Professional Documents
Culture Documents
H O S T E D BY
Alexandria University
a
Department of Civil Engineering, Birjand University of Technology, Birjand, Iran
b
Department of Computer Science, University of Sistan and Baluchestan, Zahedan, Iran
KEYWORDS Abstract This study focuses on the problem of optimum selection of tuned mass dampers (TMDs)
Structural control; parameters for buildings under seismic excitations. Vibration control of structure under seismic
Seismic excitations; excitations for reduction of structural damages and comfort/serviceability considerations is
Tuned mass damper; addressed by employing a design process based on multi-objective cuckoo search (MOCS) for
Multi-objective optimisation; simultaneous reduction of seismic responses of the structures in the terms of displacement and
Cuckoo search; acceleration of stories and tuning of TMD with a smaller mass ratio. The numerical studies are car-
Optimum design ried out on two benchmark buildings. In comparison with several other documented methods, the
performance of the MOCS for the optimal design of TMDs is evaluated. The results show that the
MOCS performs better than other strategies in term of the simultaneous reduction of maximum dis-
placements and accelerations of the structures subjected to different seismic excitations. Further-
more, the feasibility of MOCS to choose a smaller mass, stiffness, and damping is assessed. The
simulation results confirm that the MOCS is able to present a solution leading to a more practical
and economic selection of TMD parameters with a choice of a smaller mass, stiffness, and damping
so that it is capable of maintaining the desired level of the reduction of structural responses in dif-
ferent earthquake excitations.
Ó 2018 Faculty of Engineering, Alexandria University. Production and hosting by Elsevier B.V. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
TMDs are one of the oldest and most used vibration control
devices. They can be easily attached to the main system with-
* Corresponding author at: Department of Civil Engineering, Birjand out any renovation. A TMD system consists of a mass, an elas-
University of Technology, P.O. Box 97175-569, Birjand, Iran. tic spring and a viscous (or hysteretic) damper. The parameters
E-mail addresses: etedali@birjandut.ac.ir (S. Etedali), rakhshani@pgs. of TMDs have a direct effect on the response of the main struc-
usb.ac.ir (H. Rakhshani). tural system. Therefore, tuning the parameters of TMDs is one
Peer review under responsibility of Faculty of Engineering, Alexandria of the important issues for designers. In order to select optimal
University.
https://doi.org/10.1016/j.aej.2018.01.009
1110-0168 Ó 2018 Faculty of Engineering, Alexandria University. Production and hosting by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
3206 S. Etedali, H. Rakhshani
parameters of TMDs for a single degree of freedom (SDOF) 2. The governing equation of motion for a linear structure
main system under harmonic and white noise random excita- equipped with TMD
tions, several closed-form expressions have been introduced
[1–3]. Thompson [4] applied a frequency locus method to the The equation of motion of an N-degree-of-freedom linear
optimisation of the spring and damper rates for a dynamic shear-type building without TMD, subjected to earthquake
vibration absorber. Chang [5] derived formulas in closed forms ground acceleration, €
xg ðtÞ, can be written as:
for the design of TMDs subjected to both wind and earth-
quake loadings. Lin et al. [6] applied an extended random €s ðtÞ þ Cs x_ s ðtÞ þ Ks xs ðtÞ ¼ Ms rs €
Ms x xg ðtÞ ð1Þ
decrement method to reduce the dynamic responses of an where Ms , Cs and Ks are N N mass, damping and stiffness
MDOF system subjected to a seismic load. matrices, respectively. Also, xs ðtÞ, x_ s ðtÞ and x €s ðtÞ refer to
A wide application of genetic algorithm (GA) for the tuning N 1 displacement, velocity and acceleration vectors, respec-
of TMD parameters was found in the literature [7–9]. Bakre
tively. Furthermore, rs ¼ ½1; 1; ; 1T , is N 1 seismic influ-
and Jangid [10] used a numerical search method to find more
ence vector. By attaching a TMD on the top floor of the
effective TMD parameters. Leung et al. [11] used a particle
structure, the degree-of-freedom of the structure is increased
swarm optimisation (PSO) technique for optimal estimation
by one. The equation of motion of the structure equipped with
of TMD parameters in seismic- excited structures. Also, Leung
TMD, installed on the top storey, can be written as follows:
and Zhang [12] used the PSO technique for optimum tuning of
TMD for an SDOF main system. Unlike previous studies on xðtÞ þ Cx_ ðtÞ þ KxðtÞ ¼ Mrx€g ðtÞ
M€ ð2Þ
the optimal tuning of TMD where the TMD mass ratio was
where M, C and K refer to ðN þ 1Þ ðN þ 1Þ mass, damping
a preselected parameter, Marano et al. [13] optimized the
and stiffness matrices, respectively. Furthermore, xðtÞ, x_ ðtÞ
TMD mass ratio along with the other parameters on a main €ðtÞ are ðN þ 1Þ 1 displacement, velocity and accelera-
and x
system idealized as an SDOF. Assuming a harmonic load
tion vectors of storeys and TMD system, respectively. The
applied to the main system in the optimisation process, a har-
M, C, K matrices and xðtÞ vector are given by Eqs. (3)–(6).
mony search algorithm was applied to tune the optimal param-
eters of TMDs in [14]. Salvi and Rizzi [15] developed a M ¼ diag½m1 m2 mN md ð3Þ
Minimax optimisation method for tuning of TMD parameters. 2 3
Considering the seismic applications. c1 þ c2 c2 0 0
6 ..7
Recently, a new metaheuristic search algorithm, called 6 c c2 þ c3 c3 .7
6 2 0 7
cuckoo search (CS), has been developed to solve optimisation 6 .. .. .. 7
6 7
problems. Despite the good performance of CS in solving sev- 6 0 c3 . . 0 . 7
C¼6
6 .
7
7 ð4Þ
eral engineering optimisation problems [16–22], there is no
6 .. .. .. ..
research work on the optimal tuning of the TMD system using 6 . . . cn 0 77
6 . .. .. 7
CS for random excitation. Moreover, considering previous 6 . 7
4 . . . cn cn þ cd cd 5
studies in the literature review, it is observed that the optimum
design problem of TMD is treated as a single objective optimi- 0 0 cd cd
sation problem; however, it is actually a multi-objective opti- 2 3
misation problem. To the best knowledge of the authors, k1 þ k2 k2 0 0
6 7 ..
there is no study on employing the multi-objective 6 k k2 þ k3 k3 7 .
6 2 0 7
optimisation techniques for the optimum design of TMDs in 6 .. .. 7..
6 7
literature. 6 0 k3 . . 0 7 .
K¼6
6
7
7 ð5Þ
Considering this fact, a multi-objective cuckoo search is 6 .. .. .. ..
employed to find the optimum parameters of TMD for build- 6 . . . . kn 0 77
6 . .. .. 7
ings under seismic excitations. Time history analyses are car- 6 .. 7
4 . . kn kn þ kd kd 5
ried out on two benchmark buildings subjected to different
0 0 kd kd
seismic excitations. Compared with the other approaches, the
performances of the present approaches in the estimation of T
the optimal parameters of TMD are evaluated. x ¼ ½x1 x2 xN xd ð6Þ
This paper is divided into six sections in addition to the where mi , ci , ki and xi ði ¼ 1; 2; ; NÞ present the mass, damp-
introduction. The equation of motion of a linear structure ing coefficient, stiffness and horizontal displacement of the ith
equipped with TMD, subjected to an earthquake excitation, storey of the structure. Also, md , cd , kd and xd are the mass,
is introduced in Section 2. An overview of MOCS is described damping coefficient, stiffness and horizontal displacement of
in Section 3. Optimum design problem of TMD parameters is the TMD, respectively. Furthermore, r is ðN þ 1Þ 1 seismic
described in Section 4. The MOCS is also developed for the influence vector. In the case of a MDOF primary structure,
optimal design of TMD in buildings under seismic excitations mass ratio l and tuning frequency ratio f are defined as
in Section 4. Numerical studies are carried out on two follows:
benchmark buildings in Section 5. In comparison with other md
methods, the performance of MOCS in the estimation of the l¼ T ð7Þ
US;I Ms US;I
optimal parameters of TMD is also evaluated in
Section 5. At the end, the concluding remarks are presented xd
in Section 6. f¼ ð8Þ
xS;I
Optimum design of tuned mass dampers 3207
where UTS;I is the first mode shape of the primary structure, nor- to the next generation using the so-called Pareto front to
malized to have a unit component at the top story and xS;I is ensure the proper convergence of the algorithm. To do so,
the angular fundamental frequency of the primary structure MOCS algorithm uses the Pareto-optimal front diagram which
which is considered for the first mode of vibration. Also, xd indicates the nature of the trade-off between the conflicted
and nd are the angular frequency and the damping ratio of objective functions. The Pareto front is defined in terms of
the TMD and defined as: dominated and non-dominated solution (Please see the origi-
nal study [21] for more information about the Pareto front).
pffiffiffiffiffiffiffiffiffiffiffiffiffi cd
xd ¼ kd =md ; nd ¼ pffiffiffiffiffiffiffiffiffiffi ð9Þ Hence, the second rule corresponds to elitism. In the third step,
2 kd m d the worst solutions are thrown away with a probability and
new solutions are generated, according to the similarity of
3. Multi-objective cuckoo search solutions to the other solution. Therefore, it can play the role
of mutation. For cuckoo i, new solution xtþ1
i can be generated
using Lévy flight as follows:
Cuckoo search (CS) is a new metaheuristic optimisation algo-
rithm introduced by Yang and Deb [23]. The CS idea is formed xitþ1 ¼ xti þ a LevyðbÞ ð10Þ
on the obligate brood parasitism of some cuckoo species, com-
The product means entry wise multiplications and b is
bined with Lévy flight. CS applies Lévy flights, which are far
Lévy flights exponent. Also, a is the step size and can be
more efficient than the simple random walks. Lévy flight has
express as a ratio of constant a0 .
infinite mean and variance. Therefore, it can explore search
space better than the standard Gaussian process and is useful a ¼ a0 xtj xti ð11Þ
to larger moves for a global search where the new moves can
cover much more extensive regions. Also, a balance between where xtj and xti are two different randomly selected solutions.
local search and global search exploration is restored by CS. The generation of step size s samples is not trivial using Lévy
These features make its superiority over the other metaheuris- flights. A simple scheme is summarized in Eq. (12).
tic optimisation algorithms. The advantages of CS in terms of u
ease of implementation, few parameters to tune and the fast s ¼ a0 xtj xti LevyðbÞ ¼ 0:01 xtj xti 1 ð12Þ
convergence rate provide a better performance of CS in com- jvjb
passion with the existing algorithms, such as PSO and GA In addition, u and v are given by the normal distributions as
[24]. Also, Civicioglu and Besdok [17] have been reported that shown in Eq. (13).
more robust results than PSO and artificial bee colony algo-
rithms are given using CS. The CS algorithm has already been u N 0; r2u ; v N 0; r2v ð13Þ
successfully applied to a wide variety of engineering optimisa- where rv and rv are as follows.
tion problems [16–22]. In comparison with other optimisation ( )1=b
algorithms, the CS has fewer algorithm-dependent parameters. Gð1 þ bÞsinðpb=2Þ
The population size n, switching probability pa, step-size a and ru ¼ ; rv ¼ 1 ð14Þ
G½ð1 þ bÞ=2b2ðb1Þ=2
the Lévy flights exponent b are configurable parameters of CS.
The population size and the switching probability are variable In this equation, G is the standard Gamma function.
and have great effects on the algorithm performance. A bal- In essence, the second rule is an elitism strategy and passes
ance between local and global optimisation is restored by the the best solutions onto the next generation. This strategy helps
switching probability. With reducing pa, the probability for the algorithm to accelerate its converge speed. Finally, the
global optimisation is increased and vice versa [23]. third rule is a mutation operator and discards the worst solu-
In the multi-objective Cuckoo Search [21], the following tions with a probability. Eq. (15) shows how a new solution
idealized rules are used for optimizing a problem with K differ- xitþ1 is generated using a simple random walk.
ent objectives:
xitþ1 ¼ xti þ a0 Hðpa Þ xtj xtk ð15Þ
Each cuckoo puts k eggs at a time and dumps them in a ran-
domly chosen nest. Egg k represents the solution for the kth where His the Heaviside function, a0 is the step size scaling fac-
objective. tor, and is a random number with uniform distribution [22].
The best nests with high quality will be transferred to the The basic steps of the MOCS can be summarized as the pseudo
next generations. code shown in Fig. 1.
A host can discover an alien egg with a probability pa
2[0,1]. In this case, a new nest with k eggs will be built with 4. Optimum design problem of TMD parameters
respect to the similarities/differences of the eggs.
4.1. Optimum design problem of TMD parameters in the
The first rule simulates a randomization process. A new literature
solution can be randomly generated by Lévy flight. In multi-
objective optimisation problems, a diverse range of solutions In the previous studies in the literature, optimum design prob-
should be explored which is ensured by Lévy flights in MOCS lem of TMD was considered as a single objective optimisation
[21]. For each nest, there can be K solutions, which are gener- problem. For example, Hadi and Arfiadi [7] considered the H2
ated according to Eq. (10). The first rule also performs a local- norm of the transfer function of the structure as an objective
ized random permutation and can be considered as a form of function and applied GA to solve the optimum design problem
crossover. In the second rule, the best solutions are transferred of TMD parameters. Lee et al. [25] considered a performance
3208 S. Etedali, H. Rakhshani
index of structural response in the frequency domain as a sin- design based on the preselected mass ratio of TMD) with smal-
gle objective function and proposed a numerical method for ler mass, stiffness, and damping is proposed in this study. In
searching the optimum design parameters of TMD. Sadek this case, the TMD mass is considered as another objective
et al. [26] proposed an approach based on minimizing the dif- function. In this case, TMD mass is optimized along with other
ference between the damping of the first two modes of the parameters, so the design process is capable result in the
structure equipped with TMD for optimal design of TMD desired level of the reduction of the above mentioned struc-
parameters. Kaveh et al. [27] defined a weighted sum of the tural responses in different earthquake excitations. Moreover,
normalized responses in terms of maximum displacement of physical space for moving of TMD is a limit, therefore, from a
the first story and acceleration transfer function as optimisa- practical standpoint; it is desirable to limit the maximum dis-
tion criteria. They utilized the charged system search (CSS) placement of TMD. This limitation can be considered as a con-
optimisation algorithm to find the optimum parameters of straint in the defined optimisation problem. Multi-objective
TMD. optimisation algorithms are able to create a suitable trade-
off between the conflicting objectives and generate a set of pos-
sible solutions for designers which form the so-called Pareto
4.2. Optimum design of TMD parameters using MOCS
front. This advantage provides the possibility to choose the
best solution from among the presented solutions. In other
The optimum design problem of TMD parameters is a multi- words, the designer can select a possible solution that better
objective optimisation problem. Vibration control of struc- agrees with the designer’s own decisions. In this paper, two
tures under seismic excitation for reduction of structural dam- multi-objective optimisation problems, MOCS1 and MOCS2,
ages and comfort/serviceability considerations is directly and have been defined for optimum design of TMD parameters.
closely related to the displacements and accelerations of sto- For MOCS1, two non-commensurable objective functions,
ries. Therefore, an effective tuning of a TMD system for the I1 and I2, are defined. I1 is the maximum roof displacement of
seismic control of structures should be considered as having the controlled structure normalized to its corresponding value
two main objectives. The non-structural elements including in the uncontrolled structure. Similarly, I2 presents the maxi-
the architectural components (bearing walls, partitions, infill mum roof acceleration of the controlled structure normalized
walls, parapets, veneers, ceilings, door and window panes, to its corresponding value in the uncontrolled structure. The
glasses, racks, etc.), mechanical components (boilers and fur- controlled structure is the structure equipped with TMD pas-
naces, chimneys and smokestacks, tanks and pressure vessels, sive control device and the uncontrolled structure is the pri-
machinery, piping systems, etc.) and electrical components mary structure without TMD. These objective functions are
(communication systems, electrical wire ducts, electrical defined by the following equation.
motors, transformers, lighting fixtures, fire and smoke detec-
tion systems, etc.) represent a major portion of the total cost
maxt jjxr ðtÞjj maxt jjar ðtÞjj
of buildings. Loss or failure of these elements can affect the I1 ¼ ; I2 ¼ ð16Þ
safety of the occupants of the building and safety of others maxt jjb
x r ðtÞjj maxt jjb
a r ðtÞjj
who are immediately outside the building. It is notable that
the MOCS2 is proposed to achieve a practical and economical in which maxt jjxr ðtÞjj and maxt jjar ðtÞjj are the maximum roof
design of free parameters of TMD (rather than a design based displacement and acceleration of the structure equipped with
on preselected mass ratio of TMD) with smaller mass, stiff- TMD, respectively. Also, maxt jjb x r ðtÞjj and maxt jjb
a r ðtÞjj are
ness, and damping so that it is capable of maintaining the the corresponding responses in the structures without TMD.
desired level of the reduction of the above mentioned struc- For a pre-selected TMD mass, the optimal parameters of
tural responses in different earthquake excitations. Also, in TMD include the stiffness kd and damping cd of TMD are
order to achieve a more practical and economic selection of tuned in MOCS1. To achieve a practical design, a number of
TMD parameters, a free parameter of TMD (rather than a constraints should be considered on the TMD parameters
Optimum design of tuned mass dampers 3209
and maximum TMD displacement. Thus, the following opti- for each building. A general code is written in MATLAB soft-
misation problem is defined for MOCS1: ware [30] to carry out the time history analyses on the build-
Find : cd and kd ings subjected to seismic excitation.
Minimize md ; I1 ; I2 ranges of stiffness and damping for the TMD system are
assumed. Also, the maximum stroke length of the TMD is
8 adopted as 150 cm. Considering the buildings subjected to
>
<cd ðcd Þmax
ground artificial acceleration excitation, described in Section 5,
Subjected to kd ðkd Þmax ð18Þ
>
: the MOCS is employed to find the optimum parameters of
xmax ðTMDÞ XL TMD.
In order to reduce the structural responses in different Fig. 2 shows the Pareto-optimal front diagram of structural
earthquake excitations, the proposed strategies must be sub- case study 1 subjected to the band limited white noise excita-
jected to different earthquakes well. Therefore, a ground arti- tion in the case of MOCS1. It is obvious from this figure that
ficial acceleration is simulated in the optimisation procedure two objective functions, I1 and I2, are in conflict with each
by passing a Gaussian white noise process through a modified other. An appropriate trade-off between two conflicting objec-
form of the Kanai-Tajimi filter [28] with power spectral density tives is provided using MOCS. Each member of the Pareto
function as follows [29]: front can be represented by a vector in the design space. In
MOCS1, the vector that has the shortest distance to the origin,
4ng xg x I, is taken as the best tuning of TMD.
sðx Þ ¼ ð19Þ
x2 þ 2ng xg x þ x2g
where ng and xg are the ground damping and frequency,
respectively. In this study, ng ¼ 0:3 and xg ¼ 2p rad/s have
been used for numerical simulations. The output of this filter
simulates the earthquake which has been used in the optimisa-
tion procedure.
5. Numerical studies
Table 1 The optimum TMD parameters for different optimisation approaches (case study 1).
Optimum parameters GA [7] Lee et al. [25] CSS [27] MOCS1 MOCS2
md (tonne) 108 108 108 108 74.4154
cd (kNs/m) 151.5 271.79 88.697 160.5 69.3987
kd (kN/m) 3750 4126.93 4207.735 4428.7 2954.244
Optimum design of tuned mass dampers 3211
MOCS1 MOCS2
1.02 1.02
f opt 1.015 1
f opt
1.01
0.98
1.005
0.96
1
pi 2pi 3pi 4pi 5pi pi 2pi 3pi 4pi 5pi
ω (rad/s) ω (rad/s)
g g
MOCS1 MOCS2
0.12 0.085
0.08
ξ opt
ξ opt
0.115
d
d
0.075
0.11 0.07
pi 2pi 3pi 4pi 5pi pi 2pi 3pi 4pi 5pi
ω (rad/s) ω (rad/s)
g g
Fig. 4 The effect of the ground damping and frequency values of the artificial earthquake excitation on the optimum frequency and
damping ratio of TMD for the case study 1.
Fig. 5 Time history of the top floor displacement and acceler- Fig. 6 Time history of the top floor displacement and acceler-
ation subjected to the El Centro (1940) NS earthquake in the case ation subjected to the El Centro (1940) NS earthquake in the case
of MOCS1 (Case study 1). of MOCS2 (Case study 1).
3212 S. Etedali, H. Rakhshani
Table 2 Maximum displacements of stories subjected to the El Centro (1940) NS earthquake (case study 1).
Maximum displacement (m) Percentage of reduction
Story Without TMD GA [7] Lee et al. [25] CSS [27] MOCS1 MOCS2 GA [7] Lee et al. [25] CSS [27] MOCS1 MOCS2
1 0.031 0.019 0.020 0.019 0.019 0.019 38.71 35.48 40.32 38.71 37.17
2 0.060 0.037 0.039 0.036 0.037 0.038 38.33 35.00 39.67 38.33 36.53
3 0.087 0.058 0.057 0.053 0.053 0.055 33.33 34.48 39.65 39.08 36.39
4 0.112 0.068 0.073 0.068 0.068 0.071 39.29 34.82 39.11 39.29 36.65
5 0.133 0.082 0.087 0.083 0.082 0.085 38.35 34.59 37.97 38.35 35.81
6 0.151 0.094 0.099 0.095 0.093 0.098 37.75 34.44 37.09 38.41 35.06
7 0.166 0.104 0.108 0.106 0.104 0.109 37.35 34.94 36.39 37.35 34.59
8 0.177 0.113 0.117 0.114 0.112 0.117 36.16 33.90 35.65 36.72 33.91
9 0.184 0.119 0.123 0.120 0.118 0.123 35.33 33.15 35 35.87 33.04
10 0.188 0.122 0.126 0.123 0.122 0.126 35.11 32.98 34.84 35.11 32.74
TMD – 0.358 0.282 0.493 0.408 0.557 – – – – –
The mean value of the reduction 36.97 34.38 37.57 37.72 35.19
Table 3 Maximum acceleration of stories subjected to the El Centro (1940) NS earthquake (case study 1).
Maximum acceleration (m/s2) Percentage of reduction
Story Without TMD GA [7] Lee et al. [25] CSS [27] MOCS1 MOCS2 GA [7] Lee et al. [25] CSS [27] MOCS1 MOCS2
1 2.89 2.70 2.67 2.64 2.66 2.63 6.57 7.61 8.65 7.96 8.83
2 3.97 3.03 3.10 2.80 2.85 3.06 23.68 21.91 29.47 28.21 22.80
3 4.93 3.53 3.64 3.35 3.31 3.54 28.40 26.17 32.05 32.86 28.24
4 5.68 3.94 3.99 3.94 3.88 4.00 30.63 29.75 30.63 31.69 29.65
5 6.14 4.08 4.12 4.05 4.02 4.13 33.55 32.90 34.04 34.53 32.79
6 6.55 3.83 4.14 3.72 3.72 4.05 41.53 36.79 43.21 43.21 38.17
7 6.71 4.39 4.26 4.30 4.27 4.45 34.58 36.51 35.92 36.36 33.71
8 7.01 5.05 4.95 4.99 4.96 5.16 27.96 29.39 28.82 29.24 26.37
9 7.84 5.53 5.44 5.44 5.41 5.67 29.46 30.61 30.61 30.99 27.66
10 8.32 5.81 5.72 5.70 5.68 5.95 30.17 31.25 31.49 31.73 28.52
TMD – 13.94 11.66 19.04 15.54 21.69 – – – – –
The mean value of the reduction 28.65 28.29 30.49 30.68 27.67
Table 4 The percentage of reduction of the structural responses subjected to different earthquake excitations for different tuning
strategies (case study 1).
Earthquakes* Ij (j = 1, 2, 3) GA Lee et al. CSS MOCS1 MOCS2
Sylmar (1994) I1 1.26 9.50 0.23 9.39 4.22
FP component I2 8.08 17.51 8.09 19.59 11.15
I3 4.11 8.14 5.65 6.42 3.27
Sylmar (1994) I1 0.44 1.88 0.90 0.85 1.60
FN component I2 4.55 3.33 4.32 6.56 5.38
I3 5.87 5.54 6.65 7.42 5.69
Newhall (1994) I1 10.61 9.40 12.15 18.11 11.41
FP component I2 6.86 8.66 7.29 9.51 6.30
I3 10.39 9.62 11.84 12.09 9.29
Newhall (1994) I1 3.13 0.15 1.37 2.19 2.07
FN component I2 11.43 7.72 8.06 9.32 8.29
I3 8.35 5.07 5.17 5.37 5.03
Kobe (1995) I1 13.10 13.52 15.14 19.30 12.00
NS component I2 10.74 13.41 11.85 18.88 11.17
I3 16.91 18.52 18.56 19.70 14.85
Optimum design of tuned mass dampers 3213
Table 4 (continued)
Earthquakes* Ij (j = 1, 2, 3) GA Lee et al. CSS MOCS1 MOCS2
Table 5 The optimum TMD parameters for different optimisation approaches (case study 2).
Optimum parameters Sadek et al. [26] GA [7] CSS [27] MOCS1 MOCS2
md (tonne) 55.45 55.45 55.45 55.45 24.44
cd (kNs/m) 104.4 47.90 30.234 20.144 12.08
kd (kN/m) 464.1 437.90 355.758 377.80 156.41
Optimum design of tuned mass dampers 3215
MOCS1 MOCS2
0.8
0.84 0.795
f opt
f opt
0.79
0.82
0.785
0.8 0.78
pi 2pi 3pi 4pi 5pi pi 2pi 3pi 4pi 5pi
ω (rad/s) ω (rad/s)
g g
MOCS1 MOCS2
0.075
0.1
0.07
ξ opt
ξ opt
d
d
0.095
0.065
0.06 0.09
pi 2pi 3pi 4pi 5pi pi 2pi 3pi 4pi 5pi
ω (rad/s) ω (rad/s)
g g
Fig. 9 The effect of the ground damping and frequency values of the artificial earthquake excitation on the optimum frequency and
damping ratio of TMD for the case study 2.
6. Conclusions
Table 6 Maximum displacement of stories subjected to the El Centro (1940) NS earthquake (case study 2).
Maximum displacement (m) Percentage of reduction
Story Without Sadek et al. [26] GA [7] CSS MOCS1 MOCS2 Sadek et al. [26] GA [7] CSS MOCS1 MOCS2
TMD [27] [27]
1 0.041 0.036 0.034 0.031 0.030 0.035 12.20 17.07 25.37 25.71 14.64
2 0.088 0.077 0.072 0.066 0.065 0.075 12.50 18.18 25.56 25.95 14.63
3 0.129 0.113 0.105 0.095 0.094 0.109 12.40 18.60 26.67 27.28 15.41
4 0.166 0.145 0.134 0.121 0.119 0.140 12.65 19.28 27.4 28.17 15.79
5 0.197 0.172 0.160 0.143 0.141 0.165 12.69 18.78 27.41 28.18 16.06
6 0.222 0.194 0.184 0.163 0.162 0.185 12.61 17.12 26.35 26.96 16.46
7 0.252 0.219 0.210 0.186 0.186 0.199 13.10 16.67 26.06 26.20 20.91
8 0.286 0.245 0.236 0.209 0.210 0.215 14.34 17.48 26.61 26.60 24.74
9 0.313 0.266 0.258 0.228 0.230 0.231 15.02 17.57 26.54 26.46 26.14
10 0.327 0.281 0.272 0.241 0.243 0.243 14.07 16.82 25.78 25.64 25.78
TMD – 0.456 0.635 0.639 0.778 0.761
The mean value of the reduction 13.16 17.76 26.30 26.71 19.06
Optimum design of tuned mass dampers 3217
Table 7 Maximum acceleration of stories subjected to the El Centro (1940) NS earthquake (case study 2).
Maximum acceleration (m/s2) Percentage of reduction
Story Without Sadek et al. [26] GA [7] CSS MOCS1 MOCS2 Sadek et al. [26] GA [7] CSS MOCS1 MOCS2
TMD [27] [27]
1 3.06 3.06 3.05 3.01 3.00 3.01 0.00 0.33 1.63 1.80 1.57
2 2.77 2.66 2.63 2.64 2.65 2.70 3.97 5.05 4.69 4.42 2.48
3 2.98 2.71 2.78 2.78 2.79 2.81 9.06 6.71 6.71 6.39 5.56
4 2.76 2.71 2.7 2.65 2.64 2.62 1.81 2.17 3.99 4.22 5.05
5 2.93 2.69 2.67 2.61 2.61 2.61 8.19 8.87 10.92 11.00 10.94
6 3.14 2.95 2.85 2.67 2.63 2.84 6.05 9.24 14.97 16.33 9.44
7 2.92 2.45 2.34 2.15 2.10 2.51 16.10 19.86 26.37 28.10 13.98
8 3.68 2.71 2.7 2.49 2.50 2.91 26.36 26.63 32.34 32.07 20.96
9 4.7 3.25 3.2 3.00 2.98 3.74 30.85 31.91 36.17 36.52 20.38
10 5.36 3.99 3.86 3.61 3.64 4.21 25.56 27.99 32.65 32.15 21.38
TMD – 3.74 5.00 4.70 5.68 5.71
The mean value of the reduction 12.80 13.88 17.04 17.30 11.17
Table 8 The percentage of reduction of the structural responses subjected to different earthquake excitations for different tuning
strategies (case study 2).
Earthquakes* Ij (j = 1, 2, 3) Sadek et al. GA CSS MOCS1 MOCS2
Sylmar (1994) I1 12.52 19.68 26.76 28.75 24.71
FP component I2 1.66 2.20 3.59 5.38 4.19
I3 30.62 38.10 42.42 44.87 35.84
Sylmar (1994) I1 13.52 15.08 15.35 15.79 14.46
FN component I2 10.03 9.72 8.60 0.69 4.70
I3 23.73 22.73 21.72 23.60 21.35
Newhall (1994) I1 16.96 23.63 25.64 26.63 23.05
FP component I2 2.65 5.53 6.44 2.04 1.41
I3 22.42 21.00 22.66 22.58 20.24
Newhall (1994) I1 23.99 25.91 25.55 26.20 24.98
FN component I2 19.80 13.71 10.01 10.65 14.60
I3 17.87 16.51 14.64 15.50 13.48
Kobe (1995) I1 38.43 41.82 38.35 41.06 38.26
NS component I2 4.94 1.17 0.01 4.77 3.02
I3 3.18 0.70 2.49 1.39 0.35
Kobe (1995) I1 13.51 14.75 14.58 14.87 13.43
EW component I2 0.09 3.27 4.59 5.22 1.88
I3 22.26 18.54 16.42 16.64 14.56
Chichi (1999) I1 38.64 39.39 28.57 28.70 21.15
NS component I2 27.89 27.36 26.10 27.26 21.43
I3 43.22 42.67 28.88 28.47 25.05
Chichi (1999) I1 5.87 9.77 8.36 8.85 7.99
EW component I2 17.63 16.35 13.47 16.40 11.15
I3 13.49 13.29 10.83 11.56 10.60
Erzincan (1992) I1 17.75 25.52 25.16 27.96 22.91
NS component I2 26.96 24.61 23.28 24.84 19.69
I3 13.71 16.24 16.15 17.08 15.62
Erzincan (1992) I1 10.96 14.76 15.76 16.94 12.00
EW component I2 12.72 11.68 10.62 14.55 8.45
I3 24.57 25.32 23.90 24.35 20.29
Rinaldi (1994) I1 0.70 14.26 25.20 28.82 21.76
FP component I2 17.21 22.20 20.92 23.96 20.14
I3 20.91 33.85 36.36 36.89 33.00
(continued on next page)
3218 S. Etedali, H. Rakhshani
Table 8 (continued)
Earthquakes* Ij (j = 1, 2, 3) Sadek et al. GA CSS MOCS1 MOCS2
was capable of maintaining the desired level of reduction of structures: a multi-objective cuckoo search approach, ISA
structural responses in different earthquake excitations. In Trans. 67 (2017) 222–232.
other words, MOCS2 was able to present a solution leading [17] P. Civicioglu, E. Besdok, A conceptual comparison of the
to a more practical and economic selection of TMD Cuckoo-search, particle swarm optimization, differential
evolution and artificial bee colony algorithms, Artif. Intell.
parameters.
Rev. 39 (4) (2013) 315–346.
[18] A.A. Zamani, S. Tavakoli, S. Etedali, Control of piezoelectric
References friction dampers in smart base-isolated structures using self-
tuning and adaptive fuzzy proportional–derivative controllers, J.
[1] J. Den Hartog, Mechanical Vibrations, McGraw-Hill, New Intell. Mater. Syst. 28 (10) (2017) 1287–1302.
York, 1947. [19] S. Etedali, S. Tavakoli, M.R. Sohrabi, Design of a decoupled
[2] G. Warburton, Optimum absorber parameters for minimizing PID controller via MOCS for seismic control of smart
vibration response, Earthq. Eng. Struct. D 9 (3) (1981) 251–262. structures, Earthq. Struct. 10 (5) (2016) 1067–1087.
[3] G. Warburton, Optimum absorber parameters for various [20] S. Etedali, A new modified independent modal space control
combinations of response and excitation parameters, Earthq. approach toward control of seismic-excited structures, Bull.
Eng. Struct. D 10 (3) (1982) 381–401. Earthq. Eng. 15 (10) (2017) 4215–4243.
[4] A. Thompson, Optimum tuning and damping of a dynamic [21] X.-S. Yang, S. Deb, Multiobjective cuckoo search for design
vibration absorber applied to a force excited and damped optimization, Comput. Ind. Eng. 40 (6) (2013) 1616–1624.
primary system, J. Sound Vib. 77 (3) (1981) 403–415. [22] X.-S. Yang, S. Deb, Cuckoo search: recent advances and
[5] C. Chang, Mass dampers and their optimal designs for building applications, Neural Comput. Appl. 24 (1) (2014) 169–174.
vibration control, Eng. Struct. 21 (5) (1999) 454–463. [23] X.-S. Yang, S. Deb, Cuckoo search via Lévy flights, IEEE, 2009,
[6] C.-C. Lin, J.-F. Wang, J.-M. Ueng, Vibration control pp. 210–214.
identification of seismically excited mdof structure-PTMD [24] R. Rajabioun, Cuckoo optimization algorithm, Appl. Soft
systems, J. Sound Vib. 240 (1) (2001) 87–115. Comput. 11 (8) (2011) 5508–5518.
[7] M.N. Hadi, Y. Arfiadi, Optimum design of absorber for MDOF [25] C.-L. Lee, Y.-T. Chen, L.-L. Chung, Y.-P. Wang, Optimal
structures, J. Struct. Eng. – ASCE 124 (11) (1998) 1272–1280. design theories and applications of tuned mass dampers, Eng.
[8] M.P. Singh, S. Singh, L.M. Moreschi, Tuned mass dampers for Struct. 28 (1) (2006) 43–53.
response control of torsional buildings, Earthq. Eng. Struct. D [26] F. Sadek, B. Mohraz, A.W. Taylor, R.M. Chung, A method of
31 (4) (2002) 749–769. estimating the parameters of tuned mass dampers for seismic
[9] N.B. Desu, S. Deb, A. Dutta, Coupled tuned mass dampers for applications, Earthq. Eng. Struct. D 26 (6) (1997) 617–636.
control of coupled vibrations in asymmetric buildings, Struct. [27] A. Kaveh, S. Mohammadi, O.K. Hosseini, A. Keyhani, V.
Control Hlth. 13 (5) (2006) 897–916. Kalatjari, Optimum parameters of tuned mass dampers for
[10] S. Bakre, R. Jangid, Optimum parameters of tuned mass damper seismic applications using charged system search, IJST-T Civ.
for damped main system, Struct. Control Hlth. 14 (3) (2007) Eng. 39 (C1) (2015) 21.
448–470. [28] K. Kanai, An empirical formula for the spectrum of strong
[11] A.Y. Leung, H. Zhang, C. Cheng, Y. Lee, Particle swarm earthquake motions, 1961.
optimization of TMD by non-stationary base excitation during [29] S. Nagarajaiah, S. Narasimhan, Smart base-isolated benchmark
earthquake, Earthq. Eng. Struct. D 37 (9) (2008) 1223–1246. building. Part II: phase I sample controllers for linear isolation
[12] A. Leung, H. Zhang, Particle swarm optimization of tuned mass systems, Struct. Control Hlth. 13 (2–3) (2006) 589–604.
dampers, Eng. Struct. 31 (3) (2009) 715–728. [30] MATLAB, The MathWorks, Inc., Natick, Massachusetts,
[13] G.C. Marano, R. Greco, B. Chiaia, A comparison between United States, 2000.
different optimization criteria for tuned mass dampers design, J. [31] M. Singh, E. Matheu, L. Suarez, Active and semi-active control
Sound Vib. 329 (23) (2010) 4880–4890. of structures under sesimic excitation, Earthq. Eng. Struct. D 26
[14] G. Bekdasß, S.M. Nigdeli, Estimating optimum parameters of (2) (1997) 193–213.
tuned mass dampers using harmony search, Eng. Struct. 33 (9) [32] S. Etedali, S. Tavakoli, PD/PID controller design for seismic
(2011) 2716–2723. control of high-rise buildings using multi-objective optimization:
[15] J. Salvi, E. Rizzi, D. Moens, S. Jonckheere, A numerical a comparative study with LQR controller, J. Earthq. Tsunami
approach towards best tuning of tuned mass dampers, in: Proc. 11 (1) (2017) 1–23.
of 25th Int. Conf. on Noise and Vibration Engineering (ISMA [33] S. Etedali, M.R. Sohrabi, S. Tavakoli, An independent robust
2012), Leuven, Belgium, 2012, p. 19. modal PID control approach for seismic control of buildings, J.
[16] A.A. Zamani, S. Tavakoli, S. Etedali, Fractional order PID Civil. Eng. Urban 3 (5) (2013) 279–291.
control design for semi-active control of smart base-isolated