
represented in the computer model to greater detail, as compared to that used for
transverse loads such as in analyzing ground-motion induced, seismic response.
This is probably justified on grounds of the perceived discrepancy in the amount of
flexural stiffness of the deck in transverse and longitudinal deformation as com-
pared to that of the more flexible piers, bearings, restrainer hardware and
embankments, which undertake accordingly the large portion of the bridge’s
M. Fischinger • T. Isaković
Faculty of Civil and Geodetic Engineering, University of Ljubljana,
Jamova 2 1000 Ljubljana, Slovenia
e-mail: [email protected]; [email protected]
S. Pantazopoulou
Department of Civil and Environmental Engineering, University of Cyprus,
Aglantzia, Cyprus
e-mail: [email protected]; [email protected]
G. Pekcan
Department of Civil and Environmental Engineering, University of Nevada,
Reno, 1664 N. Virginia Street, Reno, NV 89557, USA
R. Pinho
Department of Structural Mechanics, University of Pavia, Via Ferrata 1, 27100 Pavia, Italy
A. Sextos
Department of Civil Engineering, Aristotle University of Thessaloniki,
54124 Thessaloniki, Greece
2 Modelling of Bridges for Inelastic Analysis 7
displacement demand. For this reason, detailed modelling of the deck stiffness is
only considered of use in evaluating the two-way action of the deck, in assessing the
effects of vertical service loads and self weight.
In dynamic analysis for evaluation of seismic response, be it linear-elastic or
nonlinear, a common approach to deck modelling is to represent the superstructure
by a spine of linear beam-column elements with lumped nodal masses, spanning
between successive nodes along the deck’s length. The spine follows the centre of
gravity of the cross section along the length of the bridge. The deck geometry is
used to evaluate equivalent sectional frame-element properties, often using stan-
dard flexural analysis in order to evaluate the gross, or cracked sectional stiffness,
EI, of the deck acting as a single beam. Note that superstructure bending can
be expected to cause or enhance already existing cracking from gravity loads and
live loads in reinforced concrete structures. In such cases the usual practice is to
assume effective or cracked stiffness properties for the moment of inertia of the
entire cross section about the transverse or y axis. While detailed cracked-section
stiffness analyses can be performed for each girder or the complete superstructure,
it is often sufficient to calculate the gross section stiffness Ig and reduce it to
Ieff ¼ 0.5Ig for reinforced concrete, while assuming no stiffness reduction in
prestressed decks. The torsional rigidity J for spine elements can also be determined
from standard mechanics principles and can be considered as fully effective as long
as the torsional cracking moment is not exceeded, at which point the torsional
stiffness significantly reduces. The modelling procedure detailed above corres-
ponds to the assumption of a plane-sections response, whereby longitudinal and
transverse behaviour only depend upon the position along the longitudinal axis – a
uniform distribution of all effects occurs in the transverse direction in this type of
idealization, and this is why it is inappropriate for gravity load analysis.
California Department of Transportation (CALTRANS) endorses the approach
of modelling the bridge deck using a spine of 3D frame elements (6 degrees of
freedom at each node, and a minimum of three nodes in each span – at quarter
points), with pertinent assignment of lumped mass. Recommended stiffness
properties of the superstructure are based on the assumption of un-cracked cross
sections since the superstructure is expected to respond linearly to seismic loadings.
Alternatively, an equivalent orthotropic elastic slab system of uniform depth is
proposed, having the same overall stiffness in each primary direction with that of
the longitudinal and transverse beams of the original deck. In seismically isolated
bridges, the superstructure is modelled as a rigid body and is designed for forces
calculated using the effective stiffness of the isolation system at the design
displacement.
An alternative to the single orthotropic slab model for the deck is a hybrid shell –
beam finite element assembly. In this approach, the deck slab is modelled with
rectangular shell elements. Girders are modelled with 3D frame elements connected
to the shell elements at each node with vertical restraints in order to account for the
contribution of the girders’ weak-axis moment of inertia to the superstructure
stiffness for transverse loading. In the longitudinal direction, the end of the girder
is attached to a spring representing the elastomer’s lateral stiffness.
Consideration of the vertical component of ground motion and the coupling

between longitudinal, vertical and transverse response that occurs in curved bridges
under any kind of excitation, underscores the limitation of the single-spine 3D
frame model when applied to general purpose dynamic analysis of bridge
superstructures. More detailed than single-element beam-type idealization is finite
element discretization or other, simplifying modifications to this approach such as
finite strip, folded plate, and finite segment analysis. For example, a box-girder
bridge superstructure may be represented by a mesh of two-dimensional shell or
plate elements. A convenient compromise is the so-called grillage-type model,
which idealizes the structure as a space-frame configuration comprising a network
of longitudinal and transverse beams. It represents a balance between accuracy and
practicability of modelling approach, such as minimum effort for input preparation,
minimal computational effort and meaningful interpretation of output. Details for
grillage-type modelling of various deck types are outlined in the following section.
2.2.1 Deck Types, Sectional Layouts and Properties
The simplest deck type is that of a compact slab, which, from among the various
section types, possesses the lowest flexural stiffness. In a beam-type approximation
for either longitudinal or transverse action, interaction between the two directions is
a significant aspect that need be accounted for in any model short of 3D solid-
element representation. Therefore, in the grillage approximation the deck is
idealized as an assembly of basic modelling units, each comprising a rectangular
horizontal frame with diagonal elements; the latter are meant to account for the
interaction between longitudinal and transverse deck action owing to Poisson’s
effects, and cannot be neglected in the absence of primary girders. Properties of
the associated elements are listed in Table 2.1, where, Ix and Iy are the flexural
moments of inertia along the longitudinal and transverse directions of the bridge,
respectively, Jx and Jy are the torsional moments of inertia along the longitudinal and
transverse directions, Id the diagonal element flexural stiffness, n is the Poisson’s
ratio, Lx and Ly the breadth and length of the grillage unit, and t is the deck thickness.
2.2.1.1 Beam and Slab Bridge Decks
In the idealization of this deck system the longitudinal axes of the analytical model
are located on the axes of the girders, whereas the slab is idealized with a series of
transverse strips, each being modelled by a beam. Member properties for the
analytical model are listed in Table 2.1, where, Ix is the flexural moment of inertia
with respect to the centroidal axis of the combined section, Lx is the width of the
idealized transverse “beam”, Ly the width of the idealized longitudinal beam (equal
to the girders’ spacing). The expressions above neglect the Poisson’s effect and
assume that in the transverse direction flexural rigidity is approximately equal to the
torsional rigidity (valid for square configuration of grillages).
Table 2.1 Properties of equivalent members of decks for deck modellinga
Deck section Slab Girders Cellular Voided
" 1:5 #
Diagonals 2 2
nðLx þ Ly Þ t3
Jd ¼
Lx Ly 24ð1 n2 Þ
Shear area 12Lx Es Ay ¼ F1 t L x
Ay ¼
Py G s
½ns I3 ð12I1 I2 þ Px I1 I3 þ ns Px I2 I3 Þ=
h i
4Py I1 I2 þ 4ns Py I2 I3 þ 12H 2 I1 I2 þ ns P2y I32
t 3
Jy Es Lx ð1 3nÞ t3 E s t3 A2 v
Jy ¼ Jy ¼ Lx Jy ¼ 2Lx H 2ds Jy ¼ 1 0:84 Lx t6
Gs 24ð1 n2 Þ Gs 12 b ns t t

Jx Es Ly ð1 3nÞ t3 t3 A2 t t3
v
Jx ¼ Jy ¼ Jg;x þ Ly Jx ¼ Ly H 1ds Jx ¼ 1 0:84 Ly
6 t 6
2 Modelling of Bridges for Inelastic Analysis
Gs 24ð1 n2 Þ b ns t
" #
Iy L2y t3 t3 t31 t3 Lx t3 h t i4
v
Iy ¼ Lx n Iy ¼ Lx Iy ¼ Lx þ ns Lx 2 þ Lx t1 h21 þ ns t2 h22 Iy ¼ 1 0:95
Lx 24ð1 n2 Þ 12 12 12 12 t
4
Ix L2 t3 Ix ¼ Ig;x þ Is;x Ix ¼ Ig;x þ Is;x P y t3 t
Ix ¼ L x n x Ix ¼ ns v
Ly 24ð1 n2 Þ 12 64
Section notation
a
Increased transverse stiffness can be obtained at locations where transverse diaphragms are present in the superstructure. Transverse diaphragms can be
modelled with transverse grillage beam elements with characteristics derived from T or I sections and an effective flange width of eight times the slab
thickness on either side of the diaphragm
9
2.2.1.2 Cellular or Box-Section Bridge Decks
Idealization procedures in the longitudinal direction are similar to those described

in the preceding case for slab-and-beam bridge decks; idealized longitudinal
members are taken as I-beams placed along the axes of beam-webs of the structure
(centrelines between adjacent cells). The top and bottom slabs of the deck are
considered as top and bottom flanges of the idealized “beam”. Since the centre of
gravity of the deck is generally in a higher position with respect to the level of the
bearings, a rigid vertical link should be introduced connecting the deck axis and the
bearings. Torsional rigidities are based on the whole deck section rather than on the
individual idealized beams. Intermediate webs and overhangs are excluded in this
calculation due to their negligible contribution; thus, Jx and Jy are given in terms of
the area enclosed by the median line of the closed cross section, A1 (Table 2.1), 2b is
the width of the entire deck, ns the shear moduli ratio of bottom and top deck
flanges, and t the section’s thickness. The torsional inertia thus estimated is then
assumed uniformly distributed across the deck in proportion to the width of the
idealized members. Flexural inertia and shear area for idealization in the transverse
direction are listed in Table 2.1, where Es is the modulus of elasticity of the deck
slab, Gs is the shear modulus of the deck slab, I1 and I2 are the moments of inertia
per unit length of the top and bottom flanges, I3 is the moment of inertia per unit
length of the real web.
2.2.1.3 Voided-Slab Decks
Idealization is based on the same principles as used in cellular bridges. Note that
voids reduce significantly the transverse rigidity of the superstructure but the shear-
lag effect which is prominent in Cellular-type decks is not as significant in this case.
The necessary expressions for idealizing the deck as a system of longitudinal and
transverse beams are also given in the table.
2.2.1.4 Effect of Plan Curvature
In the presence of deck curvature, response to longitudinal or vertical action becomes

coupled to response in the transverse direction. The practical significance of this
coupling depends on whether the deck functions as an open or a closed system.
Open systems are curved deck types that lack transverse bracing (very small torsional
stiffness), thereby being susceptible to torsional loads. A general practice is to assume
uncracked torsional stiffness properties if the anticipated torques due to seismic loads
are low (below the cracking limit), but only 5% of that value if the cracking torque
limit is exceeded. This is particularly important in highly curved superstructures,
because a result of the decks’ geometric curvature is the natural development of
a distributed torque associated with the diagram of longitudinal flexural moments.
Thus, this aspect of the behaviour need be properly reflected in the analytical model.
Closed systems can be analyzed as single-section curved beams, or using the multi-
beam idealization discussed previously for straight decks.
2.2.1.5 Effect of Prestressing on Component Verification Issues
The general practice is to use gross section properties when modelling the flexural
rigidity of prestressed deck sections, if the estimated seismic demands for the deck
are below cracking, whereas the average cracked stiffness is used in cases where the
anticipated demands exceed the cracking limit. Note however, that the prestressing
force affects primarily the deck’s stiffness through its influence on the extent of
flexural cracking. Therefore, under load reversals the stiffness after cracking greatly
depends on the location of the prestressing cable’s centroid and the direction of
bending, an aspect that may only be represented through non-linear modelling of
the deck member behaviour even in the pre-yielding range of response; linear
elastic idealization with average values cannot reproduce the implications of
stiffness variation imparted on the prestressed section by moment reversal. There-
fore, no uniform stiffness reduction is recommended for prestressed concrete box
girder sections to account for the effect of cracking, but rather a detailed moment-
curvature analysis may be needed in order to assess the actual deck section’s
flexural stiffness for nonlinear response history analysis (NRHA).
2.2.2 Role of Deck Modelling in Seismic Assessment
Either conducted through standard pushover analysis (SPA) or through NRHA in

assembling the bridge model the superstructure is usually considered to remain in
the elastic range. Due to its in-plane rigidity, it is often assumed to move as a rigid
body under seismic loads, and the entire modelling problem is reduced to modelling
of the bents with geometric constraints simulating the rigid superstructure. Nonlin-
ear modelling is then restricted to represent joints between superstructure segments,
connections with supporting bents, and to the assessment analyses of older bridges,
where the superstructure may not be protected against inelastic action by capacity
design principles. In general, the standard pushover analysis can provide satisfac-
tory predictions to the maximum displacement responses of straight bridges in the
longitudinal direction (residual displacements tend to differ even in this case, due to
the different hysteretic models used in detailed nonlinear response history analysis
as compared to the equivalent single-degree-of-freedom (SDOF) approximation of
the SPA). In some cases, response of long bridges in the longitudinal direction is
complicated by the impact between adjacent spans. When subjected to seismic
loading, movement or expansion joints allow separate frames to develop their own
characteristic dynamic response and to modify this individual dynamic response
through complex interactions that occur in the restraining devices installed in the
movement joints. For an elastic model or a plane frame inelastic static model to be
applicable, adjacent frames should have comparable stiffness and mass
characteristics in order for the structure to respond in its fundamental mode of
vibration. Unbalanced frame geometry will lead to larger relative displacements
and transfer of increased seismic force from one frame to the other. Thus, if the
linear elastic displacements to the design earthquake exceed the expansion joint
width, the impact between successive spans should be modelled as well, in the
framework of NRHA.
But the difference between SPA and NRHA is particularly relevant when the
bridge response is analyzed in the transverse direction. Note that in this case
superstructure stiffness greatly affects the seismic behaviour of the bridge: excessive
deck stiffness will equalize displacement demands on the piers leading to damage
concentration in shorter piers and end abutments and embankments; flexible decks
will lead to decoupled pier behaviour. Thus, it is important to faithfully represent in
the model the bridge deck flexibility, particularly in the case of long structures, for
two practical reasons: (a) in order to capture higher mode contributions which can
become significant for irregular systems, (b) so that the distribution of seismic
forces, which are applied along the length of the deck, follow the pattern of lateral
deflection occurring in the deck due to its own compliance. A first step in order to
obtain a reliable pattern of lateral load distribution, which may then be used in
conducting SPA on long or irregular decks is to perform first a standard deflection
analysis of the deck, whereby the deck model is loaded in the transverse direction by
the superstructure’s own gravity loads while accurately representing the support
restraints against transverse translation or rotation, if any exist. In this analysis
standard linear elastic modelling with gross sectional properties as listed in Table 2.1
are sufficient. Where vertical superstructure flexibility reduces the fixity of the top
boundary conditions of columns and piers, a reduction in the stiffness coefficient can
be determined. Stiffness contribution of non-structural elements is generally
neglected; however, their contribution to inertia forces (masses) should be considered.
2.2.3 Effects of Skew and Curvature in Plan
Other issues pertaining to deck’s seismic demand assessment are, (a) the effects
of highly skew supports on bridge response in both transverse and longitudinal
direction, (b) contribution of rotational inertia of the superstructure mass in trans-
verse vibration of the bridge (particularly relevant in the case of tall piers and
massive deck cross sections) and (c) contribution of ramp structures and inter-
actions between frames, (d) in-plan and in-elevation curvature, (e) Ductility and
plastic hinge formation in secondary deck elements (continuity slabs). Clearly these
effects may only be accurately represented through 3Dimensional models that
explicitly account for all the associated kinetic and kinematic interactions. Note
that particularly skew bridges have a tendency to rotate in the horizontal plane even
under non-seismic loading. Skewness contributes to high shear forces in central pier
columns and to an increase in deck displacement. In general, longitudinal shaking
produces transverse components of force, and vice versa. Transverse shaking

causes one end of a deck panel to bear against the adjacent deck while the opposite
end swings free in response to the seismic loading, resulting in a ratcheting type of
behaviour under cyclic loading. The non-uniformity of movement joints behaviour
(opening-pounding) necessitates their modelling using their exact geometry.
To effectively reflect in the analytical model the stiffness changes of restrainers
and shear keys that occur during cyclic loading, nonlinear spring elements with
pertinent skeleton and hysteresis rules are usually required. Nonlinear impact at the
points of contact of successive skew decks must be reflected in either longitudinal
or transverse response analysis through pertinent modelling of the interaction.
To enable nonlinear contact behaviour, NRHA appears to be the preferred option.
For this type of problem simplifications using either rigid deck or single-spine
beam-type representation of the deck are inappropriate as they generally fail to
reproduce the axial forces induced in the piers by the earthquake.
Rather, a two-dimensional grillage model for the superstructure can be effective
as long as a sufficient number of transverse restraints through idealized “beams”
spanning between longitudinal girders are provided in the model, and the applied
loads are distributed to all nodes over the bridge deck area. As described in the
relevant section, torsional stiffness properties of the deck superstructure are
distributed equally to the number of longitudinal grillage elements.
As far as in-plan curvature is concerned, the curvature can increase the transverse
stiffness of the system depending on its continuity. The forces in the expansion joints
are high and the movements potentially large, and hence the joints are very prone to
damage caused by multiple impacting in both torsional and translational modes of
response. A curved bridge without expansion joints would be extremely stiff in the
horizontal plane, particularly if fully restrained at the abutments.
Secondary deck elements such as continuity slabs form the extension of the deck
and are part of the external restraining system of the bridge. They act as a tension tie
when the deck contracts, or as a compression strut when the deck expands. During
an earthquake, the continuity slab can be either loaded by compression or tension,
due to the longitudinal movements of the deck, or even by in-plane bending, due to
the transverse response of the deck. Other secondary elements such as the link slabs
connect adjacent simple-span girders meant by design to reduce or eliminate
cracking in the concrete deck can be modelled using the cracked stiffness properties
(secant to yield).
2.2.4 Verification of Deck Deformation Demands
Uncracked element section properties are typically used in practice for verification
checks. This is conservative for force-controlled members, but can lead to unsafe
assessment of damage which is characterized by displacement – this is one reason
why reduced (cracked) stiffness properties will moderate the error in bridge assess-
ment. Furthermore, for the purpose of seismic demand estimation, structural period
should be estimated after the application of service loads.
Recent standards and guidelines for seismic assessment define the seismic input
in terms of displacement demands, where the estimation of inelastic local displace-
ments and deformation demands throughout the structure are conducted via linear
static or modal response spectrum analyses. A fundamental requirement for this
estimation is the use of a realistic estimation of the global elastic stiffness which, for
concrete structures, corresponds to using realistic values of the effective cracked
stiffness of concrete members at yielding.
When seismic response is evaluated by comparing member deformation
demands to (realistic) deformation capacities, then the demands should also be
realistically estimated: if member stiffness is taken equal to the default value of
0.5EcIc, currently recommended for force- and strength-based seismic design of
new buildings by Eurocode 8 (EC8) and US codes, then member seismic deforma-
tion demands may be seriously underestimated.
The best way to realistically estimate the effective elastic stiffness of the shear
span of a concrete member (moment-to-shear ratio at the end, Ls ¼ M/V) in a
bilinear force-deformation model under monotonic loading, is to use the value of
the secant stiffness of the shear span at member yielding EIeff ¼ MyLs/3yy where
My is the value of the yield moment in the bilinear M-y model of the shear span
and yy that of the chord rotation at the yielding end.
Generally it is much more appropriate to conduct verification checks using
detailed M-f analysis of the deck sections in the critical regions, obtained after
consideration of all the necessary interactions (effective flange width, prestressing,
confinement, prestress losses, reinforcement slip, axial load, moment, and shear
interaction) which may be estimated using web-based available freeware (e.g.,
RESPONSE, ANSR, ETOOLS). Recent research has shown that pier rotation
capacity for piers designed according to current AASHTO requirements exceeds
the value of 4% even for cases with a high axial load value (Inel and Aschheim
2004). Closed form expressions for pier rotation capacity at drift and ultimate have
been developed by Biskinis and Fardis (2007):
For circular section columns of diameter D, the chord rotation at yield is given
by the expression:

Ls þ av z Ls fy db fy
yy ¼ f y þ 0:0022 max 0; 1 þ asl pffiffiffiffi (2.1)
3 6D 8 fc
and the ultimate chord rotation is given as follows:

0:5Lpl
yu ¼ yy þ ðfu fy ÞLpl 1 (2.2)
Ls
where av is a coefficient depending on the relationship between the shear strength

and the shear corresponding to flexural yielding in the member (av =1 if the former is
higher than the latter), z is the internal lever arm of section forces, asl is a coefficient
depending on the adequacy of lap splices (asl = 0 if bar pull-out is not physically
possible, otherwise asl = 1), fy is the yield curvature, db the bar diameter, Lpl is the
plastic hinge length, and the other parameters have been defined previously.
For rectangular hollow section columns or T-shaped and double-T, these

expressions are respectively:
Ls þ av z fy db fy
yy ¼ 1:065fy þ 0:0012 þ asl pffiffiffiffi ; (2.3)
3 8 fc
and

h
yu ¼ yy þ 1:05ð1 þ 0:6asl Þ 1 0:05 max 1:5; min 10; ð0:2Þn
bw
fyw
maxð0:01; o0 ÞLs 1=3 0:2
fc 25 ars fc ð2:4Þ
maxð0:01; oÞh
where h is the section depth, and o and o’ are the mechanical reinforcement ratios
(tension /compression), rs is the volumetric ratio of transverse reinforcement, and a
is a confinement effectiveness factor (a=0 for unconfined sections).
If the plastic mechanism of the bridge response is known, member deforma-
tion demands (usually and conveniently estimated in the form of chord rotation
demands at member ends) are related to the global displacement of the superstructure
through pertinent geometric relations that are prescribed by the fundamental shape of
bridge vibration (if an equivalent SDOF analysis is conducted), or otherwise estimated
from detailed Finite Element Analysis. Thus, local deformation demands may be
evaluated, corresponding to the bridge superstructure design displacement.
For composite concrete-steel members, the section properties should be adjusted
to an equivalent concrete or steel section considering the material modular ratio.
Where different concrete strengths are used in the same element (such as different
concrete strengths in prestressed-concrete girders and cast-in-place concrete deck),
the section properties should be transformed using a similar procedure. The com-
posite densities should also be transformed to an equivalent concrete or steel
density so that mass distribution may be dependably quantified.
2.3 Bearings and Shear Keys
The overall seismic response of bridges may be influenced significantly by the

details at the abutments which may include bearings, shear keys, gap, abutment
back wall, and piles. Significant soil-pile-abutment-structure interactions may take
place that would lead to complex seismic response behaviour. Hence, a realistic
abutment model should represent all major resistance mechanisms and components,
including an accurate estimation of their mass, stiffness, and nonlinear hysteretic
behaviour. In addition, bearings may be located at the piers between cap beam and
the superstructure. In this section, the importance of accurate modelling of bearings
and shear keys as well as simple modelling approaches are presented briefly, while
other details are addressed elsewhere in this report. It is noted that representation of
realistic nonlinear force-deformation relationships of bearings and shear keys is

critically important for the nonlinear seismic response characterization of bridges.
Various types of abutment models have been developed and used by researchers,
modelling techniques and details are presented by Shamsabadi (2006), Aviram
et al. (2008) and Abdel-Mohti and Pekcan (2008), which in essence can be depicted
as shown in Fig. 2.6. However, the modelling requirements and techniques for
bearings and shear keys depend on the type of bearing and shear key, and type
of analysis. Typically, for nonlinear static and nonlinear dynamic analyses, lateral
force-deformation response must be modelled explicitly in terms of a backbone
(monotonic) and complete hysteretic rules, respectively.
2.3.1 Importance of Modelling of Bearings
Bridge bearings are mechanical systems which permit movements and transmit
various types of loads from the bridge superstructure to the substructure. While the
vertical and horizontal loads are routinely applied to bridge bearings due to gravity,
traffic, wind, and other internal sources (e.g. temperature, creep, and shrinkage),
seismic loads are relatively infrequent. Hence they were not considered consistently
in the design of bridges until late 1980s. In spite of the role of the bearings in
transmitting these loads, their importance has been often neglected (Leonhardt
1981) and the importance of accurate modelling of bearings for the seismic perfor-
mance assessment of highway bridges has been overlooked. Fortunately, the level of
awareness has increased since the invention and wide acceptance of “modern-type”
bearings primarily for seismic applications such as PTFE sliding, elastomeric, FPS,
etc. Subsequently, integrated analysis and design software that incorporate macro-
level approaches to modelling of bearings have become widely available.
2.3.2 Mechanical Bearings (Steel Bearings)
Selection of proper bearing type for a bridge design is based on many consi-
derations which include the regional seismic hazard levels. It is noted that most
of existing bridges in moderate earthquake regions are more than 50 years old. The
majority of these structures were designed without any consideration of earthquake
forces. Hence, the so-called steel bearings (Fig. 2.1) are used in bridges that are
still in service in Eastern and Central U.S., Japan, Taiwan, among others. Although
current practice requires retrofitting of existing bridges using more advanced bearing
types, e.g. isolation bearings, in zones of low to medium seismicity, existing steel
bearings may possess sufficient strength and/or displacement capacity. Regardless,
realistic force-deformation response of structural components is of fundamental
importance to the process of analysis and design. With this motivation, several
experimental and analytical studies have been conducted. Mander et al. introduced
reliable hysteretic models for sliding, fixed, and rocker, bearings. The hysteretic
Fig. 2.1 Steel bearing types

Fig. 2.2 Views of different isolators: (a) elastomeric, (b) sliding, and (c) friction pendulum
response of these bearing types was modelled combining a bilinear yielding

element and a link element using Drain-2DX software.
2.3.3 Modern Bearing Types
Modern bearings are ductile, multi-rotational, and multi-directional. Accordingly,

they can be identified under two broad categories, namely, elastomeric bearings and
high-load multi-rotational bearings (e.g. Fig. 2.2). Plain elastomeric bearing pads
(PEP), steel reinforced elastomeric bearings, and lead-rubber bearings (LRB) are
the most commonly used elastomeric bearing types. Plain elastomeric bearing pads
rely upon friction at the contact surface to resist building strains. The traditional
non-seismic elastomeric pads used in bridges for thermal movements can provide
some limited seismic protection; however their integrity during large displace-
ments might be substantially deteriorated or even destroyed due to shearing of
the elastomer or rolling of the entire bearing. Accordingly, elastomeric pads with
improved seismic performance have been developed. PEPs can be modelled with an
elasto-plastic hysteresis model (e.g. plastic –Wen– link element in SAP2000) with
an initial elastic stiffness, k0.
GA
ko ¼ (2.5)
h
Force
K2
Q
K1
-D Keff D
Displacement
Fig. 2.3 Hysteresis parameters for elastomeric and lead rubber bearings (Naeim and Kelly 1999)
where, G ¼ shear modulus of elasticity, A ¼ pad area, h ¼ pad thickness.

Yielding is assumed to take place at 100% strain, and failure is assumed to take
place at 150% strain.
The nonlinear hysteretic response of elastomeric bearings, lead-rubber bearings
as well as Friction Pendulum Systems (FPS) can be essentially modelled using the
bilinear relationship shown in Fig. 2.3. In practice, all isolation bearings can be
modelled based on the three parameters K1, K2, and Q. The elastic stiffness K1 is
either estimated from available hysteresis loops from experiments or as a multiple
of K2 for LRB and FPS bearings. The characteristic strength Q is also characterized
experimentally for rubber bearings, however, for LRBs it is given based on the
yield stress in the lead and the area of the lead plug, while for FPS bearings it is
given by the friction coefficient of the sliding surface and the load carried by the
bearing (Naeim and Kelly 1999). SAP2000 (CSI 2007) provides a special biaxial
isolator element that has coupled plasticity properties for two shear deformations,
and linear effective stiffness properties for the remaining four deformations. The
plasticity model is based on the hysteretic behaviour proposed by Park et al. (1986)
and recommended for the analysis of base-isolated systems by Nagarajaiah et al.
(1991). The accuracy of this element in SAP2000 was verified against experimental
data. Element models for friction pendulum systems are discussed in Zayas and
Low (1990) and adopted in SAP2000. Specialized elements for these bearings can
be found in other standard structural analysis packages such as GT Strudl (2006)
and are summarized in Buckle et al. (2006).
There are three common high-load multi-rotational bearing types that function in
essentially the same manner: pot bearings, disc bearings, and spherical bearings
(e.g. Fig. 2.4). These bearings are typically designed to allow superstructure rotation,
or may also allow relative translation in one or more directions. The most common
type of rotating bearing is the pot bearing while translation may be provided by
SOLE PL
PISTON ASSEMBLY
STAINLESS STEEL CL STAINLESS STEEL
TOP PLATE
GUIDE BAR GUIDE BAR

BRASS RING
MASONRY PL BASE PL NEOPRENE DISC
SOLE PL
TOP PL PISTON
CL
BRASS RING
MASONRY PL BASE PL NEOPRENE DISC
Fig. 2.4 Expansion and fixed multi-rotational bearings
elastomeric bearing pads, by polytetrafluoroethylene (PTFE) stainless steel sliders, or

by rocker bearings. The force-deformation response of these types of bearings –if not
restrained by design– can be characterized essentially with either elastic stiffness, or by
Coulomb friction. It is noted that in the latter case, the nonlinear link element for FPS
bearings in SAP2000 can be used with a zero (0) curvature, i.e. essentially flat surface.
It is commonly accepted that PTFE bearings may be used as seismic isolators, but
should only be used in conjunction with other devices that provide recentring.
2.3.4 Modelling of Shear Keys
Shear keys are commonly used in bridges with seat-type abutments to provide
transverse support for the superstructure under service loads and earthquakes.
They do not carry gravity loads, but in the event of an earthquake they are required
to transfer the lateral reactions of the superstructure to the abutment or across
movement joints. Sacrificial shear keys serve as structural fuses to control damage
in abutments and the supporting piles under extreme transverse seismic loads.
Sacrificial shear keys may be interior or exterior; exterior shear keys are usually
a b
Superstructure Superstructure
Exterior shear key
Abutment Interior shear

keys Abutment
Fig. 2.5 Schematic view of exterior and interior shear keys
recommended for new construction because they are easy to inspect and repair.
Interior shear keys are constructed within the abutment, and exterior shear keys are
provided at the sides of the superstructure (Fig. 2.5).
Typically, for the design of highway bridges in seismic zones two distinct
conditions are assumed (1) shear keys remain intact and (2) shear keys fail
completely. These two conditions are intended to establish conservative bounds
on the deformation and force response of other structural elements such as bent
columns, piles, etc. (Caltrans 2006). In view of these preconditions, computer
models of a given bridge are developed both with and without the shear keys.
The former may be provided with proper elastic stiffness associated with the shear
key design and geometry, or alternatively the transverse displacements at the
shear key location are restrained. In order to identify the state of the shear key,
force demand due to the earthquake is compared with the calculated capacity of the
shear key. Capacity of shear keys is normally calculated based on a shear friction
theory. The capacity may be also estimated using the strut and tie method (ACI
2008). Experimental studies of large-scale models of shear keys have shown that the
mode of failure may be different than the shear friction method (Silva et al. 2003) for
seismic demands below the capacity of the shear keys, they can be modelled by using
a compression-only gap element. The stiffness after gap closure would be based on
the cracked shearing stiffness of the shear key. For seismic demands beyond the
capacity of the shear key, the computer models should ignore shear keys.
Recently, there has been considerable research on the failure of plain and
reinforced concrete shear keys. Laboratory experiments conducted on the seismic
performance of shear keys designed according to current seismic design criteria
have indicated that there may be significant variability in the actual strength versus
design values (Bozorgzadeh et al. 2006; Megally et al. 2001). These observations
prompted the need to better model the complete nonlinear response characteris-
tics for the seismic response assessment of bridges. Moreover, recent analytical
studies have demonstrated that consideration of the nonlinear shear key response
may be critical for the reliable and accurate assessment of the overall seismic
response of highway bridges (Saiidi et al. 2001); particularly irregular bridges
with large skew (Goel and Chopra 2008; Abdel-Mohti and Pekcan 2008). The latter
study demonstrated that when bridges were subjected to large biaxial ground
motion excitations, complete shear key failure –almost exclusively- preceded the
unseating at the abutments of skew bridges. On the other hand, in bridges with some
of the shear keys in near failure condition, the unseating may still be prevented.
Clearly, it would only be possible to capture this response, if shear keys with
realistic hysteretic properties are explicitly modelled.
Figure 2.6b shows an idealized force-deformation envelope for exterior shear
keys. This idealized hysteretic model was demonstrated in terms of two models
which correspond to the concrete and steel components. The above mentioned skew
bridge studies followed the procedures presented by Megally et al. (2001) for the
capacity determination of external shear keys. While the modelling of this unique
hysteretic response behaviour may be accomplished relatively directly using anal-
ysis software such as Drain-2DX or Drain-3DX (Fig. 2.7), the same nonlinear
external shear-key response can be modelled using a combination of link elements
as shown in Fig. 2.6a using SAP2000, OpenSEES (Mckenna et al. 2000), or other
similar software.
Experimental cyclic response of a shear-key (Megally et al. 2001) is compared to
that of a key modelled as described above using SAP2000 in Fig. 2.8. Finally, it is
noted that the abutment transverse shear key behaviour can be modelled using the
elements depicted in Fig. 2.8. These elements can address the structural capacity of
the shear key as well as contribution of passive resistance of the abutment embank-
ment in the transverse direction as a function of relative displacement between
bridge deck and abutment. The model may include gap between shear key and deck,
and the limiting passive capacity of the embankment soil in the transverse direction.
2.4 Isolation and Energy Dissipation Devices
Several types of isolation systems are in use today and many new solutions are
continuously being proposed and investigated (Skinner et al. 1993; Higashino and
Okamoto 2006; Naeim and Kelly 1999). Most currently used isolation systems
include: (i) lead rubber bearings (LRB) (Fig. 2.9a), (ii) high-damping or added-
damping rubber bearings (HDRB/ADRB) (Fig. 2.9b), (iii) friction pendulum
bearings (FPB) (Fig. 2.9c), (iv) Combinations of either low-damping rubber
bearings (LDRB) or FPB with viscous dampers (VD), (v) Combinations of flat
sliding bearings (FSB) (Fig. 2.9d) and LDRB, (vi) Combinations of FSB and steel
yielding (SY) devices (Fig. 2.9e), (vii) Combinations of FSB, SMA (shape memory
alloy)-based re-centring devices (Fig. 2.9f) and VD. Sliding bearings used in seismic
isolation typically exploit the low friction between PTFE (Polytetrafluoroethylene
or Teflon) pads in contact with lubricated polished stainless steel surfaces.
Basically four different idealized force-displacement models can be used to
describe the cyclic behaviour of the aforementioned systems. The model para-
meters are identified in Fig. 2.9 and typical values are reported below. They have
been derived from literature review and examination of product brochures of
several seismic isolation manufactures.
The first model (see Fig. 2.9a) represents a visco-elastic behaviour. It can be
used to describe the cyclic behaviour of HDRB and low damping rubber bearings
(LDRB) (Taylor et al. 1992). The main difference between HDRB and LDRB
Fig. 2.6 (a) Components of a bridge model (Shamsabadi 2006). (b) Hysteresis rule for exterior
shear keys (Megally et al. 2001)
Fig. 2.7 Type 09 element in drain-2DX and drain-3DX
Fig. 2.8 Modelling of inelastic shear key force-deformation response

Fig. 2.9 Common isolation systems and associated force-displacement behaviours: (a) lead
rubber bearings, (b) added-damping rubber bearings, (c) friction pendulum bearings, (d) flat
sliding bearings, (e) steel yielding devices, (f) SMA-based re-centring devices
consists in the damping capacity of the elastomeric material. The elastomers (either
natural or synthetic rubber) used in LDRB usually provide viscous damping ratios
not greater than 5%. The rubber compounds used in HDRB, however, provide an
effective damping which is around 15–18% at low (25–50%) shear strains, reducing
to 12–15% for shear strains greater than 100% (Derham et al. 1985). Using the
ADRB (Dolce et al. 2003), equivalent viscous damping ratios as high as 20–25%
can be reached.
The second model (see Fig. 2.9b) represents a bilinear with hardening behaviour.
It can be used to describe the cyclic behaviour of either SY devices or LRB (Skinner
et al. 1993). The lead plug of LRB exhibits an initial shear modulus equal to
approximately 130 MPa (compared to that of rubber which typically ranges from
0.4 to 1.2 MPa at 100% shear strain) and a yield shear strain of approximately 7.7%
(Kelly 1992). The yield displacement of LRB (Dy in Fig. 2.9b) is, therefore,
approximately equal to 7.7% H, H being the height of the device. Typical heights of
LRB vary from 100 to 350 mm. As a result, yield displacements ranging from 7.7
to 26.7 mm are then expected for LRB. In practice, the cross section area of the
lead plug is around 5–10% the gross section area of rubber (Skinner et al. 1993).
As a consequence, post-yield stiffness ratios (r in Fig. 2.9b) of the order of 5–15%
are expected. The maximum displacement capacity of LRB is governed by the
allowable shear strain of rubber and by the global stability of the device under
vertical load. Hence, it is governed by the total height of rubber and the cross
section dimensions of the device (diameter D for circular bearings, side dimensions
bx and by for rectangular bearings). Typically, design displacements between
120 mm and 350 mm, corresponding to rubber shear strains of 100–120% and
lead ductility ratios of 10–20, can be assumed for LRB, although much greater
rubber shear strains (200–250%) and ductility ratios (20–40) can be cyclically
sustained (Skinner et al. 1993). The aforementioned limits lead to effective
damping ratios ranging from 15% to 25%.
The third model (Fig. 2.9c) represents a rigid-plastic with hardening behaviour,
which can be used to capture the cyclic response of both FPS (Al-Hussaini et al.
1994) and combinations of FSB and LDRB. Sliding bearings used in seismic
isolation typically exploit the low friction between PTFE pads in contact with
lubricated polished stainless steel surfaces. The dynamic friction coefficient of
PTFE- steel sliding bearings (mFR in Fig. 2.9c) depends on a number of factors,
e.g.: the sliding surface conditions, the bearing pressure, the velocity of movement
and the air temperature. The friction coefficient of lubricated PTFE-steel sliding
bearings normally varies between 2% and 5%, while increasing up to 10–12% for
pure PTFE-steel surface (Dolce et al. 2005). The post-sliding stiffness of single
concave FPB is defined as W/Rc (Al-Hussaini et al. 1994), where Rc is the effective
radius of curvature of the sliding interface and W the supported weight. Similarly,
for double concave FPB with equal friction on the two sliding interfaces, the post-
sliding stiffness is defined as W/(Rc1 þ Rc2), Rc1 and Rc2 being the radii of curvature
of the two sliding surfaces. In principle, there is no theoretical limit to the displacement
capacity of FPB, provided that a device with the required dimensions can be
manufactured. Actually, the horizontal displacement capacity of FPB is conditioned
by the acceptability of the corresponding vertical displacement and residual horizontal
displacement. Both are a function of the radius of curvature Rc. As a consequence,
limitations to the ratio between the design isolation system displacement Dd and the
radius of curvature Rc are necessary to limit vertical and residual displacements.
Reasonable values of the ratio Dd/Rc are between mFR,max (e.g. 5% for lubricated
interfaces) (Naeim and Kelly 1999) and 15%. FPB with radius of curvature, ranging
from 1 to 10 m, are commercially available (Naeim and Kelly 1999). Considering the
aforementioned parameters, effective damping ratios will be in the order of 10–20%.
The idealised cyclic behaviour of SB þ LDRB is very similar to that of LRB, except
for the recentring capacity of the systems, which relies upon the shear stiffness of
rubber (kr). The viscous damping of rubber (5%) leads to greater effective damping
ratios (hence higher non-linearity factors) compared to FPB.
a HDRB/ADRB
F=KD F=cvα α=1 α<1

K0
ξ0
D D
b LRB
K1 F=rK1D
rK1
F=f(D) (1-r)K1
Fy rK1
D D
Dy Dy
c
FPB/FSB+LDR
B
F=KD F=sign(μFRW)
μFRW
2μFRW
K0
D D
d
FSB+SMA
F=f(D) F=sign(μFRW)
F2 2μFRW
F1 F2 μFRW
F1 rK1
K1
Dy Dy D D
Fig. 2.10 Schematic force-displacement behaviour of the currently used isolation systems
As an alternative to LDRB, the SMA-based re-centring devices, designed,

engineered and tested by Dolce et al. (2000) can be used. The SMA-based devices
proposed by Dolce et al. (2000) exhibit a cyclic behaviour that can be schematized
as bilinear elastic, although some energy is also dissipated by SMA, typically
resulting in 3–5% damping. The fourth model (referred to as double flag shaped
model) derives from the combination of a bilinear superelastic behaviour, modelling
the typical force-displacement cycles of SMA-based recentring devices and a rigid-
plastic behaviour, reproducing the schematic force-displacement cycles of FSB
(see Fig. 2.10d).
Linear (a ¼ 1) or nonlinear (1 < a < 0.2) viscous models (see right hand side
of Fig. 2.9a) are used to take into account possible auxiliary viscous dampers
(Constantinou et al. 1993).
The equivalent linear modelling of systems with nonlinear mechanical
behaviour (i.e. LRB, SY, FPB, SB þ LDRB, SB þ SMA) is based on the defini-
tion of the effective stiffness and equivalent viscous damping, which accounts for
the energy dissipated by the isolation systems during the seismic excitation through
its viscous, frictional or hysteretic behaviour.
The effective stiffness (Ke) is defined as the secant stiffness to the design displace-
ment Dd. The effective stiffness is related to the effective period of vibration (TIS) of
the isolated structure. Reference can be made to the following relationships:
sffiffiffiffiffiffiffiffiffiffiffiffi
W
Ke ¼ ð1:1 0:9Þ K0 ) Tis ¼ 2p for HDRB=ADRI with 50%<g<200
g K0
(2.6)
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
K1 þ r K1 ðm 1Þ Wm
Ke ¼ ) Tis ¼ 2p for LRB/SY
m g½K1 þ r K1 ðm 1Þ
(2.7)
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
W mFR W u
u 1
Ke ¼ þ ) Tis ¼ 2p u for FPB (2.8)
Rc Dd t 1 mFR
g þ
Rc Dd
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
mFR W u
u 1
Ke ¼ K0 þ ) Tis ¼ 2p u for SB þ LDRB (2.9)
Dd t K 0 m
g þ FR
W Dd
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
F2 mFR W u
u 1
Ke ¼ þ ) Tis ¼ 2p u for SB þ SMA (2.10)
Dd Dd t F2 m
g þ FR
W Dd Dd
where:
K0 ¼ G0 ðAb =te Þ
G0 is the shear modulus of the elastomer at 100% shear strain,

Ab is the effective horizontal area of the bearing,
te is thetotal thickness of the elastomer,
m ¼ Dd Dy is the ductility ratio,
r ¼ (k2/k1) is the post-yield hardening ratio,
mFR is the friction coefficient of the sliding bearings,

Rc is the radius of curvature of FPB,
W is the total weight of the building (base floor included),
F2 is the force of SMA at the design displacement Dd,
g is the acceleration of gravity.
The equivalent viscous damping is computed based on either analytical-numerical
formulations or semi-empirical relationships, as a function of the main mechanical
parameters of the isolation system and the expected maximum displacement.
In first approximation, the equivalent viscous damping can be derived based on
the well-known Jacobsen’s equation:
Wd Wviscous þ Whysteresis þ Wfriction

xeq ¼ ¼ (2.11)
4p Ws 2p Fd Dd
in which Wd is the total energy dissipated by the isolation system in the cycle of
maximum amplitude (Dd), Ws is the strain energy stored at the design displacement
Dd and Fd the corresponding force, equal to Ke Dd
The general expression of the equivalent damping ratio given in Eq. 2.11 can be
specialized to each type of isolation system as follows:
xeq ¼ ð1:1 0:9Þ x0 for HDRB=ADRI with 50%<g<200% (2.12)
2 ðr 1Þ ð1 rÞ
xeq ¼ for LRB=SY (2.13)
p m ½1 þ r ðm 1Þ
2 mFR
xeq ¼ for FPB (2.14)
p Dd
mFR þ
Rc
2 mFR
xeq ¼ for SB þ LDRB (2.15)
p K0 Dd
mFR þ
W
2 mFR
xeq ¼ for SB þ SMA (2.16)
p F2
mFR þ
W
For viscous-elastic isolation systems (i.e. HDRB/ADRI, see Figs. 2.9b and
2.10a), the equivalent damping ratio can be assumed practically constant and
equal to the viscous damping at 100% shear strain. This implies a great flexibility
in the choice of the design displacement of the isolation system (Dd), provided that
the necessary damping ratio falls within suitable ranges.
For elasto-plastic isolation systems (i.e. LRB/SY, see Figs. 2.9a, e and 2.10b), xeq
depends on Dd through the ductility ratio. In principle, for a given design displacement
Dd, it is possible to get a great variety of damping ratio values by properly selecting
the yield displacement (Dy) and post-yield stiffness ratio (r) of the isolation system.
Fig. 2.11 Preliminary selection of isolation system type, design displacement (and pier reinforce-
ment ratio) for the (a) retrofit of existing bridges and (b) design of new bridges
For friction-based isolation systems (i.e. FPB, FSB þ LDRB, FSB þ SMA, see
Fig. 2.10c, d), xeq depends on the friction coefficient (mFR), the design displacement
of the isolation system (Dd) and the secant stiffness of the auxiliary recentring
system/mechanism to the design displacement (i.e. W/Rc, K0 and F2 /W for FPB,
FSB þ LDRB, FSB þ SMA, respectively). This latter, however, is obtained only
at the end of the design process. As a consequence, for FPS, FSB þ LDRB, and
FSB þ SMA, an iterative design procedure is required, since input and output
are mutually correlated. In other words, the design displacement of the isolation
system (Dd) cannot be arbitrarily selected; otherwise the design procedure could
fail to converge.
A preliminary selection of the isolation system type, design displacement, and
(for new bridges) optimal pier reinforcement ratio is strongly recommended. This
can be done using graphical tools similar to those reported in Fig. 2.11a, b, for
existing and new bridges, respectively. The diagrams of Fig. 2.11 show a number of
high-damping elastic spectra in the so-called ADRS (Acceleration-Displacement-
Response-Spectra) format. Basically, each isolation system type is characterized
by different damping levels. As a consequence, each type can be associated to a
different group of response spectra. The dashed lines passing through the origin of
the axis correspond to two limit values of the effective period of vibration of the
bridge with seismic isolation, equal to 3Tfb (being Tfb the fundamental period of
vibration of the bridge without seismic isolation) and 4 s, respectively. The
interceptions of such radial lines with the response spectra corresponding to the
lowest and highest values of damping ratio, define a preliminary range of possible
isolation system design displacements. On the left hand side of each ADRS diagram
of Fig. 2.11, the schematic displacement vs. acceleration relationship of the most
critical (lowest shear/flexural strength) pier of the bridge in the direction of analysis
is reported. For existing RC bridges (see Fig. 2.11a), for instance, the most critical
pier is identified by a given reinforcement ratio (r*). For new RC bridges, instead,
a suitable reinforcement ratio (1% < r0 < 4%) can be initially assumed. Considering
that, in presence of seismic isolation, piers are designed to remain elastic, only
Performance Points (PP’s) that fall inside the painted background of the ADRS
diagrams can be tentatively selected. Obviously, two further aspects affect the selec-
tion of the PP, i.e., the clearance of the joints and the isolation system displacement
capacity. Once a suitable PP has been selected, the associated damping ratio (hence
isolation system type), isolation period (hence effective stiffness) and (for new
bridges) pier reinforcement ratio are identified. They can be used in the final bridge
analysis and design.
2.5 Piers
2.5.1 Modelling Seismic Response of Columns in Reinforced

Concrete Bridges
Bridges give the impression of being rather simple structures whose seismic
response could be easily predicted. Therefore, for seismic design of the majority
of bridges, simple models and simple methods are used in the practice. Design
methods are predominantly elastic with only implicit consideration of the inelastic
response. However, recent research has shown that certain structural characteristics
may yield highly irregular and even unexpected response in the transverse direction
of bridges (Paraskeva et al. 2006; Isakovic and Fischinger 2006 and Pinho et al.
2007). It has been realised that explicit inelastic seismic analysis is needed for
unusual, complex bridges.
The dynamic response history analysis is one of the options, which could be
employed to estimate the seismic response of RC bridges more realistically, taking
into account their non-linear properties. However, a valid estimate of the seismic
response obtained by this method depends on several critical issues such as an
appropriate model of earthquake load, a sufficiently detailed numerical model of
the bridge. In addition, the efficiency of computer software and its ability to analyze
the bridge in reasonable time is critical. The modelling of the bridge piers is
particularly critical and is the focus of this section.
The modern seismic design philosophy of standard (as opposed to unusual)
bridges includes a consideration that damage of the bridge should be reduced to
the flexural yielding of columns. There are several elements, which are suitable for
modelling the non-linear behaviour of bridge columns; however, the knowledge
about their applicability in practice is limited. In general, these elements could
be classified as macro- or micro- elements. Macro-elements are different types of
beam-column elements, where the non-linear behaviour is modelled using dif-
ferent hysteretic rules (force-displacement or moment-rotation relations), which
attempt to capture overall member behaviour under cyclic loads. The basis of
development of “hysteretic” – macro-models has been primarily experimental
data. Therefore the parameters of hysteresis have clearly defined physical meaning.
This makes macro-elements relatively easy to control. Since the hysteretic rules
tend to represent the overall member behaviour, macro-models usually include

substantially less elements than micro-models. This makes macro-models simple
and more appropriate for complex dynamic non-linear analysis.
The second group of elements, micro-elements are, in general, plain (2D) or
solid (3D) finite elements. Typically, the non-linear behaviour is modelled at the
level of stress–strain relationships using predefined constitutive laws that are
generally derived based on material test data. Compared to the macro-models,
micro-models require substantially more calculations. This makes the already
complex dynamic time-history analysis even more demanding on the analyst and
the computer. Compared with the macro-models, the micro-models make the
control of results and their analysis more complex and time-consuming.
Some macro-elements, e.g. fibre elements (discussed in subsequent sections)
combine the properties of previously described types of elements. For example,
fibre element is a truss type element, where the nonlinear behaviour is defined based
on the stress–strain relationship of each fibre. It is more convenient to use macro-
models when the global response of the entire bridge is of interest. The micro-models
are more appropriate when the local responses of some details are studied, or when
the structure includes complex details for which macro-elements are not available.
In view of the above considerations this section deals only with macro-elements.
Three types of these elements: (a) beam-column element with lumped plasticity, (b)
fibre element, and (c) MVL element, have been compared using two examples of
four-span viaduct. This viaduct was originally investigated experimentally and
analytically by Pinto et al. (1996).
Columns in substandard bridges designed prior to the development of
modern seismic codes require special considerations under seismic loading.
Very often these columns have insufficient transverse reinforcement, which can
result in their shear failure and low ductility. To be able to model their response
their shear strength should be properly assessed. In this section the shear strength
of hollow box columns with relatively small amount of shear reinforcement,
which was placed inside the longitudinal reinforcement, was analyzed experimen-
tally and using the three analytical methods. Two analytical methods included
in the Eurocode 8 standard (Part 2 and 3) as well as the method developed at the
University of California (Priestley et al. 1996), San Diego were used.
Modelling bridge piers for seismic analysis and assessment can be done using
different types of numerical tools. The following list is presented in increasing
order of complexity and, perhaps, decreasing order of design applicability:
(i) Finite length plastic hinge discretization coupled with a hysteresis model.
(ii) Distributed flexibility based element with a multi-linear hysteresis model.
(iii) Fibre discretization using concrete and steel cyclic constitutive models.
(iv) FEM discretizations, either 2D or 3D, using a continuum damage cyclic model
for concrete and the Menegotto-Pinto model for steel.
A general overview of modelling strategies is provided in the next sections
complemented with few examples of application.
M
a b
10
M M+máx 3
2
9
11
My K2 1
4
8
K1
ϕmáx -
Mc Kd r K+
+
+ ϕ
ϕr ϕmáx
5
12
7
K0
13
ϕc ϕy ϕ 6 -
Mmáx
Fig. 2.12 Global section model. (a) Stiffness degradation; (b) pinching effect
2.5.2 Finite Length Plastic Hinge Model
This numerical model adopts structural beam-column elements with inelastic

behaviour consisting of an elastic central zone and two end zones with inelastic
response features. The non-linear material behaviour of such elements is therefore
concentrated in their extremities, since those are the critical regions where yielding
occurs over a short length, generally not greater than the cross section depth.
In this model, cracking along the linear middle segment is not explicitly considered.
In the inelastic end-zone cross-sections, a global section non-linear model is
adopted for simulating the reinforced concrete cyclic behaviour by means of a
representative hysteresis model. The cyclic response is based on a skeleton curve
(normally the monotonic response curve) coupled with a set of rules that control
the characteristics of the loading-unloading-reloading cyclic reversals, typically
addressing the pinching effect as well as the stiffness and strength degradation
(Takeda et al. 1970; Saiidi 1982; Costa and Costa 1987; Duarte et al. 1990;
CEB 1996).
Figures 2.12a, b show typical moment-curvature curves (M-j) used to simulate
the non-linear behaviour and the hysteretic energy dissipation characteristics for
structural seismic response analysis, including the stiffness degradation and the
pinching effect (CEB 1996). Although not shown in these figures, strength degra-
dation is also possible to be included as fully addressed in Costa and Costa model
(CEB 1996).
The response is initially linear elastic, turning into non-linear with the first
cracks opening and proceeding to the post-yielding range when the tensile rein-
forcement yields. During cyclic loading the section stiffness reduces with increase
of deformation that could be associated with strength degradation. This effect is
controlled by the stiffness degradation parameter a, according to Eq. 2.1, wherein
the reduced stiffness (Kd) is obtained from the elastic stiffness (K0) affected by the
a b
σc
kfc B
fc θm Confined Concrete
Unconfined Concrete
0.2kfc CD
A
0.002 0.002k ε20c εc
Fig. 2.13 Material behaviour models: (a) Concrete; (b) Steel
a-power function of the ratio of the maximum curvature (jmàx) reached in a given
cycle and the yielding curvature (jy).
. a

Kd ¼ K0 jy jmax (2.17)
The cyclic moment-curvature curve depicted in Fig. 2.12b illustrates the

pinching effect, controlled by the parameter b according to Eq. 2.2 and can be
observed in the reloading phases (lines 8/9 and 12/13), normally associated with
crack closure.
b
K ¼ mmax =ðjmax jr Þ jy =fmax (2.18)
Strength degradation is usually associated with stiffness degradation, which is

partially included in the model by imposing that the previous cycle maximum
curvature during a given reloading stage is reached again for a target moment
that is lower than the moment in the previous cycle. Details on this procedure are
included in Costa and Costa (1987), which defines the strength degradation
controlling parameter by a monotonically increasing function of the number of
cycles and of the non-linear incursion magnitude expressed by the ratio of a given
maximum curvature to the ultimate curvature (Wang and Shah 1987); such param-
eter is, therefore, analytically controlled by the curvature.
The monotonic (or skeleton) moment-curvature relationship is typically based
on the initial concrete cracking, the tensile steel reinforcement yielding and its
post-yielding behaviour, as well as the near peak concrete response. Based on the
section geometric characteristics, the reinforcement detailing and the material
characteristics, the skeleton curve can be obtained by fibre modelling using mono-
tonic material constitutive laws as shown in Fig. 2.13a for concrete (both uncon-
fined and confined according to Kent and Park (1971), Park et al. (1982), Kappos
(1991)) and in Fig. 2.13b for steel. Alternatively, the monotonic moment-curvature
curve can be defined as a multi-linear function of the characteristic points
Bridge B213c
2.8m
Medium Pier 2.8m
Short Pier Tall Pier 2.8m
20m 20m 20m 20m
0.16m
Deck
0.16m
1.2m
5.6m
0.12m Pier 1.6m

0.12m 0.8m
0.1m ‡
2.6m
0.8m
Fig. 2.14 B213c scaled bridge geometry
(cracking, yield and ultimate) which are explicitly determined using the procedure
described in Arêde and Pinto (1996) for rectangular or T-shape RC sections without
requiring any type of fibre discretization. In such case the “yield point” represents
the effective yield point of the moment curvature relationship rather than the first
yielding of reinforcement.
The finite length plastic hinge model was adopted for the simulation of experi-
mental studies carried out on reduced scale models of bridge piers performed at the
JRC-Ispra, Italy and reported in Guedes (1997). As part of a wider study within the
framework of the PREC8 project, the scaled bridge B213c illustrated in Fig. 2.14
was considered, featuring a deck with four spans 20 m long and three hollow section
piers with 5.6, 2.8 and 8.4 m high. Piers were fully-fixed at the base and pinned
at the deck connection. The bridge was pseudo-dynamically tested with sub-
structuring such that the deck was numerically simulated whereas the piers were
tested at the reaction wall facility of the ELSA laboratory at Ispra by imposing top
displacements and measuring the forces.
Figure 2.15 shows the comparison of experimental and analytical responses for
the tall (Fig. 2.15a) and short (Fig. 2.15b) piers (Delgado et al. 2002). Concerning
the numerical simulation, each pier was modelled by a single fixed length plastic
hinge element with the measured lateral displacement history imposed on the top.
a b
Displ. (m)
600
-0.08 -0.06 -0.04 -0.020 0.02 0.04 0.06 0.08
400 1500
200 1000
Force (kN)
Force (kN)
0 500
-200 0
Numerical
-400 Experimental -500
-600 -1000 Numerical

-0.3 -0.2 -0.1 0 0.1 0.2 0.3 Experimental
Displ. (m) -1500
Fig. 2.15 Force-displacement curves for the B213c bridge piers: (a) tall and (b) short
a b 0.02
Displ. (m)
Experimental
0.01
Numerical
0
Numerical
-0.01 MG
-0.02
0 0.5 1 1.5 2 2.5 3 3.5 4
c time (s)
0.03
Displ. (m)
0.02 Experimental
0.01 Numerical
0
-0.01 Numerical
-0.02 MG
-0.03
0 0.5 1 1.5 2 2.5 3 3.5 4
d time (s)
0.04
Displ. (m)
0.03 Experimental
0.02
0.01 Numerical
0 Numerical
-0.01
-0.02 MG
-0.03
0 0.5 1 1.5 2 2.5 3 3.5 4
time (s)
Fig. 2.16 Seismic analysis of B213c bridge. (a) Equivalent planar discretization and top-
displacement time histories for (b) medium, (c) short and (d) tall piers (experimental and numerical)
The forces and the dissipated energy of these numerical results are reasonably
close to the experimental results, thus suggesting good numerical model perfor-
mance for simulating the inelastic behaviour of piers under cyclic loads with no
significant computational cost. Therefore, this pier modelling was subsequently
adopted for the seismic analysis of the bridge in the transverse directions using an
equivalent planar discretization of the bridge as schematically shown in Fig. 2.16a,
where the deck is assumed to be linear elastic and linked to the pier top sections by
hinged, rigid and massless bars.
For the bridge numerical seismic analysis the same accelerograms were adopted
as those used in the pseudo-dynamic experimental tests. Thus, non-linear dynamic
response history analysis was performed and the results of pier top-displacements
were compared against the experimental data. Moreover, to further confirm the
model ability, these results were also compared with more refined simulations
carried out by Guedes (1997) using fibre model simulation. These comparisons
are illustrated in the displacement histories included in Fig. 2.16b, c and d for the
medium, short and tall piers, respectively, where the experimental response is
plotted in black dashed line; the numerical results are plotted in solid lines, black
for those obtained by the fixed length plastic hinge model (time series labelled
Numerical) and grey for the fibre model output (Numerical MG).
An overall observation of the above mentioned comparisons shows that the
numerical simulations provide reasonably good agreement with the experimental
response for the three piers, both in terms of frequency and of peak values. Despite
some local deviations, globally the response is quite well captured and the fact that
both plastic hinge and refined fibre modelling approaches yield similar results is
itself an additional confirmation of the good balance between computational cost
and result quality of the adopted plastic hinge model. Further details concerning
both modelling and simulation strategies and assumptions can be found in Guedes
(1997) and Delgado et al. (2002).
2.5.3 Distributed Flexibility Based Element Model
This global element model was developed (Arêde 1997) within the flexibility
formulation framework for the analysis of reinforced concrete frame structures
resorting to one-to-one discretization of structural members while duly accounting
for progressive stiffness modifications along the member length. The algorithm was
implemented in the general purpose finite element analysis code CAST3M, i.e. the
former CASTEM 2000 (CEA 2003).
The development was pursued under the requirement that only one single
element should be sufficient to describe the response of a given structural member,
for which the classical displacement based formulation is not adequate because
displacement shape functions change with member stiffness modifications caused
by cracking and yielding. Therefore, the flexibility or force based formulation was
adopted in line with previous works by Taucer et al. (1991) and Spacone et al.
(1992), which proved to be particularly adequate for the required purpose since no
displacement shape functions are required. In depth discussion can be found in
Arêde (1997).
The general flexibility formulation resorts to a number of control sections where
the behaviour is monitored and only makes use of force shape functions which
have the main advantage of being correct regardless the element damaged state.
The element flexibility matrix is obtained by integration of the flexibility distribu-
tion (known at the control sections) and an internal iterative scheme is adopted to
obtain the element restoring forces corresponding to a given set of imposed
displacements at the nodes as schematically shown in Fig. 2.16. Details on this
scheme can be found in Arêde (1997).
While the above mentioned previous works dealt with fibre section modelling
for flexibility and restoring force updating at predefined and fixed control sections,
the global element model herein addressed resorts to a global section model
Fig. 2.17 State determination for force based (or flexibility) non-linear incremental scheme.
Iterations for a given load increment: (a) at the global structure level, (b) at each element and
(c) at the control section level
(based on moment-curvature behaviour laws) prescribed just at the element nodes

and coupled with moving control sections that allow constantly monitoring the
yielded, cracked and uncracked zones of the element. This model takes into account
the section non-linear flexural behaviour for only one bending direction, by
recourse to a global model based on moment-curvature (M-j) trilinear skeleton
curves as shown in Fig. 2.17a. The model is of Takeda type, requiring three
parameters for the hysteretic behaviour control in each bending sense and allows
simulating a wide variety of situations involving stiffness degradation, pinching
effect and strength deterioration.
However, in order to make possible the control of cracking and yielding sections
for each analysis step (respectively defined as the transition sections from cracked to
uncracked zones and from yielded to cracked zones), some simplification is consid-
ered in the model, consisting in a sudden transition from the uncracked phase (branch 1
in Fig. 2.17b) to cracked behaviour (branches 2 and 3 in Fig. 2.17c) whenever any
section reaches the cracking moment ( Mþ
C or MC ) for the first time. Thus, both the
uncracked and cracked stages are ruled by secant stiffness laws, greatly simplifying
the control of sections matching such behaviour stages. After yielding, the behaviour
rules are similar to those of the existing model with slight modifications.
It is noteworthy that, beyond the above mentioned simplification, any other type
of Takeda type model is suitable for this purpose. In particular, the same section
model (Costa and Costa 1987; Duarte et al. 1990; CEB 1996) as used for the finite
Fig. 2.18 Section model – primary curve and distinct behaviour phases
Fig. 2.19 General layout of yielding and cracking sections, and distinct behaviour zones
length plastic hinge formulation could also be adopted for this distributed flexibility
global element model.
According to the above stated and as illustrated in Fig. 2.19, for each step the
element is divided into yielded, cracked, and uncracked zones wherein adequate
flexibility is assigned according to the respective section model. Control sections
consist of fixed ones (the end sections, E1 and E2, and a central section H) and moving
sections (the cracking ones, C1–C4, the yielding sections, Y1 and Y2, and, possibly,
the null moment sections, O1 and O2). Sections E1 and E2 are fully controlled by the
complete model, whereas the span section H is restricted to cracked behaviour
because no yielding is assumed to take place along the central span zone. Cracking
sections are controlled by the simplified model rules shown in Fig. 2.18b, c, depending
on whether the uncracked or the cracked side is being considered, whereas sections
Y1 and Y2 are controlled by the diagrams in Fig. 2.18c, d, respectively for their
cracked and yielded sides. Null moment sections may also be activated in case they
occur in cracked zones and are controlled by the rule of Fig. 2.18c.
Figure 2.19 refers to the case of a beam with a concentrated force applied in the
span, but it is quite apparent that the same procedure is suitable for column analysis,
with an even more simplified scheme for control section motion. This is evidenced
in Fig. 2.19 which highlights different flexibility distributions along the element
for a loading case of a beam (Fig. 2.19a) as shown in Fig. 2.20b and for an
Fig. 2.20 Examples of flexibility distributions for (a) loading and (b) unloading cases
Fig. 2.21 Top displacements: dynamic simulation of (a) low and (b) high level tests
unloading situation (Fig. 2.19b) of an element loaded only by end-section moments.

The latter case is that of major interest for bridge pier analysis since it applies
both for cantilever or bent type piers. Thus, focusing essentially on that situation,
Fig. 2.19b exemplifies the case wherein both yielded zones (the variable length
plastic hinges) unload from the post-yielding branch (thus with constant flexibility
as evidenced by the dashed line in the f(x) flexibility diagram) and for which a
linear variation is assumed between the end-section and the yielding section
flexibilities. Specific details of model development and implementation can be
found in Arêde (1997).
Because the model was initially used for building analysis, no specific studies
were carried out on the seismic response analysis of bridge piers. Nevertheless the
model was investigated in Arêde (1997) for a four-story building and satisfactory
results were obtained (Figs. 2.20 and 2.21).
2.5.3.1 Fibre Model
The fibre model is widely available beam-column element model (Taucer et al.
1991; Guedes et al. 1994; Guedes 1997; SeismoStruct 2005), which can be regarded
as a step further in the refinement of standard beam models. In fibre modelling, the
sectional stress–strain state of the elements is obtained through the integration of
Gauss node B
Section b
Gauss
Section a
node A
σ
B
ε
A
L/2 3 L/2
σ
σ = + +
ε Steel Fibres
Unconfined Confined
RC Section
Concrete Fibres Concrete Fibres
Fig. 2.22 Discretization of a typical reinforced concrete cross-section
the nonlinear uniaxial stress–strain response of the individual fibres in which the
section is subdivided, distinguishing steel, confined and unconfined concrete, as
illustrated in Fig. 2.22. The adopted stiffness-based element cubic formulation then
allows both the representation of the spread of inelasticity along the member length
as well as the implicit incorporation of interaction between axial force and trans-
verse deformation of the element. The use of a sufficient number of elements per
structural member permits the reproduction of plastic hinge, typical of members
subjected to high levels of material inelasticity. The spread of inelasticity across
the section and along the member length is thus achieved without requiring
expertise calibration of any lumped plasticity element.
Structural members are represented by means of frame elements, with finite
length and assigned cross-sections. Structural and non-structural inertia mass may
also be introduced, in either lumped or distributed fashion, whilst joint/link
elements, defined as spring-type elements joining coincident locations, can be
used to model discontinuous connections. By means of such element types, a
number of different element classes (columns, beams, walls, beam-column joints,
etc.), non-structural components (energy dissipating devices, inertia masses, etc.)
and different boundary conditions (flexible foundations, seismic isolation or struc-
tural gapping and pounding) can be represented. The fibre-discretization renders
possible a realistic modelling of the different materials, and their distribution, that
make up the cross-section of a given member. The material models may feature
different levels of accuracy and complexity; the bilinear, the Menegotto-Pinto
(1973) and the Monti-Nuti (1992) models are among the most used models for
steel, whilst concrete may be characterized by tri-linear, nonlinear with constant or
variable confinement constitutive laws (e.g. Scott et al. 1982). Many other material
constitutive laws are available in the literature. The optimal number of fibres
representing the cross section should be determined based on a series of batch
tests with different mesh sizes (Sadrossadat-Zadeh and Saiidi 2007).
Fig. 2.23 Truss analogy for a cantilever beam under significant shear forces
The fibre model herein reported is implemented in the CAST3M software

package (Guedes et al. 1994; CEA 2003) using a 3D Timoshenko type beam
element so that interactions between axial and shear forces and bending moments
can be considered. Each 3D element, with two nodes and six degrees-of-freedom
per node, is divided into other longitudinal elements (the fibres) that react to both
axial and shear displacements according to suitable constitutive laws. Each fibre is
defined by a four or three node 2D element in the transverse section. Along the
longitudinal (element) axis direction, the behaviour of the 3D elements is
integrated using one Gauss point per element and the model is formulated for a
longitudinal axis eccentric to the element gravity axis. Details on the element
model formulation can be found in several publications (Guedes et al. 1994;
Guedes 1997).
Experimental evidence concerning the cyclic response of shear dominated
elements, such as squat bridge piers, show that non-linear shear behaviour must
be considered to achieve accurate numerical response simulations. Therefore, the
above mentioned classic fibre model was coupled with a strut-and-tie model to
simulate the non-linear shear behaviour by adopting the procedure presented in
Garstka et al. (1993) and Guedes (1997) as described below.
Classic fibre models assume linear elastic behaviour for shear stress compo-
nents. This approach is acceptable when shear deformations are very small, but
needs to be augmented in analyzing short bridge columns by adding a nonlinear
shear element. The shear element uses the truss analogy (Fig. 2.23) for a cantilever
Fig. 2.24 Compatibility of displacements in the equivalent truss analogy
beam under shear forces. Thus, the non-linear shear model herein described is based
on the assumption that shear strength is provided by the “equivalent” truss made of
transverse and longitudinal steel ties and diagonal concrete struts that are formed
due to diagonal cracking.
The model relies on displacement compatibility and force equilibrium in the
cross-section. Other parameters are the constitutive laws of the materials and the
shear cracking angle y that defines the orientation of the concrete struts.
The displacement compatibility in the transverse direction is established by
analysing the deformed truss shown in Fig. 2.24 subjected to axial force P,
transverse force V and bending moment M. Two concrete diagonals are considered
in the formulation for each loading sense: one for compressive and another for
tensile forces. Detailed formulation is reported in Guedes (1997).
The non-linear shear model was adopted to simulate the response of a squat
bridge pier tested at the ELSA laboratory in Ispra, Italy (Pinto et al. 1996). Both
linear and the non-linear shear models were included in the study and the results
were compared with the experimental data.
The pier was 2.8 m high and corresponded to a 1:2.5 scale specimen where a
vertical force equal to 1.72MN was applied on the top to simulate the dead load of
the deck. The mechanical properties of materials used in the tests corresponded to
C25/30 concrete grade and B500 Tempcore steel (Fig. 2.25).
The top force-displacement diagrams are shown in Fig. 2.26, where comparison
between experimental and numerical results is made.
Figures 2.24a, b show that the non-linear shear model reproduces quite well the
narrowing and lengthening of the hysteretic force-displacement curve but the linear
shear model overestimated the dissipated energy.
a St2
St3
St2 b FIBRES
[mm] 160 480 160
160 SECTION TYPE

St1
-AS1= 1.690%
2
St1 (11250mm )
-Stirrups: φ6mm
St1 (50mm Spacing)
1280 20φδ
-Stirrup tupe:
St1 (136x492)
St2 (136x348)
St3 (136x332)
11φ14
(Cover concrete = 8mm)
160
Confined Concrete Unconfined Concrete Longitudinal Steel Shear
19φ14
Fig. 2.25 Model validation. (a) Pier cross-section and (b) Corresponding fibre discretization
a b
Force [MN] Force [MN]
2.5 2.0
P5D-Total at 2.8m P5D-Total at 2.8m
(Linear) (Non-Linear: 45␱)
.0
.0
Numerical
Numerical Experimental
Experimental
Displ. [m] Displ. [m]

-2.5 -2.5
-5.0 .0 4.0 -5.0 .0 4.0
X1.E-2 X1.E-2
Fig. 2.26 Experimental vs. numerical top force-displacement results. (a) Linear shear model and
(b) non-linear shear model (Guedes 1997)
2.5.4 Two and Three-Dimensional FEM Discretizations
In the framework of refined modelling of bridge piers, 2D or 3D FEM

discretizations can be adopted wherein both concrete and steel are simulated
using suitable constitutive laws, the former with plane or solid finite elements and
the latter with uniaxial truss elements. Within this contribution, the behaviour of
steel bar elements is ruled by the Giuffré-Menegotto-Pinto model (Menegotto and
Pinto 1973), whilst for concrete a Continuum Damage Mechanics cyclic model
(Faria et al. 1998, 2002) is adopted. Besides other stand-alone computer codes
developed by the model author and collaborators, the model is implemented in the
CAST3M package and has already proved to be suitable for seismic behaviour
analysis of RC hollow bridge piers (Faria et al. 2004) and (Delgado et al. 2007).
The Continuum Damage Mechanics model adopted for concrete is cast with a
strain-driven suitable scheme for simulating the degradation occurring under ten-
sion and compression (Faria et al. 1998) (Fig. 2.27).
Fig. 2.27 Constitutive models’ evolution under 1D loading. (a) Concrete behaviour and (b) steel
cyclic model (Faria et al. 2004)
This model has been extensively used for seismic or cyclic response studies
of RC walls (Faria et al. 2002) and bridge pier scaled models as reported in
Faria et al. (2004), Delgado et al. (2009), Arêde et al. (2009). The conclusions
generally point to good agreement between numerical and experimental results.
The cyclic global force- displacement response curves were quite well estimated by
the numerical model which also succeeded in adequately simulating energy dissi-
pation, the strut-and-tie strength mechanisms experimentally observed on shear
dominated structural members, the occurrence of localized or smeared cracks, as
well as the formation of plastic hinges.
2.5.5 Example 1 on Fiber Model Application
The study of three numerical models suitable for modelling the flexural response is
presented in this section. Simple beam-column element model with lumped plastic-
ity and fiber beam-column element model were first investigated. Then the validity
of the MVL (multiple-vertical-line), also known as multi-spring element (Saiidi
et al. 1989 and Jiang and Saiidi 1990) element was also tested. The MVL element is
a combination of the previous two types. The element consists of several springs,
which are rigidly connected at the top and the bottom of the element. The cyclic
response of each spring is controlled by hysteretic rules.
A large-scale specimen (1:2.5) of typical viaduct was analyzed in the transverse
direction. Full-scale structures consisted of a 200-m deck and three single column
bents (Fig. 2.28). The deck was pinned at the abutments. When modelling a viaduct,
the superstructure was assumed to be elastic. Abutments were modelled as infinitely
rigid. Columns were pinned at the level of the superstructure and fixed to the
footings. Seismic load was defined with the generated earthquake record, used in
the experimental studies.
a Viaduct V232 b Column cross-section
0.4m
4 x 50 m = 200 m
transv. dire. 3.2m 4.0m
of viaduct
7
7
7 0.4m
0.4m 1.2m 0.4m
A1 B2 B3 B4 A5 2.0m
Fig. 2.28 Full-scale structure
Hu
Equalized area
l Actual distribution
Idealized distribution
My
lp
Mu φp φy
φu
Fig. 2.29 Equivalent column and the definition of the equivalent plastic hinge length based on the
idealized curvature distribution
2.5.5.1 Beam-Column Reinforced Concrete Element with Lumped Plasticity
In this model part of the element between the node and the point of contraflexure
can be represented by an equivalent column (Fig. 2.29).
The drift ratio is then defined as the top displacement divided by the height
of the column. The displacement is obtained by the double integration of the
curvature along the height of the column. Typically for macro elements a number
of idealizations have been used in the application. Linear distribution of bending
moment and idealized curvature distribution are assumed. Plastic curvature is then
considered to be constant over the equivalent plastic hinge length, which has been
empirically determined (e.g. Kanaan and Powel 1973).
Although the element appears to be crude it is a realistic representation of
structural members in which plastic hinging occurs at the ends. An application
for hollow-box columns in a highway viaduct, described in previous sections is
a 800
Force [kN] b 800
Force [kN]
experiment 600 experiment 600

analysis 400 analysis 400
α = 0.5 α = 1.0
200 200
0
20 0
-60 -40 -20 - 40 60
-200
-200 -40 -20 - 20 40 60
displacement [mm] -200
-400 displacement [mm]
-400
-600
-600
-800
-800
Fig. 2.30 Shear force-displacement diagram for the initial model (a ¼ 0.5) and model using
a ¼ 1.0 (design earthquake)
c d Force [kN]
Force [kN] 1000
1000
experiment
800 800
experiment analysis
600 600
analysis α= 1.0
400 400
α= 0.5
200 200
0 0
-100 -80 -60 -40 -20-200 - 20 40 60 80 -150 -100 -50 -200- 50 100 150
-400 displacement [mm] -400 displacement [cm]

-600 -600
-800 -800
-1000 -1000
Fig. 2.31 Shear force-displacement diagram for the initial model (a ¼ 0.5) and model using
a ¼ 1.0 (high-level earthquake)
briefly presented below. The beam-column macro element used in this study was
incorporated into DRAIN-2D program (Kanaan and Powel 1973) at the University
of Ljubljana (Fajfar and Fischinger 1987). In this element tri-linear Takeda hyster-
etic rules control the response of the rotational springs. In the initial model no
“tuning” of the element parameters was done. All properties (including hardening
parameter) were calculated from first principles. The common average value of the
unloading parameter in the Takeda model (a ¼ 0.5), which determines the rate of
the unloading stiffness deterioration, was used.
The correlation between the analytical and experimental displacement response
histories, obtained by the initial model, was good and so was the modelling of the
predominantly flexural hysteretic behaviour of the tall central column (Isakovic and
Fischinger 1998). However, in the case of the design earthquake, the initial model
underestimated the actual stiffness degradation on the unloading branch for the
short column (Fig. 2.30). To account for higher stiffness degradation, unloading
parameter a ¼ 1.0 should be used in the modified model. This change improved the
calculated response in the case of the design earthquake, but not in the case of the
stronger earthquake (Fig. 2.31), indicating that different a values should be used for
different levels of response.
a End column, 2M = 5.60 m. b End column, 2M = 5.60 m.

50 800
top displacement [mm]
40 test test
initial model 600
initial model
shear force [kN]

30
400
20
10 200
0 0
-60 -40 -20 0 20 40 60
-10 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
-200
-20
-400
-30
-40
-50
time [s] top displacement [mm]
Fig. 2.32 Displacement history and shear force-displacement diagram (initial model)
These results are quite typical for the application of the macro elements. We can
conclude that the overall results have been quite good in all cases. However, it is
unrealistic to expect and claim that such, empirically based models can capture
all the details of the response in the cases for which they were not calibrated.
Nevertheless, even more refined models, like the stress–strain monitoring fibre
element cannot guarantee good (or even at least acceptable) results.
2.5.5.2 Fibre Element
Among several possibilities in program OpenSees (Mazzoni et al. 2003) the force
based non-linear beam-column element with distributed plasticity was used since it
was found to be more efficient than standard displacement-based fiber element.
Model of the column can consist of elements, which are further divided into smaller
segments with different cross-sections. The correlation between analytical dis-
placement time history response, obtained by the initial model (with typical values
of characteristic parameters), and experimentally obtained displacements was good
(Fig. 2.30). Similar to the initial beam column element with lumped plasticity, this
model failed to estimate actual stiffness degradation on the unloading branch
(Fig. 2.30b). Several improvements were necessary to obtain better results
(Fig. 2.31). Model of concrete and model of steel as well as the number of
integration points were changed. For example: the strength of the concrete in
tension was taken into account and, instead of the bilinear stress–strain relationship,
the Giufré-Menegotto-Pinto (Maekawa et al. 2003) model for steel was used, etc.
(Figs. 2.32 and 2.33).
2.5.5.3 MVL (Multiple-Vertical-Line) Element
MVL element includes also a horizontal spring to model shear behaviour (Fig. 2.34).
This element was originally proposed by Japanese researchers (Kabeyasawa et al.
1983) and later modified by Vulcano et al. 1989 and Fischinger et al. (1992).
End column, 2M = 5.60 m. End column, 2M = 5.60 m.

50 800
40 test test
600
modified model modified model

30
400
shear force [kN]

20
10 200
0 0
-10 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 -60 -40 -20
-200
0 20 40 60
-20
-400
-30
-40 -600
-50 -800
time [s]
Fig. 2.33 Displacement history and shear force-displacement diagram (modified model)
Fig. 2.34 Multiple-Vertical-Line-EleMent (MVLEM) and hysteretic rules of vertical springs
All these versions of the element were for the analysis of the unidirectional
response only. Another version of the element has been developed (Fischinger
et al. 2004) for bi-directional analysis of structures.
Each column of the viaduct was modelled with nine MVL elements. The
displacement time-history obtained with the initial model (using standard
parameters: a ¼ 1.0, b ¼ 1.5, g ¼ 1.05, d ¼ 0.50), was reasonable (Fig. 2.35).
The estimation of the stiffness degradation on the unloading branch was better than
that obtained with the previous two elements. Therefore, standard parameters were
not changed.
2.5.5.4 Comparison of the Models
It can be concluded, that all three models are suitable for modelling the global
behaviour of viaduct columns. All the initial models (using standard values of
parameters) estimated the maximum displacements, as well as the maximum
forces, quite well. Some discrepancy with the experiment was detected mostly
during the unloading phase.
End column, 2M = 5.60 m. End column, 2M = 5.60 m.

50 800
40 test test
600
initial model initial model
30
400
shear force [kN]

20
10 200
0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 -60 -40 -20 0 20 40 60
-10 -200
-20
-400
-30
-40 -600
-50 -800
time [s] top displacement [mm]
Fig. 2.35 Displacement time-history and shear force-displacement diagram (initial model)
Table 2.2 Advantages and limitations of the presented elements

Type of element Advantages Limitations
Beam-column Simple model with small number Cannot be used for the analysis of bi-
element with of elements (often one per directional response
lumped column) Unable to estimate a local response.
plasticity Non-linearity defined based on
the hysteretic rule with clear
physical meaning
Easy to control
Fiber element Able to estimate the local Relatively complex analysis
response Several iterations are necessary to
Can be used for the analysis of establish the appropriate model
bi-directional response Control of results is more complex
MVL element Relatively simple In general, several elements per
Non-linearity defined based on column are necessary to obtain
the hysteretic rule with clear acceptable estimation of the
physical meanings response
Able to estimate local response Appropriate number of elements
Can be used for analysis of bi- should be defined iteratively
directional response
The presented models differ with respect to their sophistication. It can be

concluded that although the beam-column element with lumped plasticity is the
simplest, it is quite successful in the prediction of the global response. This makes it
very suitable for the non-linear response history analysis, where the simple model is
needed to simplify the analysis, less time-consuming, and easier to control. How-
ever, when the local (e.g. deformation or stress in some parts of the column’s cross-
section) or bi-directional response is investigated, this element cannot be used. In
such cases the other two types of elements are preferred. The advantages and
limitations of the three elements are summarized in the Table 2.2.
Pier 1 Pier 2 Pier 3

2.8 m
2.8 m
2.8 m
20 m 20 m 20 m 20 m
1.2 m
5.6 m
0.12 m
0.16 m 0.8 m
0.7 m
0.12 m
0.1 m
2.6 m 1.6 m
Deck Pier
Fig. 2.36 Bridge configuration and member cross sections
2.5.6 Example 2 on Fiber Model Application
In the framework of an integrated European programme of research in support of

Eurocode 8, six bridge prototypes, representative of typical multi-span continuous
deck motorway bridges, have been designed (Pinto et al. 1996) with different
procedures for a PGA of 0.35 g, in medium soil conditions (soil type B), applying
the EC8 provisions. Corresponding large-scale (1:2.5) bridge models have then
been constructed and tested in pseudo-dynamic (PsD) fashion at the Joint Research
Centre at Ispra, Italy.
A PsD test, despite being carried out quasi-statically, employs on-line computer
calculations and control together with experimental measurement of the properties
of the real structure to provide a realistic simulation of its dynamic response.
Inertial and viscous damping forces are modelled analytically, and an earthquake
ground acceleration history is given as input data to the computer running the
pseudo-dynamic algorithm. The horizontal displacements of the controlled degrees
of freedom are calculated and then applied to the test structure by servo-controlled
hydraulic actuators fixed to the reaction wall. The PsD testing of the bridge was
performed using the sub-structuring technique, in which the piers were physically
tested and the deck was numerically simulated online. Further details can be found
in Pinto et al. (1996), Pinho (2000), and Sullivan et al. (2004).
2.5.6.1 The Bridge Model
One of the test models, labelled B213C, consists of three piers 5.6, 2.8 and 8.4 m
high and a continuous deck with four identical 20 m spans. The deck is assumed to
end at the abutments with shear-keys, but is free to rotate (Fig. 2.36). The deck-pier
connections are assumed to be hinged transmitting lateral forces but no moments.
20 Ø6 20 Ø8
14 Ø10
14 Ø14
6 Ø12
Section Type 1 Section Type 4
Fig. 2.37 Reinforcement layout
Table 2.3 Steel mechanical properties (Guedes 1997)

Yield Ultimate Yield Ultimate
Diameter (mm) strength (MPa) strength (MPa) strain (%) strain (%) Hardening
6 363.7 430.4 0.177 15.10 0.0022
8 503.4 563.0 0.244 12.30 0.0024
10 489.3 572.3 0.238 14.50 0.0028
12 558.2 646.8 0.271 12.80 0.0034
14 477.2 577.7 0.232 13.00 0.0038
Table 2.4 Summary of the pier cross section characteristics of the bridge (Guedes 1997)
Section Longitudinal Cubic concrete strength (MPa)
Pier type Height (m) steel (%) (Compressive/tensile)
Pier 1 4 14 1.15 37.0/3.1
Pier 2 1 7 0.50 41.2/3.1
Pier 3 4 21 1.15 50.5/3.1
Table 2.5 Deck cross section EA (kN) EI2 (kNm2) EI3 (kNm2) GJ (kNm2)
geometrical and mechanical
characteristics (Guedes 1997) 2.7837E þ 07 1.3544E þ 07 5.6517E þ 07 2.8017E þ 07
The piers have rectangular hollow section with 160 mm wall thickness
(Fig. 2.36). The reinforcement layout of the pier models are shown in Fig. 2.37.
The mechanical characteristics of materials (B500 Tempcore steel with E ¼ 206
GPa for longitudinal rebars and C25/30 concrete) and the mechanical
characteristics of the pier cross-sections are shown in Table 2.3, 2.4 and 2.5. The
deck is a hollow-core prestressed concrete girder 5.6 m wide, as depicted in
Fig. 2.36. In the PsD test, the deck was simulated numerically with 32 linear elastic
Timoshenko eccentric beam elements, whose mechanical characteristics are
presented in Table 2.5, where A is the cross-section area, I2 and I3 are the two
moments of inertia with respect to the local principal axes, J is the torsional constant
and E is the Young Modulus of 25 GPa. The inertia characteristics of the deck are
based on a specific weight of 25 kN/m3. The sub-structured part included a
-54
800
600 -44
400
-34
Stress (MPa)
Stress (MPa)
200
0 -24
-200
-14
-400
-4
-600
-800 6
-0.01 0.00 0.01 0.02 0.03 0.04 0.0005 -0.0005 -0.0015 -0.0025 -0.0035 -0.0045 -0.0055 -0.0065 -0.0075
Strain (mm/mm) Strain (mm/mm)
Fig. 2.38 Menegotto-Pinto steel model, with Filippou isotropic hardening (left), and nonlinear
constant confinement concrete model (right)
Rayleigh damping matrix, featuring a damping ratio x ¼ 0.016 associated to the

two lower transversal natural frequencies of the complete bridge.
At the top of each pier, an axial force N ¼ 1,700 kN was applied using actuators
to simulate gravity load. The input ground motion was represented by a scaled
accelerogram with duration of 4 s and a nominal peak acceleration of 0.875 g. Two
pseudo-dynamic tests were performed on the structure: one with the input motion
corresponding to the design earthquake and another defined on the basis of the
estimated ultimate capacity of the bridges, and thus equal to 1.2 times the design
earthquake.
2.5.6.2 Modelling the Piers in the FE Program
The computer program used in the analysis, SeismoStruct (SeismoSoft 2005), is

a fibre–modelling FE package for seismic analysis of framed structures.
The program is capable of estimating large displacement behaviour and the collapse
load of frame structures under static or dynamic loading, accounting for geometric
nonlinearities and material inelasticity. The piers have been modelled through a 3D
inelastic beam-column element capable of capturing geometric and material
nonlinearities. The number of fibres used in section equilibrium computations
was 400.
The stress–strain behaviour of the steel (Fig. 2.38) has been described by the
nonlinear model of Menegotto and Pinto (1973), as modified by Filippou et al.
(1983) to include isotropic strain hardening. This is an accurate and convenient
model due to its computational efficiency and its very good agreement with
experimental results. It utilizes a damage modulus to represent more accurately
the unloading stiffness, and has been modified and improved by Fragiadakis et al.
(2008) to attain better stability and accuracy. The concrete has been represented
through a nonlinear constant confinement concrete model (Fig. 2.38), as a good
compromise between simplicity and accuracy: it is a uniaxial nonlinear model
following the constitutive relationship proposed by Mander et al. (1988), later
modified by Martinez-Rueda and Elnashai (1997) to improve numerical stability
Table 2.6 Parameters for the Menegotto-Pinto steel model, with Filippou isotropic hardening
Parameter Sec 1 Sec 4
Modulus of elasticity (MPa) 203,000 203,000
Yield strength (MPa) 468 496
Strain hardening parameter 0.0027 0.0036
Transition curve initial shape parameter (default value) 20 20
1st transition curve shape coefficient (default value) 18.5 18.5
2nd transition curve shape coefficient (default value) 0.15 0.15
1st isotropic hardening coefficient (default value) 0.025 0.025
2nd isotropic hardening coefficient (default value) 2 2
Table 2.7 Parameters for Parameter Pier 1 Pier 2 Pier 3

the nonlinear constant
confinement concrete model Cylinder compressive strength (MPa) 31.5 35.0 42.9
Tensile strength (MPa) 3.1 3.1 3.1
Strain at unconfined peak stress (m/m) 0.002 0.002 0.002
Constant confinement factor 1.2 1.2 1.2
under large deformations. The model calibrating parameters, fully describing the
mechanical properties of steel and concrete, have been set as shown in Tables 2.6
and 2.7, where the concrete cylinder strength was estimated as being 85% of the
cubic sample strength listed in Table 2.4.
2.5.6.3 Cyclic Response
After the PsD testing of the bridges, the tall and the medium piers were tested
cyclically until failure (Pinto et al. 1996; Guedes 1997), respectively up to 230 and
150 mm of top displacement, under the imposed displacement history shown in
Fig. 2.39. Additional cyclic tests (up to 72 mm at the top of the pier) were carried out
on a short pier similar to the one tested in pseudo-dynamic fashion. These cyclic tests
on the piers are numerically reproduced herein through a static response-history
analysis, so as to enable a first check on the accuracy of the model. The numerical
reproduction of the cyclic test has been performed imposing on the piers the
displacement history resulting from the PsD test (Fig. 2.39). In addition, the steel
young’s modulus of the medium-height pier was halved, as suggested by Pinto et al.
(1996), in order to reproduce the reduction in the stiffness due to the shear damage
that this pier suffered prior to this cyclic test. No reduction in the steel properties was
applied to the tall pier because it was not damaged during the PsD test.
Figure 2.40 shows a reasonable match between the experimental and numerical
results for the medium pier. Only the reduction in strength at the very last cycle,
when failure occurs, is not perfectly captured. Similar trends were observed in the
response of the tall and short piers. It is clear that pinching has not been perfectly
captured due to absence of shear deformation modelling in these fibre elements, an
aspect still in the development stage (Ceresa et al. 2007). Had nonlinear shear
0.25
0.20
0.15
0.10
Displacement (m)
0.05
0.00
-0.05
-0.10
-0.15
-0.20
Med Pier Tall Pier Short Pier
-0.25
Fig. 2.39 Cyclic test displacement histories
0.90
Experimental
0.60 Numerical
0.30
Base Shear (MN)
0.00
-0.30
-0.60
-0.90
-0.16 -0.12 -0.08 -0.04 0.00 0.04 0.08 0.12 0.16
Displacement (m)
Fig. 2.40 Cyclic test results for the medium pier
deformations been considered, a closer match would have been achieved (e.g.
Ceresa et al. 2009), but the computer analysis time would have been longer
(Fig. 2.40).
2.5.6.4 Pseudo-dynamic Results
To analyze the pseudo-dynamic response of the piers, a complete FE model of

the case-study bridge was created (Casarotti and Pinho 2006) and the calculated
and measured results were compared in terms of displacements and forces at
the top of the piers under the second, and stronger input earthquake motion.
60 750
Experimental Numerical Experimental Numerical
45
Top Displacement (mm)
500
30
Top Shear (KN)

15 250
0 0
-15
-250
-30
-45 -500
-60
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 -750
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Time (s) Time (s)
Fig. 2.41 Medium pier top displacements (left) and top shear (right)
Table 2.8 Ratios of the Tall pier (%) Med pier (%) Short pier (%)
absolute maximum response
obtained from numerical Displacement 88 94 102
calculation to that from tests Top shear 90 95 188
Figure 2.41 shows the results for the medium pier. It can be seen that there is a good
agreement in terms of both the amplitude and the frequency content of the response.
The agreement is representative of the correlation for the other two piers. Table 2.8
lists the ratios of the maximum absolute response obtained from numerical calcula-
tion to that from the tests. It is noted that the force response of the squat pier is not
reproduced with full accuracy, whereas displacements are very well predicted. The
numerical overestimation of the action at the top of the short pier can be explained
by the fact that the fibre-based element formulation did not account for shear
deformation.
2.5.6.5 Closing Remarks
Structural behaviour is inherently nonlinear, particularly in the presence of large

displacements or material nonlinearities, the structural response can be accurately
caught only by means of nonlinear dynamic analyses. The fibre modelling approach
employed in the current work is shown to be capable of associating simplicity of use,
even for not highly experienced users. Moreover, its ability to simulate the nonlinear
dynamic response of reinforced concrete bridges to seismic loads has been proven by
simulating large-scale experimental pseudo-dynamic tests. Results of the dynamic and
modal analyses performed reveal a good agreement with the pseudo-dynamic tests,
both in terms of displacements and forces at the top of the tall and medium-height
piers. At present, shear strains across the element cross-section are not included in the
fibre-element formulation adopted, i.e. the strain state of a section is fully represented
by the curvature at centroidal axial strains alone: this approach is not accurate enough
for representing the squat pier deformation state, where shear deformations are of
relevance. In this case, despite the relevance of the shear response, the prediction of the
deformation of the squat member was still fairly good.
This section has thus illustrated how the use of simple-to-calibrate fibre structural
models can be employed to reproduce with good level of accuracy the nonlinear
structural response of continuous span bridge structures. In other words, it is believed
that such an advanced analytical tool can be readily handled within a common
engineering practice framework, provided a basic level of awareness on the decisions
that the designer has to face, discussed herein is available.
2.5.7 Analytical Modelling of Hollow Box Columns
In Sects. 2.5.4–2.5.6 several modelling methods with different levels of sophistica-

tion were described and demonstrated. Another example of the development and
application of microscopic models is presented in this section.
A series of quarter scale hollow columns were tested under cyclic loading at
Porto University (Delgado et al. 2006, 2007, 2009). These columns were analyzed
using CAST3M computer code (CEA 2003), which is a general purpose finite
element analysis program. A wide variety of non-linear elements are included in
CAST3M , particularly, a damage model developed at the Faculty of Engineering of
Porto University (FEUP) (Faria et al. 1998; Costa et al. 2005). Studies have shown
that the damage model is suitable for seismic behaviour analysis of RC bridge
piers (Faria et al. 2004). The damage model is Continuum Damage Mechanics
based constitutive model for the concrete zone discretized into 3D finite elements
incorporating two independent scalar damage variables that account for the degra-
dation due to tensile or compressive stress conditions. The Giuffré-Menegotto-
Pinto model (Giuffrè and Pinto 1970) for the cyclic behaviour simulation of the
steel reinforcement discretized via truss elements is used.
In analytical studies of the piers, the second and third repeated cycles of each
displacement level were removed to facilitate comparison with experimental data.
Only one-half of the cross section was modelled due to symmetry. In addition to
comparing the measured and calculated hysteresis curves, the strains in various
components were calculated and divided by the yield strain and are discussed in the
following sections.
2.5.7.1 Comparison of Analytical and Experimental Results
Figures 2.42 and 2.43 compare the calculated and measured force-displacement
relationships for two of the columns under moderate amplitude loading. It can be
seen that generally good correlation was observed between the measured and
calculated results.
The result of tensile damage pattern in PO1-N4 is illustrated in Fig. 2.43 for
the initial cycles in which the first shear and flexural cracks were observed along
nearly the entire pier height on the webs and concentrated at the flange bases.
The compressive strain pattern is shown in Fig. 2.44b, c for 1.43% drift ratio when
some damage was observed at the pier base during the tests. The deformed mesh for
250
200
150
Horizontal Force (kN)
100
50
-50
-100
Pier PO1-N4
DamageModel
-150
-200
-250
-30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30
Fig. 2.42 Experimental and numerical results comparison for pier PO1-N4
250
200
150
Horizontal Force (kN)
100
50
-50
-100
Pier PO1-N6
-150 DamageModel
-200
-250
-30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30
Fig. 2.43 Experimental and numerical results comparison for pier PO1-N6
a DAMT
> 0 .00:E+00
b DAMC
> 0 .00E+00
c
< 9 .87E-01 < 1 .00E+00
7 . 71E-03 7 .81E-03
5 .40E-02 5 .47E-02
0 .10 0 .10
0 .15 0 .15
0 .19 0 .20
0 .24 0 .24
0 .29 0 .29
0 .33 0 .34
0 .38 0 .38
0 .42 0 .43
0 .47 0 .48
0 .52 0 .52
0 .56 0 .57
0 .61 0 .62
0 .66 0 .66
0 .70 0 .71
0 .75 0 .76
0 .79 0 .80
0 .84 0 .85
0 .89 0 .90
0 .93 0 .95
0 .98 0 .99
Tensile damage (d = 3mm) Comp. damage (d = 20mm) Base section comp. damage
(d = 20mm)
d e EPSS
f DPSS
>-1 .23E-01 >-2 .18E+00
< 5 .65E+00 < 1 .97E+00
-7 .76E-02 -2 .1
0 .19 -2 .0
0 .46 -1 .8
0 .73 -1 .6
1 .0 -1 .4
1 .3 -1 .2
1 .5 -0 .98
1 .8 -0 .79
2 .1 -0 .59
2 .4 -0 .40
2 .6 -0 .20
2 .9 -8 .48E-03
3 .2 0 .19
3 .4 0 .38
3 .7 0 .50
4 .0 0 .77
4 .2 0 .96
4 .5 1 .2
4 .8 1 .4
5 .1 1 .5
5 .3 1 .7
5 .6 1 .9
Deformed meh(d = 20mm) Transverse reinf. strain Longitudinal reinf. strain

(d = 20mm) (d = 20mm)
Fig. 2.44 Numerical results of PO1-N4 for 1.43% drift
the drift ratio of 1.43% is shown in Fig. 2.44d, in which significant shear deforma-
tion is evident. In the results shown in Fig. 2.44e, f, it can be observed that the strain
in the transverse and longitudinal reinforcement bars under 1.43% drift ratio
exceeded the yield strain as evidenced by ductilities higher than 1. The transverse
reinforcement strain pattern shows the significant influence of the shear forces,
reflected by stirrup yielding along the entire pier height suggesting development of
a strut-and-tie shear mechanism. The longitudinal bar strains indicate yielding in
the outer bars above the foundation, but with a maximum ductility of slightly above
1 suggesting very small flexural plastic deformations. These results are in general
agreement with those observed in the experimental studies.
The calculated tensile strain ratios for the initial cycles (Fig. 2.45a) show
significant cracking at webs and flanges, mainly concentrated at the pier base.
The compressive damage pattern for 1.43% drift ratio (Fig. 2.45b and c) indicates
maximum damage near the pier base, which is consistent with observed damage that
consisted of minor concrete spalling at the corners. The deformed mesh for 1.43%
drift ratio at the top of pier PO1-N6 shows significantly less shear deformation
compared with PO1-N4 (Fig. 2.45d) because of the higher transverse steel ratio.
a DAMT
> 0.00E+00
b DAMC
> 0 . 00E+00
c
< 9.87E-01 < 6 . 65E-01
7 . 71E-03 5 . 19E-03
5 . 40E-02 3 . 63E-02
0 . 10 6 . 75E-02
0 . 15 9 . 87E-02
0 . 19 0 . 13
0 . 24 0 . 16
0 . 29 0 . 19
0 . 33 0 . 22
0 . 38 0 . 25
0 . 42 0 . 29
0 . 47 0 . 32
0 . 52 0 . 35
0 . 56 0 . 38
0 . 61 0 . 41
0 . 66 0 . 44
0 . 70 0 . 47
0 . 75 0 . 50
0 . 79 0 . 53
0 . 84 0 . 57
0 . 89 0 . 60
0 . 93 0 . 63
0 . 98 0 . 66
Tensile damage (d = 3mm) Comp. damage (d = 20mm) Base section comp. damage
(d = 20mm)
d e f EPSS
>-2 . 09E+00
EPSS
>-1 . 40E-01 < 1 . 18E+01
< 2 . 55E+00 -2 . 0
-0 . 12 -1 . 3
6 . 67E-03 -0 . 68
0 . 13 -3 . 29E-02
0 . 26 0 . 62
0 . 38 1.3
0 . 51 1.9
0 . 64 2.6
0 . 76 3.2
0 . 89 3.9
1.0 4.5
1.1 5.2
1.3 5.8
1.4
6.5
1.5
7.1
1.6
7.8
1.8
8.4
1.9
9.1
2.0
9.7
2.1
10 .
2.3
11 .
2.4
12 .
2.5
Deformed mesh (d = 20mm) Transverse reinf. strain Longitudinal reinf. strain

(d = 20mm) (d = 20mm)
Fig. 2.45 Numerical results of PO1-N6 for 1.43% drift
Compared to transverse bar strains in PO1-N4, the strains shown in Fig. 2.45e
are considerably lower due to the higher amount of transverse steel in PO1-N6. Finally,
the longitudinal rebar strain distribution shows that extent of yielding is significantly
higher than that of PO1-N4. This because the higher amount of transverse steel
reduced shear degradation and allowed for higher flexural strains to develop.
2.5.7.2 Concluding Remarks
An application of refined constitutive models for estimating the nonlinear seismic

behaviour of hollow box reinforced concrete bridge piers was described. A consti-
tutive model based on the Continuum Damage Mechanics was used for the con-
crete, incorporating two independent scalar damage variables to reproduce the
degradation under tensile or compressive stress conditions. Steel reinforcement
was discretized via 2-noded truss elements, and the corresponding behaviour was
simulated using the Menegotto-Pinto cyclic model.
The efficiency of the numerical model was demonstrated by simulating the

experimental tests performed at the LESE on reduced scale bridge pier models
tested under cyclic loads. Important characteristics of the test models were ade-
quately captured using the analytical model. The shear capacity of each pier was also
accurately captured by the model developed by Priestley et al. (1996), with shear
failure in PO1-N4, and limited flexural ductility in PO1-N6, which has twice as
much shear reinforcement. The ability of the detailed analytical model to capture
shear deformations is due to the refined element modelling it incorporated.
Distributed shear degradation and failure is generally hard to be replicated by simple
macromodels.
2.6 Modelling of Dynamic Interaction Between Piers,

Foundation and Soil
2.6.1 Pseudo-static Winkler Approach
The most commonly adopted engineering method for calculating the pseudo-static
interaction between the piles of a bridge foundation and the soil is the Winkler model
in which the soil reaction to pile movement is represented by independent (linear or
non-linear) unidirectional translational spring elements distributed along the pile
shaft to account for the soil response in the elastic and inelastic range respectively.
Although approximate, Winkler formulations are widely used not only because their
predictions are in good agreement with results from more rigorous solutions but also
because the variation of soil properties along the pile length can be relatively easily
incorporated. Moreover, they are efficient in terms of computational time required,
thus allowing for easier numerical handling of the structural inelastic response.
The corresponding mechanical parameters for the springs are frequently
obtained from experimental results (leading to P-y curves for lateral and T-z curves
for axial loading) as well as from very simplified models. A commonly used P-y
curve is the lateral soil resistance vs. deflection relationship proposed by the
American Petroleum Institute (1993):

kH
P ¼ 0:9pu tanh y (2.19)
0:9pu
where pu is the ultimate bearing capacity at depth H, y is the lateral deflection and k
is the initial modulus of subgrade reaction which is both depth and diameter-
dependent despite the fact that in many cases (i.e. NAVFAC 1982) the modulus
of the subgrade reaction is assumed to be independent of diameter.
This curve is widely used in current practice, especially in the U.S., and it is
applied also for dynamic or non-linear problems. The latter is achieved with a
simple transformation of Eq. 2.1 to a bi-linear relationship by assuming a specific
deformation Dy to enter the inelastic range (typically equal to 2.5 cm for
Fig. 2.46 Two and three dimensional finite element representation of soil-foundation-pier system
stiffness (Kappos and Sextos 2001)
cohesionless soils) and a second branch stiffness reduced to 1/4 of the initial soil
stiffness (Kappos and Sextos 2001). Alternatively to the above procedure, the static
stiffness detached from the complex dynamic stiffness matrix (as discussed in the
following section) is also used in practice.
To complete the foundation modelling, a horizontal inelastic soil spring can be used
at the top of the pile to represent the strength and stiffness provided by passive soil
resistance against the pile cap while a vertical inelastic spring is commonly used at the
pile tip to account for downward and upward capacity of the supporting soil.
For the case that particular soil layers are considered to be susceptible to
liquefaction, recent studies suggest that both the lateral subgrade reaction of piles
and the maximum reaction force of the laterally-spreading soils have to be reduced
at the corresponding locations along the pile length. This reduction factor lies in the
range of 0.1–0.2 (Finn 2005) or 0.05–0.2 (Tokimatsu 1999). An average 10%
therefore of the lateral stiffness provided along the liquefied soil layers is deemed
reasonable, however, the designer has to bear in mind that the particular assumption
is strongly earthquake magnitude dependent.
Based on the above discussion, it can be claimed that at least from a static point of
view, the use of lateral soil resistance-deflection curves (even linear) is a convenient
approach for the estimation of the dynamic characteristics of the bridge. Nevertheless,
despite the wide application of the P-y approach for the assessment of the structural
response in the design practice, there are certain limitations that have to be stressed:
(a) the uncertainty of estimating the parameters involved when load tests are not
available (especially of defining pu and k), is disproportionally high compared
to the simplicity of the approach. It is notable that although Eq. 2.19 is adopted
by both the Multidisciplinary Centre for Earthquake Engineering Research and
ATC (MCEER/ATC 2003) and the California Department of Transportation
(CALTRANS) guidelines, the proposed sets of the required subgrade moduli
differ on average by a factor of 4 (Finn 2005).
(b) the relationships available that relate the 1D (expressed in terms of modulus of
subgrade reaction k) to 2D and 3D soil stiffness (Fig. 2.46), the latter expressed
in terms of modulus of elasticity Es, and Poisson’s ratio n, are not directly
related nor verified by 2D and 3D FE analyses. As a result, a set of calibration
assumptions is required for establishing a correspondence between the Winkler
and plane-strain FE approaches (Kappos and Sextos 2001) based on the initial
formulations proposed by Vesic (1961):
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:65Es Es D 4
k ¼ Dkh ¼
12
(2.20)
ð1 v2 Þ 2 ð1 v2 ÞEp Ip
where kh is the modulus of subgrade reaction and D the pile diameter. More-
over, the transformation from one soil parameter to the other is straightforward
only in the case that they are assumed constant with the depth while a set of
additional and rather case-dependent calibrations is required to obtain agree-
ment in the inelastic range.
(c) the pile group effect is essentially neglected. Even if the piles are statically
connected using appropriate single valued springs to represent the increased
flexibility of a pile group compared to the summation of the stiffness of all
individual piles (a practice that is acceptable for static analysis), the actual
dynamic impedance and the subsequent damping are completely neglected.
(d) the extension of the above P-y relationships or other curves calibrated from
static analysis or loading testing for use in the framework of dynamic analysis is
subjective.
(e) estimating the effect of soil-structure interaction solely on the basis of the
increased foundation flexibility, is an oversimplification that may lead to
unconservative response estimates under certain circumstances (Mylonakis
and Gazetas 2000).
(f) the convenience of the (particular statically based) P-y method, often leads to the
extension of its application for the case of inelastic dynamic analysis in the time
domain. Such an extension, although tempting for special cases of structural
design (i.e. performance based design of new or retrofit of existing important
structures) leads to the misleading perception of modelling refinement without
proper understanding and consideration of the complex dynamic nature of SSI
phenomena. As a result, important aspects of the soil-foundation-superstructure
system response are hidden under allegedly ‘all-purpose’ 3D linear/nonlinear,
static/dynamic stick models.
It can be claimed therefore that as soil-foundation-structure interaction is a
multi-parametric and strongly frequency-dependent phenomenon, it inevitably
has to be seen from a dynamic point of view, through a very careful selection of
FE models, associated parameters and modelling assumptions. As a result, the
pseudo-static Winkler approach is deemed appropriate only for cases wherein:
(a) what is of interest is the identification of the dynamic characteristics of the
overall soil-foundation-superstructure system and not the actual seismic
response of the system in the time domain.
(b) a preliminary (standard or modal) pushover analysis is performed in order to

quickly assess the inelastic mechanisms that are expected to be developed in the
bridge under earthquake excitation.
(c) a response spectrum analysis is performed or a linear response history analysis
is run for a relatively low level of seismic forces.
(d) an inelastic dynamic analysis is conducted but the energy absorption is
expected to be mainly concentrated on the superstructure while the material
and radiation damping at the soil-foundation interface is a-priori judged of
secondary importance (i.e. in cases that the underlying soil formations are stiff
and uniform with depth).
2.6.2 Linear Soil-Foundation-Bridge Interaction Analysis

in the Time Domain
Currently, despite the lack of specific design guidelines on how to model, compute
and consider Soil-Structure Interaction effects in the framework of the seismic
excitation of bridges in the time domain, it is common knowledge that foundation is
flexible, dissipates energy and interacts with the surrounding soil and the super-
structure, in such a way that it both filters seismic motion (kinematic interaction or
wave scattering effect) while it is subjected to inertial forces generated by the
vibration of the superstructure (inertial interaction). This phenomenon is complex
and its beneficial or detrimental effect on the dynamic response of the structure is
dependent on a series of parameters such as (Pender 1993; Wolf 1994) the intensity
of ground motion, the dominant wavelengths, the angle of incidence of the seismic
waves, the stromatography, the stiffness and damping of soil, as well as the size,
geometry, stiffness, slenderness and dynamic characteristics of the foundation and
the structure.
The basic methods to be used in the analysis of Soil-Structure Interaction effects
that implement Finite Element Discretization are the Complete Finite Element
approach and the Substructure Method. Due to the substantial computational time
required, the Complete Finite Element approach is primarily used when the overall
2D or 3D geometry of the problem is of interest and the response of the soil-
foundation-superstructure is mainly linear elastic. It has to be noted though, that in
general, the use of Finite Element representation as a means of numerically
predicting the seismic wave propagation, is relatively inferior to more specialized
approaches such as the Finite Difference method. In order to enhance the accuracy
of the FE analysis therefore, it is deemed critical to ensure realistic boundary
conditions and optimal mesh dimensions. The first is commonly tackled with the
incorporation of Kelvin elements (separately for the horizontal and
vertical directions) which is easier to be implemented compared to other artificial
boundary conditions such as superposition, paraxial or extrapolation boundaries.
It is noted that with proper selection of model parameters, the boundary inter-
ference may indeed be eliminated, as has been shown by comparing numerical with
experimental results (i.e. Pitilakis et al. 2004). The frequency dependent constants
of the Kelvin elements can be calculated using the solution developed by Novak
and Mitwally (1988), with coefficients proposed for the two horizontal and the
vertical direction respectively. Additionally, it is important to achieve adequate
(albeit not exhaustive) mesh refinement in order to minimize the error related to
finite wave propagation. A very practical limit for the characteristic length lc of the
elements has been proposed by Lysmer and Kulemeyer (1969) and is still exten-
sively used in FE discretization.
According to the Substructure Method on the other hand, the soil-foundation-
structure system is divided into substructures, typically the superstructure, the near
field soil domain inclusive of the foundation and the far field domain. Due to
considerable computational economy the substructure method has been more
extensively used in the past either in the form of:
(a) Coupled FEM/BEM approaches (Renault and Meskouris 2004; Savidis et al.
2000 among others): the advantage of this approach is that the soil can be
discretizised only in the interaction horizon, while the boundary conditions are
consistent and hence, the wave propagation in the free-field can be accurately
calculated considering non-relaxed boundary conditions.
(b) Uncoupling and superposition of kinematic and inertial interaction (Kausel and
Roesset 1994; Mylonakis et al. 1997): The dynamic stiffness matrix of the
superstructure is attached to an additional impedance matrix representing the
underlying unbounded soil region and the superstructure is then excited by
the response history (denoted as Foundation Input Motion – F.I.M.) of a hypo-
thetical soil-foundation sub-system lacking the superstructure mass.
In case of deep (pile) foundation, the procedure can be summarized in four
independent steps:
1. Analysis of the free-field soil response (i.e. without the presence of piles) to
vertically incident S waves.
2. Analysis of the interaction of the single pile or pile group with the surrounding
soil, driven by the free-field response of step 1.
3. Computation of the dynamic impedances (“springs” and “dashpots”) at the pile
head or the pile-group cap, associated with the swaying (Rx and Ry), rocking Rry
and Rrx) and cross-swaying-rocking (Rx,ry and Ry,rx) motion of the foundation.
It is noted that specifically for pile groups, the dynamic impedance of the
foundation cannot be computed by simply adding the dynamic stiffness of the
individual piles. This pile-to-pile interaction is frequency-dependent, resulting
from waves that are emitted from the periphery of each pile and propagate to
strike the neighbouring piles (Mylonakis et al. 1997). A variety of numerical and
analytical methods have been developed (Nogami et al. 1992; Makris and
Gazetas 1992; El-Naggar and Novak 1996; Gazetas and Mylonakis 2002) to
compute the dynamic response of this interaction. Especially for the rotational
stiffness of foundations supporting bridge piers or building columns that are
expected to develop plastic hinge at their base, a non-linear moment-rotation
Fig. 2.47 Combination of soil-structure-interaction and post-yielding RC pier response
relationship can be implemented (Sextos et al. 2003a) combining the rotational

compliance of the foundation with a lumped plasticity model for the RC section.
In particular, by combining the flexibility of the non-linear pier-base inelastic
spring and the linear rotational foundation spring that was calculated in the
inertial soil-structure interaction stage, the final rotational spring (Fig. 2.47) is
derived, being characterized by a first branch (uncoupled rotational) stiffness
0
equal to=Y and a second branch stiffness =y equal to:
0 1 1
=y ¼ y
¼ (2.21)
1
=y þ Mu M
p
y Re
dyn
KHH dyn
KHM =e
þ
ð0:08Lþ0:022fyl dbl Þðfu fy Þ
dyn dyn
KHH
2
KMM K dyn dyn dyn
þKyV KHH Mu My
HM
where yp, Mu, My are the plastic rotation, the ultimate and the yield moment of
dyn dyn dyn
the pier base RC section respectively, KHH ; KMM ; KHM are the horizontal,
rocking and coupled modes of vibration terms of the dynamic stiffness matrix,
which for the case of pile groups are functions of the damping coefficients zΗΗ,
zMΗ, zMM and the dynamic interaction factors adyn dyn
ij , Kyv is the (static) rotational
stiffness component attributed to the antisymmetric vertical loading of the piles,
e ¼ H/M is the foundation eccentricity, L is the distance from the critical pier
section to the point of contraflexure, fyl is the yield strength of the longitudinal
bars and dbl is the diameter of the longitudinal reinforcement.
4. Analysis of the dynamic response of the superstructure supported on the

“springs” and “dashpots” of step 3, subjected to the kinematic pile-head motion
of step 2.
The above approach is similarly applied for the case of surface foundations
according to the corresponding four steps. A summary of closed-form solutions of
dynamic stiffness of spread footings that have been derived by regression analysis,
based on Finite- and Boundary-Element data can be found elsewhere (Mylonakis
et al. 2006a). Commonly, due to their small embedment, the cross-swaying-rocking
stiffness of the spread footings can be neglected.
In general, for each of the above analysis steps, several alternative formulations
have been developed and published in the literature, including Finite Element,
Boundary Element, semi-analytical and analytical solutions, as well as a variety
of simplified methods (reviewed by Pender 1993 and Gazetas and Mylonakis 1998).
In practice however, especially when specialized software is not available, it is
convenient that the above inertial and kinematic interaction uncoupled sub-systems
are split into two FE models using any standard structural analysis software and the
corresponding damping and stiffness coefficients proposed in the literature.
2.6.2.1 Considering the Characteristics of Seismic Ground Motion
From the four steps described above, it is believed that the highest level of
uncertainty is related to the identification of the incoming wavefield (as described
in step 1). It has long been shown that the dynamic response of both sub-systems
(i.e. massless soil-foundation and flexibly supported superstructure) is strongly
frequency-dependent, hence earthquake- dependent. Therefore, as stressed by
numerous previous studies, the study of SSI effects cannot be seen independently
from the characteristics of the incoming seismic motion. To further illustrate the
interplay between the supporting soil and foundation and the induced wavefield,
Mylonakis et al. (2006b) have extended the examination of the role played by soil
in the collapse of the Hanshin Expressway during the 1995 earthquake (Fig. 2.48).
It is shown that the combination of the substantial modification of the bedrock
motion frequency content by the underlying soil profile (apparent in Fig. 2.48) and
the altered dynamic characteristics of the superstructure (both in terms of period
elongation and enhanced fundamental mode contribution) lead to a 100% increase
of the seismic demand compared to what would have been expected using a
standard ‘fixed base’ approach.
Other studies (i.e. Sextos et al. 2003b) also verify that considering soil-structure
interaction while neglecting local site response, the importance of SSI may be
underestimated by a maximum factor of 2.5, compared to the case where the local
(1D/2D or 3D basin site effects) had been properly incorporated in the framework
of the SSI analysis.
As a result, it can be claimed that even the most state-of-the-art computing tools
for considering the complex stiffness and damping matrix at the foundation level,
when used independently of realistic consideration of the multi-layer, damped soil
1.8
11
3.1 NSPT
0 50
2.3
10
14
20
0.8
2.4
9.6 3.2
2.4
0.8
10.6
Fig. 2.48 Layout of the Hanshin expressway (Mylonakis et al. 2006b)
structure, may not only lead to erroneous results but most importantly, convey the
false impression that advanced and refined modelling is used without ensuring the
corresponding analysis reliability (Fig. 2.49).
2.6.2.2 Modelling Rotational, Directional and Multiple Support Excitation
Additionally to the importance of considering the local site response as part of

the SSI analysis, there are at least three other significant factors that affect the
excitation of the bridge:
(a) Rotational component of the foundation response: It has been shown (Kavadas
and Gazetas 1993; Mylonakis et al. 1997) that the kinematic interaction between
pile and soil, apart from filtering the high frequency content of the incident
3.5
JMA
3.0 Takatori
Fukiai
Higashi Kobe (amplified)
2.5 Motoyama (amplified)
2.0
SA : g
1.5
1.0
0.5
Tfixed TSSI
0.0
0 1 2 3
PERIOD : s
Fig. 2.49 Influence of period elongation on the attracted seismic force of the Hanshin expressway
(Fukae section) for selected response spectra of the Kobe earthquake (Mylonakis et al. 2006b)
seismic waves, induces rotations at the pile head-base of the structure interface,
especially in the case of single piles or footings with a very small number of
piles (micropiles). This is a crucial issue, commonly neglected during the
excitation of the superstructure, that has to be more thoroughly investigated.
(b) Asynchronous excitation: The spatial variation of earthquake ground motions is
another aspect that is not accounted for in the vast majority of the design cases.
This is of course a reasonable approximation for buildings or short bridges, but
may considerably alter the overall seismic response of longer bridges especially
in the case that the local site conditions vary significantly with length (Der
Kiureghian, Keshishian 1997; Shinozuka et al. 2000; Mylonakis et al. 2001;
Sextos et al. 2003b; Ates et al. 2006). This problem is very complex and various
methods have been proposed for tackling the uncertainty related to the defini-
tion of a ‘realistic’ support-dependent earthquake input, as well as for
predicting the conditions under which the effect of asynchronous input is
indeed critical for the structure. Presenting in depth the multiparametric nature
of spatial variability is beyond the scope of this book. However, it is interesting
to notice that as, in most practical situations of long structures, the foundation
soil and the subsequent foundation design varies significantly along the bridge
length, the kinematic interaction is support-dependent, and hence the founda-
tion presence is an additional source of spatial variability of ground motion. As
a result, SSI effects and the final spatial variation of the travelling earthquake
motion are strongly coupled phenomena.
(c) Direction of excitation: The importance of the angle of ground motion

incidence has been highlighted by many researchers, working primarily on
buildings and the Penzien and Watabe model (1975) that is, assuming the
translational components of ground motion uncorrelated along a well-defined
orthogonal axes of the structure. Recent research by Sextos et al. (2004; Sextos
and Kappos 2008) among others, has also shown significant discrepancy
between the response quantities produced when the seismic components are
applied along the structural axes and those resulting from the excitation of the
structure for different angles of wave incidence, especially in the case of long
and curved bridges.
2.6.3 Nonlinear Soil-Foundation-Bridge Interaction Analysis

in the Time Domain
2.6.3.1 Simulating the Non-linear Response of Soils
As the earthquake intensity increases, the behaviour of soil deposits becomes more
and more complex, while the corresponding seismic demand to the bridge super-
structure may increase substantially. Advanced FE software currently supports a
large number of sophisticated material models most of which can be incorporated in
the framework of 2D or 3D discretization. Nevertheless, the numerical simulation
of the inelastic mechanisms developing simultaneously at the soil, foundation and
structure is very demanding regarding the computational time required, whereas the
accuracy foreseen through the use of advanced models is rather counteracted by the
numerous uncertainties involved. This may be primarily attributed to the fact that:
(a) despite the significant advances in modern computer software, it is difficult to
achieve a balance between the material modelling refinement for the soil and
the structure as most computer codes do not support built-in features special-
ized for both soil volumes, and structural yielding mechanisms (RC, steel or
masonry). Most importantly, even when they do (i.e. ANSYS, ABAQUS,
ADINA among others), their simultaneous activation in the time domain may
lead to significant analysis convergence difficulties.
(b) most material constitutive laws (i.e. elastic–perfectly plastic von Mises model,
Mohr-Coulomb or the Drucker- Prager criterion) that are indeed implemented
in the available structural or general purpose analysis software, require a
number of decisions on particular coefficients to which quite often the analysis
is found to be numerically sensitive. As a result, a large number of parametric
analyses and verification studies are needed.
(c) even though general purpose numerical analysis software can now efficiently
solve non-linear boundary conditions problems (i.e. sliding at the pile-soil inter-
face or sliding/rocking at the footing soil interface), it is not by any means (and
could not easily be) ensured that the coupled material and boundary conditions
non-linearity is effectively predicted by the use of the two software features used
simultaneously.
(d) similarly, the application of complex material laws for both pseudo-static
and cyclic dynamic analysis does not lead to stable and accurate converged
solutions unless first verified by very experienced users.
One solution to the above described problem is to assess in advance the locations
of the soil-foundation-superstructure system that are expected to be subjected to the
highest level of inelastic demand and develop the FE modelling strategy accord-
ingly. As a result, it is deemed a reasonable engineering approximation for cases of
low to moderate non-linearity in the soil, to apply the kinematic and inertial
procedure described in the previous section (which, strictly speaking, is not valid
in the inelastic range) using effective shear strain-compatible soil moduli and
damping for the determination of the spring and dashpot coefficients (i.e. Makris
and Gazetas 1992).
Whenever there is evidence of significant non-linear soil behaviour, and in order
to avoid undesirable analysis complexity, it is proposed to combine refined 3D
structural models developed primarily for the inelastic analysis of the superstruc-
ture in the time domain, with simple stick foundation elements supported on
ready-to-use (and computationally non-demanding) nonlinear Winkler springs
that account for the static, cyclic, and dynamic response of foundations.
Finally, when the response of the foundation itself is also sought, (i.e. pile plastic
hinges are expected to develop below the surface), then the modelling refinement
can extend to the foundation which can again be assumed supported on analytically
derived springs that account for soil non-linearity.
Three dimensional inelastic analyses in the time domain for the complete soil-
foundation-structure system is recommended mainly for cases that the structural
response can be approximated as linear elastic and the soil plastic strains are
of interest.
2.6.3.2 Dealing with Layer-Specific Liquefaction
One of the non-linear soil effects, which are most difficult to simulate using
standard finite element software, is soil liquefaction. This is a crucial soil response
phenomenon that may lead to large and permanent ground displacements hence it
has to be properly considered at least for particular soil stratification. Nevertheless,
with the exception of user-defined modules imported in commercial general pur-
pose FE codes, the programs available for considering soil liquefaction do not
account for structural inelastic behaviour (or do not account for the superstructure at
all), with the exception of Opensees.
One possible approximation to overcome the particular problem is the
implementation of an independent site response analysis where the liquefaction
susceptible layers are appropriately accounted for through specialized software,
thus leading to depth-dependent acceleration and displacement time histories
(Kwon et al. 2009).
2.7 Modelling of Abutment-Embankment-Superstructure

Interaction
2.7.1 Simple P-y Relationships for Modelling

Embankment-Abutment Systems
Nonlinear static (pushover) analysis has become a popular tool for seismic assess-
ment of reinforced concrete (RC) bridges primarily as a means to identify the
hierarchy of failure at a low computational cost. Nevertheless, the nonlinearity
expected in bridges during strong ground motions, cannot be attributed solely to
yielding of RC sections, although these are the elements that are often purposely
designed to exhibit inelastic behaviour. However, additional material nonlinearity
mechanisms (of the foundation, approach embankment, and/or backfill soil) and
geometrical nonlinearity mechanisms (activation of control components such as
bearings, ‘stoppers’, or seismic joints) can also contribute significantly to the
nonlinearity of the overall system response. Typically, both sources of nonlinearity
affect the seismic response of a bridge; however, hysteretic response of nonlinear
materials may have higher levels of uncertainty compared to the (pre-defined)
presence of gaps and joints. Despite the importance of modelling such complex
bridge lateral boundary conditions and the existence of specific guidelines in the US
(Caltrans, ATC, MCEER) and in Europe (Eurocode 8–2) for the design of pile
foundations and abutments, only minor guidance is provided by codes for numeri-
cal modelling of the coupled soil-bridge system. Practical consideration and static
assessment of the nonlinear soil-foundation-pier-deck system may be found in
Elnashai and McClure (1996) and Kappos and Sextos (2001), among others,
while static spring-based approaches for the study of soil-abutment-deck interaction
have been long developed (Maragakis et al. 1989; Siddharthan et al. 1997; Spyrakos
and Ioannidis 2003; Mackie and Stojadinovic 2002; Shamsabadi et al. 2007) most
of which are summarized in a comprehensive report by Aviram et al. (2008). When
the problem is studied under dynamic loading (Goel and Chopra 1997; Zhang and
Makris 2001 and 2002; Kotsoglou and Pantazopoulou 2007 and 2009, among
others) the derivation of the complex and coupled dynamic stiffness matrix of the
lateral supports of the bridge is a multi-parametric and complicated process and as
such, wider scatter is observed among the solutions available in the literature.
In addition, consideration of the 3D geometric and material nonlinearity of
abutment-foundation-backfill-embankment system is often equally complex, uncer-
tain, and computationally expensive and impractical for standard bridges. As a
result, engineers resort to the simplified relationships prescribed by Caltrans
to estimate the abutment-backfill soil capacity and stiffness. Another simplified
alternative is carrying out a separate pushover analysis that can be performed for the
abutment and foundation systems in order to quantify the lateral support stiffness of
the bridge (Kappos et al. 2007). The advantage of this approach is that both soil and
(abutment and foundation) concrete nonlinear behaviour can be considered as a
Fig. 2.50 Configuration of a typical seat type abutment-embankment system studied
means to provide case-specific force–deflection (i.e. P-y) relationships that can, in

turn, be used as nonlinear spring boundary constitutive models in the pushover
analysis of the overall bridge structure.
Based on the above observations, an effort was recently made (Sextos et al.
2008) to extend the above concept by performing a set of 3Dimensional nonlinear
finite element analyses on typical California overpass abutment-embankment
systems in order to provide simplified P-y relationships of the lateral supporting
system as a function of abutment type, foundation-embankment-backfill geometry,
and soil properties. These relationships can be potentially used in cases where more
accurate data is not available.
2.7.2 Typical Bridges Studied
Six typical reinforced concrete California bridges (namely Route 14, LADWP,
W180, MGR, Adobe, and La Veta) consisting of box-girder superstructures,
seat-type abutments, and shallow pile foundations were adopted in the study by
(Sextos et al. 2008). Given the short spans and relatively high deck stiffness of
the particular structures, the embankment mobilization and the inelastic
behaviour of the soil material under high shear deformation levels is anticipated
to have a significant effect on the response of the bridge under seismic loading.
The geometry of the cases studied is illustrated in Fig. 2.50 and is also
summarized in Table 2.9.
2.7.3 Modelling of the Abutment-Foundation-Backfill-

Embankment Systems
The abutment-embankment systems were modelled in 3Dimensional space using

the finite element program ABAQUS and are presented in Fig. 2.51. It was deemed
a realistic assumption to consider the soil (backfill, embankment and foundation) as
Table 2.9 Geometrical characteristics of the six abutment types

H (m) h1 (m) h2 (m) b (m) l (m) L (m) Lww (m) Lpile (m) Dpile(m) pile group
Route 14 4.70 1.75 2.25 1.00 3.40 16.50 5.35 – – –
LADWP 4.00 1.30 2.10 1.00 2.90 13.50 4.80 – – –
W180 5.30 2.35 2.25 1.20 3.50 13.00 5.70 13.00 0.40 3 10
MGR 5.00 1.90 2.35 1.10 3.50 14.00 5.30 15.00 0.75 5 1/6 1
Adobe 4.40 1.25 2.25 1.20 2.45 13.00 6.00 15.00 0.40 2 10
La Veta 4.15 1.90 1.60 1.15 3.00 23.50 5.25 17.00 0.60 2 11
Fig. 2.51 Finite element models of the six abutment-embankment systems (Sextos et al. 2008)
the nonlinear material mechanism while the RC (abutment and pile) sections
remain linear elastic with cracked section properties reduced by a factor of 2/3
compared to their gross section stiffness. Backfill soil was considered to be well
compacted granular material according to the Caltrans guidelines despite the fact
that in practice different soil types may exist. For modelling purposes, a Young’s
modulus of 60 MPa and a friction angle of 39 were adopted for the backfill soil
type while, for simplicity, the same properties were assumed for the embankment as
well, even though backfill material is usually stiffer and stronger than the embank-
ment material. The foundation soil was parametrically modified to correspond to
competent soil (A) and poor soil (B) according to Caltrans SDC (2006). Character-
istic values of modulus of elasticity E ¼ 15 MPa, friction angle j ¼ 22o and
cohesion c ¼ 100kPa for Soil A and E ¼ 5 MPa, j ¼ 5o, cohesion c ¼ 50kPa
for Soil B were also assumed for this study while Poisson’s ratio (n) and the unit
weight (g) were taken equal to 0.3 and 20kN/m3 respectively. It is noted herein,
that in the framework of the static analysis, compatible (Das 1994) static soil
stiffness was adopted since the implementation of shear wave velocity typically
overestimates the static stiffness of the finite element volume. The Mohr-Coulomb
constitutive model implemented in ABAQUS was used to simulate the nonlinear
soil behaviour while a gradually increasing pressure was applied for the push-
over analysis as distributed normal and shear force on the abutment, along the
longitudinal and transverse directions, respectively.
Fig. 2.52 Indicative plastic strains (left) and displacements (right) of an Egnatia Highway
overcrossing (Sextos & Taskari, 2009)
Pushover curves (longitudinal direction) Pushover curves (transverse direction)

40000 40000
35000 35000 soilA

soilA 30000 soilB
30000 CALTRANS
Force (kN)
Force (kN)
soilB 25000
25000
CALTRANS
20000 20000
15000 15000
10000 10000
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Displacement (m) Displacement (m)
Fig. 2.53 Pushover curves of the Route 14 bridge abutment-embankment system along the
longitudinal (left) and transverse direction (right)
Details regarding finite element discretization can be found in Sextos et al. (2008).
The corresponding nonlinear P-y relationships were derived for the six bridges
studied and for both soil types and excitation directions. It is noted herein that the
particular approach is not affected by the potential presence of bearings, gaps,
stoppers, and joints that have to be modelled independently as part of the main bridge
structure, but rather represent solely the abutment-embankment system (Fig. 2.52).
2.7.4 Proposed P-y Relationships for Typical Abutment-

Embankment Systems and Comparison with Caltrans
Guidelines
Figures 2.53, 2.54, 2.55, 2.56, 2.57, 2.58 (Sextos et al. 2008) illustrate the pushover
curves for the longitudinal (left) and the transverse (right) direction of the six
abutment-embankment systems studied for the two Caltrans soil types (A and B).
Based on passive earth pressure tests and the force deflection results from

40000 40000
35000 soilA 35000
soilA
Force (kN)
30000 soilB 30000
Force (kN)
CALTRANS soilB
25000 25000 CALTRANS
20000 20000
15000 15000
10000 10000
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Fig. 2.54 Pushover curves of the LADWP bridge abutment-embankment system along the

40000 40000
soilA soilA
35000 soilB
35000
soilB
30000
Force (kN)
30000
Force (kN)
CALTRANS CALTRANS
25000 25000
20000 20000
15000 15000
10000 10000
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Fig. 2.55 Pushover curves of the W180 bridge abutment-embankment system along the longitu-
dinal (left) and transverse direction (right)

40000 40000
soilA soilA
35000 35000
soilB soilB
30000 CALTRANS 30000
CALTRANS
Force (kN)
Force (kN)
25000 25000
20000 20000
15000 15000
10000 10000
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Fig. 2.56 Pushover curves of the MGR bridge abutment-embankment system along the longitu-

40000 40000
soilA
35000 soilB 35000
Force (kN)
Force (kN)
25000 25000
soilA
20000 20000
soilB
15000 15000
CALTRANS
10000 10000
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Fig. 2.57 Pushover curves of the Adobe bridge abutment-embankment system along the longitu-

40000 40000
soilA
35000 35000
soilB
Force (kN)
Force (kN)
25000 25000
20000 20000
15000 15000 soilA
10000 10000 soilB
CALTRANS
5000 5000
0 0
0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25
Fig. 2.58 Pushover curves of the La Veta bridge abutment-embankment system along the
large-scale abutment testing at UC Davis, a value for the initial embankment fill
stiffness equal to 11.5kN/mm/m width of the wall was used, as recommended by the
Caltrans guidelines. The initial stiffness was adjusted proportionally to the backwall
height:
Kabut ¼ Ki w ðh=1:7Þ (2.22)
where w is the width of the backwall and (h/1.7) is the proportionality factor based
on the 1.7 m height of UC Davis abutment specimen. The ultimate abutment load
was assumed to be limited by a maximum static soil passive pressure of 239kPa.
In the transverse direction, the abutment stiffness and strength obtained for the
longitudinal direction were modified using factors corresponding to a wing wall
effectiveness and participation coefficients of 2/3 and 4/3, respectively (Maroney
and Chai 1994).
The initial stiffness and the ultimate capacity of the abutments according to
Caltrans method (2006) are also plotted for comparison on each of Figs. 2.53, 2.54,
2.55, 2.56, 2.57, 2.58. The elastic and inelastic stiffness values for all the systems as
Table 2.10 Bilinear idealization of the pushover curves derived for the longitudinal (left) and the
transverse direction (right) as normalized by the abutment width
Soil A Soil B Soil A Soil B TRANS
LONG direction Kinitial Kinitial Kinitial Kinitial direction
(MN/m2) Kinel dyA(m) Kinel dyB(m) Kinel dyA(m) Kinel dyB(m) (MN/m2)
Route 14 15.7 0.037 11.2 0.033 12.3 0.037 5.4 0.033 Route 14
4.4 3.4 2.8 1.9
LADWP 18.2 0.041 12.6 0.038 12.9 0.041 6.1 0.038 LADWP
6.1 3.0 3.3 2.6
W180 28.7 0.05 22.8 0.046 21.1 0.05 9.3 0.046 W180
10.8 6.3 4.7 2.6
MGR 24.6 0.052 19.9 0.043 18.0 0.052 8.0 0.043 MGR
8.4 5.3 4.6 2.9
Adobe 21.8 0.047 16.5 0.043 17.1 0.047 3.2 0.043 Adobe
9.6 5.9 8.1 2.1
La Veta 14.8 0.048 11.5 0.042 12.7 0.048 5.7 0.042 La Veta
5.1 3.5 4.7 3.4
well as the corresponding yield displacements dy are summarized in Table 2.10,

normalized by the width of the abutment so the resulting unit is [force]/[length]2.
As anticipated, the stiffness values (both initial and post-elastic) in the longitudinal
direction are in general larger than those in the transverse direction. Moreover, the
initial stiffness values obtained from 3D analysis are generally higher than those
derived using the Caltrans procedure. The reason should be the fact that the latter
does not account for factors such as abutment dimensions, embankment geometry,
soil properties and foundation system stiffness but is based on a single, well-
controlled, large-scale abutment test result. In addition, the 3D finite element
approach is equally related to a number of modelling uncertainties.
References
Abdel-Mohti A, Pekcan G (2008) Seismic response of skewed RC box-girder bridges. J Earthq

Eng Eng Vib, IEM and MCEER 7(4):415–426
ACI [American Concrete Institute] (2008) ACI 318–08 Building code requirements for structural
concrete. American Concrete Institute, Farmington Hills
Al-Hussaini TM, Zayas VA, Constantinou MC (1994) Seismic isolation of a multi-story frame
structure using spherical sliding isolation systems. Technical Report No. NCEER-94-0007,
National Centre for Earthquake Engineering Research, Buffalo
Arêde A (1997) Seismic assessment of reinforced concrete frame structures with a new flexibility
based element, Ph.D. thesis, FEUP, Porto. http://ncrep.fe.up.pt/web/artigos/AArede_PhD_
Thesis_Public.pdf
Arêde A, Pinto AV (1996) Reinforced concrete global section modelling: definition of skeleton
curves. Special Publication No.I.96.36, ISIS, EC, JRC, Ispra (VA), Italy
Arêde A, Vila Pouca N, Monteiro A, Delgado P, Costa A, Delgado R (2009) RC hollow-pier
modelling and shear influence on the cyclic numerical response. In: Proceedings of the
COMPDYN 2009 – ECCOMAS thematic conference on computational methods in structural

dynamics and earthquake engineering, Rhodes
Ates S, Bayraktar A, Dumanoglu A (2006) The effect of spatially varying earthquake ground
motions on the stochastic response of bridges isolated with friction pendulum systems. Soil
Dyn Earthq Eng 26:31–44
Aviram A, Mackie K, Stojadinovic B (2008) Guidelines for nonlinear analysis of bridge structures
in California. PEER Report 2008/03, University of California, Berkeley
Biskinis D, Fardis MN (2007) Effect of lap splices on flexural resistance and cyclic deformation
capacity of RC members. Beton- und Stahlbetonbau (Ernst & Sohn, Berlin), 102, 51–59
Bozorgzadeh A, Megally S, Restrepo J, Ashford SA (2006) Capacity evaluation of exterior
sacrificial shear keys of bridge abutments. J Bridge Eng 11(5):555–565
Buckle IG, Constantinou M, Dicleli M, Ghasemi H (2006) Seismic isolation of highway bridges.
Special Publication MCEER-06-SP07, Multidisciplinary Centre for Extreme Events Research,
Buffalo
Caltrans [California Department of Transportation] (2006) Caltrans seismic design criteria version
1.4. Engineering Service Centre, Earthquake Engineering Branch, Sacramento, CA.
Casarotti C, Pinho R (2006) Seismic response of continuous span bridges through fibre-based finite
element analysis. J Earthq Eng Eng Vib 5(1):119–131
CEA (2003) Manuel d’utilisation de Cast3m, Commissariat à l’Énergie Atomique, Pasquet P
CEB (1996) Costa and Costa model – RC frames under earthquake loading, Comité Euro-
International du Béton, Bulletin n 231
Ceresa P, Petrini L, Pinho R (2007) Flexure-shear fibre beam-column elements for modelling
frame structures under seismic loading – state of the art. J Earthq Eng 11(SP1):46–88
Ceresa P, Petrini L, Pinho R, Sousa R (2009) A fibre flexure-shear model for seismic analysis of
RC framed structures. Earthq Eng Struct Dyn 38(5):565–586
Computers and Structures, Inc (2007) CSI analysis reference manual for SAP2000, ETABS, and
SAFE. Computers and Structures, Inc, Berkeley
Constantinou, MC, Symans, MD, Tsopelas, P ,Taylor, DP (1993) Fluid viscous dampers in
applications of seismic energy dissipation and seismic isolation. In: Proceedings of ATC-17-1
seminar on seismic isolation, passive energy dissipation, and active control, San Francisco, CA,
pp 581–591
Costa AG, Costa AC (1987) Modelo Histerético das Forças-Deslocamentos Adequado à Anàlise
Sı́smica de Estruturas (Force-displacement hysteretic model for seismic analysis of structures).
LNEC, Lisbon
Costa C, Pegon P, Arêde A, Castro J (2005) Implementation of the damage model in tension and
compression with plasticity in Cast3m Report EUR, ISPC, CEC, JRC, Ispra (VA)
Das BM (1994) Principles of geotechnical engineering, 3rd edn. PWS Kent Publishers, Boston
Delgado P, Costa A, Delgado R (2002) A simple methodology for seismic safety assessment of
bridges. In: 12th European conference on earthquake engineering, Elsevier Science Ltd,
London, 9–13 September
Delgado P, Rocha P, Rodrigues V, Santos M, Arêde A, Vila Pouca N, Costa A, Delgado R (2006)
Experimental cyclic tests and retrofit of RC hollow piers. In: Proceedings of the 13th European
conference on earthquake engineering (13ECEE), 3–8 September, Paper N. 1205, Geneva,
Switzerland
Delgado P, Rocha P, Pedrosa J, Arêde A, Vila Pouca N, Santos M, Costa A, Delgado R (2007)
Retrofitting of bridge hollow piers with CFRP. In: COMPDYN 2007 – ECCOMAS thematic
conference on computational methods in structural dynamics and earthquake engineering,
13–16 June, Paper N. 1492, Rethymnon, Crete, Greece,
Delgado R, Delgado P, Vila Pouca N, Arêde A, Rocha P, Costa A (2009) Shear effects on hollow
section piers under seismic actions: experimental and numerical analysis. Bull Earthq Eng
7(2):377–389
Der Kiureghian A, Keshishian PE (1997) Effects of incoherence, wave passage and spatially
varying site conditions on bridge response. In: Proceedings of FHWA/NCEER workshop on
the national representation of seismic motion, Report 97–0010, New York, pp 393–407
Derham CJ, Kelly JM, Thomas AG (1985) Nonlinear natural rubber bearings for seismic isolation.
Nucl Eng Des 84(3):417–428
Dolce M, Cardone D, Marnetto R (2000) Implementation and testing of passive control devices
based on shape memory alloys. Earth Eng Struct Dyn 29:945–958
Dolce M, Cardone D, Croatto F (2005) Frictional behaviour of Steel-PTFE interfaces for seismic
isolation. Bull Earthq Eng 3(1):75–99
Dolce M, Cardone D, Marnetto R, Nigro D, Palermo G (2003) A new added damping rubber
isolator (ADRI): experimental tests and numerical simulations, Proc. 8th World Seminar on
Seismic Isolation, Energy Dissipation and Active Vibration Control of Structures, Yerevan
(Armenia), Oct 6–10
Duarte RT, Oliveira CS, Costa AC, Costa AG (1990) A non-linear model for seismic analysis, design
and safety assessment of reinforced concrete buildings. In: Koridze A (ed) Earthquake damage
evaluation & vulnerability analysis of building structures. OMEGA Scientific, Wallingford
El-Naggar MH, Novak M (1996) Nonlinear analysis for dynamic lateral pile response. J Soil Dyn
Earthq Eng 15(4):233–244
Elnashai AS, McClure DC (1996) Effect of modelling assumptions and input motion
characteristics on seismic design parameters of RC ridge piers. Earthq Eng Struct Dyn
25(5):435–463
Fajfar P, Fischinger M (1987) Non-linear seismic analysis of RC buildings: implications of a case
study. Eur Earthq Eng 1(1):31–43
Faria R, Oliver J, Cervera M (1998) A strain based plastic viscous damage model for massive
concrete structures. Int J Solid Struct 35(14):1533–1558
Faria R, Vila Pouca N, Delgado R (2002) Seismic behaviour of a RC wall: numerical simulation
and experimental validation. J Earthq Eng 6(4):473–498
Faria R, Vila Pouca N, Delgado R (2004) Simulation of the cyclic behaviour of RC rectangular
hollow section bridge piers via a detailed numerical model. J Earthq Eng 8(5):725–748
Filippou FC, Popov EP, Bertero VV (1983) Modelling of RC joints under cyclic excitations.
J Struct Eng 109(11):2666–2684
Finn L (2005) A study of piles during earthquakes: issues of design and analysis. Bull Earthq Eng
3:141–234
Fischinger M, Vidic T, Fajfar P (1992) Non-linear seismic analysis of structural walls using the
multiple-vertical-line-element model, non-linear seismic analysis and design of reinforced
concrete buildings. Elsevier, Bled, pp 191–202
Fischinger M, Isaković T, Kante P (2004) Implementation of a macro model to predict seismic
response of RC structural walls. Comput Concrete 1(2):211–226
Fragiadakis M, Pinho R, Antoniou S (2008) Modelling inelastic buckling of reinforcing bars under
earthquake loading. In: Papadrakakis M, Charmpis DC, Lagaros ND, Tsompanakis Y (eds)
Progress in computational dynamics and earthquake engineering. A.A. Balkema Publishers –
Taylor, Francis, Rotterdam
Garstka B, Kratzig WB, Stangenberg F (1993) Damage assessment in cyclically reinforced concrete
members. In: Proceedings of the second European conference on structural dynamics –
EURODYN’93, 21–23 June, Trondheim, Norway
Gazetas G, Mylonakis G (1998) Seismic soil-structure interaction: new evidence and emerging
issues. Geotechnical Special Publication 75, Geotechnical earthquake engineering and soil 12
dynamics III, ASCE, 2, 1119–1174
Gazetas G, Mylonakis G (2002) Kinematic pile response to vertical P wave seismic excitation.
J Geotech Geoenviron Eng 128:860–867
Giuffrè A, Pinto PE (1970) Il comportamento del cemento armato per sollecitazione ciclice di forte
intensità Giornale del genio civile 108(5):391–408
Goel RK, Chopra A (1997) Evaluation of bridge abutment capacity and stiffness during
earthquakes. Earthq Spectra 13(1):1–23
Goel R, Chopra A (2008) Role of shear keys is seismic behaviour of bridge crossing fault-rupture
zones. J Bridge Eng 13(4):398–408
GT STRUDL (2006) GT STRUDL version 29 user’s reference manual, Georgia Tech. Research
Corporation, Atlanta, GA
Guedes J (1997), Seismic behaviour of reinforced concrete bridges. Modelling, numerical analysis
and experimental assessment. PhD thesis, University of Porto, Porto, Portugal.
Guedes J, Pegon P, Pinto AV (1994) A fibre/Timoshenko beam element in CASTEM 2000,
Special Publication Nr. I.94.31 applied mechanics unit, Safety Technology Institute, Commis-
sion of the European Communities, Joint Research Centre, Ispra Establishment, Italy
Higashino M, Okamoto S (2006) Response control and seismic isolation of buildings. Taylor,
Francis Ltd, London/New York
Inel M, Aschheim M (2004) Seismic design of columns of short bridges accounting for embank-
ment flexibility. J Struct Eng ASCE 130(10):1515–1528
Isaković T, Fischinger M (1998) Engineering modelling for inelastic seismic response of RC
bridge columns. Int J Eng Model 11(3/4):67–72
Isaković T, Fischinger M (2006) Higher modes in simplified inelastic seismic analysis of single
column bent viaducts. Earthq Eng Struct Dyn 35(1):95–114
Jiang Y, Saiidi M (1990) 4-Spring element for cyclic response of RC columns. J Struct Eng ASCE
116(4):1018–1029
Kabeyasawa T, Shiohara H, Otani S, Aoyama H (1983) Analysis of the full-scale seven-story
reinforced concrete test structure. J Fac Eng Univ Tokyo 37(2):431–478
Kanaan AE, Powell GH (1973) A general purpose computer program for dynamic analysis of
planar structures. Report UBC/EERC-73/6, University of California, Berkeley
Kappos AJ (1991) Analytical prediction of the collapse earthquake for RC buildings: suggested
methodology. Earthq Eng Struct Dyn 20(2):167–176
Kappos AJ, Sextos AG (2001) Effect of foundation type and compliance on seismic response of
1988 RC bridges. J Bridge Eng, ASCE 6(2):120–130
Kappos A, Potikas P, Sextos A (2007) Seismic assessment of an overpass bridge accounting for
nonlinear material and soil response and varying boundary conditions. Computational methods
in structural dynamics and earthquake engineering, COMPDYN 2007, Rethymnon, Greece,
CD-ROM volume
Kausel E, Roesset JM (1994) Soil-structure interaction for nuclear containment structures.
In: Proceedings ASCE power division specialty conference, Boulder, Colorado
Kavvadas M, Gazetas G (1993) Kinematic seismic response and bending of free-head piles in
layered soil. Geotechnique 43:207–222
Kelly TE (1992) Skellerup industries, lead rubber isolation bearings: experimental properties.
Holmes Consulting Group, Auckland
Kent DC, Park R (1971, July) Flexural members with confined concrete. J Struct Div, ASCE
97(ST7):1969–1990
Kotsoglou A, Pantazopoulou S (2007) Bridge–embankment interaction under transverse ground
excitation. Earthq Eng Struct Dyn 36:1719–1740
Kotsoglou A, Pantazopoulou S (2009) Assessment and modelling of embankment participation in
the seismic response of integral abutment bridges. Bull Earthq Eng 7:343–361
Kwon O-S, Sextos A, Elnashai A (2009) Seismic fragility of a bridge on liquefaction susceptible
soil, In: 10th international conference on seismic safety and reliability, 13–17 September,
Osaka, Japan
Leonhardt F (1981) From past achievements to new challenges for joints and bearings. First world
conference on joints and bearings, vol. 1, American Concrete Institute, SP-70, pp 736–755
Lysmer J, Kulemeyer RL (1969) Finite dynamic model for infinite media. J Eng Mech Div ASCE
95:759–877
Mackie K, Stojadinovic B (2002) Probabilistic seismic demand model for typical highway
overpass bridges. In: 12th European conference on earthquake engineering, London, UK,
CD-ROM volume
Maekawa K, Pimanmas A, Okamura H (2003) Nonlinear mechanics of reinforced concrete. Spon
Press, London
Makris N, Gazetas G (1992) Dynamic pile-soil-pile interaction. Part II: lateral and seismic
response. Earthq Eng Struct Dyn 21(2):145–162
Mander JB, Priestley MJN, Park R (1988) Theoretical stress–strain model for confined concrete.
J Struct Eng ASCE 114(ST8):1804–1826
Maragakis E, Thornton G, Saiidi M, Siddharthan R (1989) A simple non-linear model for the
investigation of the effects of the gap closure at the abutment joints of short bridges. Earthq
Eng Struct Dyn 18(8):1163–1178
Maroney BH, Chai YH (1994) Seismic design and retrofitting of reinforced concrete bridges.
In: Proceedings of 2nd international workshop, Earthquake Commission of New Zealand,
Queenstown
Martinez-Rueda JE, Elnashai AS (1997) Confined concrete model under cyclic load. Mater Struct
30(197):139–147
Mazzoni S, Mckenna F, Scott MH, Fenves GL, Jeremic B (2003) Open system for earthquake
engineering simulation. (OpenSees) – command language manual. UCB, PEER, University of
California, Berkeley
MCEER/ATC (2003) Recommended LFRD guidelines for the seismic design of highway bridges,
Part I: Specifications, Part I: Commentary and appendices. MCEER/ATC49. MCEER Report
No. MCEER-03SP03, University of Buffalo, Buffalo, NY.
Mckenna F, Fenves GL, Scott MH, Jeremić B (2000) Open system for earthquake engineering
simulation. http://opensees.berkeley.edu
Megally SH, Silva PF, Seible, F (2001) Seismic response of sacrificial shear keys in bridge
abutments. Report No. SSRP-2001/23, Department of Structural Engineering, University of
California, San Diego
Menegotto M, Pinto PE (1973) Method of analysis for cyclically loaded R.C. plane frames
including changes in geometry and non-elastic behaviour of elements under combined normal
force and bending, symposium on the resistance and ultimate deformability of structures acted
on by well defined repeated loads, International Association for Bridge and Structural Engi-
neering, Zurich, Switzerland, pp 15–22
Monti G, Nuti C (1992) Nonlinear cyclic behaviour of reinforcing bars including buckling. J Struct
Eng 118(12):3268–3284
Mylonakis G, Gazetas G (2000) Seismic soil-structure interaction: beneficial or detrimental?
J Earthq Eng 4(3):277–301
Mylonakis G, Nikolaou A, Gazetas G (1997) Soil-pile bridge seismic interaction: kinematic and
inertial effects. Part I: Soft soil, earthquake. Eng Struct Dyn 26:337–359
Mylonakis G, Papastamatiou D, Psycharis J, Mahmoud K (2001) Simplified modelling of bridge
response on soft soil to nonuniform seismic excitation, ASCE J Bridge Eng 6(6):587–597
Mylonakis G, Nikolaou S, Gazetas G (2006a) Footings under seismic loading: analysis and design
issues with emphasis on bridge foundations. Soil Dyn Earthq Eng 26(9):824–853
Mylonakis G, Syngros C, Gazetas G, Tazoh T (2006b) The role of soil in the collapse of 18 piers of
Hanshin Expressway in the Kobe earthquake. Earthq Eng Struct Dyn 35(5):547–575
Naeim F, Kelly JM (1999) Design of seismic isolated structures: from theory to practice. Wiley,
Chichester
Nagarajaiah S, Reinhorn, AM, Constantinou MC (1991) 3D-Basis: nonlinear dynamic analysis
of three-dimensional base isolated structures: Part II, Technical Report NCEER- 91–0005,
National Centre for Earthquake Engineering Research, State University of New York at
Buffalo, Buffalo, NY
NAVFAC [Naval Facilities Engineering Command] (1982) DM-7.2, Foundations and earth
structures, design manual. Department of the Navy, Alexandria
Nogami T, Konagai K, Otani J, Chen HL (1992) Nonlinear soil-pile interaction model for dynamic
lateral motion. J Geotech Eng ASCE 118(1):106–116
Novak M, Mitwally H (1988) Transmitting boundary for axisymmetrical dilation problems.
J Eng Mech 114(1):181–187
Paraskeva TS, Kappos AJ, Sextos AG (2006) Extension of modal pushover analysis to seismic
assessment of bridges. Earthq Eng, Struct Dyn 35(11):1269–1293
Park R, Priestley M, Gill W (1982) Ductility of square-confined concrete columns. J Struct Div
ASCE 108(4):929–950
Park Y, Wen Y, Ang A (1986) Random vibration of hysteretic systems under bi-directional ground
motions. Earthq Eng Struct Dyn 14:543–557, Wiley, New York
Pender MJ (1993) Aseismic pile foundation design analysis. Bull N Z Natl Soc Earthq Eng
26(1):49–161
Penzien J, Watabe M (1975) Characteristics of 3D earthquake ground motions. Earthq Eng, Struct
Dyn 3:365–373
Petroleum Institute (1993) Planning, designing and constructing fixed offshore platforms –
working stress design, RP 2A-WSD, USA
Pinho R (2000) Shaking table testing of RC walls. ISET J Earthq Technol 37(4):119–142
Pinho R, Casarotti C, Antoniou S (2007) A comparison of single-run pushover analysis techniques
for seismic assessment of bridges. Earthq Eng Struct Dyn 36(10):1347–1362
Pinto AV, Verzeletti G, Pegon P, Magonette G, Negro P, Guedes J (1996) Pseudo-dynamic testing
of large-scale RC bridges, ELSA Lab, Report EUR 16378 EN
Pitilakis K, Kirtas E, Sextos A, Bolton M, Madabhushi G, Brennan A (2004) Validation by
centrifuge testing of numerical simulations for soil-foundation-structure systems. In: 13th
world conference on earthquake engineering, Vancouver, Paper No. 2772
Priestley MJN, Seible F, Calvi GM (1996) Seismic design and retrofit of bridges. John Wiley,
Sons, Inc., New York
Renault P, Meskouris K (2004) Coupled boundary element – finite element method in soil-
structure interaction analyses, advanced numerical analyses of solids and structures, and
beyond. Graz University of Technology, Graz, pp 169–181
Sadrossadat-Zadeh M, Saiidi M (2007) Pre-test analytical studies of NEESR-SG 4-span bridge
model using OpenSees. Centre for Civil Engineering Earthquake Research, Department of
Civil Engineering, Report No. CCEER-07-3, University of Nevada, Reno, Nevada
Saiidi M (1982) Hysteresis models for reinforced concrete. J Struct Div ASCE 108(ST5):
1077–1085
Saiidi M, Ghusn G, Jiang Y (1989) Five-spring element for biaxially bent RC columns. J Struct
Eng ASCE 115(2):398–416
Saiidi M, Moore R, Itani A (2001) Seismic performance of reinforced concrete bridges with
unconventional configurations. ACI Struct J 98(5):717–726
Savidis SA, Bode C, Hirschauer R (2000) Three-dimensional structure-soil-structure interaction
under seismic excitations with partial uplift. 12th world conference on earthquake engineering,
Auckland, New Zealand.
Scott BD, Park R, Priestley MJN (1982) Stress-strain behaviour of concrete confined by
overlapping hoops at low and high strain rates. ACI J 79(1):13–27
SeismoSoft (2005) SeismoStruct – A computer program for static and dynamic nonlinear analysis
of framed structures [on line]. Available from URL: http://www.seismosoft.com. Accessed 30
March 2005
Sextos AG, Kappos AJ (2008) Seismic response of bridges under asynchronous excitation and
comparison with EC8 design rules. Bull Earthq Eng 7(2):519–545
Sextos A, Kappos A, Pitilakis K (2003a) Inelastic dynamic analysis of RC bridges accounting
for spatial variability of ground motion, site effects and soil-structure interaction phenomena.
Part 2: Parametric analysis. Earthq EngStruct Dyn 32(4):629–652
Sextos A, Pitilakis K, Kappos A (2003b) Inelastic dynamic analysis of RC bridges accounting
for spatial variability of ground motion, site effects and soil-structure interaction phenomena.
Part 1: Methodology and analytical tools. Earthq Eng Struct Dyn 32(4):607–627
Sextos A, Kappos A, Mergos P (2004) Effect of soil-structure interaction and spatial variability of
ground motion on irregular bridges: the case of the Krystallopigi bridge, 13th world conference
on earthquake engineering, Paper No. 2298, Vancouver, August 2004
Sextos A, Mackie K, Stojadinovic B, Taskari O (2008) Simplified P-y relationships for modelling
embankment abutment systems of typical California bridges, 14th world conference on
earthquake engineering, Beijing, China
Sextos A, Taskari O (2009) Single and multi-platform simulation of linear and non-linear bridge-
soil systems, chapter in Coupled Site and Soil-Structure Interaction Effects with Application to
Seismic Risk Mitigation, Springer
Shamsabadi A (2006) Three-dimensional nonlinear seismic soil-abutment-foundation structure

interaction analysis of skewed bridges. PhD dissertation, Department of Civil and Environ-
mental Engineering, University of Southern California, Los Angeles, CA
Shamsabadi A, Rollins KM, Kapuskar M (2007) Nonlinear soil–abutment–bridge structure
interaction for seismic performance-based design. J Geotech Geoenviron Eng, ASCE
13(6):707–714
Shinozuka M, Saxena V, Deodatis G (2000) Effect of spatial variation of earthquake ground
motion on highway structures. Technical Report, MCEER-00-0013
Siddharthan R, El-Gamal M, Maragakis E (1997) Stiffnesses of abutments on spread footings with
cohesionless backfill. Can Geotech J 34:686–697
Silva PF, Megally S, Seible F (2003) Seismic performance of sacrificial interior shear keys.
ACI Struct J 100(2):177–187
Skinner RI, Robinson H, McVerry GH (1993) An introduction to seismic isolation. Wiley,
Chichester
Spacone E, Ciampi V, Filippou FC (1992) A beam element for seismic damage analysis, Report
EERC 92–07, Earthquake Engineering Research Center. University of California, Berkeley
Spyrakos C, Ioannidis G (2003) Seismic behaviour of a post-tensioned integral bridge including
soil–structure interaction (SSI). Soil Dyn Earthq Eng 23:53–63
Sullivan T, Pinho R, Pavese A (2004) An introduction to structural testing techniques in earth-
quake engineering, Educational Report ROSE 2004/01, IUSS Press, Pavia
Takeda T, Sozen MA, Nielsen NN (1970) Reinforced concrete response to simulated earthquakes.
J Struct Mech Div ASCE ST12:96
Taucer F, Spacone E, Filippou FC (1991) A fiber beam-column element for seismic response
analysis of RC structures, Report EERC 91–17, Earthquake Engineering Research Center,
University of California. Berkeley
Taylor AW, Lin AN, Martin JW (1992) Performance of elastomers in isolation bearings:
a literature review. Earthq Spectra 8(2):279–304
Tokimatsu K (1999) Performance of pile foundations in laterally spreading soils. Proc 2nd Int
Conf Earthq Geotechnical Eng, Lisbon, Portugal 3:957–964
Vesic A (1961) Beams on elastic foundations. Proc 5th Int Conf Soil Mech Foundation Eng, Paris
1:845–850
Vulcano A, Bertero VV, Caloti V (1989) Analytical modelling of RC structural walls. Proc 9th
WCEE, Tokyo-Kyoto, Maruzen 6:41–46
Wang M-L, Shah SP (1987) Reinforced concrete hysteresis model based on the damage concept.
Earthq Eng Struct Dyn 15:993–1003
Wolf JP (1994) Foundation vibration analysis using simple physical models. Prentice Hall,
Englewood Cliffs
Zayas V, Low S (1990) A simple pendulum technique for achieving seismic isolation. Earthq
Spectra 6(2):317–333
Zhang J, Makris N (2001) Seismic response of highway overcrossings including soil–structure
interaction. PEER Report 2001/02, University of California, Berkeley
Zhang J, Makris N (2002) Kinematic response functions and dynamic stiffnesses of bridge
embankments. Earthq Eng Struct Dyn 31:1933–1966

Modelling of Bridges For Inelastic Analysis: Saiidi@unr - Edu

Uploaded by

Copyright:

Available Formats

Modelling of Bridges For Inelastic Analysis: Saiidi@unr - Edu

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modelling of Bridges For Inelastic Analysis: Saiidi@unr - Edu

Uploaded by

Copyright:

Available Formats

Chapter 2

Modelling of Bridges for Inelastic Analysis

M. Saiid Saiidi, Antonio Arêde, Donatello Cardone, Pedro Delgado,

M. Saiid Saiidi (*)

versatility of inelastic analysis also makes the results particularly sensitive to

2.2 Superstructure (Deck)

This section addresses modelling issues of bridge decks occurring in dynamic

Consideration of the vertical component of ground motion and the coupling

2.2.1 Deck Types, Sectional Layouts and Properties

2.2.1.1 Beam and Slab Bridge Decks

2.2.1.2 Cellular or Box-Section Bridge Decks

Idealization procedures in the longitudinal direction are similar to those described

2.2.1.3 Voided-Slab Decks

2.2.1.4 Effect of Plan Curvature

In the presence of deck curvature, response to longitudinal or vertical action becomes

2.2.1.5 Effect of Prestressing on Component Verification Issues

2.2.2 Role of Deck Modelling in Seismic Assessment

Either conducted through standard pushover analysis (SPA) or through NRHA in

2.2.3 Effects of Skew and Curvature in Plan

produces transverse components of force, and vice versa. Transverse shaking

2.2.4 Verification of Deck Deformation Demands

and the ultimate chord rotation is given as follows:

where av is a coefficient depending on the relationship between the shear strength

For rectangular hollow section columns or T-shaped and double-T, these

2.3 Bearings and Shear Keys

The overall seismic response of bridges may be influenced significantly by the

realistic nonlinear force-deformation relationships of bearings and shear keys is

2.3.1 Importance of Modelling of Bearings

2.3.2 Mechanical Bearings (Steel Bearings)

Fig. 2.1 Steel bearing types

response of these bearing types was modelled combining a bilinear yielding

2.3.3 Modern Bearing Types

Modern bearings are ductile, multi-rotational, and multi-directional. Accordingly,

where, G ¼ shear modulus of elasticity, A ¼ pad area, h ¼ pad thickness.

GUIDE BAR GUIDE BAR

MASONRY PL BASE PL NEOPRENE DISC

MASONRY PL BASE PL NEOPRENE DISC

Fig. 2.4 Expansion and fixed multi-rotational bearings

elastomeric bearing pads, by polytetrafluoroethylene (PTFE) stainless steel sliders, or

2.3.4 Modelling of Shear Keys

Exterior shear key

Abutment Interior shear

Fig. 2.5 Schematic view of exterior and interior shear keys

2.4 Isolation and Energy Dissipation Devices

Fig. 2.7 Type 09 element in drain-2DX and drain-3DX

Fig. 2.8 Modelling of inelastic shear key force-deformation response

F=KD F=cvα α=1 α<1

As an alternative to LDRB, the SMA-based re-centring devices, designed,

G0 is the shear modulus of the elastomer at 100% shear strain,

mFR is the friction coefficient of the sliding bearings,

Wd Wviscous þ Whysteresis þ Wfriction

xeq ¼ ð1:1  0:9Þ x0 for HDRB=ADRI with 50%<g<200% (2.12)

2.5.1 Modelling Seismic Response of Columns in Reinforced

tend to represent the overall member behaviour, macro-models usually include

2.5.2 Finite Length Plastic Hinge Model

This numerical model adopts structural beam-column elements with inelastic

Fig. 2.13 Material behaviour models: (a) Concrete; (b) Steel

The cyclic moment-curvature curve depicted in Fig. 2.12b illustrates the

Strength degradation is usually associated with stiffness degradation, which is

xeq ¼ ð1:1 0:9Þ x0 for HDRB=ADRI with 50%<g<200% (2.12)