Professional Documents
Culture Documents
Final Published Paper
Final Published Paper
Structures
journal homepage: www.elsevier.com/locate/structures
A R T I C L E I N F O A B S T R A C T
Keywords: An improved formulation of the horizontal shear behavior of the unbonded fiber-reinforced elastomeric isolators
Analytical formulation (UFREI), which predicts both the pre-rollover and the post-rollover behavior considering the effect of the vertical
Unbonded fiber-reinforced elastomeric iso load, is proposed. A simplified and computationally inexpensive formulation of the horizontal shear behavior of
lators
the UFREI has also been proposed afterward. Further, these formulations are verified using experimental results
Rollover
Near-fault earthquakes
and compared with the results obtained using the other earlier methods in the recent literature. The horizontal
MR Damper shear behavior predicted by both of the proposed formulations is observed to fit the experimental results very
Shear behavior well. The shear behavior predicted by these formulations is observed to be more accurate than the corresponding
Vertical load predicted by the other formulations from the recent literature. A parametric study characterizing the effect of the
vertical pressure and the initial elastomeric shear modulus on the horizontal shear behavior of isolators is also
carried out. The proposed formulations are further used to model the behavior of the UFREI employed to isolate
the benchmark structure considered in the study.
Supplementary damping devices are further installed alongside the nonlinear UFREI isolation system to
mitigate the excessive displacement demand on isolators under near-fault ground motions. The isolated
benchmark structure is thus coupled with a Magneto-rheological Damper (MRD) with an improved control al
gorithm for nonlinear isolation systems in three configurations, i.e., semi-active, passive-on, and passive-off. A
comparative study of these configurations shows that MRD in passive-on mode provides the minimum isolators
displacement response, but an overall improved structural response is achieved with MRD in the semi-active
mode of operation.
1. Introduction and Takhirov [4] evaluated the compression stiffness theoretically and
further investigated the horizontal shear behavior of fiber-reinforced
Earthquakes have been a major cause of the loss of life and property elastomeric isolators (FREI) experimentally. It was also shown that the
in the history of humankind, and major of it is caused due to weak FREIs have better isolation efficiency when used unbonded to sub-
structures. Base isolation is a seismic mitigation technique that focuses structure and superstructure. These types of isolators are called
on reducing the seismic demand on the structures and/or their contents Unbonded fiber-reinforced elastomeric isolators (UFREI). Moon et al.
[1,2] by introducing an element having low horizontal stiffness and high [5] tested different fibers using vertical and horizontal loading tests and
damping between the ground and superstructure, thus increasing the found carbon fibers superior to other fibers considered, i.e., nylon and
period of the structure. Elastomeric bearings are the most common type glass fiber. Several other experimental and analytical studies [6–9] have
of base-isolator being used today. It consists of alternating elastomer been done since to check the efficiency of unbonded and bonded FREI
layers stacked with reinforcement layers (generally steel plates). Kelly under several types of horizontal and vertical loading cycles. A signifi
[3] evaluated the feasibility and mechanical properties of fiber as cant enhancement in the structure’s response as compared to that in
reinforcement in place of steel plates. It discussed the advantages of fixed-base structure and isolated structure with SREIs is seen in all the
using fiber over steel as reinforcement for its implementation in small studies. It is also seen that a properly designed and stable UFREI has
buildings in developing countries. These advantages included better performance than its bonded counterpart due to the rollover
manufacturing ease, decreased cost, and being lightweight. Later, Kelly phenomenon.
* Corresponding author.
E-mail addresses: shiv.prakash@iitb.ac.in (S. Prakash), rsjangid@civil.iitb.ac.in (R.S. Jangid).
https://doi.org/10.1016/j.istruc.2022.06.024
Received 20 April 2022; Received in revised form 4 June 2022; Accepted 8 June 2022
2352-0124/© 2022 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
435
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 2. Cantilever representation of the rollover section with (a) a follower load (PF-1); (b) a moment (PF-2).
vertical load, while the shear modulus is kept constant for the rollover √̅̅̅
portion. and, pcrit =
2πGSa
√̅̅̅ (6)
Thus, the force–deformation behavior before full rollover is given as. 2 3tr
F = F1 + 2F 2 (2) where a, b, S, and tr are the length, width, shape factor, and total elas
tomer thickness of the isolator; G is the shear modulus of the elastomer; p
The central section and rollover sections along with the forces, are
is the vertical pressure on the isolator; u is the horizontal displacement
shown in Fig. 1. Here, F1 and F2 are the horizontal force due to the
of the isolator; dRS is the shear deformation of the rollover section under
central shear portion and each of the two rollover sections, respectively,
F2. The bending deformation (dRB) in each of the rollover section is
which are further formulated as.
( ) characterized by assuming the rollover portion to be a cantilever beam
u with a representative follower load V acting on the free end as shown in
F 1 = Geff (b)(a-u) (3)
tr Fig. 2(a). The deformation of the cantilever under the follower load V
( ) can be written as.
G tur (b)dRS { π }
F2 = (4) EIeff φ′′ + Vsin φ + − φ(u) = 0 (7)
2 2
⎧ ⎡ ⎧ ⎫2 ⎤
(8)
′
⎪ ⎪ ⎪ φ (u) = 0; φ(0) = 0; φ(u) = θ; dRB = hθ
⎪
⎪ ⎪
⎨ ⎪
⎬ ⎥(
⎪
⎪
⎪
⎢
⎢ p u)
⎪
⎪ G⎣1 − ( ( ) ) ⎥ ⎦ 1 − for 0⩽u⩽1.0tr F 2 = Vsinθ (9)
⎪
⎪ ⎪
⎪ u 2 ⎪
⎪ a
⎪
⎪ ⎩pcrit 1 − a ⎭
⎪
⎪
⎨
⎡ ⎧ ⎫2 ⎤ where EIeff is the effective bending stiffness of the rollover section about
Where, Geff =
⎪
⎪ ⎪
⎪
⎪
⎪
⎪
⎪ the bottom pivot point. The effective bending stiffness is assumed to be
⎪ ⎢ ⎪ ⎪
⎪
⎪
⎪ ⎢ ⎨ p ⎬ ⎥⎥
( )
equal to the bending stiffness of the elastomer layers in this study. An
⎪
⎪ ⎢ ( ) ⎥ 1− 1.0tr for 1.0tr ⩽u⩽1.5tr
⎪
⎪ G
⎪ ⎣
⎪
⎢ 1 −
⎪p
⎪
( 1.0tr
)2 ⎪
⎪
⎥ a iterative approach, as shown in Fig. 3, is adopted to find the appropriate
⎪ ⎪ crit 1 − ⎪ ⎦
⎪
⎩ ⎪
⎩ a ⎪
⎭ follower load for each displacement increment till convergence. The
horizontal (dHR) and vertical displacement (dVR) of the rollover edge are
(5) calculated as.
436
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Table 1
Physical and material properties of the adopted isolators.
Property Foster [38] Ngo et al. (2017) [15] Ngo et al. (2017) [15] Das et al. (2015) [39] de Raaf [40] Toopchi-Nezhad (2008) [6] Osgooei et al. [41]
(A1, B1) (B2)
∫ ∫
u u
Since the inclusion of the vertical load is done on the central shear
dVR (s) = sinφ ds; dHR (s) = cosφ ds (10)
0 0 portion of the isolator, the full-rollover displacement remains almost the
same as that with the EF. Here onwards, the Proposed formulation – 1
At full-rollover displacement of the UFREI (u = uFR),
discussed above is termed PF-1 in this study for brevity.
dVR (u) ≈ h, and dHR (u) = dFR (11)
Once the full-rollover occurs, the bending deformation stops in 2.2. Proposed formulation – 2
rollover sections, and the whole deformed isolator is assumed to deform
further in shear thus, In the PF-1 above, a differential equation (Eq. (7)) needs to be solved
at each time step till convergence which often requires a cumbersome
F = F1R + 2F2,FR + 2F3 (12) computational process. If a representative external moment is taken in
An extra F3 term is included in the expression to consider the post- place of follower load (Fig. 2(b)) to cause bending deformation in the
full-rollover shear deformation of the rollover sections. For F3 calcula cantilever representation of the rollover section, a much simpler dif
tion, the rollover sections are assumed to be triangular with base and ferential equation needs to be solved at each incremental step, thus
height dimensions as dFR and h, respectively. As the full-rollover will simplifying and improving the speed of the computational process. The
always happen at a horizontal displacement higher than 1.0tr, thus the equation for bending deformation of the cantilever section under an
expression of the effective shear modulus for u > 1.0tr can directly be external moment can be written as.
used. ′
EIeff φ − M = 0 (16)
( )
u
F1R = Geff (b)(a − uFR ) (13) Ms hMu
tr φ(0) = 0; φ(s) = ; Thus, dRB = (17)
EIeff EIeff
F2,FR = F2 (at full − rollover i.e. u = uFR ) (14)
F 2 = M/h (18)
( )
G u− tur FR (b)(dFR ) The simplicity of the computational process can easily be inferred
F3 = (15) from the explicit expression of the rollover bending displacement (dRB),
2
which otherwise was found after solving a differential equation at each
Fig. 4. Comparative force–displacement plots of the experimental results with the results obtained using different shear formulations of the isolators.
437
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 5. Comparative force–displacement plots of the experimental force values with the results obtained using different shear formulations of the isolators.
time step previously. The rest of the forces are found to be similar as in resulting error is normalized with the RMS error value of the FEF to
the PF-1 explained in the above section till convergence. Here onwards, define an error-index as shown below.
the Proposed formulation – 2 discussed above is termed PF-2 in this √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
∑
study for brevity.
2
Normalized RMS Error Index (Ẽ RMS ) = ∑ (F-F EX ) (19)
(FEF -F EX )2
2.3. Validation and discussion An ̃ERMS value lesser than unity means that the shear force values
predicted by the formulation in consideration are closer to the experi
The accuracy of the above-proposed horizontal behavior formula mental results than that predicted by the EF. As it can be seen in Figs. 4
tions is further validated by comparing the predicted shear behavior and 5, the E
̃RMS value is much lesser than the unity for both the PF-1 and
with eleven experimental data sets from the recent literatures PF-2 formulations in all of the cases except for the de Raaf [40] results.
[6,15,38–41]. The source of the experimental data including the geo The bearings used by de Raaf [40] are similar to those by Toopchi-
metric and material details of the isolators for each set, are tabulated in Nezhad [6] except for the isolator’s height, which is 24 mm for the
Table 1. The experimental data sets are thus selected to include a wide former and 25 mm for the latter. The experimental horizontal force
range of isolator dimensions, vertical pressure, and elastomeric shear values at each displacement amplitude in de Raaf [40] and Toopchi-
modulus. As this study deals with the horizontal behavior of the UFREI Nezhad [6] cases are also similar except for the force value at the
under a static displacement boundary condition dealing with its rollover displacement 2tr where de Raaf [40] case has a significantly higher force
characteristics, the backbone curve of the results under dynamic dis value, thus deviating significantly from both the PF-1 trendline and PF-2
placements has been adopted from the experimental results whenever trendline. Despite this deviation in de Raaf [40] results, the error in force
applicable. values with the PF-1 and PF-2 formulation with reference to experi
mental values remains almost similar (E ̃RMS =1) to that with the EF.
2.3.1. Horizontal force–displacement plots Another deviation from the general behavioral trend of the UFREI is
The horizontal shear force in isolators at different displacements is observed in the case of the isolators adopted from Das et al. [39], where
calculated using the above-proposed methods for each isolator and a failure is observed with the EF while the other two formulations show a
loading combination, and it is further compared with corresponding positive stiffening behavior. The aspect ratio of the isolators, in this case,
experimental results from the literature. The comparative plots of the is 0.95, which is too low for it to be stable at higher displacements; thus,
force values obtained experimentally with the force values calculated the lateral behavior predicted with the EF is more accurate and nearest
using the EF, the PF-1, and the PF-2 are shown in Figs. 4 and 5. In these to the original behavior. The positive stiffening behavior with the PF-1
figures, FEX represents the experimental horizontal forces at different and the PF-2 formulations is observed due to the negative effective
displacement amplitudes adopted from the literature; FEF represents the shear modulus obtained with the formula proposed in Eqs. (5) and (13).
horizontal forces calculated using the method proposed by Van Engelen Alternatively, the shear modulus is assumed to be constant throughout
et al. [20]; FPF-1 represents the horizontal forces calculated using the PF- the displacement history in EF; thus, no such deviation is observed with
1 formulation with improved shear modulus expression and a follower the corresponding formulation. This deviation from the original
load; FPF-2 represents the horizontal forces calculated using the PF-2 behavior is very less significant though, since an aspect ratio lesser than
formulation with an external moment. Root mean square (RMS) error 2.5 is generally avoided to prevent instability [13,42].
is calculated with reference to the experimental force values, and the
438
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 6. Horizontal shear behavior of the UFREI with change in (a) vertical pressure; (b) initial shear modulus.
Although both PF-1 and PF-2 methods model the shear behavior of same for all four cases. Meanwhile, the PF-1 and PF-2 formulations
UFREIs more accurately than the EF, a direct comparison of the consider the effect of vertical pressure; thus, the force values predicted
̃RMS value for both PF-1 and PF-2 methods shows them to be almost
E using these methods are more accurate than the corresponding with the
similarly accurate. The ̃
ERMS value for the PF-1 is slightly lesser than that EF. Although the accuracy of the EF is comparable to other methods at
of the PF-2 for most cases but the PF-2 is computationally much cheaper, lower horizontal displacements, it deteriorates and over-predicts the
given the explicit expression of the dRB. The horizontal behavior calcu horizontal shear force as the lateral displacement increases due to the
lation with EF also requires solving the same differential equation at non-inclusion of the vertical pressure in the formulation.
each time step. The post-rollover behavior predicted by the PF-1 is A parametric study is carried out to assess further the effect of the
observed to lie in between both rest formulations, with the EF and the vertical load and the initial shear modulus on the horizontal behavior of
PF-2 predicting the stiffest and least stiff horizontal behavior. UFREI. The UFREI samples used by Osgooei et al. [41] are adopted for
the parametric study, and the PF-1 formulation is used to characterize its
2.3.2. Vertical load effect horizontal behavior under a static displacement boundary condition.
The effect of vertical pressure on the horizontal shear behavior of the The resultant plots with variations in vertical force and initial elasto
isolator is very evident in Foster [38] results shown in Fig. 4, where the meric shear modulus are shown in Fig. 6(a) and (b), respectively. With
experimental horizontal force at a given displacement is seen to decrease the help of these plots, it can be further inferred that a corresponding
with an increase in the vertical pressure (p (in MPa) = 0.5, 1, 2, 3). Since softening and a stiffening behavior is seen with an increase in the ver
the EF is independent of the vertical pressure, the force values are the tical pressure and the elastomeric shear modulus, respectively. The
Fig. 7. Force-displacement behavior of UFREI under Northridge (FF), 1994 earthquake ground motion using different horizontal formulations.
439
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 8. Force-displacement behavior of UFREI under Northridge (NF), 1994 earthquake ground motion using different horizontal formulations.
Table 2
Full rollover displacement predictions.
Full-rollover Displacement Foster Ngo et al. (2017) Ngo et al. (2017) Das et al. (2015) de Raaf Toopchi-Nezhad (2008) Osgooei et al.
(uFR) [38] [15] [15] [39] [40] [6] [41]
(A1, B1) (B2)
softening of the horizontal behavior with an increase in vertical load to that of the PF-1 with a higher full rollover displacement. The dynamic
increases at higher vertical loads till the isolator buckles, while the response plots of the UFREI under other ground motions from Section
stiffening behavior with elastomeric shear modulus is almost uniform. 3.2 are similar and thus not included for brevity.
The effect of both vertical pressure and initial shear modulus variations
on horizontal behavior is relatively nominal at lower horizontal dis 2.3.3. Full-rollover displacement
placements and becomes more evident at higher displacements. Apart The horizontal displacement at which the full-rollover occurs for
from the static behavior of the UFREI, the effect of vertical pressure on various isolators considered in the study for both the methods is pro
the dynamic horizontal behavior of UFREI under real earthquake ground vided in Table 2. The full-rollover displacement is predicted within a
motions is also investigated. A rigid body structural system with the 0.5% error for all the isolators. It is observed that the PF-2 predicts a
same horizontal displacement degree of freedom as the UFREI beneath it higher full-rollover displacement than the PF-1. It is also observed that
is assumed to ascertain the required vertical pressure on the UFREI. The the full rollover displacement is independent of the vertical pressure and
isolators and earthquake ground motions used in this study are adopted the initial shear modulus of the elastomer. This behavior is evident from
respectively from Section 3.1 and Section 3.2 ahead. The resultant the same full-rollover displacements obtained for the isolators taken by
force–displacement plots of the assumed UFREI with different formu Foster [38] under four different vertical loads and the same full-rollover
lations under real seismic excitations can be observed in Figs. 7 and 8. displacements obtained for the isolators taken by Ngo et al. (2017) [15]
Since the isolator displacements under the far-fault ground motions are with a different initial shear modulus of elastomers. A similar variation
generally lesser, the UFREI hasn’t reached the full rollover state in Fig. 7. in full-rollover displacement with vertical load and elastomeric shear
Meanwhile, the post-rollover behavior of the isolator can easily be seen modulus is also observed in the plots given in Fig. 6. Thus, full-rollover
in the isolator displacement plots under near-fault ground motion in displacements are seen to be only dependent on the geometrical prop
Fig. 8. The full-rollover displacement of the designed UFREI can be erties of the UFREI.
referred to from the horizontal behavior plot in Section 3.1. The initial As discussed earlier, the aspect ratio is one of the major geometrical
softening behavior with rollover is seen to be more pronounced in the properties of the isolator. It has already been established earlier that a
horizontal plots predicted by PF-1 as compared to EF, and thus, the UFREI with a very low aspect ratio becomes unstable to horizontal
hysteresis plots using PF-1 are skewed clockwise referenced to the EF displacement and a UFREI with a very high aspect ratio becomes too stiff
plots. This behavior is further seen to increase at higher displacements for isolation. Since the UFREI systems have the best efficiency during the
and decrease with higher vertical loads. The horizontal behavior of the rollover phenomenon with horizontal displacements below the full-
UFREI with the PF-2 has been omitted since the behavior is very similar rollover displacement, a parametric study of the effect of the aspect
440
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 9. Variation of Full-rollover displacement (uFR) with the Aspect Ratio of UFREI.
Table 3
Isolator properties.
Dimension (a × b × h) 313 mm × 313 mm × 125.2 mm
Rubber layer thickness 4.89 mm
Number of elastomer layers 26 (including cover layers)
Shear Modulus, G 0.5 MPa
441
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
442
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
F u = KX (26) maximum (Vmax); otherwise, the voltage is set to zero. If the applied
force is equal to the required optimal force, then voltage remains
.X = AX + BF n + L(Y − CX − DF n ) (27) unchanged.
V = Vmax H{(F u − Fn )F d } = Vmax H{(Fu − F rb − F d )F d } (28)
Here, L{⋅} is Laplace transform operator. The matrices Q and R are
relatively decided based on the weights given to the control force and
Here, H{•} is Heaviside step function.
the objective response. The matrix Q is selected based on the objective
responses getting minimized. The R matrix is decided iteratively to get
the optimum results. In this study, the top floor acceleration response
3.4. Nonlinear dynamic analysis
and the isolator displacement response constitute the objective functions
for fixed-base structure and base-isolated structure, respectively. It is
Newmark average acceleration method is used to solve Eq. (20) for
seen that MRD has better efficiency when placed on bottom floors; thus,
various devices and building configurations. Although there are better
it is placed at the basement level with isolators for better efficiency. In
non-iterative and dissipative time integrator algorithms [61] available
MRD, force is controlled by varying the voltage applied to the current
for a nonlinear dynamic system, such algorithms mainly consider the
driver. As UFREI is used as the isolation system in this study which is
stiffness nonlinearity in the system. As the behavior of the MRD
highly nonlinear in nature, it needs to be incorporated into the voltage
employed in the study is highly complex and nonlinear in nature and
control algorithm accordingly. Since residual isolator force (Frb) is
directly dependent on the structure’s responses, an iterative procedure
constant for a given displacement increment, thus control to the damper
becomes imminent. Thus, a predictor–corrector iterative method is used
is applied after excluding the Frb from the total required control force
at each step till convergence while Newmark average acceleration
(Fu). Now, if the required optimal force and the applied damper force are
method is used each time to solve the corresponding linear system
in the same direction and the latter is smaller, then the voltage is set to
(including linear isolator stiffness term as discussed earlier). For each
443
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 13. Top floor acceleration history of the benchmark structure with various configurations under (a) Kocaeli (FF), 1999; (b) Kocaeli (NF), 1999 ground motion.
Fig. 14. Maximum top floor acceleration response with different configurations under the rest of the considered ground motions.
444
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 15. Isolator displacement history of the benchmark structure with various configurations under (a) Kocaeli FF), 1999; (b) Kocaeli (NF), 1999 ground motion.
Fig. 16. Maximum isolator displacement response with different configurations under the rest of the considered ground motions.
increment in displacement, forces in the isolator and MRD are calculated ON), and Passive-off (P-OFF). Cases with the benchmark structure with
using the models proposed in the above sections. A semi-implicit Runge- MRD installed alone are also considered for comparison.
Kutta method is used to find the damper forces developed in the MRD.
The analysis is done in MATLAB [62] with the help of the SIMULINK 3.5. Benchmark Structure’s response
toolbox, and the flow chart of the whole process is shown in Fig. 12.
Several uncontrolled and controlled configurations of the structure- Three major response parameters discussed in this study are top floor
isolator-damper system are considered for the seismic analysis of the acceleration, isolator displacement, and the required damper force.
structures. The supplementary damper system used in this study consists Time-history responses with corresponding amplitudes are plotted for
of MRD in its three working modes – Semi-active (SA), Passive-on (P- one of each near-fault and far-fault earthquake for different benchmark
445
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
Fig. 17. Force developed in the MRD installed with the various configurations of the benchmark building under far-fault and near-fault earthquakes.
structure’s configurations, and the response amplitudes under the rest of MRD (P-ON) dampers in combination with the isolators. MRD (SA)
the earthquakes are plotted as a bar graph for brevity. The response- dampers are designed to control the isolator displacement in hybrid
history plots and response amplitude plots for the top floor accelera configurations, thus the least acceleration response among all hybrid
tion response are shown in Figs. 13 and 14, respectively. When semi- configurations (i.e., BI + MRD (SA), BI + MRD (P-ON), and BI + MRD (P-
active MRD is used standalone without the isolation system, it’s OFF)) is achieved with the least stiff MRD (P-OFF) attached to the iso
designed to control the top floor acceleration response of the structure lated structure. Overall, all the hybrid configurations have an improved
while it’s designed to control the isolator displacement when clubbed lower acceleration response than the corresponding in the uncontrolled
with isolators. Thus, MRD (SA) is generally seen to have minimum top configuration or configurations with MRD implemented as standalone.
floor acceleration response among all three modes, followed by MRD (P- Isolator displacement is another important response parameter of the
ON) and MRD (P-OFF), respectively. An overall best acceleration isolated structures. The exceptionally high isolator displacement
response is generally achieved with BI configuration, which increases response under near-fault ground motions can easily be seen in the time
with the introduction of other supplementary damping devices. history plots and amplitude plots shown in Figs. 15 and 16, respectively.
This stiffening behavior in isolated structures is seen due to the It can also be observed from these plots that all the hybrid configurations
introduction of the dampers, and thus, it is more prominent with the with dampers attached to the isolated structure decrease the isolator
446
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
displacement response irrespective of the fault location of the earth very low aspect ratios. Although these isolators are generally not
quake. It is observed that the isolated structure clubbed with MRD (P- used in practice to maintain a stable seismic behavior of the
ON) mode has the least isolator displacement among all the configura structure.
tions. Other modes of MRD application, i.e., MRD (SA) and MRD (P-OFF) 4) When MRD is applied standalone to the fixed-base structure, the least
are relatively less effective in reducing the isolator response than the top floor acceleration response is generally achieved with semi-
MRD (P-ON); meanwhile, they are cheaper in terms of the required active control of the damper with the top floor acceleration as the
damper force (Fig. 17). Application of MRD (P-OFF) to the isolated control objective. BI structure gives the least acceleration response
structure is cheapest in terms of the damper force and gives the highest among all the configurations considered in the study, which often
isolator displacement among all three modes. MRD in semi-active mode increases due to stiffening from the introduction of supplementary
as a supplementary damping system strikes a balance between the damping devices. Thus, MRD (P-OFF) as a supplementary damping
isolator displacement response reduction and the cost attached with the device provides the least top floor acceleration response among the
required damper force. Semi-active H2/LQG control of the MRD is hybrid configurations, followed by the MRD (SA) and MRD (P-ON)
relatively less effective in this study due to the highly nonlinear behavior consecutively as a supplementary damping device.
of the isolators. 5) MRD (P-ON) as a supplementary damping device to the isolated
structure gives the least isolator displacement response (especially
4. Conclusions under near-fault ground motions), but it is very expensive in terms of
the required control force. MRD (SA) strikes a balance between the
Base-isolated structures are generally seen to have significantly high isolator displacement response reduction and the cost related to the
isolator displacement response under near-fault earthquake ground required damper force. MRD (P-OFF) as a supplementary damping
motions. Due to this high displacement demand on the isolators, the device to the isolation system is the least expensive in terms of
shear behavior formulations of UFREI need to consider the rollover and required damper forces, and it gives the highest isolator displace
post-rollover behavior. There are very few shear formulations in the ment response among all the considered hybrid configurations.
previous literature that discusses the rollover prediction and post-
rollover behavior of the isolators in detail. Declaration of Competing Interest
In the first part of the study, an improved and simplified horizontal
shear formulation of the U-FREI for the pre-rollover and post-rollover The authors declare that they have no known competing financial
behavior which also considers the effect of the vertical load, is pro interests or personal relationships that could have appeared to influence
posed. The results are further verified and compared with experimental the work reported in this paper.
results and results obtained using older formulations. The later part of
the study investigates the response of an isolated benchmark MDOF References
structure under a wide range of earthquake ground motions, including
both near-fault and far-fault earthquakes. The benchmark structure is [1] Pellecchia D, Lo Feudo S, Vaiana N, Dion J, Rosati L. A procedure to model and
design elastomeric-based isolation systems for the seismic protection of rocking art
isolated using UFREI, and supplemental damping devices are further objects. Comp-Aided Civil Infrastr Eng 2021. https://doi.org/10.1111/
employed to restrict the high displacement demands on the isolators. mice.12775.
Three configurations of MR damper with an improved control algorithm [2] Berto L, Favaretto T, Saetta A. Seismic risk mitigation technique for art objects:
experimental evaluation and numerical modelling of double concave curved
are used as supplementary damping devices in the study. A fixed-base surface sliders. Bull Earthq Eng 2013;11(5):1817–40. https://doi.org/10.1007/
configuration of the benchmark structure is also investigated for a s10518-013-9441-8.
comparative study of the structure’s response. The conclusion which can [3] Kelly JM. Analysis of fiber-reinforced elastomeric isolators. J Seismol Earthquake
Eng 1999;2(1).
be drawn from this study are as follows: [4] Kelly JM, Takhirov S. Analytical and experimental study of fiber-reinforced elastomeric
isolators fiber-reinforced elastomeric isolators. Pacific Earthquake Research Center,
1) The force values predicted by the PF-1 and the PF-2 for the horizontal University of California, Berkeley; 2001:PEER 2001/11.
[5] Moon B-Y, Kang G-J, Kang B-S, Kelly JM. Design and manufacturing of fiber
shear behavior of the UFREI fit the experimental results very well. reinforced elastomeric isolator for seismic isolation. J Mater Process Technol 2002;
Both the formulations consider the effect of rollover and vertical load 130–131:145–50. https://doi.org/10.1016/s0924-0136(02)00713-6.
on the horizontal behavior of UFREI, and thus, the force values are [6] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Lateral response evaluation of fiber-
reinforced neoprene seismic isolators utilized in an unbonded application. J Struct
more accurate than the force values predicted by the earlier formu
Eng 2008;134(10):1627–37. https://doi.org/10.1061/(asce)0733-9445(2008)134:
lation. The PF-2 is also the least computationally expensive among 10(1627).
all the formulations. [7] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Shake table study on an ordinary low-
2) The PF-2 predicts a higher full-rollover displacement, and the full- rise building seismically isolated with SU-FREIs (stable unbonded-fiber reinforced
elastomeric isolators). Earthquake Eng Struct Dyn 2009;38(11):1335–57. https://
rollover displacement values predicted by PF-1 or EF act as a lower doi.org/10.1002/eqe.923.
bound for the actual full-rollover displacement. It is also seen that the [8] Al-Anany YM, Tait MJ. Fiber reinforced elastomeric isolators for the seismic
stiffening of the horizontal behavior starts with the initial contact at isolation of bridges. Compos Struct 2017;160:300–11. https://doi.org/10.1016/j.
compstruct.2016.10.008.
a displacement much before the full-rollover displacement. The post- [9] Soleimanlo HS, Barkhordar MA. Effect of shape factor and rubber stiffness of fiber-
rollover behavior predicted by the PF-2 is the least stiff among all reinforced elastomeric bearings on the vertical stiffness of isolators. Trends Appl
three formulations. The full rollover displacement is observed to be Sci Res 2013;8(1):14–25. https://doi.org/10.3923/tasr.2013.14.25.
[10] Koh CG, Kelly JM. A simple mechanical model for elastomeric bearings used in
independent of the vertical load on the isolators and the initial shear base isolation. Int J Mech Sci 1988;30(12):933–43. https://doi.org/10.1016/0020-
modulus of the elastomer, but it is seen to decrease with the increase 7403(88)90075-6.
in the aspect ratio of the UFREI. [11] Gent AN, Lindley PB. The compression of bonded rubber blocks. Proc Instit Mech
Eng 1959;173(1):111–22. https://doi.org/10.1243/pime_proc_1959_173_022_02.
3) The horizontal shear behavior of the isolators is seen to get less stiff [12] Gent AN, Meinecke EA. Compression, bending, and shear of bonded rubber blocks.
with an increase in vertical load on the isolators and a decrease in the Polym Eng Sci 1970;10(1):48–53. https://doi.org/10.1002/pen.760100110.
initial shear modulus of the elastomer. This behavior is relatively [13] Toopchi-Nezhad H. Horizontal stiffness solutions for unbonded fiber reinforced
elastomeric bearings. Struct Eng Mech 2014;49(3):395–410. https://doi.org/
more evident at higher horizontal displacements. The seismic
10.12989/sem.2014.49.3.395.
behavior of the UFREI also showed that the softening due to rollover [14] Gerharer U, Strauss A, Bergmeister K. Verbesserte bemessungsrichtlinien für
is more prominent in the behavior predicted by proposed formula bewehrte elastomerlager. Bautechnik 2011;88(7):451–8. https://doi.org/10.1002/
tions (i.e., PF-1 and PF-2) as compared to that with the EF. The shear bate.201101475.
[15] Van Ngo T, Dutta A, Deb SK. Evaluation of horizontal stiffness of fibre-reinforced
behavior predicted by the PF-1 and PF-2 is observed to deviate from elastomeric isolators. Earthquake Eng Struct Dyn 2017;46(11):1747–67. https://
the real behavior at higher displacements for unstable isolators with doi.org/10.1002/eqe.2879.
447
S. Prakash and R.S. Jangid Structures 42 (2022) 434–448
[16] Russo G, Pauletta M, Cortesia A. A study on experimental shear behavior of fiber- [38] Foster BAD. Base Isolation Using Stable Unbonded Fibre Reinforced Elastomeric
reinforced elastomeric isolators with various fiber layouts, elastomers and aging Isolators (SU-FREI). Published online December 2011.
conditions. Eng Struct 2013;52:422–33. https://doi.org/10.1016/j. [39] Das A, Dutta A, Deb SK. Performance of fiber-reinforced elastomeric base isolators
engstruct.2013.02.034. under cyclic excitation. Struct Control Health Monitor 2014;22(2):197–220.
[17] Pauletta M. Method to design fiber-reinforced elastomeric isolators (U-FREIs) and https://doi.org/10.1002/stc.1668.
application. Eng Struct 2019;197:109366. https://doi.org/10.1016/j. [40] de Raaf M. Experimental study of unbonded fiber reinforced elastomeric bearings;
engstruct.2019.109366. 2009.
[18] Vaiana N, Sessa S, Marmo F, Rosati L. An accurate and computationally efficient [41] Osgooei PM, Van Engelen NC, Konstantinidis D, Tait MJ. Experimental and finite
uniaxial phenomenological model for steel and fiber reinforced elastomeric element study on the lateral response of modified rectangular fiber-reinforced
bearings. Compos Struct 2019;211:196–212. https://doi.org/10.1016/j. elastomeric isolators (MR-FREIs). Eng Struct 2015;85:293–303. https://doi.org/
compstruct.2018.12.017. 10.1016/j.engstruct.2014.11.037.
[19] Vaiana N, Capuano R, Sessa S, Marmo F, Rosati L. Nonlinear dynamic analysis of [42] Toopchi-Nezhad H, Tait MJ, Drysdale RG. Testing and modeling of square carbon
seismically base-isolated structures by a novel opensees hysteretic material model. fiber-reinforced elastomeric seismic isolators. Struct Control Health Monitor 2008;
Appl Sci 2021;11(3):900. https://doi.org/10.3390/app11030900. 15(6):876–900. https://doi.org/10.1002/stc.225.
[20] Van Engelen NC, Tait MJ, Konstantinidis D. Model of the shear behavior of [43] Pauletta M, Cortesia A, Russo G. Roll-out instability of small size fiber-reinforced
unbonded fiber-reinforced elastomeric isolators. J Struct Eng 2015;141(7): elastomeric isolators in unbonded applications. Eng Struct 2015;102:358–68.
04014169. https://doi.org/10.1061/(asce)st.1943-541x.0001120. https://doi.org/10.1016/j.engstruct.2015.08.019.
[21] Jangid RS, Kelly JM. Base isolation for near-fault motions. Earthquake Eng Struct [44] Thuyet VN, Deb SK, Dutta A. Mitigation of seismic vulnerability of prototype low-
Dyn 2001;30(5):691–707. https://doi.org/10.1002/eqe.31. rise masonry building using U-FREIs. J Perform Constr Facil 2018;32(2):04017136.
[22] Panchal VR, Jangid RS. Behaviour of liquid storage tanks with VCFPS under near- https://doi.org/10.1061/(asce)cf.1943-5509.0001136.
fault ground motions. Struct Infrastruct Eng 2012;8(1):71–88. https://doi.org/ [45] Ohtori Y, Christenson RE, Spencer BF, Dyke SJ. Benchmark control problems for
10.1080/15732470903300919. seismically excited nonlinear buildings. J Eng Mech 2004;130(4):366–85. https://
[23] Panchal VR, Jangid RS. Seismic response of structures with variable friction doi.org/10.1061/(asce)0733-9399(2004)130:4(366).
pendulum system. J Earthquake Eng 2009;13(2):193–216. https://doi.org/ [46] Ehsani B, Toopchi-Nezhad H. Systematic design of unbonded fiber reinforced
10.1080/13632460802597786. elastomeric isolators. Eng Struct 2017;132:383–98. https://doi.org/10.1016/j.
[24] Bani-Hani KA, Sheban MA. Semi-active neuro-control for base-isolation system engstruct.2016.11.036.
using magnetorheological (MR) dampers. Earthquake Eng Struct Dyn 2006;35(9): [47] Housner GW, Trifunac MD. Analysis of accelerograms—Parkfield earthquake. Bull
1119–44. https://doi.org/10.1002/eqe.574. Seismol Soc Am 1967;57(6):1193–220. https://doi.org/10.1785/bssa0570061193.
[25] Wang Y, Dyke S. Modal-based LQG for smart base isolation system design in [48] Bozorgnia Y, Niazi M, Campbell KW. Characteristics of free-field vertical ground
seismic response control. Struct Control Health Monitor 2012;20(5):753–68. motion during the northridge earthquake. Earthquake Spectra 1995;11(4):515–25.
https://doi.org/10.1002/stc.1490. https://doi.org/10.1193/1.1585825.
[26] Chen P-C, Tsai K-C, Lin P-Y. Real-time hybrid testing of a smart base isolation [49] Abrahamson NA, Somerville PG. Effects of the hanging wall and footwall on
system. Earthquake Eng Struct Dyn 2013;43(1):139–58. https://doi.org/10.1002/ ground motions recorded during the Northridge earthquake. Bull Seismol Soc Am.
eqe.2341. 1996;86(1B):S93-S99. doi:10.1785/bssa08601b0s93.
[27] Lin PY, Roschke PN, Loh CH. Hybrid base-isolation with magnetorheological [50] Teisseyre R, Majewski E, Takeo M. Earthquake source asymmetry, structural media
damper and fuzzy control. Struct Control Health Monitor 2007;14(3):384–405. and rotation effects. Springer Berlin Heidelberg; 2006. doi:10.1007/3-540-31337-0.
https://doi.org/10.1002/stc.163. [51] COSMOS Virtual Data Center. www.strongmotioncenter.org. Accessed March 6,
[28] Prakash S, Jangid RS. Optimum parameters of tuned mass damper-inerter for 2022. https://www.strongmotioncenter.org/vdc/scripts/default.plx.
damped structure under seismic excitation. Int J Dyn Control. 2022. doi:10.1007/ [52] PEER Ground Motion Database - PEER Center. Berkeley.edu. Published 2014.
s40435-022-00911-x. https://ngawest2.berkeley.edu/.
[29] Ivers DE, Miller LR. Experimental comparison of passive, semi-active on/off, and [53] Dahl PR. Solid friction damping of mechanical vibrations. AIAA J 1976;14(12):
semi-active continuous suspensions. SAE Technical Paper Series. Published online 1675–82. https://doi.org/10.2514/3.61511.
November 1, 1989. doi:10.4271/892484. [54] Kwok NM, Ha QP, Nguyen TH, Li J, Samali B. A novel hysteretic model for
[30] Tseng HE, Hedrick JK. Semi-active control laws – optimal and sub-optimal. Veh magnetorheological fluid dampers and parameter identification using particle
Syst Dyn 1994;23(1):545–69. https://doi.org/10.1080/00423119408969074. swarm optimization. Sens Actuators, A 2006;132(2):441–51. https://doi.org/
[31] Dyke SJ, Spencer BF, Quast P, Kaspari DC, Sain MK. Implementation of an Active 10.1016/j.sna.2006.03.015.
mass driver using acceleration feedback control. Comput-Aided Civ Infrastruct Eng [55] Li WH, Yao GZ, Chen G, Yeo SH, Yap FF. Testing and steady state modeling of a
1996;11(5):305–23. https://doi.org/10.1111/j.1467-8667.1996.tb00445.x. linear MR damper under sinusoidal loading. Smart Mater Struct 2000;9(1):95–102.
[32] Jing C, Youlin X, Weilian Q, Zhilun W. Seismic response control of a complex https://doi.org/10.1088/0964-1726/9/1/310.
structure using multiple MR dampers: experimental investigation. Earthquake Eng [56] Spencer BF, Dyke SJ, Sain MK, Carlson JD. Phenomenological Model for
Eng Vibr 2004;3(2):181–93. https://doi.org/10.1007/bf02858233. Magnetorheological Dampers. J Eng Mech 1997;123(3):230–8. https://doi.org/
[33] Gordaninejad F, Breese DG. Heating of magnetorheological fluid dampers. J Intell 10.1061/(asce)0733-9399(1997)123:3(230).
Mater Syst Struct 1999;10(8):634–45. https://doi.org/10.1106/55d1-xaxp-yfh6- [57] Spencer BF, Yang G, Carlson JD, Sain MK. “Smart” Dampers forSeismic Protection
b2fb. of Structures:A Full-Scale Study. In: Second World Conference on Structural Control.
[34] Al-Anany YM, Tait MJ. Experimental assessment of utilizing fiber reinforced [58] Bharti SD, Dumne SM, Shrimali MK. Seismic response analysis of adjacent
elastomeric isolators as bearings for bridge applications. Compos B Eng 2017;114: buildings connected with MR dampers. Eng Struct 2010;32(8):2122–33. https://
373–85. https://doi.org/10.1016/j.compositesb.2017.01.060. doi.org/10.1016/j.engstruct.2010.03.015.
[35] Castillo Ruano P, Strauss A. An experimental study on unbonded circular fiber [59] Dyke SJ, Spencer BF, Sain MK, Carlson JD. Modeling and control of
reinforced elastomeric bearings. Eng Struct 2018;177:72–84. https://doi.org/ magnetorheological dampers for seismic response reduction. Smart Mater Struct
10.1016/j.engstruct.2018.09.062. 1996;5(5):565–75. https://doi.org/10.1088/0964-1726/5/5/006.
[36] de Raaf MGP, Tait MJ, Toopchi-Nezhad H. Stability of fiber-reinforced elastomeric [60] Spencer BF, Suhardjo J, Sain MK. Frequency domain optimal control strategies for
bearings in an unbonded application. J Compos Mater 2011;45(18):1873–84. aseismic protection. J Eng Mech 1994;120(1):135–58. https://doi.org/10.1061/
https://doi.org/10.1177/0021998310388319. (asce)0733-9399(1994)120:1(135).
[37] Strauss A, Apostolidi E, Zimmermann T, Gerhaher U, Dritsos S. Experimental [61] Chang SY. Family of structure-dependent explicit methods for structural dynamics.
investigations of fiber and steel reinforced elastomeric bearings: shear modulus and J Eng Mech 2014;140(6):06014005. https://doi.org/10.1061/(asce)em.1943-
damping coefficient. Eng Struct 2014;75:402–13. https://doi.org/10.1016/j. 7889.0000748.
engstruct.2014.06.008. [62] MATLAB. Natick, Massachusetts: The Mathworks Inc.; 2021.
448