Upload
daniele-di-luca
View
213
Download
0
Embed Size (px)
Citation preview
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
1/10
Bonded versus unbonded strip fiber reinforced elastomeric isolators: Finite
element analysis
Hamid Toopchi-Nezhad ⇑, Michael J. Tait, Robert G. Drysdale
Department of Civil Engineering, McMaster University, 1280 Main St. W., Hamilton, ON, Canada L8S 4L7
a r t i c l e i n f o
Article history:Available online 10 August 2010
Keywords:
Strip fiber reinforced elastomeric isolators
Laminated rubber bearings
Finite element analysis
Seismic mitigation
Base isolation
a b s t r a c t
This paper presents a finite element (FE) model for the analysis of strip fiber reinforced elastomeric iso-lators (FREIs) that are subjected to any given combination of static vertical and lateral loads. The model is
able to simulate both bonded and unbonded boundary conditions at the top and bottom contact surfaces
of the isolator. Compared to bonded (B)-FREIs, the FE-analysis of stable unbonded (SU)-FREIs presents
additional analysis challenges. SU-FREI refers to unbonded FREIs that exhibit stable rollover deformation
under lateral loads. Additional analysis challenges are attributed to changes in the boundary conditions
of SU-FREI as a result of rollover type deformation. To address these challenges, the utilized FE-mesh is
updated during analysis consistent with the deformed geometry of the isolator. Using the proposed FE-
model, the lateral responses of a B-FREI and a SU-FREI were evaluated. Both isolators had the same mate-
rial and geometrical properties and were subjected to identical constant vertical loading. Comparing the
lateral responses, it was found that the SU-FREI was considerably more efficient than the B-FREI as a seis-
mic isolator. In addition, the in-service stress demands on the SU-FREI were found to be significantly
lower than the B-FREI.
2010 Elsevier Ltd. All rights reserved.
1. Introduction
Fiber reinforced elastomeric isolators (FREIs) comprise alternat-
ing bonded layers of elastomer and fiber reinforcement. The elasto-
meric layers provide lateral flexibility and the primary role of fiber
reinforcement is to constrain lateral bulging of the elastomer when
the isolator is subjected to vertical compressive loads. In terms of
connection to the isolator’s top and bottom contact supports, FREIs
can be classified as either bonded or unbonded. In a bonded (B)-
FREI, two thick steel mounting plates are bonded to the outer rub-
ber layers at the top and bottom of the isolator. During installation,
the top and bottom mounting plates are bolted to the superstruc-
ture and substructure, respectively. In an unbonded FREI, the isola-
tor is placed between the substructure and superstructure without
any bonding or fastening at its contact surfaces. During an earth-
quake, the shear loads at the contact surfaces of an unbonded
isolator are transferred through friction. In a stable unbonded
(SU)-FREI, the isolator’s geometry can be selected such that it
maintains lateral stability at extreme lateral displacements.
Over the past decade, a number of experimental investigations
on individual bonded and unbonded FREIs have been conducted
[1–10]. The common outcome of these studies is that the investi-
gated FREIs have had suitable mechanical properties to be used
as seismic isolators. Recently, a shake table study confirmed the
seismic mitigation efficiency of SU-FREI systems [11]. These stud-
ies suggest that FREIs are a viable alternative for conventional steel
reinforced elastomeric isolators (SREIs).
Current seismic codes mandate that the final mechanical design
properties of isolators should be evaluated through experiment.
This mandate applies to all isolator types. Accordingly, the main
objective of preliminary design is to provide the required informa-
tion for the fabrication of the prototype isolators. For preliminary
design, both the vertical and horizontal stiffness and damping ratio
values of an isolator should be reasonably estimated. Knowing the
vertical stiffness, one can verify that the vertical frequency of the
isolator is sufficiently greater than the target lateral base isolation
frequency to eliminate any significant contribution of rocking
modes in the total response of the base isolated structure. The esti-
mation of horizontal stiffness is critical to ensure that the designed
isolator satisfies the target base isolation frequency. Additionally,
knowing the horizontal stiffness and damping ratio, one can assess
whether the demand lateral displacements lie within the permissi-
ble displacement range of the isolator.
Since the deformation characteristic of the fiber reinforcement
is different than the steel reinforcing plates, the closed-form equa-
tions, available for the stiffness solution of conventional SREIs, are
not generally applicable to FREIs. Solutions to the vertical compres-
sion and bending stiffness of strip, rectangular, and circular FREIs
0263-8223/$ - see front matter 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.compstruct.2010.07.009
⇑ Corresponding author. Tel.: +1 905 5259140/24860; fax: +1 905 5299688.
E-mail addresses: [email protected] (H. Toopchi-Nezhad), taitm@mcmas-
ter.ca (M.J. Tait), [email protected] (R.G. Drysdale).
Composite Structures 93 (2011) 850–859
Contents lists available at ScienceDirect
Composite Structures
j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c o m p s t r u c t
http://dx.doi.org/10.1016/j.compstruct.2010.07.009mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.compstruct.2010.07.009http://www.sciencedirect.com/science/journal/02638223http://www.elsevier.com/locate/compstructhttp://www.elsevier.com/locate/compstructhttp://www.sciencedirect.com/science/journal/02638223http://dx.doi.org/10.1016/j.compstruct.2010.07.009mailto:[email protected]:[email protected]:[email protected]:[email protected]://dx.doi.org/10.1016/j.compstruct.2010.07.009
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
2/10
are available [12–15]. Also, both the horizontal stiffness and buck-
ling load of strip B-FREIs have been analytically investigated
[16,17]. Given the complex deformation characteristics of FREIs,
the analytical solutions for B-FREIs typically involve sophisticated
equations. Furthermore, the literature lacks any closed-form solu-
tion for the lateral response of SU-FREIs.
Finite element (FE) is a powerful design tool that can be em-
ployed in the preliminary design of FREIs. Analysis of elastomericcomponents requires a robust FE-model that is capable of address-
ing large deformations and accounts for the nearly incompressible
behavior of the elastomer. Although FE-analysis of B-FREIs is rela-
tively straight forward, there are challenges in the FE-analysis of
SU-FREIs due to the rollover deformation of these isolators under
lateral loads. A limited number of studies report on FE-analysis
of FREIs. These studies are either limited to the vertical compres-
sion response [18], or focus on the vertical and lateral responses
of B-FREIs only [9].
The focus of this study is on the lateral response evaluation of
both B- and SU-FREIs through FE-analysis. A FE-model, capable of
simulating both conventional bonded and unbonded boundary con-
ditions at the top and bottom contact surfaces of strip FREIs, is pre-
sented. The term strip is selected as the analysis is carried out for
the unit out-of-plane length of the isolator. The primary goal of the
presented FE-analysis is to evaluate the horizontal stiffness of the
isolators at different lateral displacements. The analysis results
are also used to assess the stress (or strain) state in the isolator’s
components. In addition to developing the FE-model, an extensive
comparative study on the lateral response of a B-FREI and a corre-
sponding SU-FREI is presented.
2. Fiber reinforced isolator
Fig. 1 shows a sketch of the cross section of the FREI investi-
gated in this study. The isolator comprises 11 layers of fiber rein-
forcement layers interleaved and bonded between 12 layers of
rubber. The physical dimensions and material properties of the iso-
lator are shown in Table 1. The isolator carries a constant vertical
compression load of P = 112 N.
3. Analytical evaluation of the horizontal stiffness for strip
FREIs
3.1. B-FREIs
The main objective of the analytical solution is to evaluate the
horizontal stiffness of a B-FREI based on the properties given in Ta-
ble 1. The analytical solutions of conventional SREIs are not directly
applicable to B-FREIs due to the different mechanical characteris-
tics of fiber reinforcement as compared to the steel reinforcing
plates. In a laterally deformed SREI, the steel plates remain planar
and are nearly rigid in both tension and flexure. On the contrary,despite their very large in-plane tensile stiffness, the fiber rein-
forcement layers show some extensional flexibility with no bend-
ing rigidity. Accordingly, in a laterally deformed B-FREI, the fiber
reinforcement layers undergo warping deformation at their ends
(see Fig. 2a) as a result of the internal moment and shear that de-
velop in the isolator.
Horizontal stiffness of a strip B-FREI can be estimated using the
closed-form elastic solution available for the horizontal stiffness
evaluation of a homogeneous short vertical beam that is subjected
to axial compression and lateral shear forces. The theory extends
the Haringx theory [19] on the stability of rubber rods by account-
ing for the shear and warping deformations of the cross section. Toreplicate the boundary conditions of a B-FREI, the lower end of the
beam is assumed to be fixed against any displacement, rotation
and warping, and the upper end is constrained against rotation
and warping but allowed to move in both lateral and axial direc-
tions. Based on these assumptions, the horizontal stiffness of the
beam (or the B-FREI given in Fig. 1) can be calculated by [16];
K H ¼ Ge A
t r
P
2P ð1þP Þþb2P ðb1 þb2 Þ
ffiffiffiffiffiffiffiffiffiffiffi2qb1
p tan
ffiffiffiffib18q
q þ 2
P bP ðb1 þb2 Þ
ffiffiffiffiffiffiffiffiffiffiffi2qb2
p tan
ffiffiffiffib28q
q 1
when ð1 þ P Þkc kb P 0 ð1Þ
K H ¼ Ge A
t r P
2P ð1þP Þþb2
P ðb1 þb2 Þ ffiffiffiffiffiffiffiffiffiffiffi2
qb1p tan ffiffiffiffib1
8qq
2P b
P ðb1 þb2 Þ ffiffiffiffiffiffiffiffiffiffiffi2
qb2p tan ffiffiffiffib2
8qq
1
when ð1 þ P Þkc kb < 0 ð2Þ
where A is the plan cross section area of the isolator, P ¼ P =Ge A rep-
resents a dimensionless compression force, and coefficients b1 and
b2 are defined as follows;
b1 ¼ ½P ð1 þ P Þ þ kb kc
þ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffi½P ð1 þ P Þ þ kb kc
2þ 4P ½ð1 þ P Þkc kb
q ð3Þ
b2 ¼ ½P ð1 þ P Þ þ kb kc
þ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffiffiffiffiffiffiffi½P ð1 þ P Þ þ kb kc
2 þ 4P ½ð1 þ P Þkc kb
q ð4Þ
Parameters kb and kc , which are attributed to the cross sectionwarping of a laterally deformed B-FREI, can be estimated using
[17];
K b 20 1 2
21 ðaaÞ2
h i 1 1
210 ðaaÞ2 þ P 3
10 þ 8
525 ðaaÞ2
h in o21 2
77 ðaaÞ2 þ 3675
64kS 3 ða=t f Þ3
ð5Þ
kc
3 1 221
ðaaÞ2h i
23 463
ðaaÞ2 þ P 9 þ 26105
ðaaÞ2 þ 245S 12kða=t f Þ
3
1 277
ðaaÞ2 þ 367564kS 3 ða=t f Þ
3
ð6Þ
In Eqs. (5) and (6), aa, the extensional rigidity of the fiber reinforce-ment is given by
t e
t f
2a
R1
R12
R6
Fig. 1. Cross section of the strip FREI used in the finite element analysis.
Table 1
Material and geometrical properties of the strip FREI.
Material properties
Ge = Shear modulus of the elastomer (rubber) = 0.4 MPa
t f = Poisson’s ratio of the fiber reinforcement = 0.2E f = Young’s modulus of the fiber reinforcement = 137 GPa
Geometrical properties
2a = Width of the isolator = 70 mm
t e = Thickness of a single elastomer layer = 1.587 mmt r = Total thickness of the rubber layers = 19.044 mm
t f = Thickness of the fiber reinforcement = 0.55 mm
h = Height of the isolator = 25.094 mm
S = Shape factor (=a/t e) = 22
H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859 851
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
3/10
aa ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi12kS
a
t f
s ð7Þ
where
k ¼ð1 t2 f ÞGe
E f ð8Þ
In Eqs. (1) and (2), q ¼ ðEI Þeff =Ge At 2r indicates the ratio of the effec-
tive bending to shear rigidity of the isolator. For a homogeneous
beam (EI )eff = EI , however, for a laminated strip rubber isolator, itis defined as follows [17];
ðEI Þeff ¼ a3GeS
2 24
ðaaÞ4 1 þ
1
3ðaaÞ2
aatanhðaaÞ
ð9Þ
3.2. SU-FREIs
As shown in Fig. 2b, when a SU-FREI is laterally displaced, its
upper and lower faces roll off the contact supports and the isolator
exhibits rollover deformation. The rollover deformation results in a
non-linear load–displacement relationship. Therefore, the horizon-
tal stiffness varies with isolator lateral displacement. As such, the
analytical stiffness solution for this type of isolators is more com-
plex and Eqs. (1) and (2) are no longer applicable. Currently, noclosed-form solution has been reported in the literature to esti-
mate the horizontal stiffness of SU-FREIs.
3.3. Limitations of closed-form solutions
Any closed-form stiffness solution only serves for the prelimin-
ary design of FREIs. There are a number of features which are not
addressed by the simplified equations used in the closed-form
solutions. Among these features, one can refer to the dependency
of the rubber material properties on the amplitude of the shear
strain, which is ignored in closed-form linear solutions. Addition-
ally, due to the static nature of the solution, no information can
be obtained on the influence of rate and history of the lateral dis-
placement on the lateral response of the isolator. Analytical evalu-
ation of the effective damping of the isolator, which is an important
design parameter, is not possible. Therefore, the final design prop-
erties of any isolation device should be evaluated through experi-
mental testing on prototype samples of the isolator [20].
4. Finite element modeling
4.1. Objectives
In the FE-analysis presented in this paper, the strip isolator, un-
der a constant axial compression load, is subjected to a static lat-
eral load acting at its top support. Due to the non-linear nature
of the analysis, the lateral load is applied incrementally in multiple
steps from zero to its target value. The target lateral load is selectedsuch that a 200% t r lateral deformation in the isolator is achieved.
The primary objective of the FE-analysis is to evaluate the lateral
load–displacement relationship of a B-FREI or a SU-FREI with given
geometric and material properties as listed in Table 1. From the lat-
eral load–displacement curve, the effective (secant) horizontal
stiffness corresponding to each level of lateral displacement can
be assessed to verify whether the isolator meets the seismic isola-
tion design requirements. The second objective of the FE-analysis
is to evaluate the stress state within the isolator. Estimating the
maximum stress demand in the rubber and fiber reinforcement
layers and evaluating the required bond strength between the
alternating layers are needed for manufacturing requirements.
4.2. General description of the model
Modeling and analysis of the isolators were carried out using
MSC. Marc [21], a commercially available general purpose FE-soft-
ware package. In the model presented, the rubber layers were dis-
cretized using quadrilateral plane-strain solid elements, and 2D-
truss elements were selected to represent the fiber reinforcement.
The truss elements in the model were characterized with zero
physical thickness. A previous experimental study [5] revealed that
the ratio of width to the total thickness of isolator (i.e., the aspect
ratio), plays a crucial role in the lateral response of SU-FREIs. To ac-
count for the real physical thickness of the fiber reinforcement in
the model, and to accurately simulate the isolator aspect ratio, ri-gid gaps were defined between the top and bottom faces of the fi-
ber reinforcement and the adjacent rubber layers.
Two horizontal rigid wires were defined at the top and bottom
of the isolator to represent the contact supports. All degrees of
freedom of the bottom support and the rotational degree of free-
dom of the top support were constrained. Since both the vertical
load and lateral load were applied to the top support, it was al-
lowed to move in both the vertical and horizontal directions.
Given the available contact models in the utilized software, the
‘‘touching” contact model was selected for the SU-FREI model. In
this contact model, the nodal points of the exterior rubber ele-
ments at the contact surface were constrained in the direction nor-
mal to the contact support. No tension stress was resisted at the
contact surfaces meaning that when the compression contactstress approached zero, the nodal points were allowed to detach
from the contact support. Shear force between the contact sup-
ports and the isolator was transferred through a Coulomb friction
mechanism. To prevent slip at the contact support, the coefficient
of friction at the contact surfaces of the isolator and the rigid sup-
ports was selected to be one in the FE-model. Based on a previous
experimental study [6,7] that was conducted on square SU-FREIs
having the same characteristics as listed in Table 1, slip at the con-
tact surfaces of the isolators was found to be negligible.
The FE-model of the B-FREI was the same as the SU-FREI model,
however, the rigid supports were connected to the isolator using
the ‘‘glue” contact model in MSC. Marc [21]. This contact model
prevents any detachment or slip between the isolator and the rigid
supports by constraining the nodal points in the directions normaland tangential to the contact support.
(a) B-FREI (b) SU-FREI
Warping at
the ends of fiber
reinforcement layers
Fig. 2. Sketch of laterally deformed FREIs with different boundary conditions.
852 H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
4/10
4.3. Material models used for the rubber and the fiber reinforcement
Since the FE-analysis developed in this study serves as a tool for
the preliminary design of FREIs, it relies on a minimum set of re-
quired material data properties to minimize the cost of the preli-
minary design. As such, the presented FE-analysis employs a
Neo-Hookean model as the simplest available hyperelastic material
model for elastomeric materials. In this model, the rubber is trea-ted as an elastic isotropic material with a strain energy function
that is characterized by shear (Ge) and bulk (ke) moduli of the rub-
ber. The nominal shear modulus at 100% elongation can be pro-
vided, as routine standard product information, by many rubber
suppliers. If the rubber in an isolator performs as a nearly incom-
pressible material with a Poisson’s ratio of approximately 0.5, an
appropriate value of the bulk modulus can be estimated using
the following well-known elasticity relationship.
keGe
¼ 2ð1 þ teÞ3ð1 2teÞ
ð10Þ
The bulk modulus of rubber in the presented FE-analysis is ta-
ken as ke = 1900 MPa. With this value and the shear modulus of
Ge = 0.4 MPa (see Table 1), the use of Eq. (10) results in a Poisson’s
ratio of 0.4999 which is sufficiently close to 0.5 to simulate the
incompressible behavior of the rubber in the FE-analysis.
The fiber reinforcement in the presented FE-model is treated as
a linear elastic isotropic material with material properties given in
Table 1.
4.4. Rubber incompressibility
The incompressibility of rubber can result in serious numerical
errors if standard isoparametric elements, in which the stress state
is determined from the strain state, are used in the FE-analysis.
Since the volumetric strain in the finite elements is nearly zero
in a nearly incompressible material, determining the mean stress
or pressure based on the volumetric part of the strain is challeng-ing [22]. Additionally, standard isoparametric elements in an
incompressible material show a pathological behavior entitled vol-
umetric mesh-locking that is attributed to the inaccurate perfor-
mance of an element due to an over-constrained condition and
insufficient active degrees of freedom [23]. To avoid these prob-
lems, modern FE-techniques utilize so called mixed-methods for
incompressible materials [22]. In mixed-methods both the strains
and stresses are assumed as the unknowns. In the FE-model pro-
posed in this paper, a commonly used mixed-method developed
by Hermann [24] has been used for the rubber elements.
4.5. Large deformation and element distortion
Elastomeric isolators may undergo very large deformations un-der extreme lateral loads. Therefore, in the FE-analysis of these iso-
lators, the use of conventional total Lagrangian formulation is not
recommended. As such, in the FE-analysis presented in this study,
an updated Lagrangian approach has been used. In this approach,
the orientation of the local coordinate system is updated during
the analysis based on the deformed configuration of the element.
In a laterally deformed SU-FREI, the rollover deformation, in
addition to the large deformation of the isolator, adds to the com-
plexity of the FE-analysis. When rollover deformation occurs, one
end of the outer rubber-layers rolls off the contact support while
the opposite end is pressed against the contact support (see
Fig. 2b). Accordingly, in the FE-analysis, the rubber elements at
the pressed ends may experience severe distortion such that they
no longer accurately discretize the problem. As the lateral load isincreased, the intermediate rubber layers of the isolator may also
experience local excessive distortion. When the FE-software de-
tects any excessive distortion, it automatically terminates the anal-
ysis to preserve the accuracy of the results. To resolve this
termination, prior to excessive element distortion in the original
mesh, the FE-solution can be mapped onto a new mesh that is
adapted with the current deformed geometry of the model. The
new mesh arrangement results in a new FE-problem. The analysis
is then continued by treating the solution from the previous mesh,at the point of mapping, as the initial conditions of the new FE-
problem.
A global remeshing procedure available in MSC. Marc [21] was
used in the presented FE-analysis. The remeshing procedure uti-
lized the updated Lagrangian formulation. The remeshing criterion
was based on monitoring the deviations of the inner angles of the
elements from their undeformed configuration. The default limit of
the software was used as the threshold angle-change. For the top
and bottom rubber layers, the edge length of the quadrilateral ele-
ments was selected to be approximately 0.3 mm in the original
mesh. The element edge length for the intermediate rubber layers
was set to be approximately 0.4 mm due to the reduced distorted
pattern of deformation in these layers as compared to the outer
rubber layers. If required, the updated mesh could be finer in the
regions of high-strain gradients in the rubber layers.
The B-FREI was modeled using the same original FE-mesh that
was employed for the SU-FREI. Since the boundary conditions of
the isolator remained unchanged throughout the analysis, no
remeshing was required. The lower-order finite elements were
used in modeling of both isolators as the performance of these ele-
ments is superior to higher-order elements in simulating large dis-
tortions [25]. For both isolators, the length of the 2D-truss
elements, representing the fiber reinforcement, was selected to
be 0.28 mm.
5. Finite element model validation
5.1. B-FREI
Table 2 contains the horizontal stiffness values of the B-FREI,
which are calculated using closed-formed equations and the FE-
analysis. For conventional SREIs, when the vertical load carried
by the isolator is significantly lower than the isolator’s buckling
load, the horizontal stiffness can be calculated using the simple
formula of K H = Ge A/t r [26]. This simple formula is expected to over-
estimate the horizontal stiffness of a B-FREI due to the flexibility of
the fiber reinforcement layers that are employed in the isolator.
Given the properties cited in Table 1, under a vertical load of
P = 112 N, the control parameter ð1 þ P Þkc kb acquires a positive
value. Therefore, Eq. (1) is used to calculate the horizontal stiffness
of the B-FREI. As can be seen in Table 2, for the B-FREI considered in
this study, the stiffness value that is estimated by the simple for-
mula of Ge A/t r is in close agreement with the value calculated byEq. (1). This close agreement suggests that one can use the simple
formula, with less calculation effort, for the preliminary design of
B-FREIs. However, caution should be used as the error of employ-
ing the simple formula rises with increased flexibility of the fiber
reinforcement in the isolator.
From the FE-analysis, a linear lateral load–displacement rela-
tionship was evaluated for the B-FREI. The slope of this line was
Table 2
Horizontal stiffness (N/mm) of the B-FREI.
Analytical Finite element
Ge A/t r Eq. (1)
1.470 1.455 1.453
H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859 853
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
5/10
calculated as the horizontal stiffness of the isolator. According to
Table 2, excellent agreement is observed between both the finite
element and analytical stiffness solutions. This confirms the accu-
racy of the presented FE-model and its suitability for the prelimin-
ary design of the B-FREIs.
5.2. SU-FREI
As described earlier, both B- and SU-FREI models employed the
same FE-assemblage. However, in the SU-FREI model, no tension is
transmitted between the isolator and its top and bottom contact
supports. Due to the model similarity, validation of the FE-model
for B-FREI provided sufficient confidence that the SU-FREI model
would give satisfactory results. However, for further validation,
the lateral load–displacement curve of the SU-FREI that was ob-
tained from the FE-analysis was compared with the results from
a previous experimental research study on similar SU-FREIs [6,7].
This experimental study was conducted on square SU-FREIs having
the same characteristics as listed in Table 1.
Fig. 3 shows the lateral load–displacement relationships from
both the experimental study and the FE-analysis. It should be
noted that the FE-results in Fig. 3 are obtained by multiplying
the lateral load of the strip isolator by 70, to account for the
70 mm out-of-plane length of the square isolator. In Fig. 3, the lat-
eral deformations of the isolator are described in terms of % t r lat-
eral displacement not percent of shear strain, which is commonly
used to express the level of lateral deformation in conventional
SREIs. The normalized lateral displacement is employed as shear
deformation is not the dominant deformation mode (see Fig. 2b).
As seen in Fig. 3, reasonable agreement was found between the
experimental and FE-results at displacements ranging from 100%
to 200% t r . However, for lower displacements, the accuracy of the
FE-analysis was reduced as the presented FE-model assumed a
constant value for Ge (corresponding to 100% elongation in the rub-
ber) throughout the analysis, and neglected any strain dependency
of the rubber material.
6. Finite element analysis results and discussion
6.1. Lateral load–displacement relationship
Fig. 4 contains the lateral load–displacement relationships for
both the B- and SU-FREIs up to 200% t r lateral displacement. As
shown in this figure, while the lateral response of the B-FREI re-
mains nearly linear, the SU-FREI behaves nonlinearly due to the
rollover deformation. As a result of the rollover deformation, the
horizontal effective secant stiffness decreases with increased lat-
eral displacement. This decrease results in an increase in the base
isolated period of the isolator, increasing its seismic mitigation effi-
ciency. Accordingly, for a given FREI, unbonded application leads to
superior seismic isolation provided that lateral stability of the iso-
lator is maintained.
It has been shown that unbonded FREIs with inappropriate
geometry may exhibit lateral instability [5]. Inthedesign ofan unb-
onded FREI, the isolator must maintain positive incremental load-
resisting capacity throughout the entire range of lateral displace-
ments that are imposed on the isolator. The presented FE-model
can be employed to verify the achievement of stable rollover (SR)
deformation in the preliminary design of an unbonded FREI. As
can beseen in Fig. 4, the unbonded isolator maintained positive tan-
gent stiffness over the range of lateral displacement investigated.
In a base isolated system, the lateral shear load resisted by the
isolators is transmitted to the superstructure. According to Fig. 4,
for any given lateral displacement above 50% t r , the SU-FREI is sub-
jected to a significantly lower shear compared to the B-FREI. As the
lateral displacement is increased, the difference between the lat-
eral (shear) loads, resisted by the B- and SU-FREIs is found to in-
crease (see Fig. 4). At 200% t r displacement, the shear load that is
transmitted to the superstructure by the SU-FRE is approximately40% less than that of the B-FREI. Therefore, with the same physical
dimensions and material properties, the response attenuation of
the SU-FREI is superior to that of the B-FREI.
The secant horizontal stiffness of the SU-FREI at 200% t r dis-
placement (38 mm) is calculated to be 0.85 (N/mm)/mm. To
achieve this horizontal stiffness in the B-FREI, using the simple for-
mula of K H = Ge A/t r , the minimum required total thickness of rub-
ber layers in the isolator would be t r = 33 mm. To assemble such
an isolator with materials given in Table 1, 21 rubber layers and
20 interleaved bonded fiber reinforcement layers are required. This
implies a 75% increase in the volume of utilized materials. Addi-
tionally, the significant increase in the isolator height may force
the designer to increase the isolator’s width to prevent Euler buck-
ling [26], which may occur in the B-FREI under large lateraldisplacements.
An examination of Fig. 4 indicates that for any given level of lat-
eral load, the SU-FREI undergoes larger lateral displacement than
the B-FREI. A well recognized way of dealing with excessive lateral
displacements in elastomeric isolators is to provide supplementary
damping in the isolation system. Traditionally, a lead core or a
high-damped rubber compound is employed in order to dissipate
a larger portion of the input excitation and limit isolator deforma-
tion. Results of previous studies [2,5–7] suggest that in FREIs that
utilize sufficient relative volume of fiber to elastomer, the interac-
tion between fiber reinforcement and the rubber layers provides a
new source of energy dissipation in addition to the intrinsic damp-
ing of the elastomer. This phenomenon is more pronounced for SU-
FREIs due to the increased distorted pattern of lateral displacementas compared to B-FREIs (compare Fig. 2a and b). Even though the
Fig. 3. Lateral load–displacement relationship of the square SU-FREI; comparisonbetween finite element and experimental results.
Fig. 4. Lateral load–displacement of the bonded and stable unbonded FREIs.
854 H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
6/10
SU-FREIs are more flexible than their corresponding bonded isola-
tors, the presence of increased damping may aid in restricting ex-
treme lateral displacement under a strong earthquake event.
6.2. Stress/strain state in the isolators
For each rubber element, the local axes are denoted by Axis 1
and Axis 2, which are parallel and perpendicular to the orientation
of the fiber reinforcement layers, respectively. Fig. 5 contains the
stress contour for normal stress S 22 corresponding to 200% t r dis-
placement. The normal stress S 22 acts perpendicular to the orienta-
tion of fiber reinforcement layers. The stress contours shown in
Fig. 5 reflects the average analysis output at all integration points
of the elements.
For the B-FREI, as can be seen in Fig. 5a, the axial load is carried
through a compression core within the isolator, which acts as an
equivalent column. The cross section area of the equivalent column
is limited to the overlap region between the top and bottom faces
of the laterally deformed B-FREI. In a B-FREI, the boundary condi-tions and the point of application of the vertical load resultant at
the top and bottom of the isolator, remain constant regardless of
the level of lateral deformation. Therefore, to establish equilibrium
in the isolator, the balancing moments are generated at the top and
bottom surfaces of a laterally deformed B-FREI (see Fig. 6a). As a re-
sult, the regions outside the central compression core undergo sig-
nificant tensile stresses which are normal to the bonding interface
between the outer rubber layers and the steel mounting plates and
between the inner rubber and the fiber reinforcement layers. The
tensile regions (yellow1 triangles) and the compression core (equiv-
alent column) can be clearly seen in Fig. 5a.
Due to unbonded application, no tensile stress is transferred to
a laterally deformed SU-FREI at the isolator’s contact supports. As a
result, no balancing moment develops at the top and bottom sur-faces of a SU-FREI. As the SU-FREI is deformed laterally, one end
of the contact surface rolls off the support, and the opposite end
is pressed against the support. Therefore, as shown in Fig. 5b, the
stress S 22 at the contact surfaces ranges from its maximum value,
at the pressed end, to zero, at contact regions that are at the onset
of rolling off the support. Due to the non-uniform stress distribu-
tion, the point of application of the vertical load resultant, at each
contact surface, shifts toward the pressed corner of the isolator. As
can be seen in Fig. 6b, the offset of vertical resultant loads at the
top and bottom of the isolator produces a couple that balances
the overturning moment caused by the shear loads.
A close examination of Fig. 5a and b indicates that tension stres-
ses S 22 in the SU-FREI are negligible compared to the B-FREI. Ten-
sion regions in the SU-FREI include the outer rubber layers at
regions where the bearing has rolled off the supports, and the rub-
ber layers next to these layers where tension stresses are localizedonly near the maximum curvature in the layer. Other than these
regions, the remainder of the SU-FREI carries compression along
local Axis 2 of the rubber elements. From Fig. 5a and b, one can
conclude that the peeling stress demand on the bond between rub-
ber and fiber reinforcement layers in a B-FREI is an order of mag-
nitude higher than that of the SU-FREI.
According to Fig. 5a and b the magnitudes of peak compression
S 22 that develop in both isolators are comparable. In the B-FREI the
maximum pressure occurs at the centre of the compression core
within the isolator and remains nearly unchanged along the isola-
tor’s height. As the lateral deformation increases, the cross section
area of the compression core decreases and the magnitude of peak
pressure increases. Accordingly, similar to conventional SREIs, in
the design of a B-FREI, one should ensure that the isolator willnot encounter Euler buckling [26] at extreme lateral deformations.
In the SU-FREI, the peak pressure S 22 is locally concentrated at the
corners of the isolator where the outer rubber layers are pressed
against the contact supports. Similar to the B-FREI, the vertical load
is carried through a compression zone within the SU-FREI. How-
ever, excluding the local regions at top and bottom corners of the
SU-FREI, the peak pressure within the body of the compression
zone in the SU-FREI is significantly lower than that of the B-FREI.
Excluding the outer rubber layers, the compression zone in the
SU-FREI is subjected to a relatively more uniform stress as opposed
to the B-FREI.
Fig. 7a–d shows the distribution of the normalized stress S 22/ pnalong the length of the 6th rubber layer (R6) located at the mid-
height of the isolator (see Fig. 1), at lateral displacements of 50%,
(a) B-FREI (b) SU-FREI
Fig. 5. Contour of normal stress S 22 (MPa) in the rubber layers of the isolators at lateral displacement of 200% t r (positive values indicate tension).
(a) B-FREI (b) SU-FREI
P M
V a
V a
P
M V b
P
P
V b
Fig. 6. Free body diagram in laterally deformed FREIs with different boundaryconditions.
1 For interpretation of color in Fig. 5, the reader is referred to the web version of this article.
H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859 855
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
7/10
100%, 150%, and 200% t r , respectively. Under a vertical load of
112 N, and given isolator width of 70 mm, the nominal vertical
pressure per unit length of both isolators is pn = 1.6 MPa. According
to Fig. 7, as the lateral displacement in the B-FREI increases the
peak compression stress increases and the length of the compres-sion region decreases. However, the peak compression stress in the
SU-FREI remains approximately constant regardless the level of
lateral displacement. Additionally, no tension develops in the rub-
ber material. Since the peak pressure at the middle of the SU-FREI
is approximately 38% lower than in the B-FREI at extreme lateral
displacement of 200% t r (Fig. 7d), the likelihood of Euler buckling
in the SU-FREI is significantly lower than in the B-FREI.
Fig. 8a and b contain the contours of normal stress S 11 in the iso-
lators corresponding to 200% t r lateral deformation. The normal
stress S 11 acts parallel to the orientation of fiber reinforcement lay-
ers. An examination of Figs. 5a and 8a indicate that the overlap re-
gion between the top and bottom supports of the B-FREI is
subjected to biaxial compression. But, the regions outside the over-
lap (yellow triangles) are under biaxial tension. In the tension re-
gion, the bond between rubber and fiber reinforcement layers
must have sufficient strength to resist large biaxial tensile stressesat extreme lateral deformations.
Similar to the B-FREI, the central region of the SU-FREI resists
biaxial compression (see Figs. 5b and 8a). According to Fig. 8a
and b, the peak values of compression stress S 11 for both isolators
are similar. However, unlike the B-FREI, the peak compression
stresses in the SU-FREI are only localized at the compressed cor-
ners of the isolator. Beyond these corners, the peak compressive
stress S 11 within the central compression region of the SU-FREI is
approximately 50% lower than that of the B-FREI. The peak tensile
stresses S 11 in the SU-FREI are significantly (40%) lower than for
the B-FREI. The maximum tensile stress is localized in a few
(a) 50% t r (b) 100% t r
(c) 150% t r (d) 200% t r
Fig. 7. Distribution of normalized stress S 22/ pn along the length of the 6th bottom rubber layer of the isolators at different lateral displacement amplitudes; (Notes: nominal
vertical pressure pn = 1.6 MPa; negative stress values indicate compression).
(a) B-FREI (b) SU-FREI
2.55
1.92
1.30
0.67
0.05
- 0.58
- 1.20
- 1.83
- 2.45
- 3.08
- 3.70
1.50
0.98
0.46
- 0.06
- 0.58
- 1.10
- 1.62
- 2.14
- 2.66
- 3.18
- 3.70
Fig. 8. Contour of normal stress S 11 (MPa) in the rubber layers of the isolators (positive values indicate tension).
856 H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
8/10
interior rubber layers near the maximum curvature of the layer be-
yond the central compression zone (see Fig. 8b).
In a laterally deformed SU-FREI, the compressive stresses are
concentrated at the compressed corners of isolator (see Figs. 5b
and 8b). In these regions, the rubber material and bond between
rubber and the fiber reinforcement layers are subjected to large
biaxial compressive stresses. Nonetheless, damage in the corner
areas is not a common failure mode in SU-FREIs as the materialsustains a confined state of stress. In previous experimental studies
that were conducted on SU-FREIs [5–7], no damage or delamina-
tion was observed between the rubber and fiber reinforcement lay-
ers at the corners of the tested isolators. In these tests, the isolators
were repeatedly subjected to large lateral displacements.
The distribution of the normalized stresses S 11/ pn along the
length of the rubber layer R6 (see Fig. 1) is shown in Fig. 9a–d
for lateral displacement magnitudes of 50%, 100%, 150%, and
200% t r , respectively. As seen in these figures, overall, peak stress
values (compression or tension) in the SU-FREI are significantly
lower than those in the B-FREI. The peak compression in the SU-
FREI remains nearly constant at different lateral displacements.
As the lateral displacement increases the length of compression re-
gion within the rubber material decreases in both isolators.
For efficient operation of an elastomeric isolator, the shear
strain in the rubber material must be extremely large. Fig. 10aand b illustrates the contour of shear strain within the isolators
and Fig. 11a–d contains the distribution of shear strain along the
6th rubber layer (R6 in Fig. 1), at different lateral displacements.
As seen in Fig. 11 the peak values of shear strain in the rubber
material of both isolators are comparable. For a displacement of
200% t r (38 mm) the shear strain in the isolator can be approxi-
mated as (38 mm/25 mm) 1.5, which is in reasonable agreement
with the values given in Figs. 10 and 11d. According to Fig. 10a and
(a) 50% t r
(c) 150% t r
(b) 100% t r
(d) 200% t r
Fig. 9. Distribution of normalized stress S 11/ pn along the length of the 6th bottom rubber layer of the isolators at different lateral displacement amplitudes; (Notes: nominal
vertical pressure pn = 1.6 MPa; negative stress values indicate compression).
(a) B-FREI (b) SU-FREI
1.65
1.49
1.32
1.26
0.99
0.03
0.66
0.50
0.33
0.17
0.00
1.65
1.39
1.12
0.85
0.59
0.33
0.06
- 0.20
- 0.47
- 0.73
- 1.00
Fig. 10. Contour of shear strain in the rubber layers of the isolators at 200% t r lateral displacement.
H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859 857
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
9/10
dashed curves in Fig. 11a–d, the profile of shear strain is approxi-
mately uniform within the B-FREI. However, in the SU-FREI, uni-
form shear strains are found to occur only in the central region
of the isolator, which remains in contact with the top and bottom
supports. In the regions of the SU-FREI that are not in contact withthe supports, the shear strain is found to decrease from its peak po-
sitive value (see Fig. 10b and solid curves in Fig. 11a–d). At extreme
displacements the shear strain acquires negative values in the SU-
FREI at regions close to the ends of the rubber layer (see Fig. 11d).
The main purpose of fiber reinforcement in a FREI is to restrain
the lateral bulging of the rubber layers when the isolator is sub-
jected to vertical loads. When a FREI is loaded vertically, the rubber
layers, which are confined by the reinforcement layers, undergo
compression and the fiber reinforcement layers in turn experience
tension. At zero lateral displacement, the profile of fiber tensile
stress along the length of the reinforcement layer can be described
with a parabola with the peak value of tensile stress occurring at
the mid-length of the reinforcement layer. Lateral displacement
of an FREI can affect the fiber stress profile and alter the magnitudeand location of the peak stress developed in the fiber reinforce-
ment layer. Fig. 12 shows the variation of peak fiber tensile stress
as a function of lateral displacement. The fiber reinforcement layer
in this figure is located at the mid-height of the isolator. As can be
seen in Fig. 12, the peak fiber tensile stress in the B-FREI increases
significantly with increased lateral displacement. However, for the
SU-FREI, the influence of increased lateral displacements on the
peak fiber tensile stress is negligible.
From the FE-analysis, it is found that the fiber reinforcement
layers in the SU-FREI are generally subjected to significantly lower
tensile stresses as compared to the B-FREI. According to Fig. 12, at
200% t r displacement, the maximum fiber tensile stress at the mid-
height of the SU-FREI is nearly 40% lower than that of the B-FREI.
Also, the peak shear stress demand on the bond between fiber rein-forcement and rubber layers in the SU-FREI is lower. At 200% t r
displacement, the FE-analysis indicates that in the B-FREI the mag-
nitude of peak tensile stress for different reinforcement layers is
approximately equal. However, in comparison with the inner
layers, the peak tensile stress in the extreme reinforcement layers
of the SU-FREI is largely reduced as a result of rollover response
behavior.
7. Conclusions
Finite element analyses were conducted on two identical FREIs,
one assuming conventional bonded application (the bonded (B)-
FREI) and the other one assuming unbonded contact surfaces that
remained laterally stable (the stable unbonded (SU)-FREI). TheFE-model was found to be sufficiently accurate for the preliminary
(a) 50% t r (b) 100% t r
(c) 150% t r (d) 200% t r
Fig. 11. Distribution of shear strain along the length of the 6th bottom rubber layer of the isolators at different lateral displacement amplitudes.
Fig. 12. Influence of isolator displacement amplitude on the peak tensile stress in
the middle fiber reinforcement layer.
858 H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859
8/9/2019 Bonded versus unbonded strip fiber reinforced elastomeric isolators: FEA
10/10
design of the isolators. In addition, it could be used for effective
determination of the peak stress and strain demands on isolator’s
components. Both B- and SU-FREIs were deformed up to 200% t r (t r = total thickness of rubber layers in the isolator) lateral displace-
ment under identical constant vertical loads. Results of the FE-
analyses documented the following advantages in the lateral re-
sponse of the SU-FREI compared to that of the B-FREI:
i. Significantly lower horizontal stiffness, and hence signifi-
cantly higher seismic isolation efficiency.
ii. Considerably lower stress demand on rubber material.
iii. Negligible peeling stress demand on the bond between rub-
ber and the fiber reinforcement layers.
iv. Reduced tensile stress demand on the fiber reinforcement,
and subsequently a significantly lower shear stress demand
on bond between rubber and the fiber reinforcement layers.
There is significant potential for SU-FREIs to be cost effective
elastomeric isolators for the seismic protection of ordinary build-
ings and structures. To achieve similar seismic mitigation, a SU-
FREI requires a significantly shorter operational height than a B-
FREI. This indicates considerable material saving in the isolator.
Due to the lower in-service demands on SU-FREIs, they can be fab-
ricated using a simple manufacturing process. Additional cost sav-
ing exists due to the elimination of thick steel mounting plates,
which are typically bonded to the top and bottom faces of isolators
employed in bonded applications.
Acknowledgements
The authors would like to gratefully acknowledge the support
provided by the Ontario Ministry of Research and Innovation
(MRI), and the Natural Sciences and Engineering Research Council
of Canada (NSERC). Additionally, the authors are grateful for the
support provided by MSC Incorporation.
References
[1] Kelly JM. Analysis of fiber-reinforced elastomeric isolators. J Seismol EarthqEng 1999;2(1):19–34.
[2] Kelly JM. Seismic isolation systems for developing countries. Earthq Spectra2002;18(3):385–406.
[3] Moon BY, Kang GJ, Kang BS, Kelly JM. Design and manufacturing of fiberreinforced elastomeric isolator for seismic isolation. J Mater Process Technol2002;130(131):145–50.
[4] Kang BS, Kang GJ, Moon BY. Hole and lead plug effect on fiber reinforcedelastomeric isolator for seismic isolation. J Mater Process Technol2003;140:592–7.
[5] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Testing and modeling of squarecarbon fiber-reinforced elastomeric seismic isolators. Struct Control HealthMonit 2008;15(6):876–900.
[6] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Lateral response evaluation of fiber-reinforced neoprene seismic isolators utilized in an unbonded application.Struct Eng, ASCE 2008;134(10):1627–38.
[7] Toopchi-Nezhad H, Drysdale RG, Tait MJ. Parametric study on the response of stable unbonded fiber-reinforced elastomeric isolators (SU-FREIs). J ComposMater 2009;43(15):1569–87.
[8] Dehghani Ashkezari G, Aghakouchak AA, Kokabi M. Design, manufacturing and
evaluation of the performance of steel like fiber reinforced elastomeric seismicisolators. J Mater Process Technol 2008;197(1–3):140–50.
[9] Mordini A, Strauss A. An innovative earthquake isolation system using fibrereinforced rubber bearings. Eng Struct 2008;30(10):2739–51.
[10] Kang GJ, Kang BS. Dynamic analysis of fiber-reinforced elastomeric isolationstructures. J Mech Sci Technol 2009;23:1132–41.
[11] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Shake table study on an ordinarylow-rise building seismically isolated with SU-FREIs (stable unbondedfiber-reinforced elastomeric isolators). Earthq Eng Struct Dyn 2009;38(11):1335–57.
[12] Tsai HC, Kelly JM. Stiffness analysis of fiber-reinforced elastomeric isolators.PEER Report 2001/05, Pacific Earthquake Engineering Research Center,University of California, Berkeley, California; 2001.
[13] Tsai HC. Compression stiffness of infinite-strip bearings of laminated elasticmaterial interleaving with flexible reinforcements. Int J Solids Struct2004;41(24–25):6647–60.
[14] Tsai HC. Compression stiffness of circular bearings of laminated elasticmaterial interleaving with flexible reinforcements. Int J Solids Struct2006;43(11–12):3484–97.
[15] Pinarbasi S, Mengi Y. Elastic layers bonded to flexible reinforcements. Int JSolids Struct 2008;45(3–4):794–820.
[16] Tsai HC, Kelly JM. Buckling of short beams with warping effect included. Int JSolids Struct 2005;42(1):239–53.
[17] Tsai HC, Kelly JM. Buckling load of seismic isolators affected by flexibility of reinforcement. Int J Solids Struct 2005;42(1):255–69.
[18] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Stiffness analysis of fiber-reinforcedrubber isolators under compressive loads: a finite element approach. In:Proceedings of the 9th US national and 10th Canadian conference onearthquake engineering, Toronto, Ontario; 2010.
[19] Haringx JA. On highly compressible helical springs and rubber rods, and theirapplication for vibration-free mountings – part III. Philips Res Rep1948;4:206–20.
[20] ASCE-7. Minimum design loads for buildings and other structures, ASCE/SEI 7-05. New York, American Society of Civil Engineers; 2005.
[21] MSC. Marc. Santa Ana, CA: MSC Software Corporation; 2008.[22] Zienkiewicz OC, Taylor RL. The finite element method. The basis, vol. 1, 5th
ed.. Oxford: Butterworth-Heinemann, A division of Reed Educational and
Professional Publishing Ltd.; 2000.[23] MSC Software Corporation. Nonlinear finite element analysis of elastomers,
Technical paper. Los Angeles, CA: MSC Software Corporation; 2000.[24] Hermann LR. Elasticity equations for incompressible and nearly
incompressible materials by a variational theorem. J Am Inst AeronautAstronaut 1965;3(10):1896–900.
[25] MSC. Mrac. Theory and user information, vol. A. Santa Ana, CA: MSC SoftwareCorporation; 2008.
[26] Kelly JM. Earthquake-resistant design with rubber. 2nd ed. London: Springer;1997.
H. Toopchi-Nezhad et al. / Composite Structures 93 (2011) 850–859 859