Swing Story–Lateral Force Resisting System Connected

with Dampers: Novel Seismic Vibration Control

System for Building Structures
Liang-Jiu Jia, A.M.ASCE 1; Ping Xiang 2; Minger Wu 3; and Akira Nishitani 4
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This study proposes a vibration control system consisting of a swing story subsystem, a lateral force resisting subsystem, and
connecting dampers. The swing story subsystem comprising hinged columns connected to continuous beams is employed to resist the gravity
load, and perimeter moment-resisting frames the lateral force. Dampers are adopted to transfer lateral forces between the two subsystems and
are used to dissipate seismic energy. The main characteristic of the system is that both the fundamental and higher-mode responses can be
reduced remarkably compared with conventional moment-resisting frame structures. This paper establishes equations of motion of the pro-
posed swing story control system, and the dampers are optimally designed by a multiobjective optimization algorithm with the combination
of the stability maximization and H2 criteria. To avoid instability of the flexible swing stories, sufficient stiffness is required for the dampers,
and buckling analysis is thus conducted to determine the required minimum stiffness. Excellent vibration control effects of the proposed
system are verified through numerical analyses of three-story and nine-story benchmark structures. DOI: 10.1061/(ASCE)EM.1943-
7889.0001390. © 2017 American Society of Civil Engineers.
Author keywords: Seismic vibration control; Swing story control system; Multiobjective optimum design; Uncoupled gravity load-lateral
force resisting system; Buckling analysis.

Introduction constraints at bottom of the walls (Pan et al. 2015; Qu et al. 2012),
and the pin-supported wall can reduce drift concentration at a
During the last two decades, how to conduct response control has specific story effectively (Pan et al. 2015). For tall structures, it
become an important issue in seismic design of building structures. is necessary to use posttensioned or prestressed walls to achieve
Structural response control aims at not only earthquake resistance im- more-uniform drift ratios (Zibaei and Mokari 2014). Another sys-
provement, but also comfortability enhancement. An ideal seismic tem in which walls are connected to a gravity load resisting system
vibration control system should return to its original position after an (GLRS) with energy dissipators recently has been proposed. Differ-
earthquake with minor or no damage to its main structural elements. ent from rocking wall structures, the walls in the system are can-
Earthquake-induced damage can be mitigated by restricting plastic tilever reinforced-concrete shear walls which function as a lateral
deformation to specific energy dissipation components, which can be force resisting system (LFRS), and which are connected to the floor
checked and replaced after an earthquake where necessary (Skinner diaphragm with anchorage connectors where relative motion be-
et al. 1974). Recently, a number of seismic vibration control systems tween GLRS and LFRS is allowable (Chancellor et al. 2014;
involving movable subsystems and structural fuses have been pro- Tsampras et al. 2016; Zhang et al. 2014). In this system, floor ac-
posed (Murase et al. 2013; Pan et al. 2015; Shinozaki et al. 2008; celerations, the LFRS story shears, and the forces transferred from
Taniguchi et al. 2016; Tsampras et al. 2016; Wiebe and Christopoulos the floor system to the LFRS are all reduced compared with the
2009; Xiang and Nishitani 2014, 2015a, b, 2016; Xiang et al. 2017). corresponding seismic structure. Both of the aforementioned sys-
A rocking structure is one of these systems. A frame pin- tems are found to be effective in reduction of higher mode effects.
supported wall structure is a type of rocking structure that releases In addition, a hybrid passive control system was proposed and
preliminarily studied (Kasagi et al. 2016; Murase et al. 2013;
1 Taniguchi et al. 2016), in which a base-isolated or multiisolated
Assistant Professor, Research Institute of Structural Engineering and
(base-isolation and middle-isolation) building is connected to a
Disaster Reduction, College of Civil Engineering, Tongji Univ., No.
1239 Siping Rd., Shanghai 200092, China. E-mail: lj_jia@tongji.edu.cn stiffer free wall with oil dampers. It was found that a larger ratio
2
Assistant Professor, Dept. of Structural Engineering, College of Civil of the fundamental natural periods between the main structure and
Engineering, Tongji Univ., No. 1239 Siping Rd., Shanghai 200092, China the free wall made the system more effective, and dampers were
(corresponding author). E-mail: p.xiang@tongji.edu.cn more effective when located near the isolation story. The system
3
Professor, Dept. of Structural Engineering, College of Civil Engineer- can remarkably mitigate response at the fundamental frequency.
ing, Tongji Univ., No. 1239 Siping Rd., Shanghai 200092, China. E-mail: However, the aforementioned seismic vibration control systems
wuminger@tongji.edu.cn cannot effectively reduce both the fundamental and higher-mode
4
Professor, Dept. of Architecture, Faculty of Science and Engineering, responses simultaneously.
Waseda Univ., 3-4-1 Okubo, Tokyo 1698555, Japan. E-mail: anix@
This paper proposes an innovative seismic vibration control
waseda.jp
Note. This manuscript was submitted on March 6, 2017; approved on
system termed a swing story–lateral force resisting system. This
July 20, 2017; published online on November 22, 2017. Discussion system comprises a swing story gravity load resisting system
period open until April 22, 2018; separate discussions must be submitted (SS-GLRS), a LFRS, and dampers connecting the two systems,
for individual papers. This paper is part of the Journal of Engineering which can reduce both fundamental and higher-mode responses.
Mechanics, © ASCE, ISSN 0733-9399. In this system, the SS-GLRS and LFRS are respectively expected

© ASCE 04017159-1 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Roof

3rd

LFRS Floor slab

2nd

Continuous beam
Damper
1st

Ground
LFRS SS-GLRS Damper Hinged columns
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

(a) (b)

Fig. 1. Sketches of the proposed system: (a) plan view; (b) elevation view

to sustain vertical and horizontal forces independently, and dampers system is schematically illustrated in Fig. 1, in which the SS-GLRS
transfer the horizontal forces and provide damping and energy dis- consists of hinged columns and continuous beams. Fig. 2 is the
sipation functions. The SS-GLRS uses a beam-through frame with analytical half model of the structure in one direction, in which
hinged-column ends, in which the floors are swing movable. This is the SS-GLRS is simplified as a multilumped mass model compris-
different from the GLRS in the literature (Tsampras et al. 2016), in ing hinged columns, i.e., truss elements. Dampers between the
which the GLRS is modeled as a frame with rigid connections. The SS-GLRS and the LFRS are represented by the Voigt model.
SS-GLRS can be designed in an economic way because of reduced The deformed configuration of the system is also displayed in the
bending moment demand and nonuse of isolation bearings. Perim- figure by dashed lines.
eter frames as well as cantilever shear walls can be used as the For an m-bay n-story LFRS in two dimensions, equations of
LFRS; this study investigates only the former. The SS-GLRS and motion of the LFRS can be expressed as
LFRS are coupled by connectors employing passive, active, or
semiactive control strategies (Christenson et al. 2007). Passive MLFRS ẍLFRS þ CLFRS ẋLFRS þ KLFRS xLFRS ¼ −MLFRS rLFRS ẍg
dampers providing both stiffness and damping properties are used ð1Þ
to connect the SS-GLRS and LFRS in this study, and the stiffness
and damping parameters are optimally designed by using a multi- where MLFRS , CLFRS , and KLFRS = mass, damping, and stiffness
objective optimization method based on the H2 and stability matrices of the LFRS, respectively; xLFRS ¼ fu1 ;u2 ; :::;unðmþ1Þ ;
maximization criteria (Xiang and Nishitani 2015a, b). Numerical v1 ;v2 ; :::;vnðmþ1Þ ;φ1 ;φ2 ; :::;φnðmþ1Þ gT , where ui , vi , and φi
simulation results of two benchmark buildings designed for [I ¼ 1; 2; : : : ; nðm þ 1Þ] denote the nodal horizontal and vertical
the SAC project—a joint venture of three non-profit organizations: translational displacements and the rotational angles, respectively;
The structural Engineers Association of California (SEAOC), the
and rLFRS ¼ f1; 1; : : : ; 1; 0; 0; : : : ; 0gT .
Applied Technology Council (ATC), and California Universities |fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflfflfflffl{zfflfflfflfflfflffl}
for Research in Earthquake Engineering (CUREE)—in Los Angeles nðmþ1Þ 2nðmþ1Þ

indicate that the proposed system has great potential to achieve If KLFRS is denoted as
satisfactory and innovative vibration control performance. 2 3
Kuu Kuv Kuφ
6 7
KLFRS ¼ 6
4 Kvu Kvv Kvφ 7
5 ð2Þ
Proposed Swing Story–Lateral Force Resisting
Kφu Kφv Kφφ
Frame System

This paper focuses on the proposed system involving perimeter lat- then Eq. (1) can be equivalently replaced by the following equation
eral force resisting frames other than shear walls as LFRS. The in which only the nodal horizontal translational displacements are
retained:
LFRS kf3 mf3MLS-GLRS Mcond ü þ Ccond u̇ þ Kcond u ¼ −Mcond rẍg ð3Þ
cf3
where u ¼ fu1 ; u2 ; : : : ; unðmþ1Þ gT , r ¼ f1; 1; : : : ; 1gT ; and Mcond ,
H3

kf2 mf2 Kcond , and Ccond are the condensed mass, stiffness, and damping
matrices, respectively, and they are obtained by using the Guyan
cf2 reduction method (Genta 2009)
H2

kf1
mf1
Kcond ¼ Kuu þ Kuv Tuv þ Kuφ Tuφ ð4Þ
cf1
Mcond ¼ ½ I Tuv Tuφ  · MLFRS · ½ I Tuv Tuφ T ð5Þ

B1 B2 B3 Ccond ¼ ½ I Tuv Tuφ  · CLFRS · ½ I Tuv Tuφ T ð6Þ

Fig. 2. Schematic illustration of analytical model Tuv ¼ ðKvv − Kvφ K−1 −1 −1

φφ Kφv Þ ðKvφ Kφφ Kφu − Kvu Þ ð7Þ

© ASCE 04017159-2 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Csystem = mass, stiffness, and damping matrices, respectively, which
Tuφ ¼ −K−1 −1 −1
φφ Kφu − Kφφ Kφv ðKvv − Kvφ Kφφ Kφv Þ
−1
can be expressed as
× ðKvφ K−1
φφ Kφu − Kvu Þ ð8Þ  
Mcond
Msystem ¼ ð10Þ
Based on the Euler–Lagrange equations (Goldstein et al. 2001), diagð½mf1 ; mf2 ; : : : ; mfn Þ
equations of motion of the system can be written as
where mfi (i ¼ 1; 2; : : : ; n) represents the mass of the ith story
Msystem ẍ þ Csystem ẋ þ Ksystem x ¼ −Msystem r̄ẍg ð9Þ
 
K11 K12
where x ¼ fu1 ; u2 ; : : : ; unðmþ1Þ ; uf1 ; uf2 ; : : : ; ufn gT , where ufi Ksystem ¼ ð11Þ
K21 K22
(i ¼ 1; 2; : : : ; n) denotes the horizontal translational displacement
of the ith floor; r̄ ¼ f1; 1; : : : ; 1gT ; and Msystem , Ksystem , and where
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

" #T !
K11 ¼ Kcond þ diag kf1 =nc ; 0; : : : ; 0; kf1 =nc ; : : : ; kfn =nc ; 0; : : : ; 0; kfn =nc ð12Þ
|fflfflfflffl{zfflfflfflffl} |fflfflfflffl{zfflfflfflffl}
m−1 m−1

2 3
−kf1 =nc 0|fflffl{zfflffl}
:::0 −kf1 =nc 0 0|fflffl{zfflffl}
:::0
6 7
6 m−1 ðmþ1Þ·ðn−1Þ−1 7
6 7
6 0|fflffl{zfflffl}
:::0 −kf2 =nc 0|fflffl{zfflffl}
:::0 −kf2 =nc 0|fflffl{zfflffl}
:::0 7
6 7
6 7
6 mþ1 m−1 ðmþ1Þ·ðn−2Þ 7
KT12 ¼ K21 ¼6 7 ð13Þ
6 .. .. .. .. .. 7
6 . . . . . 7
6 7
6 7
6 0|fflffl{zfflffl}
:::0 0 −kfn =nc 0|fflffl{zfflffl}
:::0 −kfn =nc 7
4 5
ðmþ1Þ·ðn−1Þ−1 m−1

2 Pn Pn Pn 3
i¼1 mfi g i¼2 mfi g i¼2 mfi g
6 kf1 − H1
−
H2 H2 7
6 Pn Pn Pn 7
6 7
6 mfi g i¼2 mfi g mfi g .. 7
6 i¼2
kf2 − − i¼3 . 7
6 H2 H2 H3 7
K22 ¼6
6
7
7 ð14Þ
6 .. .. mfn g 7
6 . . 7
6 Hn 7
6 7
4 mfn g mfn g 5
kfn −
Hn Hn

where kfi (i ¼ 1; 2; : : : ; n) is the linear stiffness of the connecting dampers of the ith floor (Fig. 2); nc = number of the connections in one
story, which equals 2 in this study (Fig. 1); and g = acceleration of gravity
 
C11 C12
Csystem ¼ ð15Þ
C21 C22

where h iT 
C11 ¼ Ccond þ diag cf1 =nc ; 0; : : : ; 0; cf1 =nc ; : : : ; cfn =nc ; 0; : : : ; 0; cfn =nc ð16Þ
|fflfflfflffl{zfflfflfflffl} |fflfflfflffl{zfflfflfflffl}
m−1 m−1

2 −c =n 0|fflffl{zfflffl}
:::0 −cf1 =nc 0 0|fflffl{zfflffl}
:::0 3
f1 c
6 m−1 ðmþ1Þ·ðn−1Þ−1 7
6 7
6 0:::0 −cf2 =nc 0|fflffl{zfflffl}
:::0 −cf2 =nc 0|fflffl{zfflffl}
:::0 7
6 |fflffl{zfflffl} 7
6 7
6 mþ1 m−1 ðmþ1Þ·ðn−2Þ 7
CT12 ¼ C21 ¼6 7 ð17Þ
6 .. .. .. .. .. 7
6 . . . . . 7
6 7
6 7
4 0:::0 0 −cfn =nc 0|fflffl{zfflffl}
:::0 −cfn =nc 5
|fflffl{zfflffl}
ðmþ1Þ·ðn−1Þ−1 m−1

C22 ¼ diagð½ cf1 ; cf2 ; :::; cfn T Þ ð18Þ

where cfi (i ¼ 1; 2; : : : ; n) is the linear damping coefficient of the connecting dampers of the ith floor (Fig. 2).

© ASCE 04017159-3 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Optimum Design The multiobjective optimization genetic algorithm nondomi-
nated sorting genetic algorithm II (NSGA-II) (Deb et al. 2002)
Optimum design variables in the system are the stiffness and was used to solve the multiobjective optimization problem with
damping coefficients of the connecting dampers, i.e., kfi and cfi the combination of the H2 criterion and the SMC to achieve favor-
(i ¼ 1; 2; : : : ; n). The flexibly supported columns are apt to buckle able robustness of structural control (Faravelli and Casciati 2008).
if the stiffness of the connecting dampers is not large enough. To
prevent instability in the SS-GLRS, the critical buckling load of the
SS-GLRS should be larger than the combined effect of gravity Numerical Study
loads and inertial forces, which corresponds to the minimum re-
quired stiffness coefficients, kfi (i ¼ 1; 2; : : : ; n). Herein, the struc- Benchmark Buildings
tural dynamic stability problem was simplified as a static problem
by introducing Ai-distribution inertia forces as the horizontal forces Two typical steel structures, three-story and -story buildings
exerted on the floor levels, and beneficial damping forces provided designed by Brandow and Johnston Associates (Los Angeles,
California) for the SAC Phase II Steel Project for the Los Angeles,
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

by the connecting dampers were ignored in the buckling analysis.

The minimum stiffness values were selected as the lower bound of California region (Spencer and Dyke 1998) were used to evaluate
kfi (i ¼ 1; 2; : : : ; n). The axial stiffness of the beams in the corre- the seismic performance of the proposed control system. The two
sponding conventional structures connecting between the perimeter buildings are both steel moment-resisting frame structures typical
and core frames were determined as the upper bound of kfi in the United States, comprising a few seismic frames and several
(i ¼ 1; 2; : : : ; n). Damping ratios ζ i were limited to the range of gravity frames, and the gravity frames include continuous columns
[0, 1] in the optimization process, and cfi ¼ 2mf1 ωs1 ζ i , where pin-connected to beams (Elkady and Lignos 2015; Tagawa et al.
2010). These benchmark structures were selected because steel
ωs1 denotes the fundamental circular frequency of the LFRS.
perimeter moment-resisting frames (MRFs) in these buildings resist
This study used the H2 and stability maximization criteria
lateral loads and thus small cross sections are required for beams
(Xiang and Nishitani 2015a, b) for the optimum design of the sys-
and columns of the inner GLRS, to which the proposed structural
tem. Damage of nonstructural components is closely related to
system is similar. The two buildings were chosen also because they
the magnitudes of interstory drifts of the SS-GLRS, and thus
serve as benchmark structures for the SAC studies and will thus
the objective function based on the H2 criterion was set as follows:
provide a wider basis for the comparison results. In the numerical
J H2 ¼ Max:ðkHGLRSðdriftÞ1 ðjωÞk2 ; study, the two structures were modified such that the original con-
tinuous columns were replaced by hinged columns and the beams
kHGLRSðdriftÞ2 ðjωÞk2 ; : : : ; kHGLRSðdriftÞn ðjωÞk2 Þ ð19Þ were replaced by continuous beams. The three-story structure is
36.58 m by 54.87 m in plan, with a bay width each of 9.15 m
where kHGLRSðdriftÞi ðjωÞk2 (i ¼ 1; 2; : : : ; n) denotes the 2-norm in both directions, and 11.89 m in elevation, with a floor-to-floor
value of the frequency response function (FRF) magnitude of height of 3.96 m. The nine-story structure is 45.73 m by 45.73 m in
the ith interstory drift. Because the 2-norm values of the FRFs plan, with a bay width each of 9.15 m in both directions, and
of interstory drifts of every story are involved in the preceding 37.19 m in elevation; the first floor height is 5.49 m and the sec-
objective function, interstory drifts of every story need to be ond–ninth floors are each 3.96 m high. The levels of the buildings
controlled. are numbered with respect to the ground level. Two-dimensional
On the other hand, the stability maximization criterion (SMC) models in the weak north–south (N-S) direction were numerically
aims at making the free vibration response decay in the minimum analyzed, and the basement in the nine-story building was
duration. The optimization can be achieved when the degree of sta- neglected in this study. Fig. 3 shows analytical models of the
bility, Λ, is maximized. In other words, the eigenvalue with three-story building; those of the nine-story building are in the lit-
the smallest absolute value of the real parts should be located as erature (FEMA 2000; Spencer and Dyke 1998).
far as possible from the imaginary axis in the left-half complex
plane (Nishihara and Matsuhisa 1997; Xiang and Nishitani Buckling Analysis
2015a, b, 2016).
According to the SMC, the objective function is defined as To prevent instability in the flexible SS-GLRS, buckling analysis
was required to obtain the minimum stiffness coefficients of the
J SMC ¼ Max:ðReðλ1 Þ; Reðλ2 Þ; : : : ; Reðλnðmþ2Þ ÞÞ ð20Þ connecting dampers. Analytical solutions were obtained by solving
a linearized buckling analysis in which only floor gravities were
where λi ½i ¼ 1; 2; : : : ; nðm þ 2Þ are the ith eigenvalues of the considered and damping forces provided by the dampers were
system. ignored. The linearized buckling analysis problem can be solved

Fig. 3. Three-story building N-S elevation: (a) benchmark building N-S MRF; (b) analytical model for the proposed system

© ASCE 04017159-4 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 as a linear eigenvalue problem. The buckling criterion is that the distribution of the buildings. Four cases were investigated in which
stiffness matrix is singular, so that displacements are indeterminate. the standard shear coefficient C0 was set to 0, 0.2, 1.0, and 2.0,
If Ksystem is singular, then DetðKsystem Þ ¼ 0 should be satisfied. By respectively. The obtained equivalent static forces in Ai -distribution
solving DetðKsystem Þ ¼ 0, the required minimum stiffness can be were exerted on the mass points at the floor levels in the SS-GLRS,
obtained, which is the maximum value among the solutions. These and buckling analyses were carried out using ABAQUS 6.9. Truss
solutions underestimate the required minimum stiffness because the elements were adopted for the columns and the dampers. First, ini-
inertial forces that actually exist are not considered. tial values were set to the cross-sectional areas of the dampers, and
This study thus adopted a pseudostatic analysis method to solve the buckling eigenvalue of the first mode was obtained. Then the
the dynamic stability problem, in which the time varying inertia cross-sectional areas of the dampers were reduced/magnified to
forces were replaced by equivalent static forces applied at the floor their original values multiplied by the eigenvalue, and an eigen-
levels. The shear coefficient distribution (Ai -distribution) in the value buckling analysis was reconducted. These two steps were
seismic design code of Japan (Building Center of Japan 1997) repeated until the buckling eigenvalue of the first mode finally
was adopted for specifying vertical distribution of seismic loads, reached 1. The obtained axial stiffness of the dampers was the mini-
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

in which C0 and Ai are two factors concerned with the Ai - mum required stiffness. Fig. 4 shows the first buckling modes of the
distribution; C0 is the standard shear coefficient, which is set to three-story and nine-story structures in the final step when the fun-
0.2 and 1.0 for moderate and severe earthquake ground motions, damental eigenvalues reached 1. Figs. 4(a and b) show the results
respectively, and Ai is the lateral shear distribution factor, which when the cross-sectional areas of the dampers along the elevation
is determined by the fundamental natural period and the weight were equal. The axial stiffness of the dampers for the upper four
stories in the nine-story structure was surplus because no bucking
occurred there [Fig. 4(b)]. Accordingly, the ratio of the cross-
sectional area ofPthe dampersP at the kth floor to that of the first
floor was set to 9i¼k ðmfi gÞ= 9i¼1 ðmfi gÞ in the buckling analysis
for the nine-story structure [Fig. 4(c)]. Fig. 5 shows the minimum
required stiffness values obtained by the aforementioned methods.
The horizontal axis represents the standard shear coefficient C0 ,
and the vertical axis displays the magnitudes of the required mini-
mum stiffness. The data points corresponding to the horizontal axis
of 0 are the results of the case in which only the gravity loads were
considered; the analytical result was very close to the numerical
result. The minimum required stiffness slightly increased as
the horizontal force increased. The required minimum stiffness

Fig. 4. Buckling shapes of the first mode: (a) three-story (with the same
cross-sectional area of dampers); (b) nine-story (with the same cross-
sectional area of dampers); (c) nine-story (with proportional cross- Fig. 5. Relationship between required minimum stiffness and horizon-
sectional areas of dampers) tal forces: (a) three-story; (b) nine-story

© ASCE 04017159-5 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 corresponding to C0 of 2 was set as the lower bound of the design same as the solutions of the corresponding single-objective optimi-
variables kfi (i ¼ 1; 2; : : : ; n) in the optimum design. zation problems. That is to say, the square symbols correspond to
the optimal solutions for the H2 criterion, and the circle symbols on
the Pareto fronts are the optimal solutions for the SMC criterion.
Optimum Design Results and Frequency Response
The triangle symbols on the Pareto fronts are the trade-off optimal
Functions
solutions. Table 1 lists the optimum design variables corresponding
The multiobjective algorithm NSGA-II (Deb et al. 2002) was used to the symbols in the figure. Figs. 7(a–c) show the FRF magnitudes
for the optimum design, as mentioned in section “Proposed Swing from the ground acceleration to the interstory drifts, the absolute
Story–Lateral Force Resisting Frame System.” In the algorithm, the accelerations of the GLRS, and the interstory drifts of the LFRS,
probabilities of crossover and mutation were set to 0.8 and 0.5, respectively, for the three-story structure. In the figures, the dark
respectively. The population size and the number of generations lines represent the conventional steel moment-resisting frame struc-
were set to 100 and 500, respectively. The number of objective ture in which the lateral stiffness of the GLRS is ignored and the
functions and the number of decision variables were both two in
this study. For the three-story structure, the design variables kfi
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

(i ¼ 1; 2; : : : ; n) were equal and constrained to the range of

[8.59 × 106 , 5.58 × 108 ] (N=m). For the nine-story structure,
the design variables kfi (i ¼ 2; 3; : : : ; n) were set to proportional
to kf1 , which was constrained to the range of [4.20 × 107 ,
4.46 × 108 ] (N=m). The design variables cfi (i ¼ 1; 2; : : : ; n) were
set equal along the elevation for both the three-story and the nine-
story structures, and constrained to the range of ½0; 2mf1 ωs1 
(Ns=m). The multiobjective optimization problem was solved using
a computer program developed in MATLAB.
The obtained Pareto-optimal solutions, i.e., the finally selected
population size (N p ) individuals by a crowded-comparison-
operator, composed Pareto fronts (Fig. 6). The square and circle
symbols in the figure denote the boundary solutions on the Pareto
fronts, which own infinite crowding distances, and they are the

0.7
Pareto-optimal solutions
Optimal solution for JH
0.65 2
Optimal solution for JSMC
0.6 Trade-off optimal solution
2
JH

0.55

0.5

0.45
-1.4 -1.2 -1 -0.8 -0.6 -0.4
(a) JSMC

1.04
Pareto-optimal solutions
Optimal solution for JH
1.02 2
Optimal solution for JSMC
1 Trade-off optimal solution
2
JH

0.98

0.96

0.94
-0.198-0.196-0.194-0.192 -0.19 -0.188 Fig. 7. Magnitudes of frequency response functions of the three-story
(b) JSMC structural systems from ground accelerations to (a) interstory drifts of
GLRS; (b) absolute accelerations of GLRS; (c) interstory drifts of
Fig. 6. Pareto-optimal solutions: (a) three-story; (b) nine-story LFRS; ωs1 is the natural frequency of the LFRS

Table 1. Optimum Values of Stiffness and Damping Coefficients

kfi (×107 N=m) cfi (×106 Ns=m)
Structure Story H2 SMC Trade-off H2 SMC Trade-off
Three-story First–third 3.96 1.25 2.08 8.60 4.22 6.23
Nine-story First 4.45 4.20 4.20 3.44 3.12 3.34
P9 P9 P9
kth (i ¼ 2; : : : ; 9) ðm gÞ ðm fi gÞ ðm fi gÞ
P9i¼k fi × 4.45 Pi¼k
9
× 4.20 Pi¼k
9
× 4.20
i¼1 ðmfi gÞ i¼1 ðm fi gÞ i¼1 ðm fi gÞ

© ASCE 04017159-6 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Table 2. DBE Ground Motion Ensemble
EQ number Event Magnitude Year Component Site type Pulse type PGVa (cm=s) Scale factor
1 Chi-Chi Taiwan 7.6 1999 TCU045-E Far-field — 50.06 1.68
2 Chi-Chi Taiwan 7.6 1999 TCU067-E Near-field No pulse-like 92.01 1.08
3 Northridge-01 6.7 1994 LOS270 Far-field — 41.11 1.63
4 Northridge-01 6.7 1994 SYL360 Near-field Pulse-like 129.31 0.90
5 Landers 7.3 1992 CLW-TR Far-field — 43.40 1.85
6 Landers 7.3 1992 LCN260 Near-field Pulse-like 133.33 1.05
7 Loma Prieta 6.9 1989 CAP000 Far-field — 38.01 1.98
8 Loma Prieta 6.9 1989 CLS000 Near-field No pulse-like 55.94 1.57
a
For original earthquake ground motion records.
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

seismic weight of the GLRS is distributed to the LFRS, the light

lines indicate the case of the proposed structural system optimally
designed by the SMC, and the medium lines and the thick lines
respectively represent the H2 design criterion and the trade-off
solutions. Fig. 7 shows that the FRF magnitudes at the modal
frequencies were all mitigated remarkably in the proposed system
compared with the conventional system. The proposed system de-
signed by the SMC achieved the best performance for absolute
accelerations of the GLRS and interstory drifts of the LFRS. The
system using the trade-off optimal solutions provided compromise
performance between the systems based on the SMC and the H2
design criteria. There was a negative effect at frequencies lower Fig. 8. Spectral acceleration for DBE ground motion ensemble (5%
than the first modal frequency, especially for interstory drifts of damping ratio)
the first story in GLRS, indicating that relatively large interstory
drift of the first story in GLRS will occur under long-period earth-
quake ground motions. The relatively large displacement of the first
story may be further reduced by employing a larger amount of
damping in connections of the first story, which will be studied in
the future, instead of using the same amount of damping along the
elevation of the building as in the present study. For the nine-story
structure, there was a minor difference between the FRF magni-
tudes of the proposed system designed by the obtained three groups
of optimal parameters, and the magnitudes at the fundamental and
higher modal frequencies were also all mitigated significantly in the
proposed system compared with the conventional structure.

Seismic Performance Evaluation

This study subjected the benchmark buildings to unidirectional
earthquake ground motions along the horizontal direction. Eight
ground motions selected from the FEMA P-695 (FEMA 2009) far-
field and near-field sets and downloaded from the PEER NGA
(2016) database were used as input excitations in the numerical
simulations. The selected ground motions covered three different
kinds of earthquakes: short-period random phase type, mid-long–
period pulse type, and long-period fling-step type. This study also
used the famous destructive earthquakes in Landers (1992) and Fig. 9. Comparison of numerical results for response time histories of
Chi-Chi (1999), which are characterized by the pulse type, abun- the proposed three-story structure under EQ #1: (a) interstory drift dis-
dant long-period components, a high ratio of peak ground velocity placements of top story of GLRS; (b) absolute accelerations of top floor
to peak ground acceleration, and large permanent ground displace- of GLRS
ment. The eight ground-motion records were scaled to match their
acceleration response spectra to the ASCE7-10 (ASCE 2010)
design basis earthquake (DBE) spectrum for a site in Los Angeles,
California (latitude=longitude ¼ 34.04°N=118.25°W). Table 2 Fig. 9 compares the response time histories of the proposed system
gives the details of the DBE ground motions used. Fig. 8 depicts obtained by MATLAB and ABAQUS; they agree well with each
the target DBE spectrum and the acceleration response spectra of other, which verifies the accuracy of the numerical model and
the scaled DBE ensemble and their median spectrum. The structure the analysis method. Fig. 10 shows the numerical results of time
was postulated to remain within the elastic domain in the numerical history analyses for the three-story structure under simulated
analyses. ground motion EQ #1. The interstory drifts and floor accelerations
A MATLAB-based program was developed for dynamic analy- were mitigated significantly compared with the conventional sys-
ses, and finite-element analysis software ABAQUS was also used. tem both for the GLRS and the LFRS. Fig. 10(a) reveals that the

© ASCE 04017159-7 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 10. Numerical results of time history analyses of the four systems under EQ #1: (a) maximum interstory drift angles of GLRS; (b) maximum
absolute accelerations of GLRS; (c) maximum interstory drift angles of LFRS; (d) acceleration time history of top floor of GLRS

H2 criterion gave the most uniform distribution of the maximum centimeters. The maximum responses of the proposed nine-story
interstory drift angles among the structures designed according to structural system under LCN260 were larger than or close to those
the three groups of optimization results, which coincides with the of the conventional system, because the fundamental period of the
optimization objective. Fig. 10(c) shows that the interstory drifts of proposed system was close to the predominant period of LCN260
the central span and the side span of the LFRS were slightly differ-
ent due to lack of the rigid floor effect (Casciati and Faravelli 2014).
Fig. 10(d) verifies that the vibration decayed the most rapidly in the
system designed by the SMC. The system designed by the trade-off
solutions achieved a compromise control performance, which is
more suitable to be adopted in practical design. Fig. 11 compares
the energy response time histories of the proposed system designed
by the trade-off solutions and those of the conventional system
under EQ #1. The figure shows the input energy E, total damping
energy DE, kinetic energy K E, total potential energy PE, damping
energy at the dampers DEcon , and potential energy at the
dampers PEcon . The input energy of the proposed system was
smaller than that of the conventional system, and a large portion
of the input energy was consumed by the connecting dampers
in the proposed system. Figs. 12 and 13 respectively present the
maximum responses of the three-story and nine-story structures
under each ground motion over the height of the structures. The
maximum interstory drifts of the upper stories of the GLRS were
mitigated, whereas those of the first story of the GLRS were close
to or larger than those of the conventional system under some
excitations, especially for the nine-story structure. The maximum
values of the floor absolute accelerations of the GLRS and the in-
terstory drifts of the LFRS were reduced in every story compared
with the conventional system, except for the nine-story structure
under LCN260. Figs. 12(d) and 13(d) show the maximum damping
forces required for the dampers. Figs. 12(e) and 13(e) show the
Fig. 11. Energy response of the systems under EQ #1: (a) the conven-
maximum relative displacements between the GLRS and the LFRS,
tional system; (b) the proposed system designed by the trade-off solutions
which increased with the height increment and were only tens of

© ASCE 04017159-8 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 12. Maximum response quantities at each story/floor of the three-story structure under ground motion ensemble: (a) maximum interstory drift
angles of GLRS; (b) maximum absolute accelerations of GLRS; (c) maximum interstory drift angles of LFRS; (d) maximum damping forces at
connections; e) maximum relative displacements at connections; dark symbols denote the conventional system; light symbols denote the proposed
system; hollow symbols denote far-field cases; solid symbols denote near-field cases; triangles correspond to Chi-Chi, diamonds correspond to
Northridge, squares correspond to Landers, and circles correspond to Loma Prieta

and the input energy was much larger compared with the conven- story subsystem includes hinged columns connected to continuous
tional system (Fig. 14). For the three-story structure, although the beams, which are different from typical steel moment-resisting
proposed system did not show a significant benefit concerning frame structures in the United States. The dampers between the
the input energy, most of the input energy was dissipated by gravity load resisting and lateral force resisting frames were used
the connecting dampers in the proposed system (Fig. 11). Accord- to provide both stiffness and damping properties. Based on
ingly, responses of the system were much smaller than those of numerical investigations, the main findings can be summarized
the conventional system, and both the fundamental and higher- as follows:
mode responses were reduced remarkably (Fig. 7). Fig. 15 com- 1. Numerical results of the developed MATLAB program and the
pares the maximum and root-mean square (RMS) values of the commercial software ABAQUS were in good agreement, de-
responses of the nine-story structure under LCN260. Although monstrating the validity of the established analytical model
the interstory drifts of the GLRS in the proposed system were and simulation methodology in this study;
larger than those of the conventional system, the maximum values 2. For the proposed swing story vibration control system, the same
of floor acceleration and interstory drift of the LFRS were close stiffness for the dampers along the elevation can be adopted for
to or slightly larger than those of the conventional system and low-rise structures, whereas different stiffnesses for the dampers
their RMS values were still mitigated considerably in the pro- along the elevation are recommended for high-rise structures;
posed system. 3. Multiobjective optimum design was conducted for the proposed
system, and trade-off solutions can be adopted as the values of
the design parameters to achieve compromise performance;
Conclusions 4. The proposed swing story vibration control system can reduce
both fundamental and higher-mode effects;
This paper proposed a new seismic vibration control system com- 5. Through numerical simulations of three-story and nine-story
prising swing story gravity load resisting frames and perimeter benchmark structures, the proposed system was demonstrated
lateral force resisting frames connected with dampers. The swing to have a favorable vibration control effect for reducing both

© ASCE 04017159-9 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

Fig. 13. Maximum response quantities at each story/floor of the nine-story structure under ground motion ensemble: (a) maximum interstory drift
angles of GLRS; (b) maximum absolute accelerations of GLRS; (c) maximum interstory drift angles of LFRS; (d) maximum damping forces at
connections; (e) maximum relative displacements at connections; dark symbols denote the conventional system; light symbols denote the proposed
system; hollow symbols denote far-field cases; solid symbols denote near-field cases; triangles correspond to Chi-Chi, diamonds correspond to
Northridge, squares correspond to Landers, and circles correspond to Loma Prieta

interstory drift and floor absolute acceleration under different

6 T of
1 T of proposed
1
TCU045E earthquake ground motions, compared with the conventional
T of T of proposed TCU067E
conventional 3-story 1 1 steel moment-resisting frame structures; and
5 3-story structure conventional 9-story LOS270
6. In the worst case, in which the predominant period of an earth-
Energy Spectrum [m/s]

structure 9-story structure SYL360

structure CLW-TR quake ground motion is close to the fundamental period of
4 LCN260 the proposed system, although the interstory drift angles of
CAP000
CLS000
the swing story subsystem may be larger than those of the
3 conventional system, the maximum values of floor acceleration
and interstory drift of the lateral force resisting frames were
2 slightly larger than or close to those of the conventional sys-
tem and their root-mean square values were still considerably
1 mitigated.
From the viewpoint of the lifecycle cost of a building structure,
0 high economical efficiency is expected for the proposed system due
0 1 2 3 4 5 to reduced cross-sectional areas of the members in the gravity
Natural period [s] frames and a favorable seismic vibration control effect. Detailed
analysis of the economic efficiency will be carried out in future
Fig. 14. Energy spectra of eight excitations (5% damping ratio)
studies. Seismic performance of the proposed swing story vibration

© ASCE 04017159-10 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Max. response of conventional system Deb, K., Pratap, A., Agarwal, S., and Meyarivan, T. (2002). “A fast and
Max. response of proposed system
RMS response of conventional system elitist multiobjective genetic algorithm: NSGA-II.” IEEE Trans. Evol.
RMS response of proposed system Comput., 6(2), 182–197.
Elkady, A., and Lignos, D. G. (2015). “Effect of gravity framing on the
9 overstrength and collapse capacity of steel frame buildings with perim-
8
eter special moment frames.” Earthquake Eng. Struct. Dyn., 44(8),
LFRS LFRS 1289–1307.
7 Faravelli, L., and Casciati, S. (2008). “Structural control toward structural
6
GLRS GLRS robustness.” 14th World Conf. on Earthquake Engineering, International
Association for Earthquake Engineering, Tokyo.
Story

5 FEMA. (2000). “State of the art report on systems performance of steel

4
moment frames subject to earthquake ground shaking.” FEMA Rep.
355C, Washington, DC.
3 FEMA. (2009). “Quantification of building seismic performance factors.”
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

2
FEMA P695, Washington, DC.
Genta, G. (2009). Vibration dynamics and control, Springer, New Delhi, India.
1 Goldstein, H., Poole, C. P., Jr., and Safko, J. L. (2001). Classical mechan-
0 0.01 0.02 0.03 0.04
(a) Inter-story drift Ang. (rad)
ics, 3rd Ed., Addison-Wesley, San Francisco.
Kasagi, M., Fujita, K., Tsuji, M., and Takewaki, I. (2016). “Automatic gen-
9 eration of smart earthquake-resistant building system: Hybrid system of
8 base-isolation and building-connection.” Heliyon, 2(2), e00069.
MATLAB [Computer software]. MathWorks, Natick, MA.
7
Murase, M., Tsuji, M., and Takewaki, I. (2013). “Smart passive control of
6 buildings with higher redundancy and robustness using base-isolation
and inter-connection.” Earthquake Struct., 4(6), 649–670.
Floor

5
4
Nishihara, O., and Matsuhisa, H. (1997). “Design optimization of passive
gyroscopic damper (stability degree maximization).” JSME Int. J. Ser. C,
3 40(4), 643–651.
2 Pan, P., Wu, S., and Nie, X. (2015). “A distributed parameter model of a
1
frame pin-supported wall structure.” Earthquake Eng. Struct. Dyn.,
44(10), 1643–1659.
0 PEER NGA (Pacific Earthquake Engineering Research Center Next
0 2 4 6 8
2 Generation Attenuation models). (2016). “PEER ground motion data-
(b) Acc. of GLRS (m/s )
base.” 〈http://peer.berkeley.edu/nga〉 (Dec. 18, 2016).
Qu, Z., Wada, A., Motoyui, S., Sakata, H., and Kishiki, S. (2012). “Pin-
Fig. 15. Maximum and root-mean square values of the responses of the
supported walls for enhancing the seismic performance of building
nine-story structural systems under EQ #6 (LCN260): (a) interstory
structures.” Earthquake Eng. Struct. Dyn., 41(14), 2075–2091.
drift angles of GLRS and LFRS; (b) absolute accelerations of GLRS Shinozaki, Y., Komuro, T., Fujino, H., Kawamoto, S. I., and Hosozawa, O.
(2008). “Development of response control system with intensive energy
absorption devices and application to building.” AIJ J. Technol. Des.,
14(28), 453–458.
control system should be further investigated through experimental Skinner, R. I., Kelly, J. M., and Heine, A. J. (1974). “Hysteretic dampers for
studies. earthquake-resistant structures.” Earthquake Eng. Struct. Dyn., 3(3),
287–296.
Spencer, B. F., and Dyke, S. J. (1998). “Next generation benchmark control
Acknowledgments problems for seismically excited buildings.” Proc., 2nd World Conf. on
Structural Control, Vol. 2, European Association for the Control of
The study is supported by National Nature Science Foundation of Structures, Pavia, Italy, 1135–1360.
China (51608391 and 51508401), which is greatly appreciated. Tagawa, H., Macrae, G., and Lowes, L. (2010). “Continuous columns
effects of gravity columns in U.S. steel moment-resisting frame struc-
tures: Continuous column effects in steel moment frames in perspective
of dynamic stability.” J. Struct. Constr. Eng., 75(650), 761–770
References (in Japanese).
Taniguchi, M., Fujita, K., Tsuji, M., and Takewaki, I. (2016). “Hybrid con-
ABAQUS [Computer software]. Dassault Systèmes, Waltham, MA. trol system for greater resilience using multiple isolation and building
ASCE. (2010). “Minimum design loads for buildings and other structures.” connection.” Front. Built Environ., 2(Oct), 1–10.
ASCE/SEI7-10, Reston, VA. Tsampras, G., et al. (2016). “Development of deformable connection for
Building Center of Japan. (1997). “BCJ Seismic provisions for design of earthquake-resistant buildings to reduce floor accelerations and force
building structures.” Tokyo (in Japanese). responses.” Earthquake Eng. Struct. Dyn., 45(9), 1473–1494.
Casciati, S., and Faravelli, L. (2014). “Quantity vs. quality in the model Wiebe, L., and Christopoulos, C. (2009). “Mitigation of higher mode
order reduction (MOR) of a linear system.” Smart Struct. Syst., effects in base-rocking systems by using multiple rocking sections.”
13(1), 99–109. J. Earthquake Eng., 13(S1), 83–108.
Chancellor, N. B., Eatherton, M. R., Roke, D. A., and Akbas, T. (2014). Xiang, P., and Nishitani, A. (2014). “Seismic vibration control of building
“Self-centering seismic lateral force resisting systems: High perfor- structures with multiple tuned mass damper floors integrated.” Earth-
mance structures for the city of tomorrow.” Buildings, 4(3), 520–548. quake Eng. Struct. Dyn., 43(6), 909–925.
Christenson, R. E., Johnson, E. A., and Spencer, B. F. (2007). “Semiactive Xiang, P., and Nishitani, A. (2015a). “Optimum design and application of
connected control method for adjacent multidegree-of-freedom non-traditional tuned mass damper toward seismic response control
buildings.” J. Eng. Mech., 10.1061/(ASCE)0733-9399(2007)133:3(290), with experimental test verification.” Earthquake Eng. Struct. Dyn.,
290–298. 44(13), 2199–2220.

© ASCE 04017159-11 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159

 Xiang, P., and Nishitani, A. (2015b). “Optimum design of tuned mass pendulum-type nontraditional tuned mass damper.” Struct. Control
damper floor system integrated into bending-shear type building based Health Monit., 1–19.
on H∞, H2, and stability maximization criteria.” Struct. Control Health Zhang, D., et al. (2014). “Development of a floor inertia force limiting
Monit., 22(6), 919–938. anchorage system under earthquake loading.” Proc., 10th National
Xiang, P., and Nishitani, A. (2016). “Structural vibration control with the Conf. on Earthquake Engineering, Earthquake Engineering Research
implementation of a pendulum-type nontraditional tuned mass damper Institute, Oakland, CA.
system.” J. Vib. Control, 1–19. Zibaei, H., and Mokari, J. (2014). “Evaluation of seismic behavior
Xiang, P., Nishitani, A., and Wu, M. (2017). “Seismic vibration and dam- improvement in RC MRFs retrofitted by controlled rocking wall
age control of high-rise structures with the implementation of a systems.” Struct. Des. Tall Spec. Build., 23(13), 995–1006.
Downloaded from ascelibrary.org by Imperial College London on 03/20/18. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04017159-12 J. Eng. Mech.

J. Eng. Mech., 2018, 144(2): 04017159