Engineering Structures 291 (2023) 116179

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Effects of steel shim geometric characteristics and imperfections on the

behavior of unbonded elastomeric bridge bearings subjected to large
lateral displacements
Saleh Ahmadi Soleimani a, Dimitrios Konstantinidis b, *, Georgios P. Balomenos a
a
Department of Civil Engineering, McMaster University, Hamilton, ON, Canada
b
Department of Civil and Environmental Engineering, University of California, Berkeley, CA, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Unbonded laminated elastomeric bridge bearings are designed to accommodate service-level deformations, but
Laminated Elastomeric Bridge Bearings during an earthquake they can experience deformation demands that are significantly larger than those for which
Steel Shims they were designed. Because these bearings are critical elements of a bridge, they must perform reliably even
Manufacturing Imperfections
under seismic-level demands. The effects of the characteristics and condition of the steel shims on the seismic
Hysteretic Behavior
Lateral Stiffness
behavior of these bearings have received limited attention in the literature. Yet, the manufacturing process of
Damping Ratio these bearings sometimes introduces imperfections. For instance, during vulcanization, the steel shims might get
Plasticity Ratio rotated or bent. In this context, this paper investigates how manufacturing imperfections and the thickness
Damage Factor characteristics of steel shims affect the behavior of unbonded elastomeric bridge bearings subjected to successive
cycles of large lateral displacement. Finite element analysis is used to study the deformation pattern of these
bearings, their hysteretic response characteristics, and the damageability of their steel shims. It is observed that
imperfections can increase a bearing’s roll-off and risk of instability. Imperfections can also increase a bearing’s
effective lateral stiffness and damage, quantified though a specific plastic-energy-based damage factor. Plastic
damage in the shims is found to be appreciably dependent on their thickness. Under cyclic displacement, damage
manifests itself more in the form of spreading of the yielded area rather than increased peak strain ductility.
Plastic strains in the steel shims do not get close to the fracture limit, and all the bearings exhibit fairly stable
hysteretic loops under the adopted loading protocol.

1. Introduction Under service-level conditions, the lateral displacement of unbonded

elastomeric bridge bearings is usually restricted to 0.5–0.7 equivalent
Steel-laminated elastomeric bridge bearings have been historically shear strain (ESS) [i.e., imposed lateral displacement divided by total
used in bridges as reliable, cost-effective devices for accommodating rubber thickness] [2,3]. However, under strong seismic conditions, re­
service-level displacements and rotations caused mainly by thermal, tainers of bearings and shear keys might fail, and then bearings might
gravity, and traffic loads. These bearings can be classified as fully experience large shearing deformation and sliding [1,4]. This could lead
unbonded, single-side unbonded, and fully bonded, based on their con­ to permanent offset or unseating of fully unbonded bearings, as observed
nections to contact surfaces. The fully bonded bearings feature vulca­ during the 1999 Chi-Chi, 2008 Wenchuan, and 2010 Chile earthquakes
nized end plates at top and bottom, which are bolted to the bridge [5–8]. As the imposed horizontal displacement on unbonded bearings
supports. In contrast, both sides of fully unbonded bearings and one side increases, the unbonded horizontal surfaces near the corners of the
of the single-side unbonded bearings are in direct frictional contact with bearings roll off the support surfaces [9] and eventually either (i) roll­
support surfaces [1]. In single-side unbonded and fully unbonded over and then slide, or (ii) slide (without rollover), depending on the
bearings, hereinafter collectively referred to as unbonded bearings, the thicknesses of the bearings and shims and the frictional resistances at the
transferred lateral forces are limited by the frictional resistance of the bearing-support interfaces [10,11]. Herein, rollover (sometimes termed
contact surfaces. full rollover) refers to the deformation state in which the initially vertical

* Corresponding author at: Department of Civil and Environmental Engineering, University of California, Berkeley, CA 94720-1710, USA.
E-mail address: [email protected] (D. Konstantinidis).

https://doi.org/10.1016/j.engstruct.2023.116179
Received 5 January 2023; Received in revised form 9 April 2023; Accepted 17 April 2023
Available online 15 June 2023
0141-0296/© 2023 Elsevier Ltd. All rights reserved.
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

sides of the bearings become in full contact with the horizontal surfaces. is assessed based on induced von Mises stresses and plastic strain.
When they have thick steel shims and are subjected to light to moderate
axial loads and cyclic horizontal displacement equivalent to 3.0 to 4.0 2. Examined bearings and assessment approach
ESS, these bearings provide stable hysteretic shear-sliding behavior with
large slip travel and minor abrasions [11–14]. In contrast, for bearings This section provides details about the material properties, geome­
with 2-mm thin shims under relatively high compressive loads (8.3 try, as well as displacement and loading demands of the examined fully
MPa), Konstantinidis et al. [10] observed appreciable abrasion as a unbonded laminated elastomeric bridge bearings.
result of post-rollover sliding that occurred in tests at 3.0 ESS. Li et al.
[13] observed that the energy dissipation capacities of bearings under 2.1. Geometry of bearings without imperfections
displacements equivalent to 3 to 4 ESS can even become close to that of a
typical seismic isolator (e.g., lead rubber bearings). Table 1 lists the geometric properties of the three examined bearings
Because of their low lateral stiffness and relatively high energy without imperfections, which varied in terms of plan dimensions and
dissipation, some studies have proposed the use of unbonded bearings as total rubber thickness. These three examined bearings were chosen
seismic fuses devices in areas of moderate seismicity [4,15]. In this role, based on bearing designs used in the Province of Ontario, Canada, and
unbonded bearings are permitted to undergo large lateral displacements feature representative characteristics of typical bearings in practice. The
and to slide at their interfaces with substructures or superstructures, number of shims in bridge bearings in Ontario ranges from three to
resulting in period elongation and energy dissipation through friction eight, and the shape factor (defined for a rubber layer as the ratio of plan
sliding mechanisms [4]. Several experimental and numerical studies area to total lateral area free to bulge) usually ranges from 6 to 12.
have been conducted in this context in order to assess the seismic per­ Table 1 refers to B1, B2, and B3 as Bearing Groups because three
formance of bridges equipped with unbonded bearings [4,11,16]. These different shim thickness values are considered for each (i.e., 2, 3, and 4
studies demonstrated the improved seismic performance and reduced mm). Note that the minimum shim thickness permitted by AASHTO [2],
damage of the bridges equipped with unbonded bearings. However, EN 1337–3 [25], and CSA [3] is 1.6, 2, and 3 mm, respectively. Bearing
restraining systems such as shear keys and angle side retainers and Group B1 involves small, low-shape-factor bearings with a high width-
adequate support lengths or seat widths were recommended or found to-rubber-height ratio. Bearing Group B2 involves small, low-shape-
necessary for controlling the peak and residual bearing displacements factor bearings with almost the minimum width-to-rubber-height ratio
and avoiding the unseating of the bearings and span loss especially in the permitted in Ontario, respectively. Bearing Group B3 involves large,
transverse direction of bridges and at the hazard level associated with high-shape-factor bearings. Bearing Groups B1, B2, and B3 serve as
the maximum considered earthquakes. benchmarks to investigate the effects of imperfections on their perfor­
Research on the effect of shim thickness on the response and po­ mance under relatively large lateral displacement demands.
tential damage of elastomeric bearings under large lateral displacement
has been scarce. Previous studies [17,18] have shown that shim thick­ 2.2. Geometry of bearings with imperfections
ness has a minimal effect on the hysteretic response characteristic,
including the effective horizontal stiffness and damping ratio, of fully Departments/ministries of transportation perform quality control
bonded bearings but has a substantial effect on the stress distribution testing of bearings before installation, part of which includes destructive
and potential plastic deformations in the shims. Compared to fully testing. For instance, the Ministry of Transportation of Ontario (MTO)
bonded bearings, the lateral behavior of unbonded bearings might be performs destructive testing on one bearing out of 20 coming from the
more dependent on the thickness of the shims, which can promote or same lot and measures its exact geometric properties [26]. This study
suppress roll-off and thus affect the effective shear area. In addition, the examined 100 MTO destructive test reports on bearings, from which
manufacturing process of elastomeric bearings involves rubber vulca­ cases with manufacturing imperfections were identified (including bent,
nization, during which the unavoidable flow of rubber under high vertically misaligned, horizontally misaligned, and rotationally mis­
pressure and temperature can cause the shims to rotate and bend aligned shims). Preliminary FEA showed that the effects of vertically and
[19–21]. Under service-level load and displacement demands, such horizontally misaligned shims on the lateral response of bearings were
manufacturing imperfections have been found to have small effects on negligible, and thus, the present study focuses on rotationally mis­
the stresses and strains in the rubber of bearings but can considerably aligned and bent shims (see Fig. 1). The examined destructive test re­
increase the tensile stresses in shims and cause plastic damage [22–24]. ports revealed that, when imperfections were present, the amount of
To the best of the authors’ knowledge, no studies have investigated the rotation or bending of the top and bottom shims was minor compared to
effects of these types of manufacturing imperfections and of the thick­ the internal shims; in addition, all internal shims had a single type of
ness of shims on the behavior of unbonded bearings subjected to large imperfection with approximately the same intensity (magnitude). Thus,
lateral displacements. Laminate imperfections might affect the me­ in considering imperfections, this study models the top and bottom
chanical properties (e.g., effective lateral stiffness and effective damping shims without imperfections and the internal shims having equal
ratio) of the bearings and also trigger damage under cyclic lateral dis­ amount of rotation or bending. The rotational misalignment of internal
placements beyond their service level. In this context, the thickness shims is denoted by θ, as shown in Fig. 2(a),(c) for bearings with three
properties of steel laminates may also play an important role. and eight shims, respectively. Based on the examined destructive test
This paper presents a numerical study on the effects of reports, the shape of bent shims can be approximated well by a half-sine
manufacturing imperfections in the form of rotationally misaligned and curve with amplitude ρ, as shown in Fig. 2(b),(d). Note that the shear
bent shims on the behavior of fully unbonded bearings subjected to large loading is applied along the short sides of the bearings, i.e., parallel to
cyclic displacements demands. Of primary focus is how shim thickness the X-axis in Fig. 2. The maximum values of θ and ρ observed in the
affects the behavior of unbonded bearings with and without imperfec­ destructive test reports were approximately 0.03 rad and 4.5 mm,
tions. Three-dimensional finite element analysis (FEA) is conducted on respectively, for the middle shim within the three-shim bearings, while
models of bearings with various characteristics in terms of plan di­ the maximum of the average values of θ and ρ of the internal shims
mensions, slenderness, shape factor, shim thicknesses, and imperfection within the eight-shim bearings were approximately 0.025 rad and 3.5
type and intensity. The analysis considers the combined effects of axial mm, respectively [average because of slight differences among the values
load and cyclic horizontal displacement with increased amplitudes from of the internal shims]. This study uses these values for the finite element
0.5 to 2.0 ESS. Initially, the deformation patterns of bearings are studied. (FE) modelling of the three-shim and eight-shim bearings with imper­
Then, the effective horizontal stiffness and equivalent damping ratio of fections. Also, the labels θ and ρ are used later as part of the nomen­
bearings are studied at different ESS values. Finally, damage in the shims clature used to distinguish FE models with imperfections.

2
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Table 1
Geometric properties of the bearings without imperfections.
Bearing Plan dimensions Total rubber Rubber layer Steel shim Number of steel Cover rubber layer Side cover Shape
group (mm) thickness (mm) thickness (mm) thickness (mm) shims thickness (mm) thickness (mm) factor

B1 400 × 300 36 13 2, 3, and 4 3 5 6 6.6

B2 400 × 300 101 13 2, 3, and 4 8 5 6 6.6
B3 700 × 500 99 12.75 2, 3, and 4 8 5 6 11.2

Fig. 1. Photographs of cut elastomeric bearings with imperfections: (a) rotationally misaligned shims; (b) bent shims [reproduced with permission from the Ministry
of Transportation of Ontario].

Fig. 2. Imperfection configurations in bearings with three and eight shims: (a, c) rotationally misaligned internal shims; (b, d) bent internal shims.

2.3. Examined bearings

A total of 27 bearings (9 without imperfections and 18 with) are

examined in this study (see Table 2) considering different bearing ge­
ometries (i.e., B1, B2, and B3) and different shim thicknesses (i.e., 2, 3,
and 4 mm). The bearing labels in Table 2 comprise either two or three
groups of alphanumeric characters (the latter if imperfections were
present): the first group of two characters denotes the bearing geometry,
the next group of three the shim thickness, and the last group of two or
three the imperfection form (if any) and its intensity. For instance,
B2TS2ρ35 denotes B2 geometry, 2-mm-thick shims, and bending (ρ) of
all shims (except the top and bottom) by 3.5 mm.
Fig. 3. Cyclic shear displacement loading protocol.

Table 2
Bearings considered for assessing the effects of geometric imperfection on their lateral behavior.
Bearing group Imperfection Bearings Bearings Bearings
with 2-mm-thick shims with 3-mm-thick shims with 4-mm-thick shims

B1 —— B1TS2 B1TS3 B1TS4

θ = 0.03 rad B1TS2θ3 B1TS3θ3 B1TS4θ3
ρ = 4.5 mm B1TS2ρ45 B1TS3ρ45 B1TS4ρ45

B2 —— B2TS2 B2TS3 B2TS4

θ = 0.025 rad B2TS2θ25 B2TS3θ25 B2TS4θ25
ρ = 3.5 mm B2TS2ρ35 B2TS3ρ35 B2TS4ρ35

B3 —— B3TS2 B3TS3 B3TS4

θ = 0.025 rad B3TS2θ25 B3TS3θ25 B3TS4θ25
ρ = 3.5 mm B3TS2ρ35 B3TS3ρ35 B3TS4ρ35

3
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

2.4. Displacement and loading demands Table 3

Material properties of the neoprene [32] and steel [37].
The behavior of the bearings is studied under combined axial load Material Property Symbol Value
and cyclic lateral displacement. Lateral displacements are imposed on
Neoprene Shear modulus G 0.65 MPa
the bearings along their shorter sides (i.e., W in Fig. 2) or parallel to the Poisson’s ratio νr 0.4998
bridge’s longitudinal axis, i.e., the X-axis in Fig. 2, because according to
preliminary analyses, the examined imperfections cause the maximum
variations in the bearings’ lateral responses under this displacement Steel Young’s modulus Es 200 GPa
Yield stress 248 MPa
scenario. The lower and upper limits for the average pressure on the σys
Poisson’s ratio νs 0.3
bearings due to gravity loads vary among different bridge specifications
[e.g., [26–28]], but they all fall within 1–7 MPa. Also, EN 1337–3 [25],
AASHTO’s [2] design Method B, and CSA [3] permit higher pressure the corresponding 3D FE model (half bearing).
values based on a total shear strain design approach. In general, the In the FE models of the bearings, the rubber is assumed to be a
lower limit of pressure is intended for preventing bearing “walk-out” scragged neoprene with mechanical properties adopted from the study by
from the supporting area, while the upper limit is set to avoid shear Sun [32] and listed in Table 3. Sun [32] included cyclic simple-shear tests
delamination. Based on technical data provided by the MTO (personal with amplitudes up to a shear strain of 2.0. Fig. 5(a) shows one full cycle of
correspondence) for bearings used in bridges in Ontario, an average hysteresis of the scragged neoprene. The rubber is modeled as a hypere­
pressure of 5.5 MPa due to gravity loads is a reasonable value and is lastic material in conjunction with a hysteresis model. The Yeoh
therefore used for the analyses. In addition, the bearings are subjected to hyperelastic model [33] is used to characterize the nonlinear elastic
a protocol of five lateral displacement cycles of increasing amplitude, response of the neoprene rubber because of its ability to accurately
from 0.5 to 2.0 ESS (see Fig. 3). The multiple cycles of lateral dis­ represent hardening under large shear strains [21,32,34]. The 3-term
placements are imposed to assess the hysteretic characteristics of the Yeoh model in Abaqus [29] differs slightly in appearance from that of
bearings for different ESS values, and to assess the potential spread of the original work by Yeoh [33] and expresses the strain energy density
damage in their shims. It is noted that imposing displacements beyond (stored energy per unit volume) as a function of the material parameters
2.0 ESS caused some preliminary analyses to abort due to excessive C10, C20, and C30, which are related to the deviatoric part of the Cauchy-
rubber element distortions at the leading edges of the bearing. Also, Green deformation tensor, and D1, D2, and D3, which are related to the
considering previous experimental studies (e.g., Konstantinidis et al. compressibility. The hysteresis model is based on the model developed by
[10], Steelman et al. [12], Xiang and Li [14], and Wu et al. [11]), Bergström and Boyce [35] and has been used in past studies to represent
bearings begin to experience full sliding at an ESS value ranging from the hysteretic behavior of rubbers [17,23,30]. This model is defined in
1.0 to 2.6 depending on the roughness of the support surfaces, average Abaqus by four unitless parameters, namely, the stress scaling factor SSF
pressure, loading rate, and temperature. Therefore, the peak displace­ (the main parameter influencing the damping ratio), the effective stress
ment demand of 2.0 ESS considered in this study is suitable for capturing exponent m, the creep strain exponent C, and the creep parameter A. More
the effects of imperfections on the lateral response of bearings prior to details about the three-term Yeoh and the hysteretic models used can be
their full sliding, after which almost no deformation occurs in the found in Ahmadi Soleimani et al. [21] and the Abaqus user manual [29],
bearings. respectively. The bulk modulus of the considered neoprene is 340 ksi
(2345 MPa) [36], which gives the values of D1 = 0.0008529 MPa− 1, D2 =
3. Finite element modeling 0.0 MPa− 1, and D3 = 0.0 MPa− 1. A regression analysis was conducted to
calibrate parameters pertaining to the deviatoric part of deformation (i.
The software Abaqus [29] is used to model the examined bearings e., C10, C20, C30, SSF, m, C, and A) with the simple shear test data of the
given its proven capability in simulating the lateral behavior of neoprene elastomer (see Fig. 5(a)). Table 4 lists the seven calibrated pa­
unbonded elastomeric bearings [21,24,30,31]. Due to symmetry, only rameters, and Fig. 5(a) shows how well the calibrated model fits the test
half of each bearing is modelled, and appropriate boundary conditions data. The steel in the examined bearings is considered to be Grade A36,
are imposed at the nodes of the plane of symmetry, i.e., the rotations with its mechanical properties listed in Table 3, and is modeled as an
about the X- and Y-axes and the displacement in the Z direction are isotropic, elastic–plastic material using the uniaxial stress–strain
constrained. As an example, Fig. 4(a) shows the full bearing B1TS3 and behavior adopted from El-Reedy [37] (see Fig. 5(b)).

Fig. 4. (a) Bearing B1TS3 together with the finite element model of its half. Finite element models of the half of bearings: (b) B2TS3; (c) B3TS3. Sections on the
planes of symmetry of bearings: (d) B2T3Sθ25; (e) B2TS3; (f) B2TS3ρ35.

4
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

vertical direction is selected to be 1 mm in the shims, while in the rubber

it ranges from 0.8 to 2 mm. The FE models consist of a total number of
elements varying from 80,000 to 400,000, depending on the geometry.
Friction plays an important role for unbonded elastomeric bearings
because they rely on it for maintaining their position. Experimental
studies on these bearings have shown that: (a) a relatively wide range of
friction thresholds exist depending on the roughness of the contacting
surfaces [40] and the average pressure and shearing velocity
[11,12,14,41], and (b) the friction coefficient, denoted in this paper as μ,
is directly proportional to the shearing velocity and inversely propor­
tional to the normal pressure [11,12,14,41]. Li et al. [13] conducted
Fig. 5. Material behavior curves: (a) Simple shear test data of the neoprene quasi-static tests under imposed average pressures of 4–8 MPa to
(adopted from [32]) and the fitted curve; (b) Grade A36 steel stress–strain data investigate the sliding behavior of bearings in contact with steel in­
(adopted from [37]).
terfaces and observed that μ can vary from 0.15 to 0.4 depending on the
average pressure. Wu et al. [11] reported μ values ranging from 0.09 to
0.24 for bearings under testing conditions similar to those of Li et al.
Table 4
[13]. Table 5 lists expressions found in the literature for μ that depend
Parameters of the hysteretic hyperelastic model for neoprene using simple shear
test data. on one or two of the following variables: average pressure p̄, shearing
velocity vs , and cumulative slip travel δtrv for the unbonded bearing. The
Model Parameter Value
first expression appears in the European standard EN 1337–3 [25] for
Yeoh model C10 0.293 MPa unbonded bearings subjected to service-level loads. The next three ex­
C20 − 0.0281 MPa
pressions in Table 5 have been proposed by Schrage [41], Magliulo et al.
C30 0.00472 MPa
[42], and Steelman et al. [12], based on the results of quasi-static tests at
low loading rates, while the last expression, by Xiang and Li [14], is the
Hysteretic model SSF 0.551 only one that considers the combined effects of vs and p̄. The expression
A 9.28 established by Steelman et al. [12] accounts for the degradation of μ due
m 1.02 ∑
to δtrv through the term δtrv /100. For the range of displacements
C − 0.934
imposed on the examined bearings in the present study, the maximum
degradation of μ (due to δtrv ) would be negligible. In Table 5, the
expression by Xiang and Li [14] is found the most appropriate one
The shims and the rubber of the bearings are modelled in Abaqus because it considers the combined effects of vs and p̄. According to this
with C3D8I (i.e., 8-node linear continuum element with incompatible relation, μ initially increases rapidly with vs and levels off and becomes
modes) elements and C3D8H (i.e., 8-node linear continuum, hybrid almost independent of vs for vs > 20 mm/s. Under such shearing ve­
element) elements, respectively. These finite element types are selected locities and the average pressure imposed on the bearings in this study,
given their robust performance in FEA of elastomeric bearings and iso­ ( )
the second part of this expression, i.e., 1.09p̄− 0.59 − 1.02p̄− 0.72 ×
lators [31,34,38,39]. For example, hybrid elements are necessary to
e− 0.39vs , becomes very close to zero. Therefore, for seismic-level exci­
prevent ill-conditioning and volumetric-mesh-locking issues of the
tations which would impose high shearing velocities, this expression can
nearly incompressible rubber, and the addition of incompatible modes
are necessary to prevent shear locking of the shims [29]. A mesh be simplified as: μ = 1.09p̄− 0.59 . For p̄ = 5.5 MPa (i.e., the average
sensitivity study was conducted to decide about the mesh size that will pressure on the examined bearings in the present study), this equation
provide the most representative values for the output parameters of gives μ = 0.4.
interest, including the total lateral reaction forces of bearings and the The constant friction coefficient value of 0.4 is used with the Coulomb
peak von Mises stresses and plastic strains in shims at 2.0 ESS. Mesh friction model for the contact definition between the bearing and the
convergence was considered successful when the relative difference of loading plates in the FEA. The friction model is used because of its
the peak of the responses of interest became less than 1% between two implementation simplicity and its ability to provide a reasonable rep­
consecutive examined mesh sizes. In summary, the FE models of the resentation of the frictional behavior [30,43,44]. It is also noted that
bearings with B1 and B2 geometries are discretized using 60 and 35 μ = 0.4 was also reported by Xiang et al. [16] based on shake table
elements along the X- and Z-axes, respectively, while for the B3 geom­ tests on bearings in contact with steel surfaces, and it is close to the μ =
etry, 85 and 50 elements are used in these directions. Finer meshes are 0.35 value recommended by Caltrans [45] for the seismic analysis of
used close to the sides of the bearings to handle the increased distortion elastomeric bearings in contact with steel surfaces. Finally, at the inte­
there compared to the central regions. The element size along the rior steel–rubber interfaces of the bearings, continuity constraints are
imposed by merging the coinciding nodes there. This modeling
assumption is made because shear delamination at the rubber-shim in­
Table 5
terfaces has not been reported in any studies of unbonded bearings
Expressions available in the literature for the μ of elastomeric bearings (p̄ in MPa,
experiencing moderate monotonic axial loads followed by several lateral
δtrv in m, vs in mm/s).
displacement cycles with amplitude of 2.0 ESS, or even larger [9,11,12].
Friction coefficient, μ Contact Reference
In contrast to what occurs in fully bonded bearings, in unbonded bear­
surface
ings, tensile stresses at steel-elastomer interfaces (which are caused
0.1 + 0.9/p̄ Rubber- EN 1337–3 [25] mainly by the unbalanced moment arising from P-delta effect due to the
Concrete
0.05 + 0.4/p̄ Rubber- Schrage [41]
simultaneous actions of shear and compression) are limited because of
Concrete roll-off of their top or bottom surfaces [46,47].
0.1 + 0.055/p̄, for 0.14 ≤ p̄ ≤ 5 Rubber- Magliulo et al. The bearings are subjected to a compression force corresponding to
∑
Concrete [42] the average pressure of 5.5 MPa, followed by the cyclic lateral dis­
0.18 + 0.37/p̄ − Rubber- Steelman et al.
δtrv /100
placements with the protocol shown in Fig. 3, during which the
Concrete [12]
1.09p̄− 0.59
−
(
1.09p̄− 0.59
− 1.02p̄− 0.72
)
× Rubber-Steel Xiang and Li compression force is kept constant. In the compression stage, the top
e − 0.39vs [14] loading plate is displaced downward without being allowed to rotate or

5
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Fig. 6. Deformed shapes (on XY-plane of symmetry): (a) bearing B2TS2 at different ESS levels, (b-d) selected bearings at either ESS = 2.0 or ESS = –2.0.

displace horizontally while in the shearing stage, the top plate is dis­ roll over easier when it is sheared toward the left side. This can be
placed horizontally without being allowed to rotate. Finally, FEA are considered an undesirable feature in laminated bridge bearings because
carried out in small increments using a Lagrangian formulation and the further displacement beyond rollover has been experimentally observed
Newton-Raphson iteration method, and both geometric and material to cause extensive damage in the rubber [10]. Also, it is observed in
nonlinearities are included. Fig. 6(d) that the bent shims in B2TS2ρ35 make the initially vertical
surfaces on the left and right sides to be deformed slightly differently
4. Observed deformation and displacement pattern in from each other. Finally, for each examined bearing, the intensities of
elastomeric bearings roll-off in the fourth and fifth cycle at 2.0 ESS differ slightly from each
other (e.g., see the differences between the corresponding deformed
This section discusses the deformation patterns obtained by FEA of shapes in Fig. 6(a)). This is attributed to local slippage happening at the
the examined bearings. As an illustrative example, Fig. 6(a) shows the bearing–support interfaces and plastic deformation in the shims.
deformed shapes of bearing B2TS2 at four different ESS levels ranging The potential sliding and cumulative slip travel of a given bearing
from 0.5 in the first cycle to 2.0 in the fourth and fifth cycles. Similar to was assessed by tracking the positions of nodes at the centers of the
the deformed shape of bearing B2TS2 at 0.5 ESS in Fig. 6(a), all exam­ bearing’s top and bottom horizontal surfaces and of the loading plates
ined bearings experience mainly simple-shear-like deformations for ESS during the cyclic displacement-controlled FEA. The results showed that
≤ 0.5, essentially acting as bonded bearings. As ESS increases, the none of the examined bearings experience full sliding up to ±2.0 ESS.
corners of the unbonded bearing start to roll off the support surfaces. For The sliding of all examined bearings except bearings with B2 geometries
bearing B2TS2, this occurs near ESS = 1 of the first cycle. At ±2.0 ESS, and 2-mm-thick shims is limited to 1 mm at the end of 5 cycles. How­
bearings with B2 and B3 geometries and 2-mm-thick shims (i.e., B2TS2, ever, in bearings with B2 geometries and 2-mm-thick shims, the rapid-
B3TS2, B3TS2θ25, and B3TS2ρ35 in Fig. 6(c,d)) are near the rollover rate increase in roll-off amount leads to the loss of contact at two of
condition. In contrast, the amount of roll-off in bearings with either or the neighboring nodes used for slip measurements at ESS values larger
both B1 geometry and 4-mm-thick shims (e.g., B1TS4, B1TS4θ3, B2TS4, than 1.48. Thus, any further imposed lateral displacement on the top
and B2TS4ρ45 in Fig. 6(b,c)) is limited due to their low thickness and/or loading plate is then accompanied with marked increase in the hori­
their higher shim flexural resistance. Comparison of the deformed zontal distance between those two neighboring nodes. Through this
shapes of each bearing without imperfection (i.e., B1TS4, B2TS4, and mechanism, the bearing “walks” on the support surfaces by losing its
B3TS2) to its corresponding bearings with imperfections, as shown in initial contact with the support surfaces at some regions and contacting
Fig. 6(b-d), indicates that the examined imperfections (i.e., rotationally some other regions of the support surfaces, which can be regarded as the
misaligned or bent shims) do not cause a notable difference in the walking travel of the bearing. The maximum cumulative walking travel
overall deformed shapes of the bearings. However, the roll-off amount value of about 118 mm at the end of five cycles was observed for bearing
and subsequently the ESS level at which rollover happens can be B2TS2 which is still small compared to the total imposed displacements
affected by imperfections. For instance, at − 2.0 ESS, significant portions of 2828 mm on this bearing.
of the vertical sides of bearing B3TS2θ25 contact the support surfaces
(see Fig. 6(d)), while this is not the case at +2.0 ESS. This occurs because 5. Hysteretic responses of the bearings without imperfections
in bearing B3TS2θ25, the rotationally misaligned shims and the result­
ing lateral bulging of the top and bottom rubber layers make this bearing Fig. 7(a-c) illustrates the 5-cycle shear stress–ESS loops of bearings

6
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Fig. 7. Hysteretic responses of three bearings subjected to the five-cycle loading protocol: (a) B1TS2; (b) B2TS2; (c) B3TS2. The fifth-cycle hysteretic loops of all
bearings without imperfections and with geometries: (d) B1; (e) B2; (f) B3.

B1TS2, B2TS2, and B3TS2 (i.e., bearings from the B1, B2, and B3 groups, almost no effect on the hysteretic response for the B1 bearings without
without imperfections, and with 2-mm-thick shims (TS2)), and Fig. 7(d- imperfections. This is because the hysteretic response of unbonded
f) illustrates the fifth-cycle shear stress–ESS curves of all the examined bearings that experience minimal roll-off and slippage is similar to that
bearings without imperfections. In these and subsequent graphs, shear of fully bonded bearings, for which the shim thickness has been also
stress is defined as the ratio of the horizontal force to plan area when the found to have a very small effect on the bearing’s hysteretic response
bearing is undeformed. From Fig. 7(a-c), the following observations are [17]. However, in thicker bearings (e.g., B2 and B3 bearings), increasing
made: (i) The hysteretic loops maintain their overall shape under mul­ the shim thickness can increase the tangent modulus or stiffness at
tiple displacement cycles, while slight differences exist between the higher ESS values (see Fig. 7(e,f)). This is attributed to the reduction of
loops of the last two cycles, which is attributed to plastic deformations in roll-off amount in such bearings due to the higher flexural stiffness of the
the shims (which is discussed in Section Damage in the Shims), residual thick shims that resist the bearing’s roll-off. In slender bearings (having
walking displacements from the fourth cycle, and differences between low second shape factor), instability might occur before sliding. For
the roll-off intensities of the last two cycles. (ii) The bearings display instance, Konstantinidis et al. [10] observed a rapid-rate roll-off of a
relatively narrow hysteretic loops up to ±2.0 ESS, which is attributed to tested slender bearing with 1.9-mm-thick shims and a second shape
the shear deformation of the (low-damping) neoprene used and the lack factor of 1.84, which led to instability at approximately 1.5 ESS.
of any full sliding at the bearing–support interfaces. The nonlinearity in Considering the observed effects of shim thickness on the roll-off of
the responses is attributed to both strain-hardening of the rubber and bearings with B2 geometry (Fig. 6(c)) and on their tangent stiffness
roll-off of the bearings. The strain hardening in the rubber results in (Fig. 7(e)), it can be concluded that the increase in shim thickness in
increase in the tangent shear stiffness at higher shear strains (see Fig. 7 slender bearings can effectively prevent their potential instability under
(a)), which is noticeable in short bearings (e.g., bearings with B1 ge­ large displacements. However, instability is not generally a concern in
ometry) that experience minimal roll-off. In contrast, the loss of effective short bearings, and the roll-off amount in short bearings with even thin
shear area (for regions still in contact with the supports) due to roll-off, shims (e.g., bearing B1TS2) is already small at 2.0 ESS. Finally, it is
which is more noticeable in bearings with low second shape factor, i.e., noted that for the examined bearings, increasing the shim thickness
ratio of width to total rubber thickness (e.g., bearings with B2 geometry could cause only slight increases in the tensile stresses induced in very
and 2-mm-thick shims), results in the decrease of tangent shear modulus small regions of the steel-elastomer interfaces, which might potentially
at higher ESS values. For instance, the tangent shear modulus of bearing lead to local debonding at larger displacement amplitudes.
B2TS2 (see Fig. 7(b)) reaches a low value of 0.15 MPa in cycle 5 at 2.0
ESS, approaching lateral instability. Therefore, bearings with 2-mm 6. Horizontal stiffness and damping ratio of bearings without
thick shims and second shape factor of 3 (i.e., the minimum allowed imperfections
in Ontario) seem to barely avoid instability up to 2.0 ESS. The lateral
instability of a series of bridge bearings is of concern because it can This section focuses on the effective horizontal stiffness and effective
promote large lateral displacements of the bridge superstructure. When damping ratio of the bearings without imperfections to characterize
these bearings reach the roll-over condition, where the originally ver­ their hysteretic responses and quantify the effects of shim thickness on
tical faces of the bearings become horizontal, their stiffness becomes their hysteretic responses at different displacement amplitudes. The
positive and large, and any further deformation causes their lateral force effective lateral stiffness of these bearings is compared with the nominal
to reach the frictional strength of their contact interfaces, at which point horizontal stiffness, Kh = GAh /tr , where G = shear modulus of the
the bearings begin to slide. Sliding can, in turn, lead to unseating from rubber, Ah = bearing’s plan area, and tr = total rubber thickness. The
the piers or abutments. Based on the numerical study by Filipov et al. effective lateral stiffness, Keff , hereinafter referred to simply as stiffness,
[4], the lateral instability of bridge bearings is believed to rapidly move and effective damping ratio, ξeff , are calculated at the displacement
the bearings into an unseated configuration. amplitude D = ESS × tr as
The graphs in Fig. 7(d) show that increasing the shim thickness has

7
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Fmax − Fmin WD 7. Influence of imperfections on the hysteretic response of

Keff = and ξeff = (1)
2D 2πKeff D2 bearings

respectively, where Fmin = lateral force corresponding to − ESS, Fmax = Fig. 9 illustrates the 5-cycle shear stress–ESS hysteresis loops of the
lateral force corresponding to + ESS, and WD = area enclosed by the bearings with imperfections and 2-mm-thick shims. The hysteretic re­
hysteresis loop (energy dissipated per cycle). The comparison between sponses are quite similar for a bearing with imperfections (Fig. 9) and
Keff and Kh values of the bearings is done using the normalized stiffness the corresponding one without imperfections (Fig. 7). However, it is
K̄eff = Keff /Kh . observed that the hysteretic loops of bearings with rotationally mis­
K̄eff and ξeff of the examined bearings are calculated at ESS levels 0.5, aligned shims have positive stress offsets at zero ESS (see Fig. 9), which
1.0, 1.5, and 2.0 reflecting cycles 1, 2, 3, and 4 or 5, respectively. Fig. 8 are attributed to the asymmetry of these bearings. This asymmetry leads
(a) shows graphs of K̄eff against these ESS levels for the bearings without to the development of reaction shear forces when the top plate is dis­
imperfections with 2- and 4-mm-thick shims. In Fig. 8(a), each displayed placed downward without being allowed to rotate or displace horizon­
tally during the compression stage. Also, it is observed in Fig. 9(c) that
data point at the ESS level 2.0 corresponds to the calculated K̄eff at either
the tangent stiffness of bearing B3TS2ρ35 increases rapidly at about
cycle 4 or cycle 5, whichever shows the maximum variation of K̄eff with
− 2.0 ESS. This is because major regions of the initially vertical sides of
respect to its corresponding value at the ESS level 1.5. The following
this bearing contact the support surfaces due to roll-off (as shown in
observations can be drawn from Fig. 8(a): (i) For the plotted range of ESS
Fig. 6(d)), providing considerable resistance.
values, the stiffness values are always less than their corresponding Kh
To better visualize the influence of imperfections on the hysteretic
values, and the Kh formula may not provide an accurate stiffness esti­
responses, the fourth-cycle loops of different bearings are illustrated in
mate. (ii) The stiffness values range from about 55 to 100% of their
Fig. 10, from which the following observations can be drawn: (i) As
corresponding Kh values depending on the bearing geometry and the
expected from the deformation patterns shown earlier, the shim
target ESS level. This large difference is attributed to various effects,
imperfection does not cause a considerable difference in the hysteretic
including the roll-off, the flexural deformation in the shims, the applied
responses of bearings but can affect their tangent shear modulus or
axial load, the walking displacements of the bearing, the strain hard­
stiffness, especially at ESS values close to ±2.0. For instance, rotation­
ening of the rubber, and the presence of side covers all around the
ally misaligned shims lead to the reduction of tangent stiffness at 2.0 ESS
bearings. (iii) As the ESS increases from 0.5 to 2.0, bearing B1TS4 and
from 0.15 MPa in bearing B2TS2 to 0.07 MPa in bearing B2TS2θ25,
bearing B2TS2 experience the largest stiffness increase by 16% and
increasing the risk of instability. (ii) Imperfections seem to cause the
decrease by 19%, respectively. This is related to the aforementioned fact
largest increase in tangent stiffness in the short bearings (bearings with
that roll-off reduces the stiffness (more pronouncedly in tall bearings
B1 geometry). This is because the deformation pattern in short bearings
with thin shims), while the strain hardening of the rubber increases
without imperfections is essentially simple shear in the rubber, with
stiffness (more pronouncedly in short bearings with thick shims).
entirely negligible deformations in the horizontal shims, but any shim
Similar to Fig. 8(a), Fig. 8(b) shows graphs of ξeff against the ESS levels
imperfection (deviation from being completely horizontal) introduces
for the bearings without imperfections and with 2- and 4-mm-thick additional resistance associated with deforming the shims, which man­
shims. It is observed that the overall relationship between ξeff and ESS ifests itself in increased tangent stiffness. However, tall bearings (bear­
usually contrasts with the overall relationship between K̄eff and ESS, ings with B2 and B3 geometries) tend to experience large roll-off and
which is expected based on the definition of ξeff , which includes Keff in thus force their shims to flex and rotate according to the roll-off (see
the denominator. ξeff ranges from 3% for bearing B1TS4 to 9% for shims of bearings B2TS2, B2TS4, B3TS2 in Fig. 6(c,d)). In this regard,
bearing B2TS2, with both values corresponding to 2.0 ESS. In bearing the shapes of rotationally misaligned and bent shims can partially
B2TS2, the rapid increase in ξeff with the increase in ESS is attributed to conform with the roll-off shape, and thus reduce the opposition of those
not only the stiffness reduction but also the increase in energy dissipa­ shims to the roll-off. For instance, in bearing B3TS2θ25 sheared toward
tion due to extensive plastic deformations in the shims and walking of the negative X-axis (see Fig. 6(d)), the overall shape of rotationally
the bearing. Considering both graphs of Fig. 8, it is observed that misaligned shims (θ about positive Z-axis) partially conforms with the
increasing the shim thickness from 2 to 4 mm increases the stiffness by roll-off shape. This finally causes extended regions of the initially ver­
up to about 31% and decreases the damping by up to about 80%, both tical surfaces of bearing B3TS2θ25 to contact the support surfaces in
for bearings with B2 geometry at 2.0 ESS. All in all, both the effective comparison to when it is sheared toward the positive X-axis (Fig. 6(d)).
damping and effective stiffness of elastomeric bridge bearings depend on The maximum absolute horizontal force, denoted as F, is assessed
the thickness of their steel shims, and the Kh formula may not provide together with the Keff and ξeff in order to quantify the effects of imper­
accurate enough equivalent linear models of the bearings in the design fections on the hysteretic responses. In this regard, Fig. 11 shows the
and evaluation of a bridge. sensitivity of the Keff and F of all the examined bearings to the imper­
fections at different ESS levels. The top-row and bottom-row graphs
reveal the percentage changes of these two quantities with respect to
( )/
bearings without imperfections, i.e., δKeff /Keff 0 = Keff − Keff 0 Keff0
and δF/F0 = (F − F0 )/F0 , where the subscript “0” denotes lack of im­
perfections. The top-row graphs in Fig. 11 show that imperfections in­
crease the stiffness of bearings at different ESS levels. It is observed that
the percentage increase in stiffness due to imperfections can either
Keff

eff

intensify at higher ESS levels (i.e., for short bearings (see top graphs of
Fig. 11(a))) or be fairly constant at different ESS levels (i.e., for bearings
with low width-to-rubber-height ratio (see top graphs of Fig. 11(b))).
Additionally, for bearings with B1 geometry, imperfections of thicker
shims cause the largest increase in δKeff /Keff 0 , which is more pronounced
at higher ESS levels (see Fig. 11(a)). However, this might not be true for
Fig. 8. Properties of examined bearings without imperfections and with 2- and bearings with B2 and B3 geometries because of the earlier mentioned
3-mm-thick shims at different ESS levels: (a) K̄eff ; (b) ξeff . effects of shim imperfection on the roll-off of these bearings. For
instance, rotationally misaligned shims lead to the contact of major

8
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Fig. 9. Five-cycle hysteretic responses of bearings with geometries (a) B1, (b) B2, and (c) B3; all with imperfections and 2-mm-thick shims.

Fig. 10. Fourth-cycle hysteretic responses of bearings with geometries: (a) B1; (b) B2; (c) B3.

regions of initially vertical surfaces of bearing B3TS2θ25 with the sup­ the top graph of Fig. 11(c). The top-row graphs of Fig. 11 also show that
port surfaces at about − 2.0 ESS. Once this occurs, the bearing stiffens at higher ESS levels, rotationally misaligned shims cause larger increases
considerably; this stiffening is evident from the corresponding curve of in δKeff /Keff 0 compared to bent shims. In summary, imperfections could
Keff Keff
FF

Fig. 11. Sensitivity of Keff and F to the imperfections at different ESS levels for bearings with geometries: (a) B1; (b) B2; (c) B3.

9
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

increase the bearings stiffnesses up to 5% at 0.5 ESS and up to 18% at 2.0 ∑ ∑( / )

ESS. ΩE = Ωj = ESp,j EW (3)
The bottom-row graphs in Fig. 11 show that F of all bearings in­
j j

creases due to imperfections, with a maximum percentage increase at where j = index that increments over the shims; ESp,j = plastic energy
either 0.5 ESS or 2.0 ESS. For bearings with rotationally misaligned dissipation in the jth shim at the end of considered cycle; and EW = input
shims, the increase in F at 0.5 ESS corresponds mainly to the induced energy from the application of compressive load and lateral displace­
reaction force that arises in order to maintain the zero rotation and ment (i.e., the work done by external forces).
translation boundary conditions of the top plate during the compression Fig. 12 illustrates the distribution of von Mises stress, σ v , and
of these asymmetric bearings, which can be up to about 40% in bearings maximum principal plastic strain, εpl , in the shims of bearings without
with eight shims. Although F values at 0.5 ESS are relatively small by
imperfection and with 2-mm-thick shims (i.e., B1TS2, B2TS2, and
themselves, such a 40% increase in F would be at the service-level
B3TS2), in which the roll-off of corners develops rapidly as ESS in­
condition and might be an issue with regard to the serviceability limit
creases. Gray color in the von Mises stress contours denotes regions
states of bridges. The curves that correspond to bearings with bent shims
exceeding the yield stress at the indicated ESS level, and gray color in the
indicate that this form of imperfection has minor effects on F for bear­
maximum principal plastic strain contours shows the plastic regions at
ings with B2 and B3 geometries but can increase F by up to about 13%
the end of cycle 5. The following observations can be drawn from
for bearings with B1 geometry. Rotationally misaligned shims can also
Fig. 12:
increase F at 2.0 ESS up to 19%, 9%, and 23% in bearings with B1, B2,
and B3 geometry, respectively. Note that the lateral force capacity of
(i) The extreme values of von Mises stress always occur close to the
bearings not experiencing instability would be basically the frictional
left and right ends of the shims, which is attributed mainly to the
strength of the interfaces. Considering this and the observed slight effect
flexural deformations, and gray color regions (denoting the
of imperfections on the overall deformed shapes of bearings, their lateral
yielded regions) are observable first on the deformed states cor­
force capacity is not expected to be notably affected by the imperfec­
responding to 1.0 ESS in bearings B2TS2 and B3TS2 and 1.5 ESS
tions. However, any increase in the stiffness or lateral force of bearings
in bearing B1TS2.
with imperfections at an ESS level before their potential sliding would be
(ii) At ESS values higher than about 1.5, the plasticity spreads rapidly
equivalent to the increase in the force transmitted to the bridge super­
in the shims of bearings B2TS2 and B3TS2, and its extent (i.e., the
structure under seismic actions. Therefore, this might increase the
total yielded area) becomes much larger than that of bearing
vulnerability of substructure before their sliding and would be unde­
B1TS2. This is attributed to the large roll-off occurring at the
sirable if it leads to any damage to the substructure components such as
corners of bearings B2TS2 and B3TS2. By the end of cycle 5,
piers. It is noted that the effects of imperfections on ξeff were studied but
approximately 24%, 75%, and 55% of the total plan areas of all
found to be marginal, and therefore, for the sake of brevity, discussion is shims of bearings B1TS2, B2TS2, and B3TS2 are yielded. It is
kept to a minimum here. In summary, imperfections can increase ξeff by noted that the peak values of plastic strain in all the examined
up to 15% and decrease it by up to 12% depending on the geometry of bearings occur sufficiently far from the shims edges (i.e., at least
the bearing, thickness of the shims, and the ESS level. However, at a 40-millimeter distance) where local responses might suffer
considering the maximum ξeff value of 9% in the examined bearings from strain and stress concentrations.
without imperfections (see Fig. 8(b)), these relative changes are rather (iii) Shearing of these bearings from the end of cycle 2 to the end of
insignificant. cycle 5 leads to an increase in maximum principal plastic strain
by up to 850% and in plasticity extent by up to approximately
8. Damage in the shims 2500%. This shows that the plastic damage in sheared bearings
tends to spread over the shims rather than intensify in their
When unbonded bearings are subjected to large lateral displace­ critical regions.
ments, their shims can experience high flexural deformations and (iv) Comparison of the gray color areas of the von Mises stress con­
yielding due to the roll-off effect. To quantify and compare the intensity tours, that correspond to cycles 4 and 5 in each bearing, indicates
and extent of potential damage in different bearings and ascertain their that the yielded area spreads under subsequent cycles of the same
vulnerabilities to this kind of damage, this study evaluates the von Mises amplitude. For instance, the extent of yielding in bearings B1TS2,
stress, maximum principal plastic strain, and plastic energy dissipation B2TS2, and B3TS2 increases by approximately 13%, 4%, and
in the shims of the examined bearings. To this end, the distribution of 12%, respectively, when they are subjected to cycle 5. The
von Mises stress and plastic strain is initially examined for shims of observed spread of plasticity in these bearings indicates bearings
bearings without imperfections. Then, a strain-based ductility factor, με , with thin shims might finally lose their vertical stiffness or
and an energy-based damage factor, ΩE , are used to assess the intensity of become unstable if they are subjected to a large number of
plasticity and its extent, respectively, in shims of bearings with or displacement cycles even with the same amplitudes. However, a
without imperfections. The factor με is defined as notable reduction in the vertical stiffness is unlikely in a single
/ seismic event because further assessment of the examined bear­
με = εmax εys (2) ings shows that the change in vertical deflection from cycle 1 to
the end of cycle 5 is negligible. Therefore, the shims of the
where εmax = maximum total (i.e., plastic plus yield strain) principal
bearings can effectively constrain the bulging of the rubber layers
strain in the shims, and εys = yield strain of steel (i.e., 0.0012). The
even if over 75% of the total area of the shims has yielded (i.e.,
induced plastic strains were carefully examined to avoid reporting the case for bearing B2TS2). This is because the yielded regions of
spurious spikes as the maximum strain values. For example, in con­ the shims in the examined bearings are in the early stages of
ducting mesh sensitivity analyses, considerably denser meshes were strain hardening of the steel and are very far from strain
applied to verify that the peak strains at the regions of interest were not softening.
dependent on the density of the mesh, and indeed the results were very
similar, which was evidence that the stress spikes affect only negligible The horizontal axes of Fig. 13 show με of all shims in bearings
parts of the steel shims at their edges. The energy-based damage factor, without imperfections subjected to five cycles of lateral displacement.
ΩE , (proposed by Camara et al. [48] for distinguishing the damage in As can be seen, for bearings with B1 geometry, which contain three
different structural members and used by Kalfas et al. [17] for the shims, the με value of the top and bottom shims is higher than that of the
damage assessment in fully bonded bearings) is defined as

10
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

v
> y pl
v pl, max
pl, max

v
ESS =

v
ESS =

v
ESS =

v
ESS =

v
ESS =

pl
ESS =

Fig. 12. Contours of von Mises stress and plastic strain in the shims of bearings: (a) B1TS2; (b) B2TS2; (c) B3TS2.

middle shim. However, in bearings with B2 and B3 geometries, the bearings with B1 geometry is limited (see Fig. 6(b)), and the pressure
distribution of με becomes relatively more uniform along the height of caused by the axial load and moment induces additional flexural de­
the bearing; and in bearings B2TS2 and B3TS2, shims closer to the mid- formations and shear stresses in the top and bottom shims, resulting in
height experience the maximum με within that bearing. This observation maximum με within these bearings.
is attributed to two different deformation mechanisms. The first is the Comparing με values in Fig. 13 indicates that increasing the shim
roll-off, where shims flex close to their ends. The FEA shows that in thickness from 2 to 3 mm decreases the peak values of με by up to 58, 14,
bearings experiencing large roll-off (e.g., B2TS2 and B3TS2 in Fig. 12), and 27% for the bearings without imperfections with B1, B2, and B3
the shims near the bearing mid-height experience the most extreme geometries, respectively. Also, a further increase from 3 to 4 mm de­
curvatures, and thus experience the most extreme plasticity ratios. The creases the peak values of με by up to 14, 11, and 34% for the bearings.
second is lateral bulging of a rubber layer under compression. This Finally, although the με peak values within the bearings that experience
mechanism causes flexing of a shim close to its ends that arises when its large roll-off (e.g., bearings B2TS2 and B3TS2) are multiple times higher
adjacent rubber layers have different thicknesses, and it is mainly than those within the bearings with thick shims or limited roll-off (e.g.,
observed for the top and bottom shims because of the thickness differ­ bearing B1TS4), they are all much smaller than the ductility of mild steel
ence of their adjacent rubber layer and cover rubber layer. Also, when a at fracture.
bearing is sheared, a bending moment distribution develops along the Fig. 14 shows the sensitivity of damage factors ΩE and με to the shim
bearing’s height which has maximum values at top and bottom, and thickness for all the examined bearings subjected to the five-cycle
thus, the bearing experiences additional pressures on the confined loading protocol. It also provides comparisons of με and ΩE values
rubber close to the ends of the shims [17]. The roll-off amount in among bearings with and without imperfections. The following

µ µ µ
Fig. 13. Factor με at the end of the loading protocol for bearings without imperfections for bearing geometries (a) B1; (b) B2; (c) B3.

11
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

Fig. 14. Factors με and ΩE of the examined bearings with geometries (a) B1, (b) B2, and (c) B3.

observations can be drawn: (i) As the thickness of shims increases, both thicknesses, when subjected to seismic-level demands. The examined
με and ΩE decrease considerably in almost all the bearings. For example, imperfections were rotationally misaligned shims and bent shims and
an increase in shim thickness from 2 to 3 mm will decrease με and ΩE by were considered in bearings with different plan dimensions, number and
up to 73% and 92%, respectively, and a further increase from 3 to 4 mm thickness of shims, and shape factor. Abaqus was used to develop and
will decrease με and ΩE by up to 34% and 68%. (ii) Imperfections in­ analyze the 3D finite element models of the bearings. The models were
crease both με and ΩE for most of the bearings. However, for bearings subjected to axial loads equivalent to an average pressure of 5.5 MPa,
with bent shims (as an imperfection), the values of these two damage followed by five cycles of lateral displacements with increasing ampli­
factors are in most cases either very close to or even less than the cor­ tude up to 200% equivalent shear strain (ESS). The FEA results were
responding values in the corresponding bearings without imperfection. used to assess the deformation patterns and hysteretic responses of the
For bearings with B1 geometry, the examined imperfections increase με bearings and quantify the potential damage in the shims in terms of von
and ΩE by up to 72% and 564%. However, the 564% increase corre­ Mises stress, maximum principal plastic strain, and plastic energy
sponds to the increase in ΩE from a very small value for bearing B1TS3 to dissipation. The main conclusions of the study are:
still the small value of 0.03 for bearing B1TS3θ3. The examined im­
perfections also increase με and ΩE by up to 24% and 12% for bearings • Thicker shims in unbonded bearings promote a deformation pattern
with B2 geometry and by up to 25% and 63% for bearings with B3 ge­ dominated by simple shear, reducing the amount of roll-off at higher
ometry. Although the examined imperfections cause the maximum shear strains. Increasing the shim thickness can prevent the potential
percentage increase in damage in bearings with B1 geometry, the plastic instability of unbonded bearings with low second shape factor (e.g.,
damage is relatively small in these bearings. For instance, ΩE in bearings second shape factor of 3) under large displacements. In this regard,
with B1 geometry attains a value of up to 0.1, while it attains a value of slender unbonded bearings (e.g., ones with second shape factor of
up to 0.48 and 0.41 in bearings with B2 and B3 geometry, respectively. lower than 3) with thin shims (e.g., 2-mm-thick) should be avoided.
The 0.48 value corresponds to bearing B2TS2θ25 and denotes that 48% • For a given shim thickness, the amount of roll-off can be increased or
of the external work is dissipated in its shims. decreased when imperfections are present in the unbonded bearing,
Although the overall plastic damage increases when imperfections depending on the imperfection type and the shearing direction; most
are present or when thinner shims are used, and the bearing experiences notably, rotationally misaligned shims can increase the roll-off and
large roll-off, all the examined bearings exhibit fairly stable hysteretic expedite the potential rollover of the bearing. Therefore, rotationally
loops under the five-cycle loading protocol of this study (as shown in misaligned shims should be avoided in slender unbonded bearings so
Fig. 9(b)). Also, in all examined bearings, the induced peak principal as to minimize the risks of rollover and instability.
plastic strains are still far below the fracture limit of mild steel, and the • The effective lateral stiffness of unbonded bearings can be consid­
vertical deflection of all examined bearings is negligibly affected by the erably different from their nominal lateral stiffness, depending on
induced plastic damage under the five-cycle loading protocol. Thus, the the target ESS and the geometry of the bearing. Therefore, when
increased plastic damage during a strong seismic event in bearings with equivalent linear models of unbonded bearings are used for seismic
or without imperfections does not seem to cause an immediate failure design and evaluation, their effective lateral stiffness and effective
(such as shim fracturing and loss of axial load capacity), provided that damping should depend not only on the material and geometric
the plastic damage and fatigue effects in shims are small prior to the properties of rubber layers but also on the target ESS and the geo­
seismic event. metric properties of the bearings, such as shim thickness and second
shape factor.
9. Conclusions • The examined imperfections had a minor effect on the damping of
bearings but increased their effective stiffness and horizontal force
This paper investigated the behavior of unbonded elastomeric bridge by up to 18 and 40%, respectively, at ESS beyond the service level
bearings with and without imperfections and with different steel shim and before sliding. Such increases in the effective stiffness and lateral

12
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

force occur in short bearings with imperfections and slender bearings isolation. Earthq Eng Struct Dyn 2013;42:1375–94. https://doi.org/10.1002/
eqe.2277.
with imperfections, respectively.
[5] Li J, Peng T, Xu Y. Damage investigation of girder bridges under the Wenchuan
• As the loading demands on bearings increased, increase in plastic earthquake and corresponding seismic design recommendations. Earthq Eng Eng
damage in the steel shims took the form of spreading of the yielded Vib 2008;7:337–44. https://doi.org/10.1007/s11803-008-1005-6.
area more than an increase in intensity. The ductility in the shims of [6] Han Q, Du X, Liu J, Li Z, Li L, Zhao J. Seismic damage of highway bridges during
the 2008 Wenchuan earthquake. Earthq Eng Eng Vib 2009;8:263–73. https://doi.
all examined bearings was much smaller than the ductility of mild org/10.1007/s11803-009-8162-0.
steel at fracture. The yielded shims can effectively constrain the [7] Liu K-Y, Chang KC, Wu BX. The role of rubber bearings of the bridges during the
lateral bulging of rubber layers if they experience widespread plas­ 1999 Taiwan Chi-Chi earthquake. In: In: 8th US National Conference on
Earthquake Engineering; 2006, 2006,. p. 7307–16.
ticity but with relatively low ductility. Therefore, in bearings without [8] Kawashima K, Unjoh S, Hoshikuma J-I, Kosa K. Damage of bridges due to the 2010
initial plastic damage or with limited amounts of it, the induced Maule, Chile, earthquake. J Earthq Eng 2011;15:1036–68. https://doi.org/
plastic damage during a strong seismic event does not seem to lead to 10.1080/13632469.2011.575531.
[9] Mori A, Moss PJ, Cooke N, Carr AJ. The behavior of bearings used for seismic
an immediate failure. isolation under shear and axial load. Earthquake Spectra; 1999; 15:199-224.
https://doi.org/10.1193%2F1.1586038.
Future works can focus on the performance of fully bonded elasto­ [10] Konstantinidis D, Kelly JM, Makris N. Experimental investigation on the seismic
response of bridge bearings. Report No. EERC-2008/02. Earthquake Engineering
meric bearings with imperfections under seismic-level demands and also Research Center. University of California, Berkeley; 2008.
on the system-level behavior of bridges featuring bonded or unbonded [11] Wu G, Wang K, Lu G, Zhang P. An experimental investigation of unbonded
elastomeric bearings with imperfections. Of particular interests would laminated elastomeric bearings and the seismic evaluations of highway bridges
with tested bearing components. Shock Vib 2018. https://doi.org/10.1155/2018/
be to investigate the potential effects of imperfections on the tensile
8439321.
stresses induced in bonded bearings arising due to the combined actions [12] Steelman JS, Fahnestock LA, Filipov ET, LaFave JM, Hajjar JF, Foutch DA. Shear
of vertical force and shear deformation and on the vulnerability of other and friction response of nonseismic laminated elastomeric bridge bearings subject
bridge components under seismic actions. to seismic demands. J Bridge Eng 2012;18:612–23. https://doi.org/10.1061/
(ASCE)BE.1943-5592.0000406.
Data availability statement [13] Li Y, Li Q, Wu Q. Ultimate shear performance and friction sliding response of
All data, models, and code generated or used during the study appear laminated elastomeric bridge bearings. J Vibroengineering 2017;19:6301–12.
in the submitted article. https://doi.org/10.21595/jve.2017.18321.
[14] Xiang N, Li J. Experimental and numerical study on seismic sliding mechanism of
laminated-rubber bearings. Eng Struct 2017;141:159–74. https://doi.org/
10.1016/j.engstruct.2017.03.032.
CRediT authorship contribution statement [15] Luo J, Fahnestock LA, Kozak DL, LaFave JM. Seismic analysis incorporating
detailed structure–abutment–foundation interaction for quasi-isolated highway
Saleh Ahmadi Soleimani: Conceptualization, Methodology, Soft­ bridges. Struct Infrastruct Eng 2017;13:581–603. https://doi.org/10.1080/
15732479.2016.1177094.
ware, Validation, Formal analysis, Investigation, Writing – original [16] Xiang N, Alam MS, Li J. Shake table studies of a highway bridge model by allowing
draft, Visualization. Dimitrios Konstantinidis: Methodology, Writing – the sliding of laminated-rubber bearings with and without restraining devices. Eng
review & editing, Supervision, Project administration, Funding acqui­ Struct 2018;171:583–601. https://doi.org/10.1016/j.engstruct.2018.05.121.
[17] Kalfas KN, Mitoulis SA, Konstantinidis D. Influence of Steel Reinforcement on the
sition. Georgios P. Balomenos: Methodology, Writing – review & Performance of Elastomeric Bearings. J Struct Eng 2020;146:04020195. https://
editing, Supervision, Project administration, Funding acquisition. doi.org/10.1061/(ASCE)ST.1943-541X.0002710.
[18] Zhang R, Li A. Finite element study of effect of parameters on stress distribution of
steel shim in elastomeric bearings. Adv Struct Eng 2019;22:2124–35. https://doi.
Declaration of Competing Interest org/10.1177/1369433219834748.
[19] Yura J, Kumar A, Yakut A, Topkaya C, Becker E, Collingwood J. Elastomeric bridge
bearings: Recommended test methods. Report No. NCHRP 449. National
The authors declare that they have no known competing financial Cooperative Highway Research Program (NCHRP). Washington, DC:
interests or personal relationships that could have appeared to influence Transportation Research Board; 2001.
[20] Ahmadi Soleimani S, Konstantinidis D, Balomenos GP. Vision-Based Quality
the work reported in this paper. Control Testing of Elastomeric Bridge Bearings. In: Structures Congress.Reston, VA:
ASCE; 2020: 651-664.
Data availability [21] Ahmadi Soleimani S, Konstantinidis D, Balomenos GP. Nondestructive assessment
of elastomeric bridge bearings using 3D digital image correlation. J Struct Eng
2022;148:04021233. https://doi.org/10.1061/(ASCE)ST.1943-541X.0003200.
All data, models, and code generated or used during the study appear [22] Kumar A. Performance related parameters of elastomeric bearings [Ph.D.
in the submitted article. Dissertation]: Dept. of Civil, Architectural and Environmental Engineering. The
Univ of Texas at Austin 2001.
[23] Belivanis KV. Expanding the use of elastomeric bearings for higher demand
Acknowledgments applications [Ph.D. thesis]: Dept. of Civil, Architectural and Environmental
Engineering, The Univ. of Texas at Austin; 2017.
[24] Ahmadi Soleimani S, Konstantinidis D, Balomenos GP. Effect of Manufacturing
This research was partially supported by the Ministry of Trans­ Imperfections on the Service-Level Performance of Elastomeric Bridge Bearings.
portation of Ontario (MTO) through the Highway Infrastructure Inno­ J Struct Eng 2022;148:04022088.
vation Funding Program. The authors are grateful to Kris Mermigas [25] EN 1337-3. Structural Bearings-Part 3: Elastomeric Bearings. Brussels, Belgium;
2005.
(Head Engineer, Structures Office, MTO) for his committed support
[26] OPSS 1202. Material specification for bearings - elastomeric plain and steel
during this work. Any opinions, findings, or recommendations in this laminated. Ministry of Transportation of Ontario, Ontario; 2017.
paper are those of the authors and do not necessarily reflect the views of [27] IDOT. IDOT Bridge Manual. Illinois Department of Transportation, IL: Springfield;
2012.
the sponsor.
[28] NYSDOT. NYSDOT Bridge Manual. Section 12: Bridge Bearings. New York State
Department of Transportation; 2019.
References [29] Systemes D. ABAQUS 2019 Theory manual. Providence, Rhode Island: Dassault
Systemes Simulia Corp; 2019.
[30] Khaloo A, Maghsoudi-Barmi A, Moeini ME. Numerical parametric investigation of
[1] Xiang N, Goto Y, Alam MS, Li J. Effect of bonding or unbonding on seismic
hysteretic behavior of steel-reinforced elastomeric bearings under large shear
behavior of bridge elastomeric bearings: lessons learned from past earthquakes in
deformation. Structures 2020;26:456–70. https://doi.org/10.1016/j.
China and Japan and inspirations for future design. Adv Bridge Eng 2021;2:1–17.
istruc.2020.04.029.
https://doi.org/10.1186/s43251-021-00036-9.
[31] Mordini A, Strauss A. An innovative earthquake isolation system using fibre
[2] AASHTO. AASHTO LRFD bridge design specifications. Washington, DC: AASHTO;
reinforced rubber bearings. Eng Struct 2008;30:2739–51. https://doi.org/
2017.
10.1016/j.engstruct.2008.03.010.
[3] CSA. S6-19 Canadian highway bridge design code. Section 11: Joints and Bearings.
[32] Sun C. Material study of the steel reinforced elastomeric bridge bearings [Ph.D.
Toronto, Ontario; 2019.
thesis]: Dept. of Civil, Architectural and Environmental Engineering, The Univ. of
[4] Filipov ET, Revell JR, Fahnestock LA, LaFave JM, Hajjar JF, Foutch DA, et al.
Texas at Austin; 2015.
Seismic performance of highway bridges with fusing bearing components for quasi-

13
S. Ahmadi Soleimani et al. Engineering Structures 291 (2023) 116179

[33] Yeoh OH. Some forms of the strain energy function for rubber. Rubber Chem [41] Schrage I. Anchoring of bearings by friction. In: Joint sealing and bearing systems
Technol 1993;66:754–71. https://doi.org/10.5254/1.3538343. for concrete structures, world congress on joints and bearings.American Concrete
[34] Moghadam SR, Konstantinidis D. Finite element study of the effect of support Institute Niagara Falls, NY, USA; 1981.
rotation on the horizontal behavior of elastomeric bearings. Compos Struct 2017; [42] Magliulo G, Capozzi V, Fabbrocino G, Manfredi G. Neoprene–concrete friction
163:474–90. https://doi.org/10.1016/j.compstruct.2016.12.013. relationships for seismic assessment of existing precast buildings. Eng Struct 2011;
[35] Bergström JS, Boyce MC. Constitutive modeling of the large strain time-dependent 33:532–8. https://doi.org/10.1016/j.engstruct.2010.11.011.
behavior of elastomers. J Mech Phys Solids 1998;46:931–54. https://doi.org/ [43] Han L. Elastomeric bearings in high demand applications: Numerical study and
10.1016/S0022-5096(97)00075-6. experimental Validation [Ph.D. thesis]: Dept. of Civil, Architectural and
[36] Belivanis KV, Han L, Sun C, Helwig T, Engelhardt M, Tassoulas J, et al. Extending Environmental Engineering, The Univ. of Texas at Austin; 2016.
use of steel-laminated elastomeric bearings to higher demand applications. Center [44] Huang W, Xu X, Wang K, Liu W. Numerical Simulation of Steel-Laminated Bearing
for Transportation Research, The Univ. of Texas at Austin; 2015. Report No. Considering Friction Slipping. Int J Eng Technol 2018:10. https://doi.org/
0–6785-1. 10.7763/IJET.2018.V10.1052.
[37] El-Reedy M. Onshore structural design calculations: Power plant and energy [45] Caltrans. Bridge Memo to Designers. Section 7: Bridge Bearings. California
processing facilities. Butterworth-Heinemann; 2016. Department of Transportation, Sacramento, California; 1994.
[38] Moghadam SR, Konstantinidis D. Simple mechanical models for the horizontal [46] Kelly JM, Konstantinidis D. Mechanics of rubber bearings for seismic and vibration
behavior of elastomeric bearings including the effect of support rotation. Eng isolation. Chichester, UK: John Wiley & Sons; 2011.
Struct 2017;150:996–1012. https://doi.org/10.1016/j.engstruct.2017.07.079. [47] Konstantinidis D, Kelly JM. Two low-cost seismic isolation systems. In: 15th world
[39] Rastgoo Moghadam S, Konstantinidis D. Experimental and Analytical Studies on conference on earthquake engineering. Lisbon, Portugal.2012: 24-28.
the Horizontal Behavior of Elastomeric Bearings under Support Rotation. J Struct [48] Camara A, Cristantielli R, Astiz MA, Málaga-Chuquitaype C. Design of hysteretic
Eng 2021;147:04021024. https://doi.org/10.1061/(ASCE)ST.1943- dampers with optimal ductility for the transverse seismic control of cable-stayed
541X.0002962. bridges. Earthq Eng Struct Dyn 2017;46:1811–33. https://doi.org/10.1002/
[40] Muscarella JV, Yura J. An experimental study of elastomeric bridge bearings with eqe.2884.
design recommendations. Center for Transportation Research, The University of
Texas at Austin; 1995 [Ph.D. thesis]:.

14