November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

International Journal of Structural Stability and Dynamics

c World Scientific Publishing Company

Optimal Pendulum Tuned Mass Damper Design Applied to High

Towers Using Genetic Algorithms: 2DOF Modelling

Gino B. Colherinhas∗; Marcus V. G. de Morais

Universidade de Brası́lia, Campus Universitário Darcy Ribeiro
Brası́lia, DF, Brazil
ginobertollucci@hotmail.com; mvmorais@unb.br

Maura A. M. Shzu; Suzana M. Avila

Universidade de Brası́lia, Campus UnB Gama
Gama, DF, Brazil
{maura;avilas}@unb.br

Received (Day Month Year)

Accepted (Day Month Year)

High and slender towers may experience excessive vibration levels caused by both wind
and seismic loads. To avoid excessive vibrations in towers Tuned Mass Dampers (TMDs)
are low-cost alternatives of passive control. Briefly, it is a damper which transfers part
of the energy of the vibration from the main structure to itself, working as a passive
device. These devices need to be finely tuned to work as dampers, otherwise, they could
amplify structural vibrations levels. This article presents optimal parameters of a pen-
dulum TMD (PTMD) to control the structural vibrations of slender towers subject to an
external white-noise force. The tower is modelled as a single degree of freedom (SDOF)
mass-spring system via assumed-mode procedure with a pendulum attached. An owner
Genetic Algorithm (GA) toolbox is used to find the optimal parameters of a PTMD, such
as the support flexural stiffness/damping, the mass-ratio and the pendulum length. The
chosen fitness function target a minimization of the maximum frequency peaks. The re-
sults are compared to a sensibility map which contains the information of the maximum
amplitude as a function of the pendulum length and the mass ratio between the pendu-
lum and the tower. The optimal parameters can be described as a power law function of
the supporting flexural stiffness. Furthermore, a parametric analysis and a time-history
verification are performed for several combinations of mass-ratio and pendulum length.

Keywords: Structural Control; Structural Dynamics; Pendulum Tuned Mass Damper;

Optimization; Genetic Algorithm

1. Introduction
High and slender towers are part of our daily lives. These structures are designed to
help specific needs (communication towers, transmission towers, or wind turbines)
or for tourist appreciation. Its low stiffness makes them more vulnerable to wind
or seismic excitations requiring the implementation of structural control systems.

∗ Universidadede Brası́lia, Campus Universitrio Darcy Ribeiro, Faculdade de Tecnologia, Bloco

G, Asa Norte, Brası́lia, Distrito Federal, 70910-900, Brasil.

1
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

2 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

Control systems are alternative options widely studied in the last years an can
be classified as passive, active, hybrid, and semi-active control. A passive control
system consist in adding one or more devices to the main structure to absorb or
transfer part of its energy1 and does not require an external power source.2 Passive
control typical mechanisms are for example: mass dampers which controls structural
response by transferring the energy between the main structure and an auxiliary
mass; metallic dampers that dissipate energy while deforming themselves and base
isolation systems that uncouple structure moving from seismic soil vibrations. The
TMD is a passive control device composed of a mass-spring-dashpot attached to
the structure, aiming to mitigates its vibration response.3
The process of tuning a TMD consists in defining optimum values of its stiff-
ness and damping in terms of the mass ratio between the absorber and the main
structure. A TMD tuned to the first structure natural frequency reduces substan-
tially the response associated to the first mode vibration while little reducing or
even increasing the response associated to higher modes. Moreover, a single TMD is
more sensitive to discrepancies on the first natural frequency and/or damping ratio
considered on the design. These limitations can be overcome by adding more than
one damper, each one of them tuned to a different vibration natural frequency.4
The beginning of the Tuned Mass Damper (TMD) appliance to civil structures
was at the sixties on high buildings, bridges, towers and industrial chimneys to help
suppress the wind-induced vibration. Later TMDs were designed for specific con-
struction such as wind turbines5, 6 and tall steel stacks.7 The performance enhance-
ment of the vibration control of wind turbine systems was reviewed by Rahman et
al,8 showing its control strategies (passive, active, semi-active), the types of vibra-
tion control dampers (Tuned Column Damper, Tuned Mass Damper, Controllable
Fluid Dampers), and some system controllers.
A PTMD is an alternative for the TMD where the mass of the absorber is replace
by a pendulum. For small amplitudes of vibration the PTMD can be considered
linear, which is the case of this current study. Tuning a PTMD consists in find
optimum values of stiffness and damping in function of the pendulum mass and
also of the pendulum length. Adding this new variable the dimension for search
optimal combinations of pendulum parameters becomes more complex, but also
allows more possibilities for optimal pendulum configurations.
Many researches investigate the design of optimum passive control systems.9–11
In addition, Deraemaeker and Soltani12 introduced an analytical formulae for the
optimum design of the linear PTMD coupled to an undamped primary system
applying Den Hartog’s equal peak method13 to derive the optimum design. La-
van14 design a multi-objective optimization of TMDs minimizing simultaneously the
structural responses, the TMD mass and the TMD stroke. Xiang and Nishitani15
performed both experimental and numerical studies to examine seismic performance
of the pendulum-type non-traditional tuned mass damper system (PNTTMD) to
control structural vibrations, founding satisfactory control of inter-story drift and
floor absolute acceleration, requiring small movement space. Oliveira et al.16 pro-
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 3

posed a set of general dimensionless optimal parameters for a PTMD, that can be
employed in the design to control the tower vibration, subjected to deterministic or
random dynamic loads, with different mass and damping ratios. Two TMD opti-
mization criteria applied to slender structures excited by wind load are investigated
by Morga and Marano:17 the reduction of the maximum displacement at the top of
the structure; the reduction of the maximum inertial acceleration at the top of the
structure. The wind load is defined by a superposition of the mean wind velocity
and a filtered white noise.
Murthag et al.18 also uses a passive control to mitigate the vibration level of a
simplified wind turbine model which includes the Blade/ Tower interaction and a
rotationally sampled turbulence. A robust optimal design criterion for a single TMD
device analysing a case of an structural vibration control of a main system subject
to stochastic dynamic loads was proposed by Marano et al.19 The dynamic input is
represented by a random base acceleration, modelled by a stationary filtered white
noise process. Bakre and Jangid20 search for optimum TMD parameters for different
damping ratios of the main system and the mass ratio of the TMD system, using
as external force and base acceleration a Gaussian white-noise random process, by
minimizing the root mean square (RMS) responses of the relative displacement
and the velocity of main mass and force transmitted to the support. Gerges and
Vickery21 uses a PTMD to reduce the structure RMS displacement subjected to
force and accelerations simulating random excitations of seismic and wind loads as
a white noise.
Some meta-heuristic methods have been used to efficiently and quickly tuning
optimal TMD design parameters such as genetic algorithm (GA),22–24 cuckoo search
(CS) algorithm,25, 26 harmony search (HS) algorithm,27–29 bionic algorithm,30 ant
colony optimization (ACO),31 particle swarm optimization (PSO),32, 33 and hybrid
approaches such as an adaptive genetic simulated annealing method.34 In addition
Mohebbi35 uses Genetic Algorithms (GAs) for design optimal Multiple Tuned Mass
Dampers (MTMDs) to mitigate the seismic response of structures. They considered
the parameters of TMDs as variables and aimed the minimization of the maximum
structural response as an objective function while a number of constraints have
been applied on TMDs response and parameters. The TMD has also been designed
to suppress the vibration of a barge-type offshore floating wind turbine by using a
genetic optimization.5
In this work, an owner GA toolbox36 is used to find the optimal PTMD param-
eters (the support flexural stiffness/damping, the pendulum length and the mass-
ratio between the pendulum and the main structure). The chosen fitness function
targets a minimization of the maximum frequency response peaks of the tower.
The GA results are compared to a parametric analysis that relates the maximum
tower displacement as a function of the mass-ratio and the pendulum length for a
specific set of stiffness/ damping. Proving that the GA optimization can achieve
the same results as the parametric optimum cases, we extend the use of the GA
toolbox to specific sets of stiffness/ damping by governing them in power law func-
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

4 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

tions. This design process of PTMD can be a good alternative to mitigates the
vibration of slender towers in comparison of other interactive analytical/numerical
or meta-heuristic procedures because it can be applied directly in terms of the
frequency response. Thus, in addition to this method being used to find optimal
PTMD parameters in an organized fashion toward a common purpose quickly, it
can be extended for complex structures, such as models using finite elements.37 A
time-history was performed for an optimal PTMD to show graphically the tower
displacement under a white-noise force.

2. Motion equations of the 2DOF system

The high, flexible, slender tower is consider as a simplified single degree of freedom
(SDOF) system. Classically, the equivalent modal parameters (stiffness and mass)
are determined using a Rayleigh-Ritz method.38
To reduce the tower system into a SDOF we assume the separated space-time
transversal displacement function w(z, t) ' wh (z, t) = φ(z)y(t), where y(t) is the
tip transversal displacement, φ(z) = 1 − cos(πz/2H) the fundamental modal shape
assumed, z the position along tower axis, and H the tower height.
Substituting the separated space-time function w(z, t) into the equation for free
vibration of a continuous beam with a tip mass, subjected to a distributed force
fs (t), Avila et al.39 obtained the dynamic response Ms ÿ(t) + Ks y(t) = fs (t), where
the equivalents stiffness and mass of the tower are respectively (2.1) and (2.2).

π4
Ks = EI (2.1)
32H 3

M
Ms = (3π − 8) + m (2.2)
2π

where M is the tower mass, m tip mass, H tower height, E Young Modulus, I
second moment of area for bending.
Before reduced to a SDOF model (corresponds to the mode to be controlled), the
dynamic behaviour of a PTMD is coupled giving a 2DOF discrete system described
by Fig. 1.3

2.1. 2DOF System: Tower with a PTMD

The equations of motion of the 2DOF system, illustrated by the Fig. 1, are given
by Eq. 2.3.

(Ms + Mp )ÿ + Mp Lp θ̈ + Cs ẏ + Ks y = Fs (t) (2.3)

Mp Lp ÿ + Mp L2p θ̈ + Cp θ̇ + (Kp + Mp gLp )θ = 0
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 5

Fig. 1. Structure with a linear pendulum attached (2DOF) excited by a force Fs (t).

These equations can be rewritten in a matrix form as shown by the Eq. 2.4.
     
(Ms + Mp ) Mp Lp ÿ Cs 0 ẏ
2 + +
Mp Lp Mp Lp θ̈ 0 Cp θ̇
    
Ks 0 y F (t)
+ = s (2.4)
0 (Kp + Mp gLp ) θ 0
where Ms main equivalent mass; Cs : main equivalent damping; Ks : main equivalent
stiffness; Mp : pendulum mass; Cp : pendulum damping; Kp : pendulum stiffness; L:
cable length; g: gravity acceleration; Fs (t) = Fs0 eiωt : excitation modal force; y(t):
main system displacement; θ(t): pendulum angular displacement.
Considering Fs (t) = eiωt , y(t) = Hy (ω)eiωt and θ(t) = Hθ (ω)eiωt , (2.4) can be
rewritten by the linear equation system (2.5).
    
A11 A12 Hy (ω) 1
= (2.5)
A21 A22 Hθ (ω) 0
where Hy (ω) and Hθ (ω) are respectively the structure and the pendulum response
function in the frequency domain:
A11 = −(Ms + Mp )ω 2 + Cs iω + Ks ,
A12 = A21 = −Mp Lp ω 2 ,
A22 = −Mp L2p ω 2 + Cp iω + (Kp + Mp gLp ).
16
solves the linear equation system (2.5) given the Frequency Response Func-
tions (FRFs) Hy (ω) (2.6) and Hθ (ω) (2.7).
Ay0 + Ay1 ω + Ay2 ω 2
Hy (ω) = (2.6)
B0 + B1 ω + B2 ω 2 + B3 ω 3 + B4 ω 4

Aθ0 + Aθ1 ω + Aθ2 ω 2

Hθ (ω) = (2.7)
B0 + B1 ω + B2 ω 2 + B3 ω 3 + B4 ω 4
where
the pendulum response terms are:
Ay0 = Lp Mp g + Kp ;
Ay1 = iCp L2p ;
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

6 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

Ay2 = −L2p Mp ;
the structure response terms are:
Aθ0 = 0;
Aθ1 = 0;
Aθ2 = Lp Mp ;
the common terms are:
B0 = Ks (Kp + Lp Mp g);
B1 = i(Cs Kp + Cp Ks L2p + Cs Lp Mp g);
B2 = −(Kp Mp + Kp Ms + Cp Cs L2p + Ks L2p Mp + Lp Mp2 g + Lp Mp Ms g);
B3 = −iL2p (Cp Ms + Cs Mp + Cp Mp );
B4 = L2p Mp Ms .
We aim to reduce the vibrational displacement of the main structure installing
a PTMD configured by its length Lp , mass Mp , damping Cp and stiffness Kp .
The maximum displacement amplitude of main structure is given by the expression
max(Hy (ω)).

2.2. Case example

This case example follow the Murtagh et al.40 model of a steel tower with a tip
mass with realistic dimensions of a simplified wind turbine with tower mass M =
66, 253.13 kg, tip mass m = 19, 876.00 kg, height H = 60 m, and flexural rigidity
EI = 3.29 × 1010 N m2 . Using the equations (2.1) and (2.2) the generalized stiffness
Ks = 463, 671.26 N/m and mass Ms = 34, 899.60 kg are obtained. The tower
damping was considered negligible (Cs ≈ 0).
Some PTMDs are based on viscous12 or friction damping,41 but the current
study is limited to PTMDs with friction dampers. The torsional stiffness and damp-
ing are defined as Kp = 1, 247.90 kN/m, and Cp = 9, 024.90 Nms, respectively.42
Setting the same PTMD design project values used by Shzu et al.,43 ie the mass
ratio µ = Mp /Ms = 0.2636 (Mp = 9, 198.6 kg) and the pendulum length Lp = 4 m,
the tower frequency response can be computed as shown in Fig. 2.

10 -2
2DOF
X: 0.58
Without TMD
Y: 0.00538
Hy (Ω) (m)

10 -4
X: 0.432
Y: 0.0001513 X: 0.707
Y: 7.948e-05

10 -6 X: 0.526
Y: 1.848e-07

0 0.2 0.4 0.6 0.8 1 1.2

Ω (Hz)

Fig. 2. Case Example: Tower FRF with and without PTMD

November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 7

3. Parametric Study - Response Maps

A parametric study is represented in a response map form as a collection of the
maximum displacement peaks max(Hy (ω)) in function of the pendulum length Lp
and mass ratio µ. The response map represents a sort of objective function that we
want to optimize.
Figure 3 represents the response map considering the presented case example,
showing a locus with minimum vibration displacements that corresponds a valley
on the surface. This geometric locus represents optimal combinations between µ
and Lp of the 2DOF model.

Fig. 3. Response map - 3D view

In Figure 4 the response map in the superior view shows clearly the curvilinear
shape of the optimal locus. Analysing the results obtained by the response map we
selected two specific points considering the same mass ratio µ = 0.2636 referred to
the case example (Subsection 2.2):

• For Lp = 4.00 → Hy (ω) = e−8.804 = 1.51 · 10−4 m;

• For Lp = 4.36 → Hy (ω) = e−9.057 = 1.16 · 10−4 m.

Despite the pendulum length Lp = 4 m of the case example is not the optimal,
the magnitude order is quite similar. The second point selected in the response map
surface has the optimal pendulum length Lp = 4.36 for the mass ratio µ = 0.2636.
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

8 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

Fig. 4. Response map - Lp vs µ view

3.1. Influences of damping Cp and stiffness Kp

To analyse the sensitivity of the torsional damping over the dynamic effects of
the structure, two cases Cp = {0.5; 1.5} × 104 N ms are considered using the same
stiffness (Kp = 1247900N/m) and the same mass ratio (µ = 0.26). The geometric
locus obtained by both response maps have the same shape. However, the frequency
response increases as lower the damping is.
• For Cp = 0.5 × 104 N ms → Hy = 1.1 · 10−4 m.
• For Cp = 1.5 × 104 N ms → Hy = 0.7 · 10−4 m.
This behaviour extends for another values of Cp . The form and position of the
curve on the response map remains unchanged in plan Lp vs. µ as shown Fig. 5.
For the torsional stiffness, the values Kp = {0.5; 1.5} × 106 N/m are analysed using

Fig. 5. Response map for Cp = {0.5; 1.5} × 104 N ms

the same damping (Cp = 9024.9N ms) and the same mass ratio (µ = 0.26).
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 9

• For Kp = 0.5 · 106 N/m → Hy = 5.5 · 10−5 m.

• For Kp = 1.5 · 106 N/m → Hy = 13.8 · 10−5 m.
It’s concluded that the valley shifts to the right in plan Lp vs. µ. When the stiffness
Kp increases the response values Hy (ω) also increases as shown Fig. 6.

Fig. 6. Response map for Kp = {0.5; 1.5} × 106 N/m

4. Genetic Algorithm Optimization (GA)

To assist the response map analysis it was used an owner GA toolbox.36
The pendulum length Lp and the mass ratio µ were chosen as the GA opti-
mization parameters. The other parameters were defined as the same as in the case
example of subsection 2.2. The initial population IP = [Lp ; µ] is created restricting
the variables in the following ranges:
0.50 ≤ Lp ≤ 10.00
0.1 ≤ µ ≤ 0.35 (4.1)

4.1. Fitness Function

The purpose of this optimization is to minimize the frequency response peaks of
the the tower described in the Analytical 2DOF formulation (Eq. 2.6). The fitness
function minimizes Hy (ω) maximizing its inverse.
1
ff itness = , i = 1, 2, . . . , Nind (4.2)
max Hy (ω)i
where i is the chromosome of the population Nind .

4.2. Results
Once the fitness function ff itness was defined, the optimization was performed
several times to define the best parameters of the GA toolbox in order to find the
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

10 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

case of faster convergence, as follows:

• Ngen = 100, number of generations;

• Nind = 100, number of individuals in the population;
• pc = 60%, crossover probability;
• pm = 2%, mutation probability;
• pelit = 2%, elitism probability;
• pdec = 20%, decimation probability;
• Ndec = 20, step of generation for the occurence of decimation.

Three hundred best results was gathered from GA and superimposed on the
response map. It was noticed that the set of the best results from optimization
process took up almost the same place of the best parametric results, Fig. 7 over
the µ vs Lp plan. It is noticed that the optimization results are very close to the
response map valley. The GA toolbox presented a good convergence and the quality

Fig. 7. The 300 optimal results of Lp and µ over the response map

of being fast. This feature defines the GAs as versatile tools which can be used to
optimize several engineering problems.

4.3. Results Analysis

A study over the results is carried out to estimate the appropriated regressions of
the locus obtained by the response map.
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 11

The optimization process carried out 150 loops. In Figure 8 the GA results
are gathered on µ vs Lp plan and a power regressions in µ = ai Lbpi + ci form are
computed, where ai , bi , and ci are constants for each Kp .

0.24

0.22

0.2

0.18

0.16
X: 3.538
Y: 0.1401
0.14

0.12

0.1
3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5
Lp

Fig. 8. Power regression of the (µ; Lp ) optimization results for different values of Kp (log x log
scale)

These power functions presents a certain linearity in log-log scale. The designer
can easily select in Fig. 8 the optimal pendulum parameters for Lp , µ and Kp , since
the damping Cp does not affect the behaviour of these curves.

5. Optimization results validation

To validate the presented implementation, we perform a case study, a parametric
analysis to several possibilities, and a time-history analysis example.

5.1. Case Study

This case study is based on the model presented in subsection 2.2, excluding the
related pendulum parameters (Kp , Cp , µ, Lp ) which are our goals. In this Case
Study we suggest a project methodology in five steps for the design of an optimal
PTMD.
1st step: Tower dynamic analysis. It is suggested a study of the dynamic
behaviour of the tower without the PTMD. We already have made these preliminary
analysis on subsection 2.2 (see Fig. 2).
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

12 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

2nd step: See the influence of the PTMD damping. In the current study
the subsection 3.1 explain that the damping do not change the geometric locus
position of the response map over the plan µ × Lp , but affects directly the FRF
amplitude (when damping increases the response decreases). Resuming, the damp-
ing do not affect the selection of other pendulum parameters. We set an arbitrary
damping Cp∗ = 9024.9 N ms (same value of the example case).
3nd step: See the influence of the PTMD stiffness. In subsection 3.1 it’s
concluded that the stiffness affects the response amplitude over the plan µ×Lp . It’s
also concluded that exists many cases of optimal PTMD designs. Lower amplitude
responses are found for lower stiffness values. We select Kp∗ = 0.50 × 106 N/m
because it is our lower stiffness data.
4nd step: Select the other PTMD parameters The selection of the pendu-
lum length and the mass ratio are consequence of the stiffness selected by using the
power regression curves. Different optimal values can be selected by respecting the
designer preferences and design restrictions. The mass ratio µ∗ = 0.14 is arbitrarily
selected and the length L∗p = 3.54 m is according to the power regression curve for
Kp∗ = 0.50 × 106 N/m (Fig. 8).
5nd step: Tower with a PTMD dynamic analysis. Using the designed
PTMD optimal values (Kp∗ = 0.50 × 106 N/m, Cp∗ = 9024.9 N ms, µ∗ = 0.14,
L∗p = 3.54 m) the 2DOF FRF can be seen in Fig. 9. Two variations of this case

10 -2
2DOF
X: 0.58
Without TMD
Y: 0.00538
X: 0.664
Hy (Ω) (m)

10 -4 Y: 4.536e-05

X: 0.46
Y: 4.728e-05

10 -6
X: 0.529
Y: 8.295e-07
0 0.2 0.4 0.6 0.8 1 1.2
Ω (Hz)

Fig. 9. Case Study: Tower FRF with and without PTMD

study are performed to study the sensibility of the length Lp and the mass ratio µ.
First we set the pendulum length to Lp = 0.8 × L∗p m and Lp = 1.2 × L∗p m
maintaining the optimal Kp∗ , Cp∗ , and µ∗ values. The comparison between the FRF
case study and pendulum length variations are shown in Fig. 10. Then we set the
mass ratio to µ = 0.8 × µ∗ m and µ = 1.2 × µ∗ m maintaining the optimal Kp∗ ,
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 13

10 -2 Without TMD
2DOF (K *p ,C*p , µ * , 0.8×L*p )

2DOF (K *p ,C*p , µ * , L *p )
10 -3
2DOF (K *p ,C*p , µ * , 1.2×L*p )
Hy (Ω) (m)

10 -4

10 -5

10 -6

10 -7
0 0.2 0.4 0.6 0.8 1 1.2
Ω (Hz)

Fig. 10. Comparison between the case study and pendulum length variations

Cp∗ , and L∗p values.The comparison between the FRF case study and mass ratios
variations are shown in Fig. 11.

10 -2 Without TMD
2DOF (K * ,C* , 0.8×µ * , L * )
p p p

10 -3 2DOF (K *p ,C*p , µ * , L *p )

2DOF (K *p ,C*p , 1.2×µ * , L *p )

Hy (Ω) (m)

10 -4

10 -5

10 -6

10 -7
0 0.2 0.4 0.6 0.8 1 1.2
Ω (Hz)

Fig. 11. Comparison between the case study and mass ratios variations

Over all these variations we conclude that designed case really has the minimum
response amplitude peaks. When we change the values of µ or Lp one of the peaks
increase and the another decrease, reducing its fitness value.
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

14 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

5.2. Parametric Validation

Maintaining the same damping, Figures 12 and 13 shows the FRFs varying the
stiffness (Kp = [0.50; 0.75; 1.00; 1.25; 1.50; 1.75; 2.00] × 106 N/m) for both mass
ratios µ = 0.1 and µ = 0.2.

6
µ=0.1; K p =0.50×10 ; L p =3.98
10 -4 X: 0.479 X: 0.655
Y: 0.0001533 Y: 0.000151 µ=0.1; K p =0.75×10 6 ; L p =4.77

µ=0.1; K p =1.00×10 6 ; L p =5.43

X: 0.479 X: 0.654
µ=0.1; K =1.25×10 6 ; L =6.00
Y: 4.376e-05 Y: 4.382e-05 p p

µ=0.1; K =1.50×10 6 ; L =6.54

10 -5 p p
Hy (Ω) (m)

6
µ=0.1; K p =1.75×10 ; L p =7.02
µ=0.1; K p =2.00×10 6 ; L p =7.48

10 -6
X: 0.541
Y: 1.192e-06

X: 0.541
Y: 3.373e-07
-7
10
0 0.2 0.4 0.6 0.8 1 1.2
Ω (Hz)

Fig. 12. Parametric validation: Comparison of the Frequency Response Function for µ = 0.1

6
X: 0.439 µ=0.2; K p =0.50×10 ; L p =3.13
10 -4 X: 0.677
Y: 0.0001665 Y: 0.0001554 µ=0.2; K =0.75×10 6 ; L =3.72
p p

X: 0.439 µ=0.2; K p =1.00×10 6 ; L p =4.23

X: 0.678
Y: 5.371e-05 Y: 4.634e-05 µ=0.2; K p =1.25×10 6 ; L p =4.65

µ=0.2; K p =1.50×10 6 ; L p =5.04

10 -5
Hy (Ω) (m)

µ=0.2; K p =1.75×10 6 ; L p =5.37

µ=0.2; K p =2.00×10 6 ; L p =5.75

X: 0.515
10 -6 Y: 5.586e-07

X: 0.512
Y: 1.684e-07

10 -7
0 0.2 0.4 0.6 0.8 1 1.2
Ω (Hz)

Fig. 13. Parametric validation: Comparison of the Frequency Response Function for µ = 0.2

Table 1 shows the amplitude response peaks for the combinations of µ, Kp , and
Lp .
Analysing the results of Figs. 12, 13, and Tab. 1 we conclude that the control
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 15

Table 1. Maximum amplitude responses in function of the stiffness

Kp × 106 (N/m) 0.50 0.75 1.00 1.25 1.50 1.75 2.00

µ = 0.1
max |Hy (ω)| × 10−5 (m) 4.38 6.27 8.13 10.00 11.78 13.35 15.33
Lp (m) 3.98 4.77 5.43 6.00 6.54 7.02 7.48
µ = 0.2
max |Hy (ω)| × 10−5 (m) 5.37 7.50 9.34 11.67 13.72 15.28 16.65
Lp (m) 3.13 3.72 4.23 4.65 5.04 5.37 5.75

frequency zone increases when mass ratio µ increase and lower response amplitudes
are found for lower values of Kp .
The minimum response peak in the Tab. 1 is max |Hy (ω)| = 4.38 × 10−5 m
(Kp = 0.50 × 106 N/m, Cp = 9024.9 N ms, µ = 0.1, and Lp = 3.98 m).
For each value of µ the natural frequencies considering stiffness’s variations are
the same but the amplitude response peaks increases when the pendulum stiffness
increases.

5.3. Time History

The behaviour of a white noise wind effect are performed for a five hour time history
duration considering a 1000N forcing amplitude. Figure 14 shows the time history
of a tower with and without the PTMD by using one of the optimal response cases:
Kp = 0.50 × 106 N/m, Cp = 9024.9 N ms, µ = 0.1, and Lp = 3.98 m.

6
uncontrolled
controlled
4

2
y(t) (cm)

-2

-4

-6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
t (h)

Fig. 14. Time history of the tower controlled and uncontrolled by a PTMD

The effectiveness of the optimal PTMD passive control type can be compared.
While the tower effective RMS without the control becomes yrms−uncontrolled =
2.205cm, the pendular control system can reach yrms−controlled = 0.170cm. PTMD
can reduce over than 92% of the effective RMS tip displacement comparing to the
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

16 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

case without control.

6. Conclusions
The GA toolbox allows an easy identification of a geometric locus that contains
optimal PTMD parameters using an analytical 2DOF solution of a tower with a
PTMD. A response maps sensibility study identifies the influence of the stiffness and
damping of the pendulum over the frequency response peaks of the tower. Using
Genetic Algorithms, power regression curves carried of optimal data are created
in function of the pendulum stiffness, length and mass. A design methodology are
suggested allowing the selection of optimal pendulum configurations. We conclude in
the parametric validation section that when we increase the mass ratio between the
pendulum and tower, the zone control also increases and increasing the pendulum
stiffness maintaining the same mass ratio will increase the FRF response peaks. A
time history using a behaviour of a white noise wind effect shows that the PTMD
can reduce the tip tower vibration over than 92%. The tower life time can be
extended by reducing its vibration and in consequence its fatigue.

Acknowledgments
The authors acknowledge CNPq (Brazilian Scientific Conseil) and MCTI (Indus-
trial, Science and Technology Ministry) by financial support (Process 406895/2013-
9) referent to scientific project edital MCTI/CNPq 74/2013 Capacitação laborato-
rial e formação de recursos humanos em Energia Eólica (Linha 1) for the project
entitled ”Dinâmica de Aerogeradores: Modelagem e Simulação”.

References
1. M. A. Lackner and M. A. Rotea, Passive structural control of offshore wind turbines,
Wind Energy 14(3) (2011) 373–388.
2. G. W. Housner, L. A. Bergman, T. K. Caughey, A. G. Chassiakos, R. O. Claus, S. F.
Masri, R. E. Skelton, T. T. Soong, B. F. Spencer and J. T. P. Yao, Structural control:
Past, present, and future, Journal of Engineering Mechanics 123(9) (1997) 897–971.
3. T. T. Soong and G. F. Dargush, Passive energy dissipation systems in structural
engineering chichester, 1997 isbn 0-471-96821-8, Journal of Structural Control 6(1)
(1999) 172–172.
4. R. B. Carneiro, S. M. Avila and J. L. V. de Brito, Parametric study on multiple
tuned mass dampers using interconnected masses, International Journal of Structural
Stability and Dynamics 08(01) (2008) 187–202.
5. E.-M. He, Y.-Q. Hu and Y. Zhang, Optimization design of tuned mass damper for
vibration suppression of a barge-type offshore floating wind turbine, Proceedings of the
Institution of Mechanical Engineers, Part M: Journal of Engineering for the Maritime
Environment 231(1) (2017) 302–315.
6. G. M. Stewart and M. A. Lackner, The impact of passive tuned mass dampers and
windwave misalignment on offshore wind turbine loads, Engineering Structures 73
(2014) 54 – 61.
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 17

7. N. Areemit and P. Warnitchai, Vibration suppression of a 90-m-tall steel stack by us-

ing a high-damping tuned mass damper, in Proceedings of the Eighth East Asia-Pacific
Conference on Structural Engineering and Construction, Nanyang Technological Uni-
versity, Singapore 2001.
8. M. Rahman, Z. C. Ong, W. T. Chong, S. Julai and S. Y. Khoo, Performance en-
hancement of wind turbine systems with vibration control: A review, Renewable and
Sustainable Energy Reviews 51 (2015) 43 – 54.
9. B. Keshtegar and S. Etedali, Novel mathematical models based on regression analysis
scheme for optimum tuning of tmd parameters, Journal of Solid and Fluid Mechanics
6(4) (2016) 59–75.
10. D. Orlando and P. B. Goncalves, Hybrid nonlinear control of a tall tower with a
pendulum absorber, Structural Engineering and Mechanics 46(2) (2013) 153–177.
11. H.-C. Tsai and G.-C. Lin, Optimum tuned-mass dampers for minimizing steady-
state response of support-excited and damped systems, Earthquake Engineering &
Structural Dynamics 22(11) (1993) 957–973.
12. A. Deraemaeker and P. Soltani, A short note on equal peak design for the pendulum
tuned mass dampers, Proceedings of the Institution of Mechanical Engineers, Part K:
Journal of Multi-body Dynamics 231(1) (2017) 285–291.
13. H. J. P. Den, Mechanical vibrations. fourth edition. mcgraw-hill, new york, 1956. 67s.
6d., The Journal of the Royal Aeronautical Society 61(554) (1957) p. 139139.
14. O. Lavan, Multi-objective optimal design of tuned mass dampers, Structural Control
and Health Monitoring 24(11) (2017) e2008–n/a, e2008 stc.2008.
15. P. Xiang and A. Nishitani, Structural vibration control with the implementation of a
pendulum-type nontraditional tuned mass damper system, Journal of Vibration and
Control 23(19) (2017) 3128–3146.
16. F. d. S. Oliveira, A. Gomez, S. M. Avila and J. Brito, Design criteria for a pendulum
absorber to control high building vibrations, International Journal od Innovations in
Material Science and Engineering (IMSE) 1(2) (2014) 82–89.
17. M. Morga and G. C. Marano, Optimization criteria of tmd to reduce vibrations gen-
erated by the wind in a slender structure, Journal of Vibration and Control 20(16)
(2014) 2404–2416.
18. P. J. Murtagh, A. Ghosh, B. Basu and B. M. Broderick, Passive control of wind
turbine vibrations including blade/tower interaction and rotationally sampled turbu-
lence, Wind Energy 11(4) (2008) 305–317.
19. G. C. Marano, S. Sgobba, R. Greco and M. Mezzina, Robust optimum design of
tuned mass dampers devices in random vibrations mitigation, Journal of Sound and
Vibration 313(3) (2008) 472 – 492.
20. S. V. Bakre and R. S. Jangid, Optimum parameters of tuned mass damper for damped
main system, Structural Control and Health Monitoring 14(3) (2007) 448–470.
21. R. R. Gerges and B. J. Vickery, Optimum design of pendulum-type tuned mass
dampers, The Structural Design of Tall and Special Buildings 14(4) (2005) 353–368.
22. N. B. Desu, S. K. Deb and A. Dutta, Coupled tuned mass dampers for control of
coupled vibrations in asymmetric buildings, Structural Control and Health Monitoring
13(5) (2006) 897–916.
23. G. C. Marano, R. Greco and B. Chiaia, A comparison between different optimization
criteria for tuned mass dampers design, Journal of Sound and Vibration 329(23)
(2010) 4880 – 4890.
24. R. Frans and Y. Arfiadi, Designing optimum locations and properties of mtmd sys-
tems, Procedia Engineering 125(Supplement C) (2015) 892 – 898, Civil Engineering
Innovation for a Sustainable.
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

18 Gino B. Colherinhas, Marcus V. G. de Morais, Maura A. M. Shzu, Suzana M. Avila

25. A. H. Heidari, S. Etedali and M. R. Javaheri-Tafti, A hybrid lqr-pid control design

for seismic control of buildings equipped with atmd, Frontiers of Structural and Civil
Engineering 12 (Mar 2018) 44–57.
26. S. Etedali and N. Mollayi, Cuckoo search-based least squares support vector machine
models for optimum tuning of tuned mass dampers, International Journal of Struc-
tural Stability and Dynamics 18(02) (2018) p. 1850028.
27. B. Keshtegar and S. Etedali, Nonlinear mathematical modeling and optimum design
of tuned mass dampers using adaptive dynamic harmony search algorithm, Structural
Control and Health Monitoring 25(7) (2018) p. e2163, e2163 STC-16-0282.R1.
28. H. Zhang and L. Zhang, Tuned mass damper system of high-rise intake towers op-
timized by improved harmony search algorithm, Engineering Structures 138(Supple-
ment C) (2017) 270 – 282.
29. S. M. Nigdeli and G. Bekdaş, Optimum tuned mass damper design in frequency
domain for structures, KSCE Journal of Civil Engineering 21 (Mar 2017) 912–922.
30. R. Steinbuch, Bionic optimisation of the earthquake resistance of high buildings by
tuned mass dampers, Journal of Bionic Engineering 8(3) (2011) 335 – 344.
31. A. Farshidianfar and S. Soheili, Ant colony optimization of tuned mass dampers for
earthquake oscillations of high-rise structures including soilstructure interaction, Soil
Dynamics and Earthquake Engineering 51(Supplement C) (2013) 14 – 22.
32. A. Leung and H. Zhang, Particle swarm optimization of tuned mass dampers, Engi-
neering Structures 31(3) (2009) 715 – 728.
33. M. Shahi, M. R. Sohrabi and S. Etedali, Seismic control of high-rise buildings equipped
with atmd including soil-structure interaction effects, Journal of Earthquake and
Tsunami 12(03) (2018) p. 1850010.
34. R. Xu, J. Chen and X. Zhu, A hybrid approach for parameter optimization of multiple
tuned mass dampers in reducing floor vibrations due to occupant walking: Theory and
parametric studies, Advances in Structural Engineering 20(8) (2017) 1232–1246.
35. M. Mohebbi, K. Shakeri, Y. Ghanbarpour and H. Majzoub, Designing optimal mul-
tiple tuned mass dampers using genetic algorithms (gas) for mitigating the seismic
response of structures, Journal of Vibration and Control 19(4) (2013) 605–625.
36. G. B. Colherinhas, Genetic Algorithm Toolbox and Its Applications in Engineering
(portuguese version) (Master’s dissertation in Mechanical Sciences of the University
of Brasilia, 2016).
37. G. B. Colherinhas, M. A. Shzu, S. M. Avila and M. V. de Morais, Wind tower vi-
bration controlled by a pendulum tmd using genetic optimization: Beam modelling,
Procedia Engineering 199 (2017) 1623 – 1628, X International Conference on Struc-
tural Dynamics, EURODYN 2017.
38. L. Meirovitch, Analytical methods in vibrationsMacmillan series in advanced mathe-
matics and theoretical physics, Macmillan series in advanced mathematics and theo-
retical physics (Macmillan, 1967).
39. S. M. Avila, M. A. M. Shzu, W. M. Pereira, L. S. Santos, M. V. G. Morais and Z. J. G.
Prado, Numerical modeling of the dynamic behavior of a wind turbine tower, Journal
of Vibration Engineering and Technologies 4(3) (2016) p. 249 257.
40. P. Murtagh, B. Basu and B. Broderick, Simple models for natural frequencies and
mode shapes of towers supporting utilities, Computers & Structures 82(20) (2004)
1745 – 1750.
41. L. L. Chung, L. Y. Wu, K. H. Lien, H. H. Chen and H. H. Huang, Optimal design
of friction pendulum tuned mass damper with varying friction coefficient, Structural
Control and Health Monitoring 20(4) (2012) 544–559.
42. A. L. Z. Gómez, Vibration Control of Structures Subjected to Dynamic Loads Using
November 10, 2018 15:27 WSPC/INSTRUCTION FILE ws-ijssd

Optimal Pendulum Tuned Mass Damper Design Applied to High Towers Using Genetic Algorithms: 2DOF Modelling 19

Tuned Pendulum-Shaped Mass Dampers (portuguese version) (Master’s dissertation

in Structures and Civil Construction, University de Brasilia, 2007).
43. M. A. M. Shzu, M. V. G. Morais, Z. J. G. del Prado and S. M. vila, Finite Element
Analysis of a Wind Turbine Tower with a Pendulum Tuned Mass Damper (DINAME
2015 - XVII International Symposium on Dynamic Problems of Mechanics, ABCM,
RN, Brazil, Frebruary 22-27, 2015).