(2015) Modal Analysis of A Linked Cantilever Flexible Building System - Online - IMPRIMR

You might also like

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

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/283024945

Modal Analysis of a Linked Cantilever Flexible


Building System

Article in Journal of Structural Engineering · October 2015


Impact Factor: 1.5 · DOI: 10.1061/(ASCE)ST.1943-541X.0001250

CITATION READS

1 42

2 authors:

K.T. Tse Song Jie


The Hong Kong University of Science and Tec… The Hong Kong University of Science and Tec…
48 PUBLICATIONS 206 CITATIONS 9 PUBLICATIONS 13 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Song Jie
letting you access and read them immediately. Retrieved on: 30 June 2016
Modal Analysis of a Linked Cantilever
Flexible Building System
K. T. Tse 1 and Jie Song 2
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

Abstract: This study proposes a modal analysis for a simplified model of a linked building system, i.e., a system consisting of two adjacent
tall buildings connected through links such as skybridges or skygardens. The simplified model was developed by modeling each building as a
cantilever beam and allowing for the effects of varying link mass, axial stiffness, bending stiffness, and link location. The analytical solutions
for the modal properties of the simplified model were then derived by solving the characteristic equations in the related boundary value
problem. After validating the accuracy of the analytical solution by comparing it with the associated finite-element model (FEM), the effects
of the link parameters on the modal properties, and the wind-induced responses of the system were examined by using these analytical
solutions. Meaningful results were obtained and discussed. The analytical solution and results can be used for rapidly evaluating the modal
properties of the linked building systems and for optimally designing the link during the preliminary design stage. DOI: 10.1061/(ASCE)ST
.1943-541X.0001250. © 2015 American Society of Civil Engineers.
Author keywords: Analytical solution; Modal property; Linked building system; Structural coupling; Structural linkage; Analysis and
computation.

Introduction and the application of the codes of practice. To this end, it is worth-
while to use a simplified analytical model to investigate the dynamic
High-risebuildings located near each other are increasingly being modal properties of linked building systems.
designed as linked building systems, i.e., as systems consisting of Although to the best of the authors’ knowledge, few related
buildings connected horizontally through links, such as skybridges studies have been conducted for the analytical modeling of linked
and skygardens. The Petronas Twin Towers in Kuala Lumpur, building systems in which the link is a structural element, such as a
Malaysia, and the Marina Bay Sands Hotel in Singapore are two skybridge or a skygarden, some other simplified analytical models
examples of this trend. In addition to serving as an additional passage have been proposed for a coupled system consisting of adjacent
and an evacuation route in the event of a fire or other emergency (Lee buildings connected by using supplemental damping devices
et al. 2010), from a structural point of view, the link for connecting (Christenson et al. 2006; Bhaskararao and Jangid 2007; Cimellaro
adjacent high-rise buildings could reduce the maximal structural re- and Lopez-Garcia 2011; Sandoval et al. 2012; Dym 2012). Such a
sponse and the possibility of pounding between adjacent buildings coupled system is essentially similar to the linked building system.
caused by strong seismic and wind excitations because the coupling Accordingly, reviewing these models for the coupled system is ad-
effect of the link tends to synchronize the vibration of the buildings vantageous. A widely used model of the coupled system simplifies
(Westermo 1989; Lim et al. 2011). To evaluate the dynamic response each building in the system to a lumped mass, i.e., as a single-
of the linked building system, the modal properties of the system degree-of-freedom (SDOF) system. For example, for two buildings
must be determined because they form the foundations for dynamic connected by using control devices, Zhu et al. (2001) modeled each
response analysis. The FEM is highly convenient in considering building as a SDOF system for deriving the optimal parameters of
structural details and is thus a practical approach for determining the control device for a coupled system subjected to white noise
the modal properties of tall buildings. However, modeling tall build- ground excitation. In addition, the SDOF model was extensively
ings by implementing a simplified analytical model, such as a con- used for investigating the seismic performance of the coupled sys-
tinuum beam model, is equally welcome and useful in structural tem in analytical form (Iwanami et al. 1996; Bhaskararao and
design, particularly at the initial design stage (Gurley et al. 1994; Jangid 2006, 2007; Kim et al. 2006; Sandoval et al. 2012). By using
Li et al. 1994; Iwanami et al. 1996; Miranda 1999). This is not only a similar simplified model, Richardson et al. (2013) derived closed-
because a simplified model is timesaving and easily modified, but form solutions for the displacement response of a coupled system
also because it can provide general principles for structural behavior subject to seismic excitation, and determined the optimal design
in mathematical form, which is of great benefit for the initial design of the supplemental damping device. In most of these studies,
the supplemental damping device that connected the two adjacent
1
Assistant Professor, Dept. of Civil and Environmental Engineering, buildings was usually simulated as a spring-dashpot system con-
Hong Kong Univ. of Science and Technology, Hong Kong, P.R. China. nected to the lumped mass of each building. The lumped mass
E-mail: timkttse@ust.hk of each building was usually assumed to be concentrated at one
2
Ph.D. Candidate, Dept. of Civil and Environmental Engineering, specific location along the building height (e.g., the top of the
Hong Kong Univ. of Science and Technology, Hong Kong, P.R. China
building); in other words, the link was situated precisely at the lo-
(corresponding author). E-mail: jsongaa@ust.hk
Note. This manuscript was submitted on April 7, 2014; approved on
cation of the lumped building mass of each building. Consequently,
December 9, 2014; published online on January 27, 2015. Discussion per- the model cannot be applied directly to realistic linked building
iod open until June 27, 2015; separate discussions must be submitted for systems in which the link could be designed for any arbitrary
individual papers. This paper is part of the Journal of Structural Engineer- location or in which two buildings are connected through several
ing, © ASCE, ISSN 0733-9445/04015008(12)/$25.00. separate links over the building height.

© ASCE 04015008-1 J. Struct. Eng.

J. Struct. Eng.
In addition to the mentioned SDOF model, some other studies
have adopted the continuum beam model for simulating tall C D
buildings to obtain analytical solutions for the coupled system. kb
Compared with the SDOF model, the continuum beam model
can reflect more structural properties of the system. Furthermore, ka m k a
the beam model can be used to address the system in which the link L

is placed at any location or even the case of two adjacent buildings A B H


Tower Tower uL
connected by several control devices at different locations. For in- A B m h
m
stance, by modeling each building as a cantilever beam, Gurley EI EI
et al. (1994) derived closed-form solutions for the transfer function
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

and response of a system consisting of two buildings connected z z


by using a spring-dashpot system. On the basis of the closed-form
solution, parametric analysis was conducted to minimize the wind- u A,uC uB,u D
induced response. Luco and De Barros (1998) then extended the
model and applied it to a case of two buildings connected by using
viscous dampers uniformly distributed over the building height. In Fig. 1. Sketch of a linked building system and the simplified linked
addition, Miranda (1999) and Miranda and Reyes (2002) simplified beam system
a single multistory building as a model consisting of a combination
of a flexural cantilever beam and a shear cantilever beam connected
through axially rigid links to obtain analytical solutions for the Simplified Model of a Linked Building System
maximal lateral deformation of the building subjected to ground
motions. The mentioned continuum beam models successfully re-
Assumptions
solved the particular problem in the research and established a
sound foundation for studying the analytical solutions of linked Fig. 1 shows an example of a typical linked building system and
building systems. However, additional refinements must be intro- the related simplified continuum beam model. Two uniform and
duced to these models to enable them to be applied directly to the identical cantilever beams are adopted in modeling the twin build-
linked building system. First, the link mass is usually neglected in ings (i.e., Towers A and B). Because tall buildings are usually
these models. Although this might be acceptable for the mass of the dominated by flexible deformations, the two cantilever beams
damper that connects two buildings because of its light weight, the are assumed to be sufficiently slender so that the shear deformation
link mass in the linked building system (e.g., skybridges or skygar- and rotary inertia can be neglected. Consequently, each building in
dens) is usually considerable and cannot be disregarded, especially the system can be modeled as a single uniform Euler-Bernoulli
when the link span is large and the link extends over several stories. beam. Accordingly, the subbeams that are divided by the link
In addition, the aforementioned models consider only the axial (i.e., Subsections A, B, C, and D shown in Fig. 1), can also be
stiffness of the link because the damper typically does not provide modeled as Euler-Bernoulli beams. The axial deformation of tall
substantial bending stiffness. By contrast, the bending stiffness of buildings is usually substantially smaller than the bending deflec-
the link in linked building systems could be significant. Therefore, tion so that the axial deformation of the two cantilever beams can be
the simplified analytical model for the linked building system neglected. To clearly show the external structural coupling caused
should consider the effects of mass, axial stiffness, and bending by the link, the internal structural coupling caused by the eccentric-
stiffness of the link. Furthermore, except for the continuum beam ity between the mass center and the elastic center of the cross sec-
model in Miranda (1999) and Miranda Taghavi (2005), each build- tion of each cantilever beam is eliminated by assuming that the
ing in most of the other continuum beam models is usually modeled mass center and the elastic center are coincident with the geometric
as a shear cantilever beam to represent low-rise frame structures center of the cross section. For simplicity, the authors focused on
and buildings. In view of the fact that buildings in linked building only the bending deflection in directions uA and uB , as shown in
systems are substantial, tall, and flexible, it is more beneficial to Fig. 1. According to the degree of freedom of the beam (i.e., the
model the building in the system as a cantilever Euler-Bernoulli transverse deformation), the link is modeled as a lumped mass mL
beam (Yang et al. 2004; Chen et al. 2010). connected to the two towers through two translational springs with
The objectives of this study are (1) to derive the closed-form axial stiffness ka in the horizontal plane and a rotational spring with
solutions for the modal properties of a simplified analytical model bending stiffness kb in the vertical plane, as shown in Fig. 1. The
for a linked building system, allowing for the effects of varying link damping of the link is ignored by assuming that the link is rigidly
mass, axial stiffness, vertical bending stiffness, and link location; connected to the two towers, and its damping is relatively insignifi-
and (2) to investigate the effect of the links on the modal properties cant compared with the damping of two towers.
of the system. A simplified analytical model for the linked building
system is first presented. The free-vibration problem of the simpli-
Governing Equations of the Problem
fied model is then formulated mathematically as a boundary value
problem. Closed-form solutions for the modal properties are de- The free-vibration equation of motion of each subbeam is
rived by solving the characteristic equations. After validating the expressed as (Rao 2007)
accuracy of the closed-form solutions by comparing it with the re-
∂ 4 ui ðz; tÞ ∂ 2 ui ðz; tÞ
lated FEM model, the effects of the link parameters on the modal EI 4
þm ¼0 ð1Þ
properties of the system are examined comprehensively. The ana- ∂z ∂t2
lytical solutions are then applied to an example of a linked building where E and I = Young’s modulus and the moment of inertia of the
system to show the effects of the link on wind-induced responses. cross section of the cantilever beam, respectively; m = mass per unit
Regarding the application, the results and findings can be used for length of the cantilever beam; and ui ðz; tÞði ¼ A; B; C, and D) =
the preliminary design of linked building systems and for a rapid transverse deflection of the subbeam i at height z and instant time
evaluation of the modal properties of linked building systems. t. For the link, the equation of motion can be expressed as

© ASCE 04015008-2 J. Struct. Eng.

J. Struct. Eng.
mL üL ðh; tÞ þ ka ½uL ðh; tÞ − uA ðh; tÞ − ka ½uB ðh; tÞ − uL ðh; tÞ ¼ 0 Let XðtÞ ¼ expðjwtÞ (j is imaginary unit), and substituting
ð2Þ Eqs. (16) and (17) into Eqs. (1) and (2) yields
φiv 4
i ðzÞ − ai φðzÞ ¼ 0 ð18Þ
where uL = horizontal displacement of the link; and h = link
elevation. −mL ω2 φL þ ka ½φL − φA ðhÞ − ka ½φB ðhÞ − φL  ¼ 0 ð19Þ
The boundary and the continuity conditions of the linked beam
system are given as where

uA ð0; tÞ ¼ uA0 ð0; tÞ ¼ 0 ð3Þ mω2


a4 ¼ ð20Þ
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

EI
uB ð0; tÞ ¼ uB0 ð0; tÞ ¼ 0 ð4Þ The general solution for Eq. (18) is

M C ðH; tÞ ¼ M D ðH; tÞ ¼ 0 ð5Þ φi ðzÞ ¼ Ci;1 cosðazÞ þ Ci;2 sinðazÞ


þ Ci;3 coshðazÞ þ Ci;4 sinhðazÞ ð21Þ
V C ðH; tÞ ¼ V D ðH; tÞ ¼ 0 ð6Þ
where Ci;j ði ¼ A; : : : ; D; j ¼ 1; : : : ; 4Þ are the arbitrary integra-
tion constants to be evaluated from the boundary, continuity,
uA ðh; tÞ ¼ uC ðh; tÞ ð7Þ and force equilibrium conditions. By using Eq. (21) and substitut-
ing Eqs. (16) and (17) into Eqs. (3)–(10) and Eqs. (12)–(15), 16
uA0 ðh; tÞ ¼ uC0 ðh; tÞ ð8Þ equations can be obtained with respect to the aforementioned 16
integration constants Ci;j , as shown in Appendix I.
uB ðh; tÞ ¼ uD ðh; tÞ ð9Þ Substituting Eq. (21) into the equation of motion of the link
Eq. (19) yields
uB0 ðh; tÞ ¼ uD0 ðh; tÞ ð10Þ CA;1 cos λh β þ CA;2 sin λh β þ CA;3 cosh λh β þ CA;4 sinh λh β

where the prime (′) denotes the partial derivative with respect þ CB;1 cos λh β þ CB;2 sin λh β þ CB;3 cosh λh β þ CB;4 sinh λh β
 
to height z; H = height of each cantilever beam; and M and V = λm 4
bending moment and shearing force of the cross section, respec- þ β − 2 φL ¼ 0 ð22Þ
λka
tively, which are calculated by
where
∂2u ∂M ∂3u
M ¼ −EI V¼ ¼ −EI 3 ð11Þ mL h
∂z2 ∂z ∂z β ¼ aH λm ¼ λh ¼
mH H
In addition, the shear force and moment equilibrium at the ka H 3 kb H
two joints between the link and beams (shown in Fig. 2) are λka ¼ λkb ¼ ð23Þ
EI EI
expressed as
where λm ; λh ; λka , and λkb = ratios of the mass, location, axial stiff-
V A ðh; tÞ − V C ðh; tÞ − ka ½uL − uA ðh; tÞ ¼ 0 ð12Þ ness, and bending stiffness of the link to the associated parameter of
each cantilever beam, respectively.
V B ðh; tÞ − V D ðh; tÞ þ ka ½uB ðh; tÞ − uL  ¼ 0 ð13Þ

M A ðh; tÞ − kb ½4uA0 ðh; tÞ þ 2uB0 ðh; tÞ − M C ðh; tÞ ¼ 0 ð14Þ Natural Frequencies and Mode Shapes

M B ðh; tÞ − kb ½2uA0 ðh; tÞ þ 4uB0 ðh; tÞ − M D ðh; tÞ ¼ 0 ð15Þ Solution Scheme
As can be seen from the foregoing derivation, there are 17 homog-
The general solutions for Eqs. (1) and (2), subjected to the enous linear equations in Appendix I and Eq. (22), with respect
conditions in Eqs. (3)–(10) and Eqs. (12)–(15), can be obtained to 17 constants fCA;1 ; CA;2 ; CA;3 ; CA;4 ; CB;1 ; CB;2 ; CB;3 ; CB;4 ; CC;1 ;
by separation of variables CC;2 ; CC;3 ; CC;4 ; CD;1 ; CD;2 ; CD;3 ; CD;4 ; φL gT , which can be simpli-
ui ðz; tÞ ¼ φi ðzÞXðtÞ ð16Þ fied into a condensed matrix form
A17×17 fCg17×1 ¼ 0 ð24Þ
uL ¼ φL XðtÞ ð17Þ
The elements of matrix A17×17 are presented in Appendix II. To
obtain nontrivial solutions for Eq. (24), a necessary and sufficient
MC MD
condition for the homogeneous system requires the determinant of
VC VD
matrix A17×17 to equal to 0, which results in the following charac-
ka(uL−uA) kb(4u'A+2u'B) teristic equation:
ka(uB−uL) kb(2u'A+4u'B detðA17×17 Þ ¼ 0 ð25Þ
VA VB MA MB The matrix A17×17 is so large that it is impractical to expand the
determinant explicitly. Therefore, the explicit characteristic equa-
Fig. 2. Force and moment equilibrium at the link joint
tion for solving each frequency is not provided here. In addition,

© ASCE 04015008-3 J. Struct. Eng.

J. Struct. Eng.
obtaining an explicit expression for root β of the characteristic fre- cantilever beams vibrate in the same direction) or out-of-phase
quency equation is difficult because the equation is typically tran- (i.e., the two towers have opposing motions). These two types of
scendental. Root β, however, can be numerically determined mode shapes can be formulated in terms of the integration constants
by Eq. (25), the details of which are provided later. By solving
this equation numerically, a series of roots β k [or frequencies in-phase CA;j ¼ CB;j ; CC;j ¼ CD;j j ¼ 1; : : : ; 4 ð26Þ
ωk ðk ¼ 1,2; 3; : : : Þ], can be obtained. The corresponding mode
shape can then be calculated by substituting ωk into Eq. (24) out-of-phase CA;j ¼ −CB;j ; CC;j ¼ −CD;j j ¼ 1; : : : ; 4
and solving the vector {C}.
Because of the identity of the two beams and the symmetrical ð27Þ
arrangement of the system, the mode shapes of the two beams in By substituting Eq. (26) into Eq. (24), the characteristic equa-
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

the system are evidently either in-phase (i.e., the two linked tion for the in-phase mode can be expressed as

2 38 9
0 0 − cos β − sin β cosh β sinh β 0 > CA1 >
>
> >
>
6 0 7 > CA2 >
6 0 0 sin β − cos β sinh β cosh β 7>>
>
>
>
>
6 7>> >
>
6 7
0 7> > >
6 f1 f2 − cos λh β − sin λh β − cosh λh β − sinh λh β >
<
CC1 >
>
=
6 7
6 f4 f1 sin λh β − cos λh β − sinh λh β − cosh λh β 0 7
6 7> C2 > ¼ 0
C ð28Þ
6 7>> >
6 f5 f6 β cos λh β β sin λh β −β cosh λh β −β sinh λh β 0 7 > CC3 >
>
6 7>> >
>
6 3 3 3 7>>
>
>
>
>
4 f7 f8 −β sin λh β β cos λh β −β sinh λh β −β 3 cosh λh β λka 5>> CC4 >
>
>
: >
;
2f 1 2f2 0 0 0 0 l φL
where
f 1 ¼ cos λh β − cosh λh β f2 ¼ sin λh β − sinh λh β
f 2 ¼ sin λh β − sinh λh β f4 ¼ − sin λh β − sinh λh β
ð29Þ
f 5 ¼ βf3 þ 6λkb f 4 f6 ¼ βf 4 þ 6λkb f1
f 7 ¼ β 3 f2 − λka f1 f8 ¼ β 3 f 3 − λka f 2

λm β 4
l¼ −2 ð30Þ
λka
Similarly, by substituting Eq. (27) into Eq. (24), the characteristic equation for the out-of-phase modes can be simplified as
2 38 9
0 0 − cos β − sin β cosh β sinh β 0 > CA1 >
>
> >
60 0 7 >C > >
6 0 sin β − cos β sinh β cosh β 7>>
>
>
A2 >
>
6 7>> >
>
6f f2 − cos λh β − sin λh β − cosh λh β − sinh λh β 7
0 7> > CC1 >
>
6 1 >
< >
=
6 7
6 f4 f1 sin λh β − cos λh β − sinh λh β − cosh λh β 7
0 7 CC2 ¼ 0 ð31Þ
6
6 7>>
>
>
>
6 f9 f10 β cos λh β β sin λh β −β cosh λh β −β sinh λh β 0 7 > CC3 >
> >
>
6 7> >
6 7>>
>
>
>
4 f7 f8 3
−β sin λh β 3
β cos λh β 3
−β sinh λh β −β 3 cosh λh β λka 5>> CC4 >
>
>
>
: >
;
0 0 0 0 0 0 l φL

f9 ¼ βf3 þ 2λkb f4 f 10 ¼ βf4 þ 2λkb f1 ð32Þ discussion on a special case in which the link is at the top of the two
towers (i.e., for a case in which λh ¼ 1). In this case, Subsections C
In this type of mode, the lumped link mass is evidently static, and D will disappear, and the total constant vector {C} can be
because φL is equal to 0, based on the last part of Eq. (31). shortened to a vector with nine elements as fCA;1 ; CA;2 ; CA;3 ; CA;4 ;
In this way, two independent characteristic equations for the two CB;1 ; CB;2 ; CB;3 ; CB;4 ; φL gT . The system of homogenous linear
types of mode are determined by Eqs. (28) and (31). The two as- equations regarding vector {C} can then be readily formed as
sociated frequency equations can then be derived by requiring the
determinants of the associated coefficient matrix in Eqs. (28) and
A9×9 fCg9×1 ¼ 0 ð33Þ
(31) to be zero.

where A9×9 is the new matrix of the case in which λh ¼ 1, the el-
Special Case
ements of which are presented in Appendix III. Similarly, nontrivial
The above derivations mainly concern the general situation where solutions for Eq. (33) necessitate that the determinant of matrix
the link is at an arbitrary location. This section presents a further A9×9 is equal to 0

© ASCE 04015008-4 J. Struct. Eng.

J. Struct. Eng.
detðA9×9 Þ ¼ 0 ð34Þ CA;1 ¼ −CB;1 ¼ −CA;3 ¼ CB;2 ¼ 1
f3
Expanding Eq. (34) reveals that this determinant is factorable CA;2 ¼ −CB;2 ¼ −CA;4 ¼ CB;4 ¼ −
f4
and yields the following two characteristic equations:
φL ¼ 0 ð42Þ
2λka ðf 1 f 6 − f 2 f 5 Þ þ lðf5 f8 − f6 f7 Þ ¼ 0 ð35Þ
Thus, when the axial stiffness of the link is 0 (i.e., λka ¼ 0), both
and frequency equations in Eqs. (39) and (41) are reduced to
1 þ cos β cosh β ¼ 0, which corresponds to the frequency equation
f9 f8 − f10 f7 ¼ 0 ð36Þ for a single cantilever beam. This is what is expected because, in the
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

case in which λka ¼ 0 and λkb ¼ 0, the towers are actually not con-
nected through the link and will vibrate independently. Therefore,
These two characteristic equations are concise, through which their modal properties are identical to those of a single canti-
the frequency for the system with a top link can be readily deter- lever beam.
mined. After the frequency is obtained, the corresponding mode
shape (i.e., the constant vector {C}) can be determined by substi-
tuting the frequency into the associated characteristic equation. Solution Procedure
If CA;1 is set to a value of 1, in terms of which the remaining co-
As mentioned in the previous section, expanding the frequency
efficients are to be expressed, the vector fCg9×1 in Eq. (33) can
equation of the simplified model of the linked building system
then be determined as
is impractical when the link is not at the top because the coefficient
matrix A is exceedingly large. Moreover, the frequency equation of
CA;1 ¼ CB;1 ¼ −CA;3 ¼ −CB;3 ¼ 1
the system is transcendental so that there is no explicit expression
f5 for the root β. Although an explicit expression for the frequency
CA;2 ¼ CB;2 ¼ −CA;4 ¼ −CB;4 ¼ −
f6 cannot be obtained, the frequency and mode shape can be deter-
2 f2 f5 − f1 f6 mined through numerical calculation by using Eq. (24) or Eqs. (28)
φL ¼ ð37Þ and (31). A flow chart of the solution procedure for calculating the
l f6
modal properties is presented in Fig. 3. By using the derived ana-
lytical characteristic equations, the proposed procedure can be used
and
for effectively determining the modal properties by applying the
half-interval method once the link parameters (λm , λh ; λka , and
CA;1 ¼ −CB;1 ¼ −CA;3 ¼ CB;3 ¼ 1 λkb ) are given. As shown in Fig. 3, ntotal is the total number of
f9 f modes considered in the calculation (i.e., the required number of
CA;2 ¼ −CB;2 ¼ −CA;4 ¼ CB;4 ¼ − ¼− 7
f 10 f8
φL ¼ 0 ð38Þ
Input parameters of the link: m,
ka, kb, h
Thus, the analytical solutions for the frequencies and mode
shapes of the simplified model with a top link are derived analyti-
cally by using Eqs. (35)–(38). It can be seen that the mode shapes of Define total frequency number
ntotal s
the two beams determined by Eq. (37) are identical (i.e., in-phase),
and those determined by Eq. (38) are opposite (i.e., out-of-phase),
which echo the two types of mode shapes defined by Eqs. (26) Initial condition:
and (27). n=0, a=0, b=a+
In this special case, if the vertical bending stiffness of the link is
neglected (i.e., λkb ¼ 0), then the frequency characteristic equation
and the mode shape coefficients of the in-phase mode can be further
n<ntotal end
simplified as

λm λka βðsin β cosh β − cos β sinh βÞ


þ ðλm β 4 − 2λka Þð1 þ cos β cosh βÞ ¼ 0 ð39Þ
n=n;
det[A(a)]det[A(b)]
a=b;
0
b=a+
CA;1 ¼ CB;1 ¼ −CA;3 ¼ −CB;3 ¼ 1
f3
CA;2 ¼ CB;2 ¼ −CA;4 ¼ −CB;4 ¼ −
f4
Find the solution of Eqs.(24), (28), or (31) in the
f2 f3 − f1 f4
φL ¼ ð40Þ range of [a,b] using the half-interval method
f4

and those of the out-of-phase mode can be rewritten as n=n+1; a=b; b=a+

β 3 ð1 þ cos β cosh βÞ − λka ðcos β sinh β − sin β cosh βÞ ¼ 0 Fig. 3. Flow chart of computation solution procedure for determining
ð41Þ the modal properties of the linked building system

© ASCE 04015008-5 J. Struct. Eng.

J. Struct. Eng.
Table 1. Comparison of β Calculated by the Characteristic Equations (CE) and by the FEM Model for Cases with Different Link Parameters (λh , λm , λka and
λkb )
Mode
λh λm λka λkb Method 1 2 3 4 5 6
1.0 0.01 1 0.1 FEM 1.9913 2.0487 3.7671 4.7455 4.8224 7.8825
CE 1.9914 2.0487 3.7671 4.7454 4.8223 7.8815
0.6 0.05 10 0.01 FEM 2.3819 2.4575 4.4243 4.7964 4.9552 7.9143
CE 2.3819 2.4575 4.4243 4.7963 4.9551 7.9143
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

modes). In each loop, the frequency increment Δω should be suf- of the system. Therefore, the effects of these four parameters on the
ficiently small (e.g., 0.001) to ensure that no root is missed in each in-phase frequency and mode shapes are examined.
interval ½a; b. Fig. 4 displays the variations of the first in-phase relative fre-
quency rω1 with the ratios of link mass λm and location λh . As
shown in Fig. 4, the first in-phase relative frequency rω1 decreases
Numerical Validation linearly as λm increases. And the larger the value of λh (i.e., the
higher the link is), the larger the reduction. This is a corollary
To validate the accuracy of the characteristic equations for deter-
of the additional link mass at a higher location, making the system
mining the modal properties of the simplified beam model, the
flexible. However, in the considered range, the first in-phase fre-
frequencies of the simplified model were also calculated by using
quency of the system is nearly constant when λh < 0.4 or
the related FEM model as a reference for two cases. On the basis of
λm < 0.02. This suggests that when the link elevation is lower than
the simplified model shown in Fig. 1, the corresponding FEM
0.4 H or the link mass is smaller than 0.02 of the total building
model was formed by the FEM software package ANSYS 14.0,
in which the properties of the link (i.e., ka ; kb , and mL ) were de- mass, the link mass has no apparent influence on the first in-phase
termined according to the ratios of λh ; λm ; λka , and λkb in each frequency.
case. The structural vibration frequency ω was then directly The variations of the first in-phase relative frequency rω1 with
provided by the FEM model, through which the associated nondi- the axial stiffness ratio λka and the bending stiffness ratio λkb are
mensional value of β for each case was determined by Eqs. (20) presented in Fig. 5. Evidently, rω1 is completely independent of
and (23). The values of β computed by using the FEM model were λka , indicating that the axial stiffness of the link does not affect
compared with those calculated by using the characteristic equa-
tions for the first six modes, as listed in Table 1. For the two cases,
the frequencies calculated by using the characteristic equations are
in good agreement with those determined by using the FEM model,
particularly for the low-order frequencies, thereby demonstrating a 1.1
high accuracy of the characteristic equations.
1
rω 1

0.9
Effect of Link on Modal Properties
0.8
Bu using the derived characteristic equation for the modal proper- 0
ties of the simplified model, a sensitivity analysis is conducted to 0.7 0.05
0.2 0.1
examine the effects of the link parameters (i.e., the mass, location, 0.4 0.15 λm
0.6
axial stiffness, and bending stiffness) on the modal properties of the λh 0.8
1 0.2
system. To show the effect quantitatively and allow the results to be
applied to other similar cases, a nondimensional relative frequency Fig. 4. Variations of the first in-phase relative frequency rω1 with λh
is introduced, rω;k , defined as and λm ðλka ¼ 0.1; λkb ¼ 0.001Þ
 2
ωk βk
rω;k ¼ ¼ ð43Þ
ωsingle;k β single;k

where ωsingle;k = kth frequency of the associated single cantilever 1.8


beam; and β single;k ¼ kth root of the characteristic frequency
1.6
equation for the single cantilever beam, i.e. the kth root for
1 þ cos β cosh β ¼ 0. 1.4
ω1

As derived in the section “Natural Frequencies and Mode 1.2


r

Shapes,” the in-phase and out-of-phase modes of the linked build- 1


ing system are independent; therefore, the effects of the link on
0.8
these two mode types are analyzed separately in this study. To con- 3
10 2
serve space, this paper presents only the effects of the link param- 10
2
1 10
1 0 10
eters on the first in-phase and out-of-phase modes. 10 -1 10
λka 0 -2 10
10 -1 -3 10 λkb
10 10
Modal Properties of the In-Phase Mode
Fig. 5. Variations of the first in-phase relative frequency rω1 with λka
For the in-phase mode, Eq. (28) shows that all of the link param-
and λkb ðλm ¼ 0.01; λh ¼ 1.0Þ
eters (i.e., λm ; λh ; λka , and λkb ) seem to affect the modal properties

© ASCE 04015008-6 J. Struct. Eng.

J. Struct. Eng.
1 1 1 1

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6

h/H
h/H
0.4 0.4 0.4 0.4
λm=0.001 λm=0.001
λm=0.005 λm=0.005
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

0.2 λm=0.01 0.2 0.2 0.2 λm=0.01


λm=0.05 λm=0.05
λm=0.1 λm=0.1
0 0 0 0
0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1
(a) tower 1 tower 2 (b) tower 1 tower 2

Fig. 6. Effect of link mass on the first in-phase mode shape (λh ¼ 1): (a) λkb ¼ 0.01; (b) λkb ¼ 10

the first in-phase frequency of the system. This is because for the the effect of link location, Fig. 8 shows that when the link is high,
in-phase mode, the two towers vibrate in the same direction and at the mode shape of the system deviates significantly from that of a
the same magnitude, so the two translational springs representing single beam, whereas when the link is low (e.g., λkb < 0.4), the link
the axial stiffness of the link move synchronously with the two tow- has no significant effect on the mode shape of the system, even for a
ers and do not exert any effect on the system. Although the effect of link with a large bending stiffness.
λkb is also very small when λkb < 0.01, it significantly affects the
in-phase frequency when λkb > 0.01. As shown in Fig. 5, the first
in-phase relative frequency rω1 increases remarkably with λkb when Modal Properties of the Out-of-Phase Modes
λkb changes from 0.01 to 10. However, the frequency remains As shown in Eqs. (31) and (42), the modal properties of the out-of-
nearly constant when λkb is larger than 10. In other words, for val- phase mode are unrelated to the link mass, because the lumped link
ues of λkb higher than 10, further increasing the bending stiffness of mass is static in this type of mode. Therefore, only the effects of
the link has no apparent effect on the first in-phase frequency of the link location, axial stiffness, and bending stiffness are examined for
linked building system. the out-of-phase mode.
The effects of the link parameters on the first in-phase mode The variations of the first out-of-phase relative frequency rω1
shape of the system are displayed in Figs. 6–8. Fig. 6 shows that with λh and λka are shown in Fig. 9. Clearly, the out-of-phase fre-
the link mass barely affects the mode shape. However, the other two quency rω1 increases significantly with λka , particularly when the
parameters (i.e., bending stiffness and location of the link) exert link is high (e.g., λh > 0.6). For example, a link with a λka value of
apparent effects on the first in-phase mode shape. As shown in 100 can raise rω1 to approximately 4 when the link is at the top
Fig. 7, the mode shape of the system is similar to that of a single (i.e., λh ¼ 1), which indicates a substantial constraining effect
cantilever beam when λkb is small (i.e., 0.01), whereas the mode of axial stiffness of the link on the out-of-phase mode. However,
shape changes its curvature when λkb is large. For instance, when the surface of rω1 becomes relatively flat when λka is larger than
λkb is larger than 10, the mode shape approximates that of a shear 100, suggesting that the effect of the axial stiffness of the link is
beam, instead of an Euler-Bernoulli beam, because of the con- already considerably near its limit when λka is equal to 100. Hence,
straining effect of the bending stiffness of the link. Regarding a further increase in λka cannot lead to a significant increase in the

1 1 1 1

0.8 0.8 0.8 0.8

0.6 0.6 0.6 0.6


h/H
h/H

0.4 0.4 0.4 0.4


λkb=0.01 λh=0.2
λkb=0.1 λh=0.4
0.2 0.2 λkb=1 0.2 0.2 λh=0.6
λkb=10 λh=0.8
λkb=100 λh=1
0 0 0 0
0 0.5 1 0 0.5 1 0 0.5 1 0 0.5 1
tower 1 tower 2 tower 1 tower 2

Fig. 7. Effect of bending stiffness of link on the first in-phase mode Fig. 8. Effect of link location on the first in-phase mode shape
shape (λm ¼ 0.01, λh ¼ 1) (λm ¼ 0.01, λkb ¼ 10)

© ASCE 04015008-7 J. Struct. Eng.

J. Struct. Eng.
1 1
8
λka=0.1
6 0.8 0.8
λ =1
ka
λka=10
r
ω1 4 λka=100
0.6 0.6
λ =1000
2 ka

h/H
0
3 0.4 0.4
10 2
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

10
1 1
λka10 0 0.6
0.8
0.2 0.2
10 0.4
-1 0.2 λh
10

Fig. 9. Variations of the first out-of-phase relative frequency rω1 with 0 0


-1 -0.5 0 0 0.5 1
λh and λka ðλkb ¼ 0.001Þ tower 1 tower 2

Fig. 11. Effect of the axial stiffness of the link on the first out-of-phase
mode shape (λh ¼ 1, λkb ¼ 0.001)
frequency. Regarding the effect of link location, for cases with a
large value of λka ð> 1Þ, rω1 shows an obvious peak when λh is
approximately 0.8. It is interesting that this is consistent to that re-
ported by Abramovich and Hamburger (1992), although their work system with a top link (i.e., λh ¼ 1), increasing the ratio λka trig-
is for a single cantilever beam with an in-span elastic support. For gers a gradual change in the mode shapes of the system, from a
the out-of-phase mode, each building in the linked building system rough straight line to a curve. A link with a λka value of 1,000 even
(LBS) behaves like a single cantilever beam with an elastic support renders the mode shape to become similar to a sine wave, which
at the link elevation because the link mass is static in the mode. deviates significantly from that of a single beam. Fig. 12 shows the
Therefore, in terms of the first-out-phase mode, the axial stiffness variation of the first out-of-phase mode shape with the bending
of the link can be used completely when the link is located at 0.8 H. stiffness of the link. Compared with the effect of the axial stiffness,
Fig. 10 shows the variations of the first out-of-phase frequency that of the bending stiffness on the first out-of-phase mode shape is
with the location and bending stiffness of the link. As shown in not so drastic. As shown in Fig. 12, the mode shapes exhibit var-
Fig. 10, similar to the effect of λka , increasing λkb raises the out- iations as λkb increases, but they are relatively small. Even for a link
of-phase frequency. In addition, the gradient of the surface of rω1 with a value λkb of 100, the mode shape still does not differ sig-
along λkb is relatively small when λkb is larger than 20. This in- nificantly from that of a small λkb .
dicates that the effect of bending stiffness of the link is already con- The effect of the link location on the first out-of-phase mode
siderably near its limit when λkb equals 20. Compared with the shape is shown in Fig. 13. Fig. 13(a) shows that for a link with
a large axial stiffness (e.g., λka ¼ 100), the first out-of-phase mode
limit of the frequency increase caused by the axial stiffness shown
shape remains close to that of a single cantilever beam when the
in Fig. 9, the limit caused by the bending stiffness is relatively
link is low (e.g., λh < 0.6), whereas the mode shape deviates sig-
small. This suggests that in terms of the increase in the out-of-phase
nificantly from that of a single beam when the link is at a high
frequency, the effect of the bending stiffness of the link is smaller
elevation. For instance, when the link is at the top (i.e., λh ¼ 1),
than that of the axial stiffness. Although the surface of rω1 in Fig. 10
the mode shape is very similar to a sine wave, in which the maximal
also exhibits a peak, the peak location differs from that shown in
value occurs at the middle elevation. For a link with a large bending
Fig. 9. As shown, rω1 shows peaks at λh ¼ 0.5, suggesting that for stiffness (e.g., λkb ¼ 50), the link location has a significant influ-
the out-of-phase frequency, the bending stiffness of a link at 0.5 H ence on the mode shape. As shown in the figure, the mode shapes
has the greatest capacity to stiffen the system.
The effects of the link stiffness on the first out-of-phase mode
shape are presented in Figs. 11 and 12. Fig. 11 shows that, for the 1 1

λkb=0.01

0.8 λkb=0.1
0.8
λkb=1
3
λkb=10
2.5 0.6 λkb=100
0.6
h/H

2
ω1
r

1.5 0.4 0.4

1
0.2 0.2
2
10 1
10
0
λkb 10 1
-1 0.8 0 0
10 0.6 -1 -0.5 0 0 0.5 1
0.4
10
-2 0.2 λh tower 1 tower 2

Fig. 10. Variation of the first out-of-phase relative frequency rω1 with Fig. 12. Effect of the bending stiffness of the link on the first out-of-
λh and λkb ðλka ¼ 0.01Þ phase mode shape (λh ¼ 1, λka ¼ 0.01)

© ASCE 04015008-8 J. Struct. Eng.

J. Struct. Eng.
1 1 1 1
λh=0.2
λh=0.4 λh=0.2
λ =0.6 λ =0.4
0.8 0.8h 0.8 0.8h
λh=0.8 λh=0.6
λh=1.0 λh=0.8

0.6 0.6 λ =1.0


0.6 0.6h

h/H
h/H 0.4 0.4 0.4 0.4
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

0.2 0.2 0.2 0.2

0 0 0 0
-1 -0.5 0 0 0.5 1 -1 -0.5 0 0 0.5 1
(a) tower 1 tower 2 (b) tower 1 tower 2

Fig. 13. Effect of the link location on the first out-of-phase mode shape: (a) λka ¼ 100, λkb ¼ 0.001; (b) λka ¼ 0.01, λkb ¼ 50

for cases with a lower link exhibit remarkable differences from from 0.25 to 1.56 Hz when λh changes from 0.05 to 0.8. This
those with higher links. When the link is low (e.g., λh < 0.5), the causes a 50% reduction in the maximal standard deviation of
mode shape is similar to that of a single cantilever beam and is the displacement response because the wind force energy decays
concave. However, when the link is high (e.g., λh > 0.5), the mode significantly with increase of frequency. After exceeding a value
shape is convex because of the constraining effect of the bending of 0.8 for λh, although the frequency of the first out-of-phase mode
stiffness of the link. decreases slightly, the structural response of the system is still very
small. Generally, the link can reduce the wind-induced response of
the system by 50% if the link elevation is higher than 0.6 H.
Application on an Example of a Linked Allowing for the fact that the value of λka could be different
Building System when the link is designed in another form or the connection be-
tween the link and buildings is elastic, it is worthwhile to examine
For application, a typical linked building system, consisting of two the frequencies and wind-induced responses of the system with the
identical reinforced concrete buildings connected through a 2-story same two buildings but with different values of λka . Thus, the value
link, is adopted. Each building in the system is a 40-story high-rise of λka is set to be within the range of 10−3 –103 , and the variations
building with a uniform cross section of 30 × 30 m and a height of the first two frequencies and the maximal standard deviation of
of 140 m. The mass per unit length m is 1.8 × 105 kg=m, and the displacement responses of the system are shown in Fig. 15.
the stiffness EI of each building is 2.35 × 1013 Nm2 . The first three It can be observed that the first in-phase frequency remains con-
frequencies of each building are 0.250, 1.565, and 4.383 Hz. The stant for all λka because the first in-phase frequency is completely
space between the two buildings is 15 m (i.e., the link span is 15 m). independent of λka , as shown in Fig. 5. The first out-of-phase fre-
For simplicity, the link is designed as a reinforced beam-slab quency, however, increases from 0.25 to 0.94 Hz when λka changes
system. The mass mL , axial stiffness ka , and the vertical bending from 0.1 to 100. This increment in the frequency gives rise to a 50%
stiffness kb of the link are 2.92 × 105 kg, 4.8 × 109 N=m, and reduction in the standard deviation of the displacement response of
8.64 × 108 N · m, respectively. Accordingly, the values of the ra-
tios λm , λka , and λkb are 0.01, 840, and 0.006, respectively. It is
clear that the large value of λka changes the modal properties of 1st in-phase frequency
the system, as shown in Figs. 9 and 11. Although the values of 1st out-of-phase frequency
λm and λkb are low in this case, and hence have no significant effect response
on the response, the values of λm and λkb could be high in other
1.6 55
cases in which the link extends over several floors or the link is
σ of displacement response (mm)

composed of stiff structures, such as steel megatrusses. 1.4 50


To demonstrate the overall effectiveness of the link, the standard 1.2 45
frequency (Hz)

deviations of the top displacement responses of the linked building


1 40
system are calculated under external wind force excitation. The
wind forces on the system were measured by a wind tunnel pressure 0.8 35
experiment at the CLP (China Light and Power) Power Wind/Wave
0.6 30
Tunnel Facility at the Hong Kong University of Science and Tech-
nology, details of which are given in (Song and Tse 2014). Fig. 14 0.4 25
shows the variations of the first two frequencies (i.e., the first in- 0.2 20
phase and out-of-phase modes) and maximal standard deviation of 0 0.2 0.4 0.6 0.8 1
the displacement responses of the linked building system with the λh h/H
location of link in terms of λh . As shown in Fig. 14, the frequency
Fig. 14. Variations of the first two frequencies and the maximum stan-
of the first in-phase mode remains nearly constant for all λh be-
dard deviation of the displacement responses of the system with
cause of the small values of λm and λkb . Nevertheless, the high
λh ðλka ¼ 840Þ
value of λka raises the frequency of the first out-of-phase mode

© ASCE 04015008-9 J. Struct. Eng.

J. Struct. Eng.
1st in-phase frequency
and bending stiffness. The mode shape for a case with a high link
1st out-of-phase frequency
could be distinct from that for a case with a low link.
response
The application of the analytical model on an example of a
1.6 55
linked building system shows that, compared with the case in

σ of displacement response (mm)


1.4 50 which a link is absent, the wind-induced responses of the linked
1.2 45
building system are decreased. Specifically, a link with λka ¼
100 and λh > 0.6 can reduce the response by approximately 50%.
frequency (Hz)

1 40 In addition, further increasing the value of λka beyond 100 has


0.8 35 no significant effect on the reduction of the building response.
This implies that the optimal design values of the link in terms
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

0.6 30
of building response reduction are λka ¼ 100 and λh > 0.6 in the
0.4 25 example case.
0.2 20 The simplified continuum beam model is capable of describing
-3 -2 -1 0 1 2 3
10 10 10 10 10 10 10 most features of the linked building system, particularly the
λka coupling effect of the link. Therefore, the analytical solution for
the simplified model can be directly applied to the corresponding
Fig. 15. Variations of the first two frequencies and the maximum stan- linked building system. From an engineering perspective, the ana-
dard deviation of the displacement responses of the system with lytical solutions and the results of the sensitivity analysis are useful
λka ðλh ¼ 1Þ for rapidly estimating the modal properties of a linked building sys-
tem. In addition, the analytical solutions and the results presented in
this paper can be used for establishing an optimal design of linked
building systems.
the system. The first out-of-phase frequency and structural response
is no longer highly sensitive to λka when λka > 100, indicating
that further increasing λka has no significant effect on the structural
Appendix I. Sixteen Homogenous Equations for the
response. Boundary, Continuity, and Force Equilibrium
On the basis of this discussion, it is obvious that the originally Conditions
designed link (λm ¼ 0.01, λka ¼ 840, λkb ¼ 0.006) can reduce the
wind-induced response of the system by approximately 50% when The 16 homogenous equations for the boundary, continuity, and
the link elevation is higher than 0.6 H. In addition, after a certain force equilibrium conditions are shown in the following as
value of λka (i.e., 100), further increasing λka has no significant Eqs. (44)–(62):
effect on the reduction of the response. It is more effective to
redesign the original link as a link with a smaller stiffness (but with CA;1 þ CA;3 ¼ 0 ð44Þ
λka still larger than 100), which can reduce the construction cost
and still provide nearly the same effect on the modal properties and
the wind-induced responses of the system. CA;2 þ CA;4 ¼ 0 ð45Þ

CB;1 þ CB;3 ¼ 0 ð46Þ


Conclusions
This study proposed a simplified continuum beam model consisting CB;2 þ CB;4 ¼ 0 ð47Þ
of two cantilever beams connected horizontally by using a spring-
mass system to investigate the effects of the link on the modal prop-
erties of a linked building system. The analytical solutions for the −CC;1 cos β − CC;2 sin β þ CC;3 cosh β þ CC;4 sinh β ¼ 0 ð48Þ
modal properties of the system were then derived, and an efficient
computation procedure for calculating the modal properties was
established. The high accuracy of the analytical solutions for the CC;1 sin β − CC;2 cos β þ CC;3 sinh β þ CC;4 cosh β ¼ 0 ð49Þ
modal properties was validated by comparing them with those cal-
culated by using FEM analysis. By using the analytical solutions,
the effects of the link parameters on the modal properties of the −CD;1 cos β − CD;2 sin β þ CD;3 cosh β þ CD;4 sinh β ¼ 0 ð50Þ
system were comprehensively investigated.
It is shown that the effects of the link on the modal properties CD;1 sin β − CD;2 cos β þ CD;3 sinh β þ CD;4 cosh β ¼ 0 ð51Þ
differ between the in-phase and out-of-phase modes. For the in-
phase mode, whereas the axial stiffness of the link does not affect
the frequency substantially, the link location, mass, and bending CA;1 cos λh β þ CA;2 sin λh β þ CA;3 cosh λh β þ CA;4 sinh λh β
stiffness do. In addition, the in-phase mode shape is sensitive to
the link location and bending stiffness. For the out-of-phase mode, − CC;1 cos λh β − CC;2 sin λh β − CC;3 cosh λh β
both the axial stiffness and bending stiffness significantly increase − CC;4 sinh λh β ¼ 0 ð52Þ
the frequency. Moreover, the level of increase is highly related to
the link location. The results show that the axial stiffness tends to
have the largest effect on the frequency when the link is located at − CA;1 sin λh β þ CA;2 cos λh β þ CA;3 sinh λh β þ CA;4 cosh λh β
approximately 0.8 H, and the bending stiffness has the largest ef- þ CC;1 sin λh β − CC;2 cos λh β − CC;3 sinh λh β
fect when the link is at approximately 0.5 H. The out-of-phase
mode shape is highly sensitive to the link location, axial stiffness, − CC;4 cosh λh β ¼ 0 ð53Þ

© ASCE 04015008-10 J. Struct. Eng.

J. Struct. Eng.
CB;1 cos λh β þ CB;2 sin λh β þ CB;3 cosh λh β þ CB;4 sinh λh β a5,11 ¼ a6,12 ¼ a7,15 ¼ a8,16 ¼ cosh β ð66Þ
− CD;1 cos λh β − CD;2 sin λh β − CD;3 cosh λh β
a5,12 ¼ a6,11 ¼ a7,16 ¼ a8,15 ¼ sinh β ð67Þ
− CD;4 sinh λh β ¼ 0 ð54Þ

a9,1 ¼ −a9,9 ¼ a10,2 ¼ −a10,10 ¼ a12,5 ¼ −a12,13 ¼ a13,6


− CB;1 sin λh β þ CB;2 cos λh β þ CB;3 sinh λh β þ CB;4 cosh λh β
¼ −a13,14 ¼ a17,1 ¼ a17,5 ¼ cos λh β ð68Þ
þ CD;1 sin λh β − CD;2 cos λh β − CD;3 sinh λh β
− CD;4 cosh λh β ¼ 0 ð55Þ
a9,2 ¼ −a9,10 ¼ −a10,1 ¼ a10,9 ¼ a12,6 ¼ −a12,14 ¼ −a13,5
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

p1 CA;1 þ p2 CA;2 þ p3 CA;3 þ p4 CA;4 − 2CB;1 λkb sin λh β ¼ a13,13 ¼ a17,2 ¼ a17,6 ¼ sin λh β ð69Þ

þ 2CB;2 λkb cos λh β þ 2CB;3 λkb sinh λh β þ 2CB;4 λkb cosh λh β


a9,3 ¼ −a9,11 ¼ a10,4 ¼ −a10,12 ¼ a12,7 ¼ −a12,15 ¼ a13,8
þ CC;1 β cos λh β þ CC;2 β sin λh β − CC;3 β cosh λh β
¼ −a13,16 ¼ a17,3 ¼ a17,7 ¼ cosh λh β ð70Þ
− CC;4 β sinh λh β ¼ 0 ð56Þ

a9,4 ¼ −a9,12 ¼ a10,3 ¼ −a10,11 ¼ a12,8 ¼ −a12,16 ¼ a13,7


p1 CB;1 þ p2 CB;2 þ p3 CB;3 þ p4 CB;4 − 2CA;1 λkb sin λh β
¼ −a13,15 ¼ a17,4 ¼ a17,8 ¼ sinh λh β ð71Þ
þ 2CA;2 λkb cos λh β þ 2CA;3 λkb sinh λh β þ 2CA;4 λkb cosh λh β
þ CD;1 β cos λh β þ CD;2 β sin λh β − CD;3 β cosh λh β
a11,1 ¼ a14,5 ¼ p1 a11,2 ¼ a14,6 ¼ p2
− CD;4 β sinh λh β ¼ 0 ð57Þ
a11,3 ¼ a14,7 ¼ p3 a11,4 ¼ a14,8 ¼ p4 ð72Þ

q1 CA;1 þ q2 CA;2 þ q3 CA;3 þ q4 CA;4 − CC;1 β 3 sin λh β


a11,5 ¼ a14,1 ¼ −2λkb sin λh β a11,6 ¼ a14,2 ¼ 2λkb cos λh β
þ CC;2 β 3 cos λh β − CC;3 β 3 sinh λh β − CC;4 β 3 cosh λh β
a11,7 ¼ a14,3 ¼ 2λkb sinh λh β a11,8 ¼ a14,4 ¼ 2λkb cosh λh β
þ λka φL ¼ 0 ð58Þ
ð73Þ
3
q1 CB;1 þ q2 CB;2 þ q3 CB;3 þ q4 CB;4 − CD;1 β sin λh β
a15,1 ¼ a16,5 ¼ q1 a15,2 ¼ a16,6 ¼ q2
þ CD;2 β 3 cos λh β − CD;3 β 3 sinh λh β − CD;4 β 3 cosh λh β
a15,3 ¼ a16,7 ¼ q3 a15,4 ¼ a16,8 ¼ q4 ð74Þ
þ λka φL ¼ 0 ð59Þ

where a15,9 ¼ a16,13 ¼ −β 3 sin λh β a15,10 ¼ a16,14 ¼ β 3 cos λh β

p1 ¼ −β cos λh β − 4λkb sin λh β a15,11 ¼ a16,15 ¼ −β 3 sinh λh β a15,12 ¼ a16,16 ¼ −β 3 cosh λh β

p2 ¼ −β sin λh β þ 4λkb cos λh β ð75Þ

p3 ¼ β cosh λh β þ 4λkb sinh λh β a15,17 ¼ a16,17 ¼ λka a17,17 ¼ l ð76Þ


p4 ¼ β sinh λh β þ 4λkb cosh λh β ð60Þ
The other elements of matrix A17×17 that are not mentioned in
3 the previous equations are zero.
q1 ¼ β sin λh β − λka cos λh β
q2 ¼ −β 3 cos λh β − λka sin λh β
q3 ¼ β 3 sinh λh β − λka cosh λh β Appendix III. Element a i;j of Matrix A9×9
q4 ¼ β 3 cosh λh β − λka sinh λh β ð61Þ The element ai;j of matrix A9×9 in Eqs. (33) and (34) are as
follows:
h ka H 3 kb H
β ¼ aH λh ¼ λka ¼ λkb ¼ ð62Þ a1,1 ¼ a1,3 ¼ a2,2 ¼ a2,4 ¼ a3,5 ¼ a3,7 ¼ a4,6 ¼ a4,8 ¼ 1 ð77Þ
H EI EI
a5,1 ¼ a6,5 ¼ p1 a 5 ,2 ¼ a 6 ,6 ¼ p 2

Appendix II. Element a i;j of Matrix A17×17 a5,3 ¼ a6,7 ¼ p3 a5;4 ¼ a6,8 ¼ p4 ð78Þ

The element ai;j of matrix A17×17 in Eqs. (24) and (25) are as a5,5 ¼ a6,1 ¼ −2λkb sin λh β a5,6 ¼ a6,2 ¼ 2λkb cos λh β
follows:
a5,7 ¼ a6,3 ¼ 2λkb sinh λh β a5,8 ¼ a6,4 ¼ 2λkb cosh λh β
a1,1 ¼ a1,3 ¼ a2,2 ¼ a2,4 ¼ a3,5 ¼ a3,7 ¼ a4,6 ¼ a4,8 ¼ 1 ð63Þ
ð79Þ
a5,9 ¼ a6,10 ¼ a7,13 ¼ a8,14 ¼ − cos β ð64Þ
a7,1 ¼ a8,5 ¼ q1 a 7 ,2 ¼ a 8 ,6 ¼ q 2
−a5,10 ¼ a6,9 ¼ −a7,14 ¼ a8,13 ¼ sin β ð65Þ a7,3 ¼ a8,7 ¼ q3 a 7 ,4 ¼ a 8 ,8 ¼ q 4 ð80Þ

© ASCE 04015008-11 J. Struct. Eng.

J. Struct. Eng.
a9,1 ¼ a9,5 ¼ cos β a9,2 ¼ a9,6 ¼ sin β Iwanami, K., Suzuki, K., and Seto, K. (1996). “Vibration control method
for parallel structures connected by damper and spring.” JSME Int. J.
a9,3 ¼ a9,7 ¼ cosh β a9,4 ¼ a9,8 ¼ sinh β ð81Þ Ser. C Dyn. Control Rob. Des. Manuf., 39(4), 714–720.
Kim, J., Ryu, J., and Chung, L. (2006). “Seismic performance of
a 9 ,9 ¼ l ð82Þ structures connected by viscoelastic dampers.” Eng. Struct., 28(2),
183–195.
The other elements of matrix A9×9 that are not mentioned in the Lee, D. G., Kim, H. S., and Ko, H. (2010). “Evaluation of coupling-control
previous equations are zero. effect of a sky-bridge for adjacent tall buildings.” Struct. Des. Tall
Special Build., 21(5), 311–328.
Li, Q. S., Cao, H., and Li, G. Q. (1994). “Analysis of free vibrations of
Downloaded from ascelibrary.org by Hong Kong University of Sci and Tech (HKUST) on 02/10/15. Copyright ASCE. For personal use only; all rights reserved.

Acknowledgments tall buildings.” J. Eng. Mech., 10.1061/(ASCE)0733-9399(1994)


120:9(1861), 1861–1876.
The work described in this paper was supported by the Research Lim, J., Bienkiewicz, B., and Richards, E. (2011). “Modeling of structural
Grants Council of the Hong Kong Special Administrative Region, coupling for assessment of modal properties of twin tall buildings with a
China. Thanks also go to the staff of the CLP Power Wind/Wave skybridge.” J. Wind Eng. Ind. Aerodyn., 99(5), 615–623.
Tunnel Facility at HKUST for their assistance in this project. Luco, J. E., and De Barros, F. C. (1998). “Optimal damping between
two adjacent elastic structures.” Earthquake Eng. Struct. Dyn., 27(7),
649–659.
References Miranda, E. (1999). “Approximate seismic lateral deformation demands in
multistory buildings.” J. Struct. Eng., 10.1061/(ASCE)0733-9445
Abramovich, H., and Hamburger, O. (1992). “Vibration of a uniform (1999)125:4(417), 417–425.
cantilever Timoshenko beam with translational and rotational springs Miranda, E., and Reyes, C. J. (2002). “Approximate lateral drift demands
and with a tip mass.” J. Sound Vib., 154(1), 67–80. in multistory buildings with nonuniform stiffness.” J. Struct. Eng.,
ANSYS 14.0 [Computer software]. Canonsburg, PA, ANSYS. 10.1061/(ASCE)0733-9445(2002)128:7(840), 840–849.
Bhaskararao, A. V., and Jangid, R. S. (2006). “Harmonic response of Miranda, E., and Taghavi, S. (2005). “Approximate floor acceleration de-
adjacent structures connected with a friction damper.” J. Sound Vib., mands in multistory buildings. I: Formulation.” J. Struct. Eng., 10.1061/
292(3), 710–725. (ASCE)0733-9445(2005)131:2(203), 203–211.
Bhaskararao, A. V., and Jangid, R. S. (2007). “Optimum viscous damper Rao, S. S. (2007). Vibration of continuous systems, Wiley, Hoboken, NJ.
for connecting adjacent SDOF structures for harmonic and stationary Richardson, A., Walsh, K. K., and Abdullah, M. M. (2013). “Closed-form
white-noise random excitations.” Earthquake Eng. Struct. Dyn., equations for coupling linear structures using stiffness and damping
36(4), 563–571. elements.” Struct. Control Health Monit., 20(3), 259–281.
Chen, Y., McFarland, D., Wang, Z., Spencer, B., Jr., and Bergman, L. Sandoval, M. R., Ugarte, L. B., and Spencer, B. F. (2012). “Study of
(2010). “Analysis of tall buildings with damped outriggers.” J. Struct.
structural control in coupled buildings.” Proc., 15th World Conf. on
Eng., 10.1061/(ASCE)ST.1943-541X.0000247, 1435–1443.
Earthquake Engineering, National Information Centre on Earthquake
Christenson, R. E., Spencer, B. F., Jr., Johnson, E. A., and Seto, K. (2006).
Engineering, Kanpur, India.
“Coupled building control considering the effects of building/connector
configuration.” J. Struct. Eng., 10.1061/(ASCE)0733-9445(2006)132: Song, J., and Tse, K. (2014). “Dynamic characteristics of wind-excited
6(853), 853–863. linked twin buildings based on a 3-dimensional analytical model.”
Cimellaro, G. P., and Lopez–Garcia, D. (2011). “Algorithm for design Eng. Struct., 79, 169–181.
of controlled motion of adjacent structures.” Struct. Control Health Westermo, B. D. (1989). “The dynamics of interstructural connection to
Monit., 18(2), 140–148. prevent pounding.” Earthquake Eng. Struct. Dyn., 18(5), 687–699.
Dym, C. L. (2012). “Approximating frequencies of tall buildings.” J. Struct. Yang, J. N., Agrawal, A. K., Samali, B., and Wu, J. C. (2004). “Benchmark
Eng., 10.1061/(ASCE)ST.1943-541X.0000656, 288–293. problem for response control of wind-excited tall buildings.” J. Eng.
Gurley, K., Kareem, A., Bergman, L. A., Johnson, E. A., and Klein, R. E. Mech., 10.1061/(ASCE)0733-9399(2004)130:4(437), 437–446.
(1994). “Coupling tall buildings for control of response to wind.” Zhu, H. P., Wen, Y. P., and Iemura, H. (2001). “A study on interaction con-
Proc., Sixth Int. Conf. on Structural Safety and Reliability (ICOSSAR), trol for seismic response of parallel structures.” Comput. Struct., 79(2),
AA Balkema Publishers, Rotterdam, Netherlands, 1553–1560. 231–242.

© ASCE 04015008-12 J. Struct. Eng.

J. Struct. Eng.

You might also like