Engineering Structures 126 (2016) 174–186

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Investigations of the application of gyro-mass dampers with various

types of supplemental dampers for vibration control of building
structures
Reza Mirza Hessabi ⇑, Oya Mercan
Department of Civil Engineering, University of Toronto, Toronto, ON, Canada

a r t i c l e i n f o a b s t r a c t

Article history: A gyro-mass damper (GMD) is an inertia-based passive control device. It has a gear assembly that ampli-
Received 23 October 2015 fies the rotational inertias developed in the gears and generates a resultant resisting force that is propor-
Revised 22 July 2016 tional to the relative acceleration at the end terminals of the GMD. The amplification provided by the gear
Accepted 23 July 2016
assembly can be adjusted by changing the gear masses or the gear ratios of the compound gears.
Although similar inertia-based devices have been successfully used for vibration mitigation of motor
vehicles and optical tables, there are only a few studies that investigated their application in building
Keywords:
structures. This number is even lower for the particular type of inertial damper that has been considered
Gyro-mass damper
Nonlinear viscous damper
in this study, i.e., GMD. Unlike other types of inertial dampers, the supplemental energy dissipation com-
Vibration control ponent of GMDs is not built-in to the device and can be independently attached as an external compo-
Seismic performance nent. This allows the design engineers to use this cost-effective device and select any available energy
Passive control device dissipation device to use in parallel. In this study, using a small-scale GMD, by considering the rotational
inertias of the intermediate gears, characteristic equation which describes the relationship between the
applied relative acceleration and the resulting resisting force is derived and experimentally verified. For
the introduction of GMDs into building structures, three different configurations are evaluated: (i) stand-
alone GMD, (ii) GMD-brace system, and (iii) GMD–Viscous damper–Brace (GVB) system. The structure-
GMD interaction, considering these three configurations, is investigated in frequency domain and in time
domain through energy balance equations and time history analyses. It is shown that by selecting the
system parameters properly, GVB systems with nonlinear viscous dampers can effectively improve the
seismic behaviour of the structure. This is discussed in more detail when the effects of the damper non-
linearities, as well as the various GMD equivalent mass, brace stiffness, damping values and selected
ground motions are investigated. The key findings related to the design, implementation and perfor-
mance considerations of these systems are provided.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction gear assemblies. There is a long history of development, design,

The current direction of structural engineering research aims to
achieve sustainable and resilient design alternatives that must go
beyond life safety. These alternatives include earthquake protec-
tive systems with passive, active and hybrid (semi-active) control
techniques. Passive control techniques are perhaps the most
widely employed as they require no external energy supply and
are easier to set-up and maintain. This paper studies a recently-
developed mechanical device [1], namely a gyromass damper
(GMD), which is an inertia-based passive control device that uses
⇑ Corresponding author at: Department of Civil Engineering, University of
Toronto, 35 St. George Street, Toronto, Ontario M5S1A4, Canada.
E-mail address: reza.hessabi@mail.utoronto.ca (R. Mirza Hessabi).
modeling, analysis and application of various mechanical devices
that use gear assemblies. However, their first practical application
as a passive control system was reported in early 2000s by Smith
[2], where the term ‘‘inerter” was used for them. They have since
been used to suppress vibrations of optical tables [3] and cables
[4], used in Formula 1 racing car suspension systems, under the
name of ‘‘J dampers” [5], and subsequently in motorcycles, trains
and regular cars [6–8]. In most of these studies, the dynamics of
inerters was usually studied by using the electrical circuit analogy.
Although there has been considerable research focused on the use
of inerters in mechanical systems, their application in civil engi-
neering and especially structural dynamics is limited.
In comparison to other types of inertia-based devices, GMD has
a simpler design and can be relatively inexpensive. By increasing

http://dx.doi.org/10.1016/j.engstruct.2016.07.045
0141-0296/Ó 2016 Elsevier Ltd. All rights reserved.
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
the inertial terms of the system, GMDs increase the period of vibra-
tion of the structure and which is similar to the effect of base iso-
lation systems. However, unlike base isolation systems in which
the excessive deformation of the isolation layer is problematic
for the structural response, the gears in a GMD are fixed and they
spin freely around their axes. It is also common in the literature to
draw a comparison between various types of inertial dampers and
TMDs and several previous studies have shown the similarities
between these two control systems [9]. TMDs installed on the roof
of the structures provide an alternative passive control strategy
[10]. However, it should be noted that there are several major dif-
ferences between these two control strategies. Similar to GMDs,
TMDs modify the response of the system by changing the inertial
force component of the equation of motion. However, TMDs
require a relatively large mass and thus a large space for their
installation. The equivalent mass of GMDs can be modified by
changing the gear ratios and employing multiple compound gears.
Thus they can provide similar level of force while occupying a
much smaller space. Moreover, by design, TMDs are in resonance
with the main structure, and they undergo large displacements
which need to be accommodated. As mentioned before, the free
spinning of the gears of the GMDs does not impose any such
problems.
In general, three types of inertial dampers have been proposed
and tested experimentally, i.e. hydraulic, ball-screw and rack-and-
pinion type of devices [11]. In 2011, Wang et al. [12] proposed a
device in which forces were translated by hydraulic means. In
the second type, a ball-screw mechanism is used. This configura-
tion is very similar to the original inerter proposed by Smith [2]
and thus in comparison to the other two types, has been studied
more extensively. In these devices, the rotary ball-screw converts
the axial movement into rotary movement thereby the axial veloc-
ity is amplified and applied to a viscous material [13,14]. Ikago
et al. [15] called this type of inertia-based dampers as Tuned Vis-
cous Mass Dampers (TVMDs). They derived a closed-form opti-
mum design equation for the TVMD vibration control system to
minimize the peak amplitude of the resonance curve in undamped
structures and verified their numerical simulation by conducting
shake table tests. More information about the development and
verification of TVMDs can be found in other publications of these
authors [16] including the real-life application of these devices in
a steel building in Japan [17].
In the third type, using a rack and assembly of gears, the relative
translation of the terminals is transformed into rotation of the
gears [1]. In this group of inertial dampers, the equivalent mass
of the device can be easily adjusted and unlike TVMDs, there is
no viscous damper component inside these dampers. The lack of
the necessity of employing a viscous medium inside the damper
leads to reduced maintenance requirements. In addition, the sim-
ple assembly of this type of inertial dampers makes them an effec-
tive alternative solution for the vibration control of building
structures in various parts of the world [18,19].
Unlike the other two types, the performance of the rack-and-
pinion inertial dampers has still not been well studied and thus
the current study investigated this type of inertial dampers in more
detail. The device considered in the current study is similar to the
damper in the study of Saitoh [1] and hence the same terminology,
i.e. ‘‘gyro-mass damper”, is used herein. Saitoh used a GMD to mit-
igate the lateral displacements of base isolators. In a similar appli-
cation, Wang et al. [20] studied the effects of the application of
these dampers to limit the horizontal and vertical displacements
of the building base. However, it should be emphasized that both
of these two studies are purely numerical and GMDs are only
employed at the base level of the structure.
This study is arranged as follows: with the objective of illustrat-
ing the desirable dynamic properties of stand-alone GMDs,
175
Section 2 summarizes the governing equations of motion for a
stand-alone GMD and then to verify the derived equations, a pro-
totype of this device is tested experimentally. In order to better
investigate the performance of GMDs, energy terms are provided
as quantitative measures. The main advantage of the energy for-
mulation is the replacement of vector quantities, such as displace-
ments, velocities and accelerations by the corresponding scalar
quantities. Despite their benefits, due to geometric limitations
and alignment issues, it would be very difficult, if not impossible,
to design a GMD that spans along the entire height of a story diag-
onally to work based on relative floor accelerations. In Section 3,
the benefits and shortcomings of different configurations for the
introduction of GMDs into real structures are studied. The limita-
tions of a configuration in which the GMD is attached to the struc-
ture through V braces, and the dynamic properties of the resulting
system, considering also the flexibility introduced by the brace, are
studied. As it is discussed in Section 3, an energy dissipation com-
ponent (which does not have to be necessarily a linear viscous
damper) can be added externally to the GMD-brace system. The
effects of different parameters on the behaviour of this practical
configuration with a GMD–Viscous damper–Brace (GVB) are inves-
tigated. The behaviour of a Single-Degree-of-Freedom (SDOF) sys-
tem with the GVB control system is evaluated under different
seismic loading scenarios and for various equivalent mass and stiff-
ness values, and damping mechanisms. A numerical example is
included in Section 4 that demonstrates the properties of the pro-
posed configuration. Finally, the key findings of the paper are pro-
vided in Section 5.
2. Dynamic behaviour of GMDs
2.1. Stand-alone GMDs
In many ways, gears in rotating systems act similarly to levers
in translating systems. The GMD prototype used in this study is
shown in Fig. 1. Fig. 1(a) shows a GMD element subjected to a time
varying relative axial force f. It consists of two simple (gears 1 and
6) and two compound (gears 2, 3, 4 and 5) gears. Free Body Dia-
grams (FBDs) of the gears are also shown in Fig. 2. The forces acting
on each of these gears are determined by considering the equilib-
rium between the gears in contact. It should be pointed out that,
unlike the study of Saitoh [1], the rotational inertias of the inter-
mediate gears are also considered in the following derivation. By
neglecting the mass of the rack and using the FBDs in Fig. 2, the
equations of motion for each of the gears can be written as Eqs.
(1)–(4).
J 1 €h1
f 1 r 1 þ f 1
2 r 1 ¼ 0 ð1Þ

J 2 €h2
J 3 €h3 þ f 1
2 r 2
f 3
4 r3 ¼ 0 ð2Þ
J 4 €h4 þ J 5 €h5 þ f 5
6 r5
f 3
4 r 4 ¼ 0 ð3Þ

J 6 €h6 þ f 5
6 r 6 ¼ 0 ð4Þ
where ri and Ji denote the radius and mass moment of inertia of the
ith gear, respectively. In the above equations of motion, an ideal
efficiency is assumed and the damping effects of gear friction or
backlash are neglected. It can be shown that, the geometric rela-
tionship between the arc lengths of the gears results in the follow-
ing relations between the angular rotations of each of the gears and
the first gear:
€h2 ¼ €h3 ¼ r 1 €h1 ð5Þ
r2
176 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186

Fig. 1. (a) Side-view, and (b) dimensions of the gears of the GMD considered in this study.

Fig. 2. FBD of the gears of the GMD considered in this study.

€h4 ¼ €h5 ¼ r 1 r 3 €h1 ð6Þ Parameter b in Eq. (10) is called the equivalent mass and for the
r2 r4 considered GMD configuration, it is equal to 12 ðmg1 þ mg2 þ
N 22 ðmg3 þ mg4 Þ þ N 22 N 23 ðmg5 þ mg6 ÞÞ. Thus, the force generated in a
€h6 ¼ r 1 r 3 r 5 €h1 ð7Þ
r2 r4 r6 GMD is proportional to the relative translational acceleration with
b as the proportionality constant. It can be shown that by using
After combining Eqs. (1)–(7), the following relationship can be two compound gears with the gear ratio of 5:1, a GMD similar to
€1 and the tangential force in the first gear
derived between h what is shown in Fig. 1 can generate an equivalent inertial force
"
2
2
2 # that is 1302 times larger than the inertial force generated by trans-
J1 J2 J3 r3 r3 r3 r5
f1 ¼ þ þ þ J 4 þ J 5 þ J 6 r 1 €h1 lation of an object with the same mass. Also, note that the equiva-
r 21 r 22 r 22 r2 r4 r2 r4 r2 r4 r6 lent mass is independent from the radii of the gears and can be
ð8Þ adjusted by changing the gear ratios and/or the gear masses.

The gear ratios of the compound gears 2 and 3 are defined as

N 2 ¼ r3 =r2 and N 3 ¼ r5 =r4 , respectively. It can be shown that the
number of teeth on a gear is proportional to the radius of its pitch
circle, which means that the ratios of the number of teeth of the
corresponding gears can also be used for the gear ratios inter-
changeably. As shown in Fig. 1(a), the transitional displacement
of the rack (u) is related to the rotation of the first gear (u ¼ r 1 h1 )
and the tangential force in the first gear is equal to f . Thus, by
replacing the mass moment of inertia Ji with their respective
expressions, the force-acceleration relationship in the GMD can
be rearranged as Eq. (9),

h i the built-in controller is able to impose predefined (e.g., har-
1
f ¼ mg1 þ mg2 þ N22 ðmg3 þ mg4 Þ þ N22 N23 ðmg5 þ mg6 Þ u€
2
where mgi shows the mass of the ith gear. Eq. (9) be summarized as
€
f ¼bu
2.2. Experimental validation and the effects of friction
In this section, experimental results for a small-scale prototype
are included to verify the derived equations from the previous
section. Fig. 3 shows the experimental setup. The experiments
were performed using a Quanser shaker II located at the Univer-
sity of Toronto. The shaker consists of a servo-motor driving a
12.7 mm lead screw. The lead screw drives a circulating ball
nut, which is coupled to the 457  457 mm2 table. The table itself
slides on low friction linear bearings on two ground hardened
shafts and has a 76.2 mm stroke. Under displacement control,
ð9Þ monic) displacements. The setup also has a 22.2 N (5 lb) load cell
between the shake table and the GMD that can measure the
forces.
The damper is fixed on the table and the rack of the damper is
ð10Þ placed on greased sliders to minimize friction and is connected to
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186 177

Fig. 3. Experimental setup for the stand-alone GMD.

the load cell. An external LVDT is used to measure the horizontal

displacements of the rack and these measurements are checked
against the readings of the internal displacement encoders of the
shaker. Special attention was given to keep the horizontal align-
ment of the GMD. The proper alignment of the GMD in real-
building applications is also essential. More discussion on this is
provided in Section 3.2.
It has been shown before that the performance of inertial dam-
pers can be affected by nonlinearities [15,21]. As discussed by
Smith and Wang [6], the experimental data matches better with
the theoretical predictions when the effects of friction forces (f f )
are considered. In this study, to measure the friction (f f ), gears
are removed and a low-frequency sinusoidal input displacement
with the frequency of 0.1 Hz and amplitude of 10 mm is applied.
As it is shown in Fig. 4(a), f f is almost a square wave with the direc-
tion opposite to the sign of the velocity and the amplitude of about
ðf f Þmax ¼ 0:77 N. Eq. (11) can be used to consider the friction. Fig. 5. Measured force vs measured displacement for a GMD with b = 29.87 kg.

_
f f ¼ ðf f Þmax  signðuÞ
In Fig. 4(b) measured forces for a GMD with an equivalent mass of
b = 29.87 kg are compared with the theoretically predicted forces.
The applied sinusoidal displacement for all the cases in this section
had the amplitude of 15.24 mm (0.6 in) and frequency of 1.0 Hz.
Therefore, the maximum amplitude of the predicted force is equal
to ð0:01524 mÞ  ð2p  1Þ2  b or 0:60b. A closer look at Fig. 4(b)
shows that Eqs. (10) and (11) predict the values of the forces with
an acceptable accuracy and the difference between the maximum
measured and predicted forces is close to the value of ðf f Þmax .
ð11Þ In Fig. 5, measured forces are plotted against the applied dis-
placements. From Eq. (10), it is expected that the restoring GMD
forces be proportional to the applied accelerations. In this test,
since the applied displacements are sinusoidal, the force-
displacement relationship can be expressed as
f ¼
ð2p  1Þ2 b  u or f ¼
11:79u, which shows a negative slope.
The experimental results are compared to this line. It should be
noted that the enclosed area in this plot can mainly be attributed
to the presence of frictional forces.
Table 1 shows the comparison between the measured and pre-
dicted forces under the imposed displacement time history. In this

Fig. 4. (a) Friction forces for the GMD, (b) measured and predicted forces of a GMD with b = 29.87 kg.
178 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
Table 1
Comparison between forces of the stand-alone GMDs with various values of
equivalent mass.
Equivalent mass, b (kg) 17.37 20.49 23.62 29.87
Maximum predicted force (N) 11.20 13.06 14.94 18.69
Maximum measured force (N) 9.95 13.89 15.52 21.03
table, mass of the last gear was changed to verify the accuracy of
the predicted GMD forces. The self-mass of the larger and smaller
gears are 27.2 gr and 0.91 gr, respectively and the equivalent mass
of the damper without any extra mass is 17.37 kg. When 2, 4 and 8
extra masses (i.e., additional 10, 20 and 40 g, respectively) are
introduced to the last gear, the equivalent mass increases to
20.49 kg, 23.62 kg and 29.87 kg, respectively. The value of the max-
imum predicted force of 0:6b plus the friction force (ðf f Þmax ) is
shown in the first row of Table 1 and can be compared with the
maximum measured forces for each damper.
Depending on the type of material, size of the components,
design of the connections, tightness of the fasteners, type of the
lubricants, etc. the value of friction forces may vary from one dam-
per to another. Nevertheless, as it is shown in this section, the fric-
tion force can constitute a large portion of the resisting force and
whenever its value is available, it should be considered in the anal-
yses. The friction force for each particular device should be mea-
sured separately and the proposed procedure in this section can
be used to measure this force.
2.3. Dynamic effects of GMDs on SDOF systems
To obtain the equation of motion for the GMD-structure system,
the equation of motion of an SDOF system with a known mass M,
stiffness K and inherent damping C (and damping ratio of n) needs
to be modified. As shown in Eq. (10), the force developed in a GMD
is proportional to the relative acceleration applied at the end
nodes. The new equation of the motion for a system equipped with
a GMD is rewritten as Eq. (12).
ðM þ bÞ€x þ C x_ þ Kx ¼
M€xg ð12Þ
By dividing both sides of Eq. (12) by M + b, it can be rewritten as,
€x þ ðC=½M þ bÞx_ þ ðK=½M þ bÞx ¼
ðM=½M þ bÞ€xg ð13Þ
As it can be seen in Eq. (13), after normalizing the differential
equation, the coefficients of the acceleration term on the left hand
side of the equation is equal to one. A closer look at this equation
reveals that the natural period of vibration of the SDOF system plus
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
the stand-alone GMD is equal to T n = ðM=½M þ bÞ. Thus, with the
addition of the GMD the period of the system increases.
Further comparison of the other dynamic parameters of Eq. (13)
with the corresponding values of the original system also reveals
that the damping ratio of the SDOF system with the GMD is equal
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
to n  ðM=½M þ bÞ. This explains why when the GMD is intro-
duced with b greater than 0, the damping ratio of the GMD-
structure system is always less than that of the original system.
Finally, the effective force on the right side of equation of motion
of the original SDOF system is reduced by ðM=½M þ bÞ. Energy
measures will be discussed in the next section to show the effects
of each of these changes.
3. Investigation of different configurations for the application
of GMDs in real structures
3.1. GMD-brace configuration
As shown in the previous sections, stand-alone GMDs have the
potential of enhancing the seismic performance of structures by
increasing the fundamental period of vibration of the system. How-
ever, due to geometric limitations and alignment requirements, it
would be very difficult, if not impossible, to design a GMD that
spans along the height of a story diagonally to work based on the
relative floor accelerations. Instead other configurations, such as
the V brace shown in Fig. 6 should be considered to attach them.
The braces serve as extenders and will transfer the displacement
and acceleration of the main mass to the end terminal of the GMD.
Starting with the equation of motion of this system in time
domain, Eq. (14), can be written to describe the GMD-brace
system.
          
M 0 €x C 0 x_ K þ kb
kb x M
þ þ ¼
€xg
0 b u€ 0 0 u_
kb kb u 0
ð14Þ
where x is defined in Eq. (12) and u is defined as the displacements
of the node that connects the GMD to the braces. A closer look at Eq.
(14) reveals that this equation is similar to the equation of motion
of a 2-degree-of-freedom system with an additional stiffness
element.
To study the dynamic behaviour of the system shown in Fig. 6,
the transfer function T GMD
brace can be written in the frequency
domain as:
XðjxÞ
T GMD
brace ðjxÞ ¼
X g ðjxÞ
"

1 #
1
1 1
¼
M x
Mx þ Cjx þ K þ
2 2
þ
kb
bx2
ð15Þ
pffiffiffiffiffiffiffi
where j ¼
1 and the lateral stiffness of the braces is shown by kb .
Note that since kb is the horizontal stiffness of the braces, brace
geometry should be taken into account when this parameter is cal-
culated. At his stage, two dimensionless parameters can be defined
to describe the system in Fig. 6. Equivalent mass ratio is defined as
l = b/M, and the ratio of the brace stiffness to that of the structure is
defined by b ¼ kb =K. Fig. 7 shows the magnitude of the transfer
function T GMD
brace for different frequencies.
As it can be seen in this figure, the addition of the GMD-brace
system separates the resonant peak of the original SDOF system
into two distinct frequencies. It can be seen that for each b value,
the higher-frequency resonant peak of the controlled system is
consistently larger than the resonant peak of the original system.
As a result, in comparison to the performance of the original SDOF
system, the response of the GMD-brace system is consistently
worse. This is particularly critical when the governing frequency
of the applied load coincides with the higher-frequency resonant
peak of the GMD-brace system.
This is a significant observation for the real-life application of
GMDs into building structures. In order to have a practical installa-
tion configuration for GMDs that does not suffer from the negative
effects of the GMD-brace systems, in the next section, the use a
supplemental damping device in parallel with the GMD-brace sys-
tem will be investigated.
Fig. 6. Model of an SDOF structure with a GMD-brace system.
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186 179

By using the relative displacements, the transfer function

approach can be employed directly to find the frequency response
of the SDOF system equipped with the GVB systems. This transfer
function, named as T, can be written in the frequency domain as:
 2

xxn
TðjxÞ ¼  2
  ð16Þ
1
xxn þ 2 xxn j þ Gðl; b; c; x; xn Þ

where
 2
lb x
xn
Gðl; b; c; x; xn Þ ¼  2   ð17Þ
Fig. 7. Frequency response of the SDOF with a GMD-brace system.
l x
xn
b
2c xxn j

It should be noted that the frictional effects in the GMD are

3.2. GVB systems
To alleviate the shortcomings of the GMD-brace system, it is
necessary to control the higher-frequency resonant response of
the system. One way to achieve this goal is by introducing damping
devices such as linear or nonlinear viscous dampers. Using the sim-
ilarities between TMDs and GMDs, Lazar et al. [9] and Giaralis and
Taflanidis [22] proposed new models, called TID and TMDI, respec-
tively. These studies ([9,22,23]) considered a viscous damper in
parallel with the braces. For the same reasons that were explained
for stand-alone GMDs (i.e., geometric limitations and alignment
issues), this configuration would not be practical in real applica-
tions. In another study, Ikago et al. [15], proposed a model in which
the inertial component was arranged in series with the braces and
in parallel with a viscous damper. However, the inertial component
in their model is a TVMD (and not a GMD), where a rotary ball-
screw mechanism was added to a regular linear viscous damper.
Fig. 8(a) shows a one bay-one story frame equipped with the
configuration that is used in the current study (which will be called
a GVB) and the corresponding mass-spring-dashpot oscillator is
provided in Fig. 8(b). This configuration has originally been pre-
sented by Saitoh [1], where he called it a Model II system and
investigated the effectiveness of this configuration for a different
target structure. Despite the similarities between GVBs and
TVMDs, the main advantage of using GVBs is that the supplemental
energy dissipation component is not built-in to the device and can
be attached as an external component. This will allow the design
engineers to select from a wide range of available energy dissipa-
tion devices to use in parallel with the GMD. In addition, it should
be noted that in both of the studies by Ikago et al. and Lazar et al.,
the damping of the primary structure was neglected which is not
the case in Fig. 8.
neglected in Eq. (16). A new dimensionless parameter c is defined
as the ratio of the viscous damper coefficient over the correspond-
ing value of the original structure or c ¼ cb =C. Note that when l = 0
or b ¼ 0, Gðl; b; c; x; xn Þ becomes zero, the transfer function
becomes that of the original SDOF system. Performance of an SDOF
structure equipped with a GVB system is influenced by the selec-
tion of the lateral brace stiffness, i.e. kb . A combination of the
parameters b, l; c and n results in a modified dynamic response
of the system. Thus, a proper selection of a combination of these
variables needs to be made during the design of the control system.
In order to address this issue and select an optimum set of
parameters, [15] used the fixed-point method and derived a
closed-form solution for the design of a structure subjected to har-
monic excitation. Based on their study, first, the desired equivalent
mass ratio (l) should be specified.
Then, using this parameter, b and c can be calculated from Eqs.
(18) and (19):
kb l
b¼ ¼ ð18Þ
K 1
l
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
c l 3l
c¼ b¼  ð19Þ
C 2 ð1
lÞð2
lÞ
qffiffiffi qffiffiffiffiffiffiffi
From Eq. (18) it can be shown that xb ¼ xn b
¼ xn 1
,
l 1
l
qffiffiffiffi
where xb ¼ kbb :
Due to the similarities between TMDs and GMDs, same com-
mon mass ratio values can be used for the equivalent mass ratio
(l) in GVB design. However, as it can be seen in Eqs. (18) and
(19), kb and cb are both functions of this parameter and thus it is
necessary to discuss the effects of l on the response of the system.
In Fig. 9, the frequency response of SDOF systems with and without

Fig. 8. (a) Model of an SDOF structure with a GVB system, (b) the SDOF oscillator with the GVB system.
180 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
GVB control systems are compared for different values of l. Two
main observations can be made by looking at Fig. 9. First, even
for small values of l, a significant reduction of the peak response
can be obtained. As it can be seen in this figure, although increasing
the values of l leads to lower vibration amplitudes, the rate of this
decrease reduces for the effective mass ratios larger than 5%. The
second observation is related to the effects of the assumption
about the damping of the primary structure. For the derivation of
b and c parameters (i.e., Eqs. (18) and (19)), it has been assumed
in [15] that the original structure is undamped. In other words,
by assuming a value of zero for n, the control system properties
are calculated so that the separated resonant peak values become
equal. However, Fig. 9 shows that this assumption is only valid
for smaller values of l and the obtained response for a damped pri-
mary structure does not have the same resonant peak values for
larger equivalent mass ratios.
To the best of the authors’ knowledge, all of the previous studies
on this subject (such as TVMDs) have considered a linear viscous
damper. In order to demonstrate that the real application of GMDs
is not limited to GVB systems with linear viscous dampers, a non-
linear viscous damper with a power factor of 2 is chosen here.
Since, these devices behave nonlinearly, Eq. (16) can no longer be
applied to them. Thus, numerical models in OpenSees were used
and nonlinear time history analyses were carried out. The numer-
ical simulation results showed that Eq. (18) can be successfully
used to design the stiffness of the braces, however, the damping
coefficient should be reduced. From the numerical simulations it
was also found that Eq. (20) can be used to determine the damping
coefficient of the nonlinear viscous damper (cbn Þ:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3l
cbn ¼ l  bkb 
1
l
A comparison between the frequency responses of the displace-
ments of structures controlled with a TID system, a TVMD and a
GVB configuration with a nonlinear viscous damper with a power
factor of 2 is shown in Fig. 10. As it can be seen the responses
are comparable, however, the required damping coefficient of the
nonlinear viscous damper is 10 times smaller than those of the
other two systems.
Time history analysis results of structures with and without the
GVB (with a nonlinear viscous damper) subjected to the NWH090
ground motion record (Northridge earthquake), can also be used to
confirm this result. The properties of this ground motion record are
provided in Table 2. The results are shown in Fig. 11. It is clear in
this figure that the GVB system reduces the maximum displace-
ment value effectively and the control system is particularly effi-
cient around the 8th second. In addition, the mean displacements
and accelerations of the primary structure are reduced after the
introduction of the GVB system.
Fig. 9. Frequency response of the main mass for various values of
Fig. 10. Frequency response of the main mass for TID, TVMD and GVB with a
nonlinear viscous damper systems (l = 10% and n = 2%).
3.3. Energy balance equations
When a structure is equipped with damping devices or with a
seismic isolation system, its dynamic response is modified. A thor-
ough understanding of this modified dynamic behaviour is neces-
sary to ensure the safety and effectiveness of the overall system.
However, due to the inherent uncertainty associated with earth-
quake ground motions, this task becomes more challenging when
seismic loading is considered.
One available approach to achieve this goal and study the per-
formance of damping devices is the use of energy balance equa-
tions. The main advantage of the energy formulation is the
replacement of vector quantities, such as displacements, velocities
ð20Þ and accelerations by the corresponding scalar quantities. During an
earthquake, input energy caused by ground motion enters into the
structure and then gets converted into different energy forms in
the system. As the structure vibrates, its structural elements
deform and absorb strain energy. When it stops moving at the
end of a cycle of vibration, the kinetic energy gets transferred into
strain energy. The vibration of the structure can therefore be visu-
alized as a constant transfer of kinetic energy into strain energy
and vice-versa. These energy forms are eventually dissipated in
the forms of either damping energy or hysteretic energy (if pre-
sent). The energy formulation is obtained by integrating the work
done by each system element over an increment of global struc-
tural displacement [24].
Considering a general multi-degree-of-freedom (MDOF) system
which is equipped with stand-alone GMDs and is excited at the
base by a horizontal earthquake ground motion, the governing dif-
ferential equations of motion can be written as shown in Eq. (21).
½Mf€xðtÞga þ ½CfxðtÞg
_ €
r þ ½KfxðtÞgr þ ½BfxðtÞgr ¼ 0 ð21Þ
where ½M, ½C, ½K and ½B are mass, viscous damping, stiffness and
GMD equivalent mass matrices, respectively. Also, subscripts r
and a denote the absolute or relative fxg, fxg _ and f€xg which are
the vectors of global accelerations, velocities and displacements at
time t. Note that the absolute energy formulation is used here
because according to Uang and Bertero [25], the relative energy
approach does not consider the rigid body translation of the struc-
ture. The relationship between the absolute and relative displace-
ments, velocities, and accelerations are provided in Eqs. (22)–(24),
respectively:
fxga ¼ fxgr þ fxg g ð22Þ
fxg _ r þ fx_ g g
_ a ¼ fxg ð23Þ
l (n = 2%). f€xga ¼ f€xgr þ f€xg g ð24Þ
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186 181

Table 2
List of the earthquake ground motions considered in this study.

Event NGA # Record component Near/far field Mag. Rjb (km) PGA (g) PGD (cm) Pulse-like
1 Loma Prieta 0799 SFO000 FF 6.93 58.60 0.236 3.96 No
2 0738 NAS180 FF 71.00 0.268 5.36 Yes
3 Imperial Valley 0169 H-DLT352 FF 6.53 22.00 0.351 19.61 No
4 Kobe 1107 KAK090 FF 6.90 22.50 0.345 9.60 No
5 Chi-Chi, Taiwan 1487 TCU047-N FF 7.62 35.00 0.413 22.24 –
6 TCU047-E FF 0.301 51.12
7 Kern County 0015 TAF021 FF 7.36 38.42 0.156 9.48 –
8 Landers 0838 BRS000 FF 7.28 34.90 0.132 20.10 Yes
9 N. Palm Springs 0532 CLJ000 FF 6.06 78.10 0.021 0.33 No
10 Northridge 1044 NWH090 NF 6.69 5.90 0.583 17.98 Yes
11 Loma Prieta 0779 LGP090 NF 6.93 3.90 0.587 24.55 Yes
12 Imperial Valley 0183 H-E08140 NF 6.53 3.90 0.602 32.21 Yes
13 H-E08230 NF 0.454 35.33
14 Chi-Chi, Taiwan 1231 CHY080-E NF 7.62 2.70 0.968 18.61 No
15 CHY080-N NF 0.902 34.00
16 Tabas, Iran 0143 TAB-TR NF 7.35 2.00 0.852 94.49 No
17 TAB-LN NF 0.836 37.71
18 Duzce, Turkey 1611 1058-E NF 7.14 0.20 0.111 9.40 No

Fig. 11. Response of original and controlled SDOF systems, (a) displacements, (b) accelerations (Northridge earthquake (NWH090 record)).

Z
where f€xg g is the vector of the ground motion accelerations at time
SE ¼ fxgTr ½Kfdxgr ð29Þ
t. Integrating Eq. (21) gives:
Z Z Z
fx€ga ½Mfdxga þ fxg
T
_ Tr ½Cfdxgr þ fxgTr ½Kfdxgr Z
f€xga ½Mfdxg g
T
Z IE ¼ ð30Þ
þ f€xgr ½Bfdxgr
T

By comparing the coefficients of the input acceleration terms on

Z
the right hand sides of the equation of motion of an uncontrolled
¼ f€xga ½Mfdxg g
T
ð25Þ
SDOF system and Eq. (12), it can be seen that the input acceleration
of the system with a GMD is ðm=½m þ bÞ times less than the corre-
Hence, the absolute energy balance for a linear system at time t
sponding input acceleration for the uncontrolled system. Note that
is
the ðm=½m þ bÞ ratio is always a positive value equal or smaller
KEðtÞ þ DEðtÞ þ SEðtÞ ¼ IEðtÞ ð26Þ than one. However, as it is shown in Eq. (30), the introduction of
the stand-alone GMDs in structures, only modifies the kinetic
where kinetic (KE), damping (DE), strain (SE) and input (IE) energies
energy and does not directly reduce the input energy of the system.
are defined as,
Z Z On the other hand, the introduction of the GMDs in GVB configura-
KE ¼ f€xga ½Mfdxga þ
T
f€xgr ½Bfdxgr
T
ð27Þ tion (with a linear or nonlinear viscous damper) reduces the input
energy of the system considerably. The effects of the addition of
Z the nonlinear viscous damper in the GVB configuration of Sec-
DE ¼ _ r ½Cfdxgr
fxg
T
ð28Þ tion 3.2 and the significant reduction of the input energy are also
shown in Fig. 12.
182 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
Fig. 12. Comparison of input energy time histories of SDOF systems with/without
GMDs (l = 10%).
In order to better quantify the effectiveness of the device from
energy perspective (based on Eq. (26)), a new performance index
(a1 ) is defined here, where the maximum input energy of the con-
trolled structure is divided by the input energy of the original SDOF
system:
ðMaximum IEÞControlled SDOF system
a1 ¼ ð31Þ
ðMaximum IEÞOriginal SDOF system
A comparison between the peak values of the plots in Fig. 12
shows that the introduction of the stand-alone GMD does not
change the input energy considerably but reduces the maximum
input energy for the system with a GVB system with a nonlinear
viscous damper significantly. This effect is more considerable than
the effects of employing a stand-alone viscous damper or using a
GMD-brace system. Using Eq. (31), the corresponding value of a1
for the latter control system is 0.17. This indicates that the input
energy of the controlled system with a nonlinear viscous damper
in a GVB configuration can be reduced by as high as 83%.
Fig. 13. Acceleration and displacement spectra under Loma Prieta earthquake for SDOF systems with/without GMDs (l = 10%).
Fig. 14. Acceleration and displacement spectra under Northridge earthquake for SDOF systems with/without GMDs (l = 10%).
3.4. Effects of the applied earthquake ground motion
To examine the effects of the uncertainties of the applied
ground motion on the response of the SDOF-GVB systems, maxi-
mum displacement and maximum acceleration are used as the
performance criteria here. The spectra for each of these parameters
are generated by changing the natural period of vibration of the
original SDOF system and plotting the corresponding maximum
value of the parameters. Using this procedure, for a certain ground
motion, the performance criteria can be plotted as a function of the
elastic natural period of the original SDOF system (Fig. 13). The first
objective is to find the range of the structural period for which the
response is sensitive to the applied ground motion. Two ground
motion records are chosen for the analyses in this section: a far-
field ground motion record (i.e., Loma Prieta earthquake -
SFO000 record) and a near-field ground motion record (i.e., North-
ridge earthquake - NWH090 record). More details about both of
these records can be found in Table 2. A closer look at Figs. 13
and 14 reveals that for structures with a natural period of vibration
of greater than 1.5 s, the accelerations are relatively unaffected by
the addition of GVB control system. On the other hand, it can be
seen that these systems reduce the maximum acceleration of the
system consistently when the natural period is less than 1.5 s.
Moreover, from these figures, it can be concluded that the GVB
control system is effective in reducing the maximum displacement
of the structure under these two ground motions.
To investigate the effects of different input earthquake ground
motions, a set of 18 different historical ground motions were
selected from the PEER-NGA database [26], and time history anal-
yses were performed and the same procedure was repeated. The
details of these ground motions are reported in Table 2. In this
table, similar to the study of Taniguchi et al. [27], the selected
earthquakes are divided into two categories of near- and far-field
ground motions, indicated as NF and FF, respectively. Each ground
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
Fig. 15. Performance index a1 for the controlled systems subjected to different
ground motions listed in Table 2 (l = 0.10).
Fig. 16. Portal frame supporting a satellite transmission unit [24].
motion record is applied to SDOF systems with different initial per-
iod of vibrations equipped with a GVB system and the resulting
peak a1 value is recorded as the performance criterion.
In Fig. 15, a1 is shown for the sample system of Section 3 under
the 18 selected earthquake ground motions. In this figure, the tri-
angles show the results for the far-field ground motions and the
diamonds illustrate the results for the near-filed records.
Based on the results in Fig. 15, no clear distinction can be
observed for the effectiveness of the control strategy for far-field
Fig. 17. Effects of b and c on the maximum displacement of a GVB system with a linear viscous damper.
183
or near-field earthquakes. Similar to many other control systems,
the performance of the proposed GVB system depends on the prop-
erties of the applied ground motion.
In general, it can be concluded from Fig. 16 that the introduc-
tion of the GVB system consistently enhances the response of the
SDOF system by reducing the seismic input energy. This control
strategy is particularly beneficial when the period of vibration of
the structure resonates with the dominant frequency of the
earthquake.
4. Numerical examples
It is shown in Section 3.2 that GVB systems provide a practical
solution for incorporating GMD in real structures and Eqs. (18)–
(20) can be used to design the parameters of these systems. Since
the project size, availability of the required device and the quantity
of it affect the unit price of each of viscous dampers, a generalized
cost-benefit analysis to investigate the effects of the damper selec-
tion on the eventual cost saving may not be very accurate. Still, the
numerical examples presented in this section can be used to exam-
ine the effects of each design parameter and to qualitatively com-
pare the size and performance of each component for these GVB
systems with some other passive control techniques.
For this purpose, a benchmark portal frame which supports a
satellite transmission unit is chosen from [24] (Fig. 16). The main
mass of the frame is equal to 10 tons and each of the supporting
columns has an EI equal to 370 kN m2 (corresponding to a period
of vibration of 1.10 s). System identification tests on the frame
indicated an inherent damping ratio of 2% of critical.
4.1. Parametric analysis
To examine the sensitivity of the response of a GVB system to
the variation of its design parameters, the performance of the SDOF
system shown in Fig. 16 is investigated in this section. As it is
demonstrated in Fig. 9, the equivalent mass ratio (l) has the max-
imum effects on the behaviour of the system when the primary
structure is in resonance with the applied ground motion. It can
be concluded that with the increase of l, the peak accelerations
and displacements of the system reduce. However, the effects of
b and c on the performance of the system are more complex. For
a GVB system with a linear viscous damper, the fixed-point
method from [15] can be used and Eqs. (18) and (19) can be
184 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
Fig. 18. Effects of b and c on the maximum acceleration of a GVB system with a linear viscous damper.
employed to determine the optimum b and c. In Figs. 17 and 18, a
fixed value of 0.10 is chosen for l and parameters b and c are chan-
ged. The normalized maximum displacements for a harmonic exci-
tation with an input frequency that is equal to the natural period of
the primary system are shown in Fig. 17(a). In this figure, the val-
ues of b are plotted on the x axis, the values of c are on the y axis
and the values of the ratio of the maximum displacement of the
GVB system over the maximum displacement of the uncontrolled
system are on the z axis. When b and c are zero, this ratio becomes
one. The contours of the relations between these parameters are
shown in Fig. 17(b). Similar procedure is repeated for the normal-
ized maximum accelerations and the results are shown in Fig. 18.
From these figures, it can be concluded that the combination of
b and c can result in different system responses and it emphasizes
the necessity for appropriate selection of a combination of these
variables during the design procedure. It should be noted that
the solid circle markers in Figs. 17(b) and 18(b) show the optimum
values obtained from Eqs. (18) and (19) used in the design. As it
can be seen in these figures, the design values are located within
the regions with the lowest contour values indicating highest
region of response reduction.
The results for a GVB system with a nonlinear viscous damper
are also shown in Fig. 19(a–d). Once again, b is determined from
the structure was subjected to the El Centro ground motion. In this
section, in order to investigate these effects on a GVB with a non-
linear viscous damper and also to increase the number of consid-
ered ground motions, El Centro ground motion record [15] is
used. Under the El Centro earthquake ground motion, the maxi-
mum displacement of this system was determined to be
127 mm. In this study, six different passive control strategies are
used to reduce this displacement by at least 25%. Details of these
control systems for reducing the maximum displacement of the
SDOF system by at least 25% are shown in Table 3. Design param-
eters for SDOF systems with only a viscous damper, a brace or a
GMD are found by solving the corresponding equations of motion.
The stiffness and viscous damping terms for the TMD with an
actual mass of 0.28 ton are found by using the following equations.
These design equations are obtained for random base acceleration
loading where the optimization criterion is selected to be the min-
imum root mean square value of relative displacement of the pri-
mary structure. More details about these design equations can be
found in [24].

 !
k 1
l2
ktmd ¼  mtmd ð32Þ
m ð1 þ lÞ2
sffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
Eq. (18). However, the damping coefficient value of the nonlinear
ktmd l 1
l4
viscous damper (cbn) is lower than the GVB system with the linear
viscous damper and Eq. (20) is used to design this parameter. This
parameter is normalized by dividing it by the damping coefficient
of the primary structure (i.e., cn = cbn/c).
Moreover, since it is not feasible to optimize cbn for both dis-
placement and acceleration simultaneously, Fig. 19(b) and (d)
show that the design value of cbn from Eq. (20) leads to an optimal
acceleration reduction and a satisfactory displacement reduction.
4.2. Qualitative comparisson with other control techniques
In this section, the proposed design for GVB is compared to
other commonly used passive control devices. In 2012, Ikago
et al. [15] studied the influence of the properties of the ground
motion on the response reduction of the TVMDs. They concluded
that the effectiveness of these control systems degrades for long-
period ground motion containing fewer resonant components
and in particular they observed that TVMDs are less affected when
cbn ¼ 2 mtmd    ð33Þ
mtmd 4ð1 þ lÞ 1
l2
Finally, Eqs. (18) and (19) are used to design the GVB system
with a linear viscous damper and Eqs. (18) and (20) are used to
design the system with the nonlinear damper. The control systems
in Table 3 are designed for the same value of the target maximum
displacement, i.e., they all limit the maximum displacement of the
system to less than 95 mm.
Among these, the SDOF system equipped with a stand-alone
GMD has the smallest maximum acceleration value. However, as
was mentioned before, because of the geometric limitations and
alignment issues, the introduction of a stand-alone GMD would
not be practical and the introduction of a GMD through braces con-
tribute to inferior dynamic performance. Instead, the GVB systems
are used in Table 3 and they result in a satisfactory performance. It
should be emphasized here that the required damping coefficient
of the nonlinear viscous damper for the GVB system is about 7
times less than the system with only a viscous damper. This ratio
 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186 185

Fig. 19. Effects of b and c on the maximum displacements and accelerations of a GVB system with a nonlinear viscous damper.

Table 3
Details of the considered control techniques.

Control technique Design parameter Maximum displacement (mm) Maximum acceleration (g)
Original uncontrolled SDOF system – 127 0.63
Only a viscous damper c0 = 6.65 kN.s/m 95 0.50
Only a brace k0 = 290 kN/m 95 0.62
Only a TMD mt = 0.28 ton 95 0.52
ct = 0.26 kN.s/m
kt = 8.6 kN/m
Only a GM damper b0 = 2.70 ton 95 0.41
GVB system (with a linear viscous damper) b = 0.41 ton 95 0.57
cb = 0.60 kN.s/m
kb = 14.1 kN/m
GVB system (with a nonlinear viscous damper) b = 1.68 ton 95 0.53
cbn = 0.93 kN.s/m
kb = 66.3 kN/m

is 11 for the linear viscous damper. Thus, although in general
changing the velocity exponent in the nonlinear viscous damper
does not generally influence the price of the device considerably,
the smaller damping coefficient in the nonlinear viscous damper
leads to the selection of a damper with smaller maximum force
limit which in turn reduces the cost of the project significantly.
Similarly, the required stiffness of the brace in the GVB systems
is about 20 times less than the corresponding value for the
brace-only control technique. Most importantly, unlike the TMD
system, there is no need to place a real 0.28 ton mass on the struc-
ture. Also, in comparison to the stand-alone GMD system, the
application of the GVB system also requires a smaller equivalent
mass value. In fact, using Eqs. (9) and (10), it can be shown that
with the available metallic gears in the market, two compound
gears with the gear ratios of 5:1, where the larger gears have a
pitch diameter of 6 in can provide the required 1.77 ton equivalent
mass.
5. Conclusions
Unlike the other types of inertia-based devices, the performance
of the rack-and-pinion inertial dampers has still not been well
studied. Thus the current study investigated gyro-mass dampers
as the representative of this type of inertial dampers in more
details. The potential benefits of the application of GMDs in struc-
tures as a passive control strategy are numerous. GMDs can elon-
gate the fundamental vibrational period of the structures and
potentially improve the seismic performance. In this paper a
mechanical prototype was proposed and the governing equations
for that stand-alone GMD were derived and it was shown that
the resisting force in this device is proportional to the relative
acceleration at its end terminals. The ratio between the force and
acceleration in these equations is called the equivalent mass. This
parameter was shown to be independent of the radii of the individ-
ual gears and it can be adjusted by changing the gear ratios or the
gear masses. The derived equations were verified through a series
of experiments using a Quanser II shaker located at University of
Toronto. It was shown that there was a good agreement between
the experimental results and the theoretical equations. Although
stand-alone GMDs have the potential to enhance the seismic per-
formance of structures, it was concluded that due to geometric
limitations and alignment issues they could not be introduced into
structures as stand-alone devices. Their installation through braces
was shown to be problematic as the GMD-brace-structure system
has a deteriorated dynamic performance. Therefore, the advan-
tages of using the configuration where the GMD is placed in series
186 R. Mirza Hessabi, O. Mercan / Engineering Structures 126 (2016) 174–186
with a brace and in parallel with a viscous damper (called GVB sys-
tem) were shown. The effects of the equivalent mass ratio were
evaluated in more details and it was shown that lower values of
equivalent mass ratio may be sufficient for achieving the desired
performance. It was also shown that GVB systems with lower
equivalent mass ratios are less sensitive to the inherent damping
of the structure. Unlike other types of inertial dampers, the supple-
mental energy dissipation component of a GVB is not built-in to
the device and can be independently attached as an external com-
ponent. This will allow the design engineers to select from a wide
range of available energy dissipation device to use in parallel with
the GMD. In this study, a nonlinear viscous damper is used and a
simplified design formula is proposed to carry out the preliminary
design of this control system. Through numerical analyses and by
analyzing the input energy spectrum of structures with GVB sys-
tems, it was found that these systems provide an economical alter-
native to enhance the dynamic response of structures. GVBs can be
used at multiple degrees of freedom of the structure and the corre-
sponding equivalent mass values can be optimized to improve the
dynamic response of the multi-story structure. The optimal design
and performance of GVBs in multi-story structures and in struc-
tures with nonlinearities as well as the application of GMDs as part
of an active control system are the topics of the future research.
Acknowledgements
The authors would like to thank Dr. Constantin Christopoulos,
who provided the shake table for the experiments, and Ali Ashasi
Sorkhabi and Alan Gu for their assistance during experiments.
The financial support for this study from NSERC Discovery (Grant
371627-2009) and the start-up funds from the University of Tor-
onto is gratefully acknowledged. Any opinions, findings, conclu-
sions and recommendations expressed here are those of the
authors and do not necessarily reflect the views of the sponsors.
References
[1] Saitoh M. On the performance of gyro-mass devices for displacement
mitigation in base isolation systems. Struct Control Health Monit 2012;19
(2):246–59. http://dx.doi.org/10.1002/stc.419.
[2] Smith MC. Synthesis of mechanical networks: the inerter. IEEE Trans Autom
Control 2002;47(10):1648–62. http://dx.doi.org/10.1109/TAC.2002.803532.
[3] Wang FC, Wu SY. Vibration control of an optical table employing mechatronic
inerter networks. J Vib Control 2014. 1077546314528365. DOI: 10.1177/
1077546314528365..
[4] Lazar IF, Neild SA, Wagg DJ. Vibration suppression of cables using tuned inerter
dampers. Eng Struct 2016;2016(122):62–71. http://dx.doi.org/10.1016/j.
engstruct.2016.04.017.
[5] Chen MZQ, Papageorgiou C, Scheibe F, Wang F-C, Smith MC. The missing
mechanical circuit. IEEE Circuits Syst Mag 2009;9(1):10–26. http://dx.doi.org/
10.1109/MCAS.2008.931738.
[6] Smith MC, Wang FC. Performance benefits in passive vehicle suspensions
employing inerter. Vehicle Syst Dynam 2004;42(4):235–57. http://dx.doi.org/
10.1109/CDC.2003.1272954.
[7] Evangelou S, Limebeer DJN, Sharp RS, Smith MC. Mechanical steering
compensators for high-performance motorcycles. J Appl Mech-Trans ASME
2007;74(2):332–46. http://dx.doi.org/10.1115/1.2198547.
[8] Jiang JZ, Matamoros-Sanchez AZ, Goodall RM, Smith MC. Passive suspensions
incorporating inerters for railway vehicles. Vehicle Syst Dynam Int J Vehicle
Mech Mobility 2012;50(Suppl. 1):263–76. http://dx.doi.org/10.1080/
00423114.2012.665166.
[9] Lazar IF, Neild SA, Wagg DJ. Using an inerter-based device for structural
vibration suppression. Earthquake Eng Struct Dynam 2013. http://dx.doi.org/
10.1002/eqe.2390.
[10] Marano GC, Greco R. Optimization criteria for tuned mass dampers for
structural vibration control under stochastic excitation. J Vib Control 2011;17
(5):679–88. http://dx.doi.org/10.1177/1077546310365988.
[11] Papageorgiou C, Houghton NE, Smith MC. Experimental testing and analysis of
inerter devices. J Dyn Syst Meas Contr 2008;131(1):1–11. http://dx.doi.org/
10.1115/1.3023120.
[12] Wang FC, Hong MF, Lin TC. Designing and testing a hydraulic inerter. Proc Inst
Mech Eng Part C: J Mech Eng Sci 2011;225(1):66–72. http://dx.doi.org/
10.1243/09544062JMES2199.
[13] Hwang JS, Kim J, Kim YM. Rotational inertia dampers with toggle bracing for
vibration control of a building. Eng Struct 2007;29(6):1201–8. http://dx.doi.
org/10.1016/j.engstruct.2006.08.005.
[14] Saito K, Yogo K, Sugimura Y, Nakaminami S, Park K. Application of rotary
inertia to displacement reduction for vibration control system. In: 13th world
conference on earthquake engineering. Canada: Vancouver, BC; 2004 [Paper
No. 1764].
[15] Ikago K, Saito K, Inoue N. Seismic control of single-degree-of-freedom
structure using tuned viscous mass damper. Earthquake Eng Struct Dynam
2012;41(3):453–74. http://dx.doi.org/10.1002/eqe.1138.
[16] Watanabe Y, Ikago K, Inoue N, Kida H, Nakaminami S, Tanaka H, Sugimura Y,
Saito K. Full-scale dynamic tests and analytical verification of a force-restricted
tuned viscous mass damper. Proceedings of the 15th world conference on
earthquake engineering, 2012 [Lisbon, Portugal].
[17] Sugimura Y, Goto W, Tanizawa H, Saito K, Nimomiya T. Response control effect
of steel building structure using tuned viscous mass damper. In: Proceedings
of the 15th world conference on earthquake engineering [Lisbon, Portugal].
[18] Mirza Hessabi R, Mercan O. Application of gyro-mass dampers to mitigate the
seismic failure in soft first story buildings. In: Proceedings of the 2015
structures congress. Portland (USA): American Society of Civil Engineers
(ASCE); 2015. p. 2032–43. DOI: 10.1061/9780784479117.175.
[19] Mirza Hessabi R, Zhou J, Mercan O. Seismic response investigations of
nonlinear single-degree-of-freedom systems equipped with gyromass
dampers. In: Proceedings of the 25th Canadian congress of applied
mechanics (CANCAM) [London, Canada].
[20] Wang FC, Hong MF, Chen CW. Building suspensions with inerters. Proc Inst
Mech Eng, Part C: J Mech Eng Sci 2010;224(8):1605–16. http://dx.doi.org/
10.1243/09544062JMES1909.
[21] Wang FC, Su WJ. Inerter nonlinearities and the impact on suspension control.
In: Proceedings of the 2008 American control conference. Seattle, USA. DOI:
10.1109/ACC.2008.4586992.
[22] Giaralis A, Taflanidis AA. Reliability-based design of tuned mass-damper-
inerter (TMDI) equipped multi-storey frame buildings under seismic
excitation. In: Proceedings of the 12th international conference on
applications of statistics and probability in civil engineering, ICASP12
[Vancouver, Canada].
[23] Takewaki I, Murakami S, Yoshitomi S, Tsuji M. Fundamental mechanism of
earthquake response reduction in building structures with inertial dampers.
Struct Control Health Monit 2012;19(6):590–608. http://dx.doi.org/10.1002/
stc.457.
[24] Christopoulos C, Filiatrault A. Principles of supplemental damping and seismic
isolation. Milan, Italy: IUSS Press; 2006.
[25] Uang CM, Bertero VV. Evaluation of seismic energy in structures. Earthquake
Eng Struct Dynam 1990;19(1):77–90. http://dx.doi.org/10.1002/
eqe.4290190108.
[26] Next generation attenuation relationships – strong motion database, Pacific
Earthquake engineering research center. Berkeley, CA (USA): Regents of the
University of California; Retrieved from <http://ngawest2.berkeley.edu>.
[27] Taniguchi T, Der Kiureghian A, Melkumyan M. Effect of tuned mass damper on
displacement demand of base-isolated structures. Eng Struct 2008;30
(12):3478–88. http://dx.doi.org/10.1016/j.engstruct.2008.05.027.