Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Structures 27 (2020) 962–974

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

Torsional response reduction of plan-asymmetric vertical seismic isolation T


by appropriate distribution of viscous and viscoelastic dampers

Reza Milanchiana, , Mahmood Hosseinib
a
Department of Structural Engineering, Mahabad Branch, Islamic Azad University, Mahabad, Iran
b
Structural Engineering Research Center, International Institute of Earthquake Engineering and Seismology (IIEES), Tehran, Iran

A R T I C LE I N FO A B S T R A C T

Keywords: In this article, torsional control has been studied in the plan-mass asymmetric vertical seismic isolation (VSI)
Vertical seismic isolation system considering the three-dimensionally analytical model. In the 3D VSI, a structure is partitioned by vertical
Torsional control isolation planes into two inner and outer substructures interconnected by links for lateral response reduction. In
Non-classical damping an ideal analytical model of plan-mass asymmetric VSI, plan-wise distribution of damping and stiffness have
Viscous and visco-elastic dampers
been provided for more torsional control. Seismic responses of the non-classically 3D VSI were obtained by
Runge-Kutta method
Response history analyses
Runge-Kutta numerical method, considering a large number of isolation ratios. In this investigation, the mass
and stiffness ratios of substructures are defined with respect to the original structure which provides a novel
method of graphic presentation for analysis results. In this manner, contour graphs create comprehensive and
worthwhile insight into the 3D VSI behavior and can also serve as a design tool for the preliminary design of the
system. The results show that in the case that the inner substructure is the flexible one, torsional control is nearly
effective in both substructures satisfactorily, regardless of plan-wise damping distribution. However, in the
second case that is the outer flexible substructure, torsional control is nearly effective in the stiff substructure
while the response in flexible substructure is satisfactory just in a limited range of isolation ratios.

1. Introduction either viscous or viscoelastic dampers can be used in the story plan in
an appropriate layout. Torsional control in the 3D VSI and the effects of
The viscous dampers have been considered as preferred supple- the plan-wise distribution of damper parameters is the issue that has
mentary energy dissipating devices for structural vibration control, investigated analytically in this study. For this purpose, by assigning
especially in innovative lateral displacement control strategies. The non-symmetric interconnecting damper parameters it has been tried to
feasibility of installing these dampers in different locations and im- reduce the torsional effects as well as excessive resulting displacement.
plementing those in new seismic design approaches have made these In this investigation, the mass and stiffness ratios of substructures have
dampers widespread in the building industry. In the vertical seismic been defined with respect to the original structure which provides a
isolation (VSI), an individual structure is partitioned intentionally into novel method of graphic presentation for analysis results. In this
two dynamically different substructures interconnected by viscous or manner, contour graphs create comprehensive and worthwhile insight
viscoelastic links. In this way, lateral response decreases in each sub- into the 3D VSI behavior and can also serve as a design tool for the
structure when compared to the original non-partitioned structure and preliminary design of the system. For seismic analyses, accelerograms
the performance of substructures is improved. So, we can expect that of six selected earthquakes have been employed.
both substructures meet at least higher performance level than the Before presenting the main body of the study, first in the following
considered one for the original structure. section a literature review is presented with regard to a) the use of
In a previous study by the authors, a comprehensive investigation dampers for seismic torsional response reduction, b) seismic response
was performed on the displacement control features of the VSI im- reduction through interconnecting two adjacent structures by dampers,
plementing viscous and viscoelastic links [1]. That study was limited to and a special state of that, called ‘mass isolation’. Then the analytical
planar structures, however, in practice, in seismic design of building model for the 3D plan-asymmetric VSI system is introduced, and the
structures torsion is a serious concern. To apply the idea of VSI to three- governing equations of motion are developed. In the next stage seismic
dimensional (3D) buildings, with consideration of torsional effects, response analyses of the 3D VSI, with various mass and stiffness ratios


Corresponding author.
E-mail address: Milanchian@iau-mahabad.ac.ir (R. Milanchian).

https://doi.org/10.1016/j.istruc.2020.07.009
Received 16 December 2019; Received in revised form 27 June 2020; Accepted 4 July 2020
2352-0124/ © 2020 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

are presented, and finally, the concluding remarks are expressed. the complex modal superposition method to determine dynamic char-
acteristics, mainly modal damping ratio, and modal frequency, of
2. Review of the related literature linked adjacent buildings by discrete visco-elastic and viscous dampers.
Parametric studies were performed to identify optimal parameters of
2.1. Seismic torsional response reduction dampers for achieving the maximum modal damping ratio or the
maximum response reduction of adjacent buildings. The analytical
Several studies have been conducted so far on control of torsion by formulas for determining optimum parameters of Maxwell model-de-
employing dampers in an individual non-partitioned structure. Goel [2] fined fluid dampers used to link two adjacent structures were derived
studied the effects of supplemental damping on the edge deformations using the principle of minimizing the averaged vibration energy of ei-
in asymmetric-plan systems. He identified the parameters that char- ther the primary structure or the two adjacent structures under a white-
acterize supplemental viscous damping and its plan-wise distribution, noise ground excitation [12]. A large body of work on coupled building
and then investigated the effects of these parameters on edge de- control is based primarily on determining optimal parameters of con-
formations, and showed that supplemental damping reduces the edge necting elements. Patel and Jangid [13] utilized numerical approaches
deformations to about one-third of the original building. That study was to determine the optimal parameters for connection elements.
limited to one-story buildings subjected to a single input earthquake. In Sharma and Jangid [14] investigated seismic responses of base-
continuation of that study, Goel [3] investigated the response of single- isolated benchmark building with hysteretically modeled variable
story asymmetric-plan buildings with supplemental viscous damping to sliding isolators. They showed that variable sliding isolators performed
harmonic ground motion using modal analysis techniques and showed better than conventional friction pendulum systems due to their varying
that most modal parameters, except dynamic amplification factors, are characteristic. Bharti et al. [15] examined the effectiveness of MR
affected insignificantly by the plan-wise distribution of supplemental damper for seismic response mitigation of adjacent multistory buildings
damping. under coupled building control scheme, involving passive-off, passive-
Lin and Chopra [4] examined the plan-wise distribution of fluid on, and semi-active control strategies. Zhu et al. [16] selected two
viscous dampers on the response of linearly elastic, one-story, asym- optimization criteria to minimize the vibration energy of the primary
metric-plan systems. They showed that asymmetric distributions of structure and to minimize the vibration energy of both structures.
supplemental damping would be more effective, around two times for a Closed-form equations were derived for solving the vibration control
fixed damping ratio, in reducing the response compared to symmetric problem of connecting two adjacent structures where they were cou-
distribution. Although the results of that study are widely applicable to pled via stiffness and damping elements connected in parallel by Ri-
an ensemble of ground motions characterized by the selected design chardson et al. [17]. Nigdeli and Bekdas [18] studied using optimally
spectrum, they are restricted to elastic one-story systems subjected to a Tuned Mass Dampers (TMD) to decrease the displacement vibrations of
single component of ground motion. Lin and Chopra [5] investigated structures in adjacent structures and two different designs of TMD were
the earthquake response of asymmetric one-story elastic systems with investigated. A specific type of VSI called ‘mass isolation’ has been
nonlinear viscous and viscoelastic dampers. The findings of that in- proposed by Ziyaeifar [19], in which the structural system is divided by
vestigation indicate that the earthquake response of the asymmetric a vertical isolation plane into two parts. One of them is called the mass
systems with nonlinear viscous or viscoelastic dampers can be esti- subsystem, which contains the majority of the system’s mass, and the
mated with sufficient accuracy for design applications by analyzing the other is called the stiffness subsystem, which contains the majority of
same asymmetric systems with all nonlinear dampers replaced by en- the system's stiffness. Ziyaeifar et al. [20] introduced two degrees of
ergy-equivalent linear viscous dampers. freedom model to improve the accuracy of the structural mechanism
Kim and Bang [6] proposed a strategy for finding the appropriate that supports the reaction forces of large energy dissipating devices.
plan-wise distribution of viscoelastic dampers based on the modal That model was an extension of the Maxwell three-element type model
characteristics of a structure-damper system to minimize the torsional in structural dynamics. Behnamfar et al. [21] proposed a method for the
responses of the structure. They found that appropriate plan-wise dis- rapid estimation of the dynamic response of connected structures based
tribution of supplemental dampers can effectively reduce the torsional on the concepts of kinematic and inertial interactions. They could re-
response of an asymmetric structure and that the viscoelastic dampers duce the time of design process by analyzing the structures separately
turn out to be more effective than viscous dampers. García et al. [7] considering the effect of the adjacent system. Zhao et al. [22] in-
investigated the torsional balance of elastic asymmetric structures with vestigated the energy‐dissipating performance of dampers between
viscoelastic dampers. In that work, the analytical and experimental adjacent retrofitted buildings. They indicated that the en-
response of stiffness and mass-eccentric structures subjected to different ergy‐dissipating approach can reduce the seismic response, as a result,
ground motions were studied. They observed that optimal damper ec- decrease the strengthening cost and time. Milanchian and Hosseini [23]
centricity values tend to increase linearly as the stiffness or mass ec- thoroughly studied the VSI technique with the nonlinear viscous dam-
centricities increase, and that response reduction factors ranging from pers regarding to substructure displacement and acceleration response
1.5 to 3 are possible with a small capacity damper. Moreover, viscoe- reduction. The results did not indicate a preferable performance of
lastic dampers are equally effective in controlling lateral–torsional nonlinear viscous dampers over linear viscous dampers.
coupling of torsionally flexible as well as stiff structures.
3. Analytical model of the 3D mass-asymmetric VSI
2.2. Seismic response reduction in the VSI
In the 3D VSI, the original structure is partitioned by vertical planes
In this section, the studies related to the interactional effects of three-dimensionally into two interactional inner and outer sub-
connected structures is presented. This idea was first introduced by structures. The interconnection is carried out by a set of viscoelastic
Klein et al. [8] for the purpose of wind-induced vibration reduction in links in the longitudinal and transversal direction as shown in an
large structures. Since then, numerous studies have been performed to idealized model of the 3D VSI in Fig. 1. The Kelvin model is employed
investigate connection strategies for reducing the response of adjacent to represent the viscoelastic damper for which the stiffness and the
structures. Xu et al. [9] have determined the optimal values for viscous damping coefficients are kli and cli respectively.
dampers connecting two adjacent structures of different heights It is assumed that the original structure is a square plan single-story
through parametric analysis. An extensive parametric study was carried building of side d , with mass asymmetry in the Y direction, and parti-
out to find optimum damper properties for adjacent buildings of dif- tioning takes place in a symmetric manner so that the side size of the
ferent stiffness ratios and different heights. Zhang and Xu [10,11] used inner substructure is d1. Assuming m , m1 and m2 are the masses of the

963
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 1. The plans of a single-story original structure and its mass-asymmetric 3D VSI.

original, the inner and the outer substructures respectively, vertical ⎡ m1 0 − m1. e1⎤ ⎡ m2 0 − m2 . e2 ⎤
isolation leads to m = m1 + m2 in which m1 = αm1 m and M1 = ⎢ 0 m1 0 ⎥, M2 = ⎢ 0 m2 0 ⎥
m2 = (1 − αm1) m . The masses of the two substructures are assumed to ⎢− m1. e1 0 IO1 ⎥ ⎢− m2 . e2 0 IO2 ⎥ (3)
⎣ ⎦ ⎣ ⎦
be proportional to the isolated plan areas, and therefore, the mass ratio
of the inner substructure αm1 = (d1/ d )2 . It is noted that subscripts 1 and where IO1 and IO2 are the torsional mass moment of inertia of the roof of
2 denote the inner and outer substructures from now on. The global the two substructures, which are calculated as IO1 = I 'O1 + m1. e12 and
coordinate system is considered to have its origin at the center of the IO2 = I 'O2 + m2 . e22 . I'O1 and I'O2 are the mass moment of inertia relative
building plan before applying the earthquake excitations. In the ori- to the corresponding center of mass. System stiffness matrix attributed
ginal structure, lateral stiffness in X and Y directions are provided to isolated substructures can be written as:
symmetrically by the exterior frames as k x = k y = k , and applying
vertical isolation leads to k = k1 + k2 , in which k1 = αs1 k and K 0⎤ ⎡ k1 0 0 ⎤ ⎡ k2 0 0 ⎤
K=⎡ 1 , K1 = ⎢ 0 k1 0 ⎥, K2 = ⎢ 0 k2 0 ⎥
k2 = (1 − αs1) k , where αs1is the stiffness ratio of the inner substructure. ⎣ 0 K2 ⎦ ⎢0 0 k ⎥ ⎢0 0 k ⎥
In the considered 3D model (Fig. 1), the linking dampers are placed ⎣ θ1 ⎦ ⎣ θ2 ⎦ (4)
between the internal sides of the outer substructure and the external where kθ1 and kθ2 are the rotational stiffness of the substructures, in
sides of the inner substructure. Besides, it is assumed that the system is which elements providing the lateral stiffness are assumed to be dis-
subjected to earthquake excitation in the X direction, so that different tributed symmetrically in the plan, and their values are calculated as
damping properties can be assigned to the dampers in this direction for kθ1 = 4. (k1/2). (d1/2)2 and kθ2 = 4. (k2/2). (d/2)2 . Stiffness matrix related
torsional control of substructures, that is, clx1 + clx 2 = cl and to viscoelastic link dampers, KL , is given by:
klx1 + klx 2 = kl . Whereas in the Y direction dampers with identical
properties are used symmetrically, that is cly1 = cly2 = cl/2 and ⎡ klx 0 klxθ ⎤
d d
kly1 = kly2 = kl/2 . KL = ⎡ K d − Kd ⎤, K d = ⎢ 0 kly 0 ⎥
Assuming the roof as a rigid diaphragm, there will be two transla- ⎢
⎣ − K K ⎥
⎦ ⎢ ⎥
⎣ klθx 0 klθ ⎦ (5)
tional and one rotational DOFs for each one of the substructures. So, the
displacement vector for the 3D VSI will be as Viscoelastic links provide rotational stiffness coefficient about the Z
u = {u x1, u y1, uθ1, u x 2 , u y2 , uθ2 }T . The mass eccentricity of both inner -axis which is equal to klθ = 4(kl/2)(d1/2)2 . klθx and klxθ are the values of
and outer substructures is considered to be only in the Y direction rotational-translational stiffness coefficients due to the asymmetric of
(ex = 0, e y ≠ 0 ). It is considered that the eccentricities for the sub- viscoelastic links, which are calculated as klθx = (klx1 − klx 2 )(d1/2) .
structures (e1, e2 ) remain the same concerning their plan dimensions. System damping matrix based on inherent Rayleigh damping of the
The governing equation of motion of the multi-story 3D VSI subjected substructures, C , and damping matrix attributed to viscoelastic links,
to earthquake excitations is described by: CL , are expressed as:

Mu¨ + (C + CL ) u̇ + (K + KL ) u = −Mru¨ g (t ) (1) Cin − 1 0 ⎤ Cd − Cd ⎤


C=⎡ , CL = ⎡

⎣ 0 Cin −2 ⎥
⎦ ⎢
⎣ C
− d Cd ⎥⎦ (6)
where M , K , and C are respectively the mass, stiffness and damping
matrices of the system with the size of 6n , which n is the number of in the aforementioned matrices, C d is a diagonal matrix and is defined
stories of the 3D VSI. CL and KL are also damping and stiffness matrices in accordance with viscoelastic links as:
related to visco-elastic added link dampers. u¨ g (t ) is the ground accel-
eration vector and r is the earthquake influence vector for the hor- ⎡ clx 0 clxθ ⎤
izontal excitation. For the case of 1-story 3D VSI, the influence vector is C d = ⎢ 0 cly 0 ⎥
⎢c ⎥
as r = {1, 0, 0, 1, 0, 0}T . System mass matrix can be expressed as: ⎣ lθx 0 clθ ⎦ (7)

M 0 ⎤ in which clθ is the value of rotational damping coefficient, resulting


M=⎡ 1 from links’ damping in X and Y directions about the Z -axis and is
⎣ 0 M2 ⎦ (2)
calculated as clθ = 4(cl/2)(d1/2)2 . clθx and clxθ are the values of damping
in which M1 and M2 are the mass matrices of substructures, which by coefficients, resulting from the asymmetry of viscoelastic links about
considering DOFs at the center of stiffness of the two substructures, and the center of rotation and are calculated as
e1 and e2 as their eccentricities, are expressed as: clxθ = clθx = (clx1 − clx 2 )(d1/2) .

964
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

4. Seismic response of the 3D VSI indicates unequal values for clx1 and clx2 as well as klx1and klx2 , respec-
tively above and below the X -axis in Fig. 1. Values less than 0.5 for ud
4.1. Basic features of the 3D VSI and us shift the center of damping and stiffness of the links above the X-
axis to proximity of the center of mass.
To clarify the basic concepts of the 3D VSI, a model of the 1-story Apparently, ranges of the link parameters, including damping and
torsional building is considered. In this mass-asymmetric sample model, stiffness should be reasonable. These parameters for a linear viscoe-
the story mass has been assumed to be 100 ton and the lateral stiffness lastic link can be expressed in a more appropriate manner with respect
to be 10,000 kN/m in longitudinal and transverse directions. This to the original structure specifications as:
stiffness value has been assigned such that the maximum drift, de-
termined by the static equivalent lateral force procedure in the high cl kl
ξl = , βl =
seismic zone, is limited to the seismic design code specified value. To 2mωn k (11)
study the seismic performance of the 1-story 3D VSI, it was subjected to
in which cl , kl are the sum of link damping coefficients and stiffness
different earthquake excitations and its response histories were ob-
values, respectively in the longitudinal and transversal direction, and ξl
tained by a series of time history analysis (THA) cases. In this paper,
and βl are the damping and stiffness ratios which are defined in terms of
features of the 3D VSI are studied assuming the linear elastic behavior
critical damping and lateral stiffness of the original structure. In
of isolated structures. The maximum displacement of the mass center of
structures with added devices, it is generally accepted that the max-
the rigid roof, 〈Xmax 〉, as well as its rotation about the Z -axis, 〈θmax 〉,
imum added damping is a value of 25% of the structure critical
were considered as the main seismic performance measures of the 3D
damping (Lee and Taylor [24]). In the previous studies conducted by
VSI in series of THA.
the authors, it was indicated that by a damping ratio in the range of
Comparisons of maximum responses of each substructure of the 3D
5% ≤ ξl ≤ 20%, the VSI could lead to acceptable results. On this basis, a
VSI, with those of the original structure, were considered as the seismic
moderate damping ratio of ξl = 10% was considered as an added
performance evaluation criteria. Referring to definitions of αm1 and αs1it
damping parameter in the present study.
can be stated that in all cases in which αm1 > αs1 the inner substructure
It should be noted that in the 3D VSI by viscous or viscoelastic links,
is the flexible one and the outer substructure is the stiff one, and vice
the damping matrix is of non-classically damped type and assumption
versa. For monitoring the displacement and rotation of the two flexible
of Rayleigh damping is no longer valid. In the obtained mass-asym-
and stiff substructures Eqs. (8) and (7) were used respectively for dis-
metric governing equation of motion, it is obvious that the torsional
placement and rotation response ratios.
response of the substructures is related to the features clθx and klθx of
〈Xmax 〉Stiff 〈Xmax 〉Flexible viscoelastic links. Therefore, it seems possible to control the torsional
RXS = RXF =
〈Xmax 〉Original 〈Xmax 〉Original (8) response of the 3D VSI by appropriate distribution of links’ damping in
the plan of the structure. In this study, since the ordinary modal time
〈θmax 〉Stiff 〈θ 〉 history analysis cannot be used due to the non-classical nature of the
R θS = RθF = max Flexible
〈θmax 〉Original 〈θmax 〉Original (9) damping, to solve numerically the governing coupled differential
equations of motion, a program was developed by the authors in
For a comprehensive study, different mass and stiffness ratios were MATLAB environment based on the fourth-order Runge-Kutta method.
assigned to either substructure of the 3D VSI, and the analyses were To reduce the order of governing differential equations the State Space
carried out for different parameters of the connecting links. This in- Vector technique was used in the developed program.
vestigation was mainly focused on the plan-wise distribution of links’ Regarding that the characteristics of input earthquakes are influ-
damping and stiffness values for torsional control. For the mass-asym- ential in the results of any response history analysis, so a wide range of
metric system, shown in Fig. 1, two parameters were considered for earthquakes was selected. The earthquakes were selected in such a way
damping and stiffness distribution of the longitudinal links as: that their peak spectral accelerations fall in various time periods. As
cl x1 kl x1 shown in Fig. 2, these records have been classified into two different
ud = , us = groups as the ones whose peak spectral accelerations lay in the period
cl x kl x (10)
range of less than 0.5 s, as short-period earthquakes, called Group-1
The values of ud = 0.5 and us = 0.5 are expressing the symmetric form now on, and those ones whose peak spectral accelerations lay in
damping and stiffness distribution in links in the X direction at either the period range of greater than 0.5 s, as long-period earthquakes,
side of the roof. The values other than 0.5 for ud and us result in called Group-2 from now on. All the selected earthquakes were scaled
asymmetric damping and stiffness of the links in the X direction, which with respect to peak ground acceleration to the value of 0.35 g, which is

Fig. 2. Spectral accelerations of the selected short-period earthquakes, (a), and long-period earthquakes, (b).

965
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

the design base acceleration in a very high seismic hazard zone ac- of the link parameters, and it is also recognized by the response ratio of
cording to the Iranian code of practice (Standard 2800). unity. In the lower triangle area of the graph, that is related to the stiff
substructure, response ratios are considerably less than unity almost in
4.2. Performance states of the 3D VSI all earthquakes and the majority of mass and stiffness ratios. So, it is
certain that the vertical isolation is effective in the stiff substructure,
For investigating the performance of 3D VSI a set of values from and the response ratio for this substructure is less dependent on the
0.05 to 0.95 with an increment of 0.05 was considered for the stiffness links’ parameters. In the upper triangle area of the graph which is re-
and mass ratios, αs1 and αm1. For each pair of these two ratios (total of lated to the flexible substructure, response ratios are more dependent
361 pairs) a series of THA were conducted by using six selected ac- on mass and stiffness ratios and also the links’ parameters. On this basis,
celerograms. Also in each case, three different values of 0.3, 0.4 and 0.5 from now on the majority of discussion on the 3D VIS is focused on the
were assigned to distribution parameters of links, in two states of vis- flexible substructure.
cous and viscoelastic. On this basis in total 361 × 6 × 6 = 12,996 According to the figures, three states can be differentiated in the
analysis cases were conducted. The output of these analysis cases for seismic behavior of the 3D VSI, which can be called ‘interactional’,
the two substructures, including the displacement and rotation ratios, ‘mass isolation’ and ‘control mass’. From the VICGs, it can be realized
defined by Eqs. (7) and (8), involves plenty of numerical results which that for the low and moderate link damping values, the area which
their presentation needs some special means. In a previous work of the corresponds to the response ratios less than unity is first formed in the
authors [1], a novel method of graphic presentation was suggested for middle band of the upper triangle parallel to the main diagonal. The
illustrating the large volume of results, called Vertical Isolation Con- trend of formation of the middle band with minimum response ratios
tour/Color Graph (VICG), which uses the Cartesian coordinate system indicates the proper performance of interactional state in response re-
as its skeleton. In this article, only the two response ratios, particularly duction of the 3D VSI. By increasing the interconnecting links’
rotational response ratios have been focused on. damping, the area with response ratios less than unity spreads toward
In the suggested VICG the horizontal and vertical axes of the the vertical and horizontal edges of the upper and lower triangle. As the
Cartesian coordinate system are considered respectively for presenting area corresponding to the minimum response ratios approaches the
the values of stiffness and mass ratio, respectively, and the color con- mentioned edges, mass isolation state evolves. Finally, there is a small
tour graphs demonstrate the seismic response or any other desired area of interactional states at the left lower corner of the upper triangle
output. In such graphs, the results corresponding to the stiff sub- (little values of αs1 and αm1), which could be interpreted as control mass
structure are in central symmetry with those of the flexible sub- state. Results show that by using the 3D VSI technique, displacement
structure, as the center point of the graph corresponds to αm = 0.5 and response reduction is achievable; up to 50% in flexible substructure and
αs = 0.5. To describe the main features of the introduced VICGs the also even more in stiff substructure.
results of displacement response control case of the 3D VSI, including
RXF and RXS , are presented for a case of the mass symmetric plan 4.3. Torsional control in the 3D VSI with viscous links
(ex = 0, e y=0) . In this case, the damping distribution parameters of links
was selected as ud = 0.5, so that the considered damping was dis- The torsional moments at floor levels of a 3D VSI are developed due
tributed equally between the dampers at either side of the roof, and to the eccentricities of mass centers with respect to stiffness centers of
therefore, no torsional response was expected in this case. In Figs. 3 and two substructures. In the present study, these eccentricities were as-
4 the displacement response ratios of isolated structures interconnected sumed to be e y = 20% in the original as well as the isolated structures.
by viscous links are demonstrated in the VICG presentations. With this specified eccentricity the system will have plan torsional ir-
For brevity, in the case of short-period earthquakes, the results are regularities according to most of the common codes of practice.
presented for a low damping value of ξl = 5% (Fig. 3) and in the case of Response calculations were conducted for three different damping
long-period earthquakes, the results are presented for a moderate distribution parameters i.e. ud = 0.3, 0.4 and 0.5. It is also noted that
damping value of ξl = 10% (Fig. 4). One of the main advantages of this ud = 0.5 is the symmetric damping distribution case and ud = 0.3 is the
presentation is creating the possibility of observing the results of both asymmetric case in which the center of mass coincides with the center
stiff and flexible isolated substructures in one individual graph. Regions of links’ damping.
in the VICG in which response ratio is less than unity (shown in light to It was mentioned in the previous section that in the VICGs, the re-
dark blue in the figures) represent the mass and stiffness ratios by which sults corresponding to the stiff substructure are in central symmetry
response reduction is achieved. with those of the flexible substructure, as the center point of the graph
By looking at the VICGs, one can get worthwhile insight into the 3D corresponds to αm = 0.5 and αs = 0.5. However, it must be noted that
VSI. In all of the presented graphs, the main diagonal that separates the this symmetry is true only in the case in which there is no eccentricity
results of flexible and stiff substructures from each other is independent of mass and stiffness in the original structure as well as both inner and

Fig. 3. Displacement response ratios of isolated structures interconnected by viscous links with ξl = 5% for the three earthquakes of Group-1.

966
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 4. Displacement response ratios of isolated structures interconnected by viscous links with ξl = 10% for the three earthquakes of Group-2.

Fig. 5. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in IFOS case subjected to RSN88_SFERN earthquake.

Fig. 6. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in IFOS case subjected to RSN165_IMPVALL earthquake.

outer substructures. In fact, in the defined mass-asymmetric 3D VSI, the main diagonal that separates the results of flexible and stiff sub-
although each pair of points corresponding to commutative mass and structures from each other is not recognized by the response ratio of
stiffness ratios are in central symmetry in the VICG, the property of unity. This fact can be clarified by paying attention to the explanation
torsional dynamic characteristics of two inner and outer substructures, given above on the asymmetry of the VICGs for the case of torsional
corresponding to that pair of points, is not commutative. So, the re- response ratios of the inner and outer substructures. Regions in the plot
sponse ratios of two substructures cannot be presented in one single area in which response ratio is less than unity represent the mass and
VICG. Actually, the case in which the inner and outer substructures are stiffness ratios by which rotational response is decreased in comparison
respectively the flexible and stiff ones (inner flexible and outer stiff - with the original structure. It is observed that, as expected, by assigning
IFOS case) is quite different from the case in which the inner sub- more damping to the dampers which are closer to the centers of mass
structure is stiff and the outer one is flexible (ISOF case). In Figs. 5–7, more torsional response reduction is achieved (ud = 0.3 compared to
rotational response ratios related to the IFOS case are presented in ud = 0.5). By comparing the VICGs in Figs. 5–7, it is evident that the
VICGs for only three ones of the six selected earthquakes for brevity. torsional response is not so much different from one input earthquake
It is observed in Figs. 5–7, contrary to displacement response ratios, to another. Going through the graphs, there are some common

967
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 7. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in IFOS case subjected to RSN1044_NORTH earthquake.

recognizable features better to be considered. For the close values of αs1 of damping the area corresponding to control mass is grown sig-
and αm1 (areas of the VICGs laying in the main diagonal), in which the nificantly by that one can use this concept more easily. In this idea, a
dynamic characteristics of the two inner and outer substructures are smaller portion of the building can act as the control mass for the other
almost the same, the response ratios are all more than unity. This im- larger portion. These VICGs have been assessed Quantitatively in Sec-
plies that applying close values of αs1 and αm1 leads to the improper tion 5.
performance of the 3D VSI in torsional buildings. Accordingly, using
distant values for αs1 and αm1 (areas far from the main diagonal of the
VICGs) is more appropriate for the 3D VSI in torsional buildings. 4.4. Torsional control in the 3D VSI with viscoelastic links
Figs. 8–10 demonstrate the response ratios in the mass asymmetric
3D VSI in the ISOF case. Torsional control in the mass asymmetric 3D VSI with viscoelastic
An overview of the VICGs, presented in Figs. 8–10 clarifies that the links is investigated in this section (Figs. 11–16). Similar to viscous
torsional control in the ISOF case is not as effective as the IFOS case. In links, the eccentricity of mass to the center of stiffness is assumed to be
the symmetric distribution of damping which was specified by ud = 0.5, e y = 20% in both original and isolated structures. The moderate prac-
and at the lower triangle area which relates to the inner or stiff sub- tical values of damping and stiffness ratios, which are ξl = 10% and
structure results, it is observed that response reduction and torsional βl = 0.05, have been introduced as viscoelastic links parameters in both
control are effective for the majority of mass and stiffness ratios. While, directions of the structures. Seismic response history analyses have
in the upper triangle which relates to the outer or flexible substructure, been conducted for three different links damping and stiffness dis-
response ratios vary in different regions and only in the limited area tribution parameters of ud = 0.3, 0.4, 0.5 and us = 0.3, 0.4, 0.5. Assess-
attributed to mass control the torsional response reduction is notice- ment of the torsional performance of the mass asymmetric 3D VSI has
able. been performed in two cases of ISOF and IFOS.
Looking at Figs. 8–10, one can also recognize that the asymmetric Observing the torsional control of the two inner and outer sub-
plan-wise distributions of damping (ud = 0.3 and ud = 0.4 ) have dif- structures in the presented VICGs indicates that the general trend in the
ferent influences on the responses of the two substructures. Torsional response of the mass asymmetric 3D VSI with viscoelastic links is si-
response reduction by appropriate plan-wise damping distribution has milar to the one with viscous links. However, it is not expected to have
been more effective in the flexible substructure than the stiff sub- more effective torsional control by using viscoelastic dampers. As it was
structure. According to the VICGs, the plan-wise damping distribution stated in the investigation of the isolated structures displacement re-
has led the area range with the desired level of response reduction sponse, large link stiffness values generally decrease the efficiency of
extends to the upper triangle, yet it has caused the desired area in the the linking dampers [1]. This issue is also valid in the mass asymmetric
lower triangle to be limited. Based on the results shown in Figures 3D VSI.
Figs. 8–10 it can be claimed that by plan-wise appropriate distribution

Fig. 8. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in ISOF case subjected to RSN88_SFERN earthquake.

968
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 9. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in ISOF case subjected to RSN165_IMPVALL earthquake.

Fig. 10. Rotational response ratios of isolated structures, interconnected by viscous links with ud values in ISOF case subjected to RSN1044_NORTH earthquake.

Fig.11. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in IFOS case subjected to RSN88_SFERN earthquake.

5. Quantitative investigation of the 3D VSI for practical purposes of this number only (361 − 19)/2 = 171 are the independent combi-
nations indeed.
Although the VICGs provide informative and useful insight into the A comparison of the number of desired isolation ratios combinations
behavior and response control of the 3D VSI qualitatively, for practical with the total number of considered ones is an appropriate criterion for
purposes the quantitative and numerical analysis of isolation behavior evaluating the efficiency of the 3D VSI. This criterion is called
is of great importance. Actually, response ratios in the VICGs are scat- Acceptance Ratio in this study and is defined as follows:
tered in ranges both larger and less than unity, thus, for every presented
VICG, the pair values of α s1 and α m1 were considered in such a way that 〈No. of desired isolation ratios combinations〉
Acceptance Ratio =
the response ratios would be less than unity in both substructures. Since 〈Total no. of independent combinations〉
in this study the values of these two ratios varied between 0.05 and (12)
0.95 with an increment of 0.05 (19 values for each ratio), the total
number of combination pairs of the two ratios was 19×19 = 361. Out Summary of the results related to the mass asymmetric 1-story 3D
VSI with viscous dampers, obtained from the conducted analyses for all

969
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 12. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in IFOS case subjected to RSN165_IMPVALLearthquake.

Fig. 13. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in IFOS case subjected to RSN1044_NORTH earthquake.

Fig. 14. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in ISOF case subjected to RSN88_SFERN earthquake.

selected earthquakes, are presented in Table 1 for the IFOS case of average can be decreased up to 20 percent in the flexible substructure
isolation and in Table 2 for ISOF case. in comparison to the symmetric distribution, while in the ISOF case this
These results include the average values of isolation ratios as well as response reduction is around 10 percent in average. However, for the
the mean and Standard Deviation values of the satisfactory response stiff structure, no significant torsional response reduction is achieved by
ratios in both stiff and flexible substructures for different link para- the plan-wise distribution of damping, even though the response values
meters. The average values of the response ratios, corresponding to the in stiff substructure are always satisfactory.
desired combination pairs is a good measure of seismic response re- By looking at AR values in the tables, it is evident that in most of the
duction. Moreover, the Standard Deviation of isolation ratios is an in- applied earthquakes the plan-wise distribution of damping and stiffness
dication of the VSI reliability, thereby, a larger value provides a wide has not led to significant torsional response reduction in comparison to
range of isolation ratios selection for the VSI. symmetric distribution, and the torsional control in the 3D VSI can be
According to presented results, it is observed that in the IFOS case considered effective even with a symmetric distribution of damping and
by the plan-wise distribution of damping, the torsional response in stiffness. This matter is one of the main features of the 3D VSI.

970
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Fig. 15. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in ISOF case subjected to RSN165_IMPVALL earthquake.

Fig. 16. Rotational response ratios of isolated structures, interconnected by viscoelastic links with ud values in ISOF case subjected to RSN1044_NORTH earthquake.

Table 1
Resulting Acceptance Ratio (AR) and mean and Standard Deviation values of the desired response ratios in IFOS case.
RSN88_SFERN RSN1197_CHICHI

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.31 0.70 0.58 0.45 0.84 0.3 mean 0.33 0.68 0.52 0.46 0.94
SD 0.21 0.21 0.13 0.24 SD 0.22 0.21 0.21 0.28
0.4 mean 0.32 0.70 0.64 0.42 0.83 0.4 mean 0.34 0.69 0.57 0.44 0.94
SD 0.21 0.20 0.11 0.24 SD 0.22 0.20 0.19 0.27
0.5 mean 0.32 0.71 0.73 0.41 0.83 0.5 mean 0.34 0.70 0.62 0.41 0.91
SD 0.21 0.19 0.11 0.24 SD 0.22 0.19 0.16 0.27

RSN160_IMPVALL RSN165_IMPVALL

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.32 0.70 0.49 0.48 0.84 0.3 mean 0.30 0.70 0.59 0.45 0.80
SD 0.21 0.20 0.21 0.27 SD 0.21 0.20 0.18 0.25
0.4 mean 0.32 0.71 0.54 0.46 0.82 0.4 mean 0.31 0.71 0.66 0.41 0.77
SD 0.22 0.19 0.21 0.27 SD 0.21 0.20 0.18 0.23
0.5 mean 0.33 0.73 0.61 0.43 0.79 0.5 mean 0.31 0.71 0.66 0.41 0.71
SD 0.22 0.17 0.20 0.26 SD 0.21 0.20 0.18 0.23

RSN1231_CHICHI RSN1044_NORTH

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.31 0.72 0.75 0.47 0.75 0.3 mean 0.30 0.71 0.61 0.43 0.77
SD 0.22 0.20 0.20 0.19 SD 0.21 0.20 0.17 0.24
0.4 mean 0.31 0.74 0.83 0.42 0.57 0.4 mean 0.30 0.72 0.69 0.39 0.74
SD 0.25 0.19 0.16 0.20 SD 0.22 0.20 0.14 0.23
0.5 mean 0.38 0.78 0.86 0.41 0.23 0.5 mean 0.31 0.75 0.79 0.33 0.67
SD 0.33 0.20 0.09 0.24 SD 0.22 0.17 0.10 0.20

971
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Table 2
Resulting Acceptance Ratio (AR) and mean and Standard Deviation values of the desired response ratios in ISOF case.
RSN88_SFERN RSN1197_CHICHI

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.18 0.53 0.77 0.64 0.40 0.3 mean 0.28 0.64 0.68 0.57 0.84
SD 0.11 0.19 0.18 0.22 SD 0.17 0.22 0.16 0.22
0.4 mean 0.20 0.53 0.81 0.63 0.36 0.4 mean 0.30 0.65 0.72 0.54 0.87
SD 0.16 0.21 0.15 0.22 SD 0.19 0.22 0.14 0.24
0.5 mean 0.23 0.53 0.85 0.62 0.30 0.5 mean 0.31 0.65 0.76 0.51 0.88
SD 0.22 0.24 0.12 0.21 SD 0.20 0.22 0.12 0.25

RSN160_IMPVALL RSN165_IMPVALL

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.28 0.65 0.71 0.61 0.85 0.3 mean 0.12 0.50 0.70 0.58 0.29
SD 0.18 0.22 0.19 0.24 SD 0.07 0.18 0.21 0.25
0.4 mean 0.29 0.66 0.75 0.57 0.87 0.4 mean 0.11 0.48 0.76 0.54 0.24
SD 0.19 0.22 0.16 0.25 SD 0.05 0.18 0.18 0.25
0.5 mean 0.30 0.66 0.79 0.54 0.88 0.5 mean 0.09 0.46 0.79 0.49 0.17
SD 0.20 0.22 0.14 0.27 SD 0.04 0.15 0.14 0.23

RSN1231_CHICHI RSN1044_NORTH

ud QTY αs1 αm1 RFlex. R Stiff AR ud QTY αs1 αm1 RFlex. R Stiff AR

0.3 mean 0.11 0.40 0.74 0.69 0.19 0.3 mean 0.18 0.55 0.75 0.62 0.30
SD 0.06 0.14 0.20 0.18 SD 0.16 0.21 0.21 0.24
0.4 mean 0.10 0.39 0.81 0.65 0.13 0.4 mean 0.21 0.55 0.81 0.60 0.27
SD 0.13 0.18 0.14 0.20 SD 0.21 0.24 0.17 0.24
0.5 mean 0.30 0.51 0.89 0.71 0.08 0.5 mean 0.25 0.58 0.83 0.53 0.17
SD 0.34 0.32 0.10 0.15 SD 0.26 0.27 0.13 0.20

Fig. 17. Partitioning of the 3D 5-story structure into inner and outer substructures using viscous link dampers.

Table 3 6. Application to a realistic multi-story torsional building with


Dynamic characteristics of the original and isolated substructures in the 5-story irregular plan
building model.
Mode/Story No. T (sec ) M (ton) M  (ton) K  (kN / m) αs αm In the realistic multi-story presented model, the 3D VSI concept was
investigated. As previously mentioned, in this study the efficiency of the
Original Structure 1.038 2536 1954 71,634 – – VSI technique is inspected in comparison to the original structure. So,
Outer Structure 0.96 1798 1383 58,704 0.82 0.71 the original steel moment frame structure has been designed according
Inner Structure 1.41 737 560 11,099 0.155 0.287
to AISC 360-10 provisions. It was assumed the considered building was
located in a very high seismic hazard zone with the design base ac-
celeration of 0.35 g according to the Iranian code of practice (Standard
2800). Gravity loadings were applied for a building with residential
occupancy. The box sections were considered for the column elements
and the I sections for beam elements as well.

972
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

Table 4
Response values of original and isolated substructures in the realistic 5-story 3D VSI model and the corresponding response ratios.
RSN88_SFERN

Original Inner Outer RFlex. RStiff

Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. Rot. Dis. Rot.

Story-5 7.09 0.0033 6.02 0.0026 5.90 0.0022 0.85 0.79 0.83 0.68
Story-4 5.76 0.0027 4.85 0.0021 4.86 0.0019 0.84 0.77 0.84 0.69
Story-3 4.10 0.0020 3.42 0.0015 3.51 0.0014 0.84 0.78 0.86 0.72
Story-2 2.45 0.0012 1.99 0.0009 2.12 0.0009 0.81 0.76 0.86 0.73
Story-1 0.88 0.0004 0.67 0.0003 0.75 0.0003 0.76 0.73 0.85 0.75

RSN160_IMPVALL

Original Inner Outer RFlex. RStiff

Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. Rot. Dis. Rot.

Story-5 13.21 0.0066 10.55 0.0043 8.66 0.0034 0.80 0.65 0.66 0.51
Story-4 11.01 0.0054 7.74 0.0028 7.33 0.0028 0.70 0.52 0.67 0.52
Story-3 8.08 0.0039 5.05 0.0028 5.46 0.0021 0.62 0.72 0.68 0.52
Story-2 4.81 0.0024 3.23 0.0021 3.42 0.0013 0.67 0.88 0.71 0.53
Story-1 1.60 0.0008 1.22 0.0008 1.31 0.0005 0.76 0.93 0.82 0.53

RSN1197_CHICHI-E

Original Inner Outer RFlex. RStiff

Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. Rot. Dis. Rot.

Story-5 ‘ 0.0127 15.17 0.0070 15.60 0.0078 0.72 0.55 0.74 0.61
Story-4 16.36 0.0105 12.50 0.0053 12.56 0.0064 0.76 0.51 0.77 0.61
Story-3 11.06 0.0079 8.37 0.0043 9.27 0.0050 0.76 0.54 0.84 0.63
Story-2 6.69 0.0050 4.94 0.0029 5.60 0.0032 0.74 0.57 0.84 0.64
Story-1 2.40 0.0018 1.78 0.0011 1.98 0.0012 0.74 0.58 0.83 0.64

RSN165_IMPVALL

Original Inner Outer RFlex. RStiff

Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. Rot. Dis. Rot.

Story-5 23.73 0.0171 18.40 0.0102 16.41 0.0095 0.78 0.59 0.69 0.56
Story-4 20.04 0.0146 15.21 0.0076 12.87 0.0080 0.76 0.52 0.64 0.55
Story-3 14.87 0.0108 10.82 0.0054 9.25 0.0060 0.73 0.50 0.62 0.55
Story-2 9.24 0.0067 6.19 0.0032 5.77 0.0038 0.67 0.48 0.62 0.56
Story-1 3.34 0.0024 1.98 0.0011 2.11 0.0014 0.59 0.45 0.63 0.57

RSN1231_CHICHI

Original Inner Outer RFlex. RStiff

Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. (cm) Rot. (rd) Dis. Rot. Dis. Rot.

Story-5 33.70 0.0296 25.96 0.0124 29.69 0.0176 0.77 0.42 0.88 0.60
Story-4 27.78 0.0245 20.48 0.0100 23.94 0.0145 0.74 0.41 0.86 0.59
Story-3 20.15 0.0178 13.93 0.0075 16.95 0.0106 0.69 0.42 0.84 0.60
Story-2 12.30 0.0108 7.91 0.0046 10.07 0.0065 0.64 0.43 0.82 0.60
Story-1 4.38 0.0039 2.75 0.0017 3.57 0.0023 0.63 0.43 0.82 0.60

In an irregular plan structure, the isolation planes are applied to separate cantilevers in both substructures. This approach provides a
partition the building into two inner and outer substructures. In this 3D situation for implementing the dampers in the beneath of the cut-off
VSI both of the substructures remain irregular. For this special model beams. From the practical point of view, it is necessary to consider
case, the properties of the frame elements are assumed to be unchanged sufficient space in the cut-off points for the movement of two sub-
after applying vertical isolation planes. Consequently, the inner and structures.
outer substructures include respectively the flexible and stiff sub- The dynamic characteristics of the 5-story building model and iso-
structures. Yet, in the preliminary practical design of the 3D VSI, it is lated substructures neglecting the interconnecting links are shown in
applicable to assign intentionally the main stiffness to the inner and Table 3. According to the calculated periods, the inner substructure is
outer substructures for appropriate performance according to the pre- the flexible one and the defined mass and stiffness ratios were obtained
sented VICGs. Providing vertical isolation between two newly designed as (αs1 = 0.155 and αm1 = 0.287 ). The isolation ratios imply that the in-
inner and outer adjacent structures and implementing links properly to teractional state is the dominant performance of the 3D VSI. According
the system interactively are indeed engineering issues. Various methods to previously conducted studies, employing the links with an inter-
can be raised for applying vertical isolation considering practical mat- mediate of practical damping values has led to the desired response
ters, here as shown in Fig. 17, the isolation planes were set in the mid of reduction in the interactional state of the VSI performance [1]. In the
the smallest spans of the plan, converting each cut-off beam into two multi-story model, similar to the method employed in the 1-story

973
R. Milanchian and M. Hosseini Structures 27 (2020) 962–974

model, estimation of the practical values of links damping and stiffness and stiffness, the torsional control in the 3D VSI can be considered
were taken into account in terms of effective modal mass, M ∗ and ef- effective.
fective modal stiffness, K ∗, in the first mode of the vibration. So, the - Large link stiffness generally decreases the damping effects of link
moderate practical modal damping value of about 10% of the original dampers. This issue is also valid in the plan-mass asymmetric 3D VSI
5-story model was calculated as ccr = 2M1∗ ω1, and thus as the torsional control has not been improved by the viscoelastic
c = 2 × 1954 × 6.054 × 0.1 = 2365.9(kNsec/m) . The specified damping dampers.
value distributed symmetrically among the implemented dampers in
each reference direction. Partitioning of the 3-D 5-story structure using Declaration of Competing Interest
viscoelastic link dampers has been demonstrated in Fig. 17.
Response history calculations for the selected earthquakes have The authors declare that they have no known competing financial
been conducted for the original and isolated structures. The maximum interests or personal relationships that could have appeared to influ-
story displacement and torsion in response history of each multi-story ence the work reported in this paper.
VSI system and that of the original structure together were the basis for
calculating the response ratio and evaluation of the structural perfor- Acknowledgment
mance. The results of response ratios for flexible and stiff substructures
in different stories are presented in Table 4. The PEER Ground Motion Database was very useful in selecting the
From the results in Table 4, it can be indicated that the 3D VSI desired ground motion, so the writers acknowledge the Pacific
decreases the torsion and displacement response of substructures ef- Earthquake Research Center for their valuable help to the academic
fectively in the irregular plan building. Torsional control up to 50 community.
percent in the stiff substructure and 30 percent in the flexible one could
be achieved. The results show that the response ratios in a realistic References
multi-story building conform with the response ratios obtained from the
VICGs in the corresponding isolation ratio of the multi-story model and [1] Milanchian R, Hosseini M, Nekooei M. Vertical isolation of a structure based on
so the VICGs could be used as a design tool for the preliminary design of different states of seismic performance. Earthquakes Struct 2017;13(2):103–18.
the VSI system. [2] Goel RK. Effects of supplemental viscous damping on seismic response of asym-
metric-plan systems. Earthquake Eng Struct Dyn 1998;27(2):125–41.
[3] Goel RK. Seismic behaviour of asymmetric buildings with supplemental damping.
7. Conclusion Earthquake Eng Struct Dyn 2000;29(4):461–80.
[4] Lin WH, Chopra AK. Understanding and predicting effects of supplemental viscous
damping on seismic response of asymmetric one-storey systems. Earthquake Eng
In this study, an ideal 3-dimensional plan-mass asymmetric 3D VSI Struct Dyn 2001;30(10):1475–94.
model was used to investigate the displacement and torsional control of [5] Lin WH, Chopra AK. Asymmetric one-storey elastic systems with non-linear viscous
the isolated structures. In the presented model, plan-wise distribution of and viscoelastic dampers: Earthquake response. Earthquake Eng Struct Dyn
2003;32(4):555–77.
visco-elastic dampers parameters was provided. Considering the ob-
[6] Kim J, Bang S. Optimum distribution of added viscoelastic dampers for mitigation of
tained equation of motion, it was differentiated whether the flexible torsional responses of plan-wise asymmetric structures. Eng Struct
structure is the inner substructure or the outer one. The response his- 2002;24(10):1257–69.
[7] García M, Juan C, Almazán JL. Torsional balance of plan asymmetric structures with
tory analyses conducted in different parameters and mass and stiffness
viscoelastic dampers. Eng Struct 2007;29(6):914–32.
ratios for several earthquakes and the VICGs implemented to explore [8] Klein R, Todaro A, Finne I. Investigation of a method to stabilize wind induced
the torsional behavior of the 3D VSI. Besides, the efficiency of the 3D oscillations in large structures. American Society of Mechanical Engineers; 1972.
VSI on the torsional control of multi-story buildings was examined in a [9] Xu Y, He Q, Ko J. Dynamic response of damper-connected adjacent buildings under
earthquake excitation. Eng Struct 1999;21(2):135–48.
5-story realistic model. On this basis, the following remarks can be [10] Zhang W, Xu Y. Dynamic characteristics and seismic response of adjacent buildings
stated as the conclusions of this study: linked by discrete dampers. Earthquake Eng Struct Dyn 1999;28(10):1163–85.
[11] Zhang W, Xu Y. Vibration analysis of two buildings linked by Maxwell model-de-
fined fluid dampers. J Sound Vib 2000;233(5):775–96.
- In the 3D VSI, it is practical to select a wide variety of mass and [12] Zhu H, Xu Y. Optimum parameters of Maxwell model-defined dampers used to link
stiffness ratios with different link damping values. By using larger adjacent structures. J Sound Vib 2005;279(1):253–74.
values of link's damping, the sensitivity of isolation ratios to input [13] Patel C, Jangid R. Seismic response of dynamically similar adjacent structures
connected with viscous dampers. IES J Part A: Civ Struct Eng 2010;3(1):1–13.
earthquakes characteristics decreases. Applying this technique, dis- [14] Sharma A, Jangid R. Seismic response of base-isolated benchmark building with
placement response reduction up to 50% in flexible substructure and variable sliding isolators. J Earthquake Eng 2010;14(7):1063–91.
even more in stiff substructure is achievable. [15] Bharti S, Dumne S, Shrimali M. Seismic response analysis of adjacent buildings
connected with MR dampers. Eng Struct 2010;32(8):2122–33.
- The 3D VSI can be generally considered an effective isolation [16] Zhu H, Ge D, Huang X. Optimum connecting dampers to reduce the seismic re-
technique for torsional response control as well as displacement sponses of parallel structures. J Sound Vib 2011;330(9):1931–49.
response control. The interaction between two inner and outer [17] Richardson A, Walsh KK, Abdullah MM. Closed-form equations for coupling linear
structures using stiffness and damping elements. Struct Control Health Monit
connected substructures has been led to satisfactory performance.
2013;20(3):259–81.
- In the first case that the inner substructure is flexible one (IFOS [18] Nigdeli SM, Bekdas G. Optimum tuned mass damper approaches for adjacent
case), torsional control is nearly effective in both viscous and vis- structures. Earthquakes Struct 2014;7(6):1071–91.
coelastic link dampers and almost in the majority of isolation ratios. [19] Ziyaeifar M. Method of mass isolation in seismic design of structures. Proceeding of
the 12th World Conference on Earthquake Engineering. Paper. 2000.
The torsional response could be controlled up to 50 percent in the [20] Ziyaeifar M, Gidfar S, Nekooei M. A model for mass isolation study in seismic design
stiff substructure and 30 percent in the flexible one. of structures. Struct Control Health Monit 2012;19(6):627–45.
- In the second case that the outer substructure is flexible one (ISOF [21] Behnamfar F, et al. A method for rapid estimation of dynamic coupling and spectral
responses of connected adjacent structures. Struct Des Tall Spec Build
case), torsional control is nearly effective in the stiff substructure 2016;25(13):605–25.
while the response in flexible substructure is satisfactory just in a [22] Zhao X, Bao L, Xue B. An energy-dissipating approach for adjacent retrofitted
limited range of isolation ratios. structures. Struct Des Tall Spec Build 2018;27(15):e1508.
[23] Milanchian R, Hosseini M. Study of vertical seismic isolation technique with non-
- The plan-wise distribution of links damping and stiffness has not led linear viscous dampers for lateral response reduction. J Build Eng 2019.
to significant torsional response reduction in comparison to sym- [24] Lee D, Taylor DP. Viscous damper development and future trends. Struct Des Tall
metric one and even with the symmetric distribution of damping Build 2001;10(5):311–20.

974

You might also like