Journal of Sound and Vibration 366 (2016) 62–80

Contents lists available at ScienceDirect

Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi

Study on vibration suppression based on particle damping

in centrifugal field of gear transmission
Wangqiang Xiao a,n, Jiani Li a, Sheng Wang b, Xiaomeng Fang a
a
Department of Mechanical and Electrical Engineering, Xiamen University, Xiamen 361005, China
b
Beijing Institute of Electronic System Engineering, Beijing 100854, China

a r t i c l e in f o abstract

Article history: Though particle damping technology has been applied to vibration suppression in steady
Received 7 September 2015 state, there are few reports about the study of particle dampers in centrifugal fields because
Received in revised form of its nonlinear damping performance and complex mechanism. Introducing particle
20 November 2015
damping technology into gear transmission will effectively reduce the vibration from gear
Accepted 5 December 2015
engaging, especially for harsh working conditions, such as high temperature and oil lubri-
Handling Editor: L.G. Tham
Available online 31 December 2015 cation. In this paper, we have explored the mechanism of gear excitation and determined the
relationship between the rotational speed and gear's modal parameters in centrifugal fields. A
Keywords: mechanical model of the particle damper based on the discrete element method (DEM) in
Gear transmission
centrifugal fields has been established. Furthermore, the DEM model has been verified by
Particle damping
comparing simulation data with experimental data. Based on the model, we have discussed
Discrete element method
Modal analysis the particle damper's energy dissipation mechanism in centrifugal fields, as well as the cal-
Vibration suppression culation method of energy dissipation. Moreover, the influence of the particle size on energy
dissipation characteristics has been analyzed. The results can provide theoretical guidance for
vibration and noise reduction of the gear transmission.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction

As gear transmission tending to be high-speed, heavy-duty, and high-reliability, more severe demands for the dynamic
performance of the gear transmission are put forward. The vibration caused by gear transmission is the main source of the
noise for most machinery and equipment. Therefore, to figure out how to control the vibration and noise of gear trans-
mission will exert a significant influence on the precision, performance and life of the machinery, the protection for
operators alike [1].
Currently, studies on passive vibration control technique are mainly focused on frictional dampers and viscoelastic
dampers [2–6], the material of which is very sensitive to temperature. When the temperature exceeds 60 °C, the damping
material is susceptible to creep or brittle rupture, which leads to the degradation or failure of the damping performance [6].
As a result, the application of these dampers will be greatly restricted by high-temperature and oil-lubrication working
conditions.
Particle damping [7–10] technology is a passive vibration control technology, which is based on the theory of energy
dissipation mechanism. Damping effect is mainly caused by the inelastic collision and friction of particles filled in the cavity

n
Corresponding author.

http://dx.doi.org/10.1016/j.jsv.2015.12.014
0022-460X/& 2015 Elsevier Ltd. All rights reserved.
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 63

[11–14], realizing the function of vibration reduction for the mechanical system. The advantages of particle damping include
a prominent damping effect, resistance to heat and harsh environment, little modification to the original structure, and little
associated mass, etc. [15–22]. Introducing particle damping technology into gear transmission will effectively reduce the
vibration from gear engaging, especially for harsh working conditions, such as high temperature and oil lubrication.
However, relevant research has been scarcely reported.
The main vibration source of gear transmission is the time-varying stiffness [23–24] motivation on tooth surface, which
then being transmitted to the external part. Take gear reducer as an example, whose vibration path is: the time-varying
stiffness motivation on tooth surface (vibration source) - lightening holes - shaft - bearing - bearing box - enclosure.
According to the rule that the shorter the distance between the part adopting particle damping technology with the
vibration source, the better the damping performance is, applying particle damping technology to the lighting holes will
achieve the optimal damping effect in the gear transmission system.
Currently, studies about particle damping technology are mostly specific to steady state [25–29]. However, the gear pair
is under high rotational speed while working [30] and the particulate media in the lightening holes will bear strong cen-
trifugal loads. Particles are piled up in the end remote from the center of rotation, which leads to the difference of the
dynamic response and energy dissipation characteristics of particulate media from that in steady state. Hence, it is to be
explored whether the particle damping technology is able to reduce vibration and noise effectively in centrifugal fields.
Besides, it is necessary to study the energy dissipation mechanism of particles in centrifugal fields, as well as analyze the
influence rule of the damping particles' parameters on the damping performance of gear transmission. The research results
will provide the theoretical foundation and design criteria for the application of the vibration reduction in gear
transmission.
At present, dynamic analysis method of particle dampers are regression model method [31], restoring force surface
method [32], power input method [33] and neural network method [34]. Though these simplified models have made great
achievements, they are hard to be applied to centrifugal fields, due to the particular energy dissipation mechanism of
friction and inelastic collision in gear transmission. In recent years, with the establishment of the theoretical foundation for
particulate media and the improvement of high-performance computing technique, it has become possible to utilize the
discrete element method (DEM) [7,35–39] to study the mechanism of particulate media's energy dissipation from both
macroscopy and microscopy, which may hopefully contribute to the analysis in centrifugal fields.
In this paper, we have adopted DEM as the numerical simulation method to calculate the energy dissipation of the
particle damper [21,38,40,41]. As first step, the model of the gear pair has been established as the carrier of the particle
damper. Secondly, under different rotational speed and load, the dynamic analysis of the gear pair has been carried on to get
the performance of the centrifugal field of gear transmission. Then, the value of energy dissipation of particle dampers with
different particle sizes in varied centrifugal fields has been calculated based on DEM. The paper outline is organized as
follows. First, the particle damping model in centrifugal fields of gear transmission is developed in Section 2 based on
Hertz–Mindlin soft-sphere model. Section 3 focuses on the calculation of the rotational speed and acceleration of the gear
pair by analyzing the model of time-varying mesh excitation on tooth surface. Section 4 discusses the influence of the

Fig. 1. Schematic depiction of the DEM model.

64 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

particle size on the value of energy dissipation and Section 5 verifies the simulation results with the experimental data.
Finally, the main conclusions of the work are summarized in Section 6.

2. Particle damping model in centrifugal field of gear transmission

2.1. Governing equations

DEM is a numerical method used to calculate the mechanical behavior of granular media system. It can divide the
particulate media into a number of discrete units, and then analyze the system according to Newton's second law. Dynamic
(static) relaxation method is applied to loop iteration, updating the data of all the units at each time step and simulating the
microscopic movement of each unit in the system [42].
For the modeling gear, we simplify the model as a single freedom mass-spring-damper system to a container filled
spherical particles. The schematic depiction of the DEM model is shown in Fig. 1. The body forces on the particles inside the
container due to gravitational and centrifugal forces are

gz ¼  g0
(1)
gy ¼  ω2 R

where R is the displacement between the particle and the rotation axis of the gear and ω is the particle's rotational speed.
The governing equation of the system is given as
X
Me z€ þCe z_ þ Ke z ¼ F þ Me gz (2)

where Me ; Ce ; Ke are the equivalent mass matric, damping matric and stiffness matric of the system under certain rotational
speed (Note that such parameters change with rotational speed due to the change of the natural frequency, thus we can get
equivalent parameters for each rotational speed.), z; z_ ; z€ respectively represents the displacement, velocity and acceleration
P
matrix of the gear system, F is the sum of all the contact forces in the system between particles and the container.
For the mechanical model between particles, governing equations for particle i at a given time point are as follows:
ki 
X 
mi p€ i ¼ Fnij þ Ftij þ mi ðgy þgz Þ (3)
j¼1

X
ki
Ii φ
€ ij ¼ Tij (4)
j¼1

where mi is the mass of the given particle, I i is the inertia moment, gis the ground acceleration, p€ i ; φ€ ij are the vectors of the
translational acceleration and the angular acceleration, Fnij ; Ftij are the normal and tangential contact force between particle i
and particle j (If particle i collides with the container, j represents the inner wall of the enclosure.). The acting force does not
act on the barycenter, but on the contact point between the two particles. Tij represents the torque acting on particle i, ki
represents the number of particles in contact with the particle i.

Fig. 2. Contact model.

 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 65

2.2. Contact model

The contact model is used for quantitatively calculating the normal contact force and tangential contact force when
particle collision occurs. Taking both calculation speed and accuracy into consideration, the Hertz–Mindlin soft-sphere
model is adopted. The contact force could be divided into two categories, as shown in Fig. 2. The left part in Fig. 2 shows the
interaction between particles and the inner wall of the enclosure, and the right part shows the physical contacts between
particles. Each particle motion includes translation and rotation. When two particles collide with each other, the relative
motion between the contact entities can be broken up into normal motion and tangential motion: the normal motion can be
simplified as a spring-damping component, while the tangential motion can be simplified as a spring-damping component
and a sliding frictional component.
Equations of normal motion are given as
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
< kn  pd δn þ 2ζ pd mi kn  pd δ_ n
> δn ¼ r i di ðparticle  damperÞ
F nij ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi   (5)
> _
: kn  pp δn þ 2ζ pp mi þ mj kn  pp δn δn ¼ r i þ r j  P i P j  ðparticle  particleÞ
mi mj

where δn ; δ_ n are the relative displacement and velocity between particle i and particle j or the inner wall of the enclosure, di
is the distance between the given particle and the wall, P i is particle's position vector.
Based on previous studies, ζ is the damping ratio, which is determined by the stiffness k and the damping coefficient c.
kn  pp represents for the normal stiffness between particles, while kn  pd stands for the normal stiffness between the particle
and the damper wall. Similarly, ζ pp is the damping ratio between particles and ζ pd is the damping ratio between the particle
and the damper wall.
Equation of tangential motion is given as

F tij ¼  μs F nij δ_ t =jδ_ tj (6)

Fig. 3. Direction of the contact force acting on each particle (single hole).

Fig. 4. Particles in the gear and the direction of the forces acting on them.
66 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 5. The Gear model.

Fig. 6. The first-order gear displacement nephogram of gear 1 under different rotational speeds.

where μs is the friction coefficient between particles or between the particle and the damper wall. δ_ t is the tangential
relative velocity of particle.
Fig. 3 shows the direction of the contact force acting on each particle in one of the lighting holes under different con-
ditions. The direction of arrows represents for the direction of forces. Particles in Fig. 3(a) are under random excitation,
where the particles have random collision and contact forces are in all directions. The gear in Fig. 3(b) is just about to rotate
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 67

and the directions of particles' contact forces have tendency to follow the direction of gear's motion. Fig. 3(c) is when the
gear rotates steadily and the main way to dissipate energy is rolling friction. Under such circumstance, the direction of
contact force acting on each particle is the same with the direction of gear's motion. Fig. 4 shows the particles in the gear
and the direction of the forces acting on them as well as their motion.

2.3. Calculation of equivalent parameters

(1) Calculation of the equivalent mass

In this work, the damping gear is shown in Fig. 5 (gear 1). The gear is made of 40Cr. There are 8 lightening holes in the
web of the gear, with the diameter of 15 mm. In addition, there are four bolt holes in the web, for the use of fixing the plate
cover in order to seal the lightening holes. Here we take advantage of the inherent lightening holes of the gear as the
containers of the particles, thus there is no need to add attached containers. As a result, it is equivalent to attach containers
with the mass of 0. If we attach additional containers to the gear, the mass of the containers in this case is not zero, as shown
in Fig. 5 (gear 2).
The equivalent stiffness of gear 1 and gear 2 are given as

K e ¼ 4π 2 f Ii ðM e þ M i Þ
2
(7)
with ƒ I  1 , ƒ I  2 the first mode natural frequency (Hz), M 1 , M 2 the dampers' mass.
Solving this equation and we get the simplification of the equivalent mass:
2 2
f I2 M 2  f I1 M 1
Me ¼ 2 2
(8)
f I1 f I2
Analyze the modal by ANSYS software, and get the first mode frequency under different rotational speed. The dis-
placement nephogram of gear 1 is shown in Fig. 6. Details of the finite element model created with the ANSYS software are:
Element type: Brick 8 node 185, node number: 39,930, unit number: 194,728.
We can also get the first-order gear displacement nephogram of gear 2 by finite element analysis. The first mode fre-
quency obtained from gear displacement nephogram increases correspondingly with the increase of the rotational speed.
We can get the relationship of the two in Fig. 7. The points in Fig. 7 are the simulation data from finite element analysis,
while the curves are fitted values.
According to the polynomial fitting in MATLAB, first-order natural frequency of gear 1 and 2 can be given as
(
ƒ I  1 ¼ 3:447  10  9 ω3 þ5:371  10  6 ω2 þ 2:322  10  4 ω þ7:296
(9)
ƒ I  2 ¼ 3:649  10  9 ω3 þ5:512  10  6 ω2 þ 2:795  10  4 ω þ7:555

The equivalent mass of the gear at different rotational speed is derived by Eq. (8) and Eq. (9) as
M e ¼  5:735ω3 þ 5:633  10  6 ω2 þ 2:223  10  4 ω þ 1:636 (10)
Some equivalent mass values are given in Table 1.

(2) Calculation of the equivalent elastic and damping coefficients

Equations of the equivalent elastic and damping coefficients are as follows:

K e ðωÞ ¼ 4π 2 f I ½M e ðωÞ þ M
2

C e ðωÞ ¼ 4πζ ƒ I ½M e ðωÞ þ M (11)

qffiffi
1ε 3
where the damping ratio is given by the equation: ζ ¼ π 5 [11] (ε is the coefficient of restitution), supposing the
material's coefficient of restitution is 0.75 here, putting ζ and the data in Table 1 into Eq. (11), we get the equivalent elastic

Fig. 7. Gear's first mode frequency under different rotational speed.

68 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Table 1
The equivalent mass of the gear at certain rotational speed.

ω=rpm 0 100 200 300 400 500 600 700 800

M e =kg 1.643 1.699 1.852 2.062 2.273 2.438 2.546 2.586 2.486

Table 2
The equivalent elastic and damping coefficients of the gear at certain rotational speed.

ω=rpm 0 100 200 300 400 500 600 700 800

Ke/(kN/m) 3.8502 4.0425 4.5937 5.4240 6.3902 7.3352 8.1792 8.8345 8.9740
Ce/(N s/m) 0.1591 0.1658 0.1845 0.2115 0.2411 0.2675 0.2886 0.3023 0.2987

and damping coefficients of the gear at different rotational speed. Some values at certain rotational speed are shown in
Table 2.
We can also get the fitted values of K e and C e , the equivalent elastic and damping coefficient of the gear can be put as a
function of the rotational speed:
(
K e ¼  2:157  10  8 ω3 þ 2:613  10  5 ω2  6:834  10  4 ω þ 3:85827
(12)
C e ¼  7:226  10  10 ω3 þ7:962  10  7 ω2 þ4:677  10  7 ω þ 0:1588

(3) Calculation of particle parameters

In this paper, we adopt the soft-ball model and do not take surface deformation of particles and the loading history of
contact force into consideration, which will result in less computational efforts in practical engineering problems. The
elastic and damping coefficients of the soft ball model are related to elasticity modulus, Poisson's ratio and other parameters
of the particulate material. The normal elastic coefficient is determined by the Hertz contact theory:
!
2  1 
4 1  νi 2 1  νj Ri þ Rj  1=2
kn ¼ þ (13)
3 Ei Ej Ri Rj

where E and ν are the Young's elastic modulus and Poisson's ratio of the particulate material, R is the radius of the particle (If
particle i collides with the container, j represents the inner wall of the enclosure. In this case, the inner wall of the enclosure
is regarded as having an infinite radius.). And the subscript i and j represent for the given particle i and j.
The tangential elasticity modulus can be expressed as
ks ¼ γ kn (14)
ð1  νi Þ=Gi þ ð1  νj Þ=Gj
where γ is proportional coefficient: γ ¼ [14] (with G the shear modulus of the particulate material).
ð1  0:5νi Þ=Gi þ ð1  0:5νj Þ=Gj
In this paper, we use the same particles for each simulation. Therefore, kn and ks can be simplified as
8 pffiffiffiffi
< kn ¼ 3ð1 2R E
 ν2 Þ
(15)
: ks ¼ γ kn ¼ ð1  νÞ kn
ð1  0:5νÞ

For the case that the spring oscillator with the mass of m is in the critical damping state, the mechanical energy
attenuation will be in its fastest pace, where the normal and tangential damping coefficients can be expressed as
( pffiffiffiffiffiffiffiffiffi
cn ¼ 2 mkn
pffiffiffiffiffiffiffiffiffi (16)
cs ¼ 2 mks

3. Model of time-varying mesh excitation on tooth surface

3.1. Virtual prototype model

To explore particle dampers' general energy dissipation law in centrifugal fields, we need to define the boundary con-
ditions of the dampers by analyzing gear's dynamics at varied rotational speed and load. Based on the multi-rigid body
system theory and Lagrange dynamics equation, the dynamic equations are established using virtual prototype.
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 69

Fig. 8. Virtual prototype with all bindings and loads.

Table 3
Load class and its corresponding torque.

Load class 1 2 3 4 5 6 7

Driven gear's torque /(N m) 2 4 8 16 34 60 96

The first step is to carry out gear's parametric modeling. Modeling data can be exchanged between Solidworks and
Adams by the format of IGS. The transmission ratio of the gear pair is 1:1, the modulus of the driving and driven gear is 4.5,
the tooth number is 24, the pressure angle is 20° and the addendum coefficient is 1. The material of the driven and driving
gear is 40Cr, elastic modulus is 206 GPa, density is 7800 kg/m3, and Poisson ratio is 0.28.
Next step is to set related constraints and load for the gear transmission system. Bindings added on the gear pair are:
revolution joints of the driving and driven gear, the motivation between the driven and driving gear. Load to be imposed
are: a rotating motion excitation on the driving gear as the input terminal and a load on the driven gear as the output
terminal. In order to simulate the starting process of the motor, we employ STEP function as the driving function, for STEP
function can avoid sudden load change. An established virtual prototype with bindings and load is given in Fig. 8.
As to the selecting of the contact function, in order to achieve a better way to calculate the unilateral motivation on the
gear, we choose the IMPACT Function based on stiffness and damping coefficients considering various disturbances when
gear meshing.
Next, we adopt integrator GSTIFF, format I3, with the accuracy of 1.0E 003, and set the simulation time as 1 s, time step
as 0.0001 s. The results are outputted by Adam's post-processing module.
In order to find out the dynamics of the driven gear under different load, we divide the load exerted on the driven gear
into 7 classes. The values are shown in Table 3.

3.2. Analysis of rotational speed

Fig. 9 shows the driven gear's rotational speed curve at different load when the driving gear's rotational speed is fixed.
The rotational speed of the driven gear increases with the start of the motor during 0–0.1 s. After 0.1 s, the rotational speed
remains approximately a constant value. The volatility is a result of the impact from gear meshing. The rotational speed of
the driving gear is set as 3600°/s and the rotational speed of the driven gear fluctuates around 3600°/s, to be precise, around
3587.9°/s at third-class load, and 3598.6°/s at seventh-class load. And the fluctuation is greater as the increase of the load.
To make the fluctuation clearer, we get the rotational speed of the driven gear within the time of a single tooth
engagement at third-class load and seventh-class load in Fig. 10. We can get the information from the figure that the
dynamics of the gear at different load is basically the same. The rotational speed of the driven gear fluctuates around 3600°/
s. Nevertheless, the volatility at different load differ from each other significantly: the range of the angular velocity at
seventh-class is from 3470°/s to 3830°/s, much higher than that at third-class load, which is from 3505°/s to 3640°/s. The
result shows that as the load increases, the fluctuation of gear's rotational speed increases correspondingly and as a result,
deformation and vibration of the gear will enlarge.
70 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 9. Rotational speed of the driven gear (a) at third-class load and (b) at seventh-class load.

3640 3850

speed
speed 3800
3620

3750
3600

3700
3580
3650
3560
3600

3540
3550

3520 3500

3500 3450
0.324 0.3245 0.325 0.3255 0.326 0.3265 0.327 0.3275 0.328 0.3285 0.4105 0.411 0.4115 0.412 0.4125 0.413 0.4135 0.414 0.4145
Time(sec) Time(sec)

Fig. 10. Rotational speed fluctuation within a single tooth engagement period (a) at third-class load and (b) at seventh-class load.

Fig. 11. Angular acceleration of the driven gear (a) at third-class load and (b) at seventh-class load.
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 71

Fig. 12. Angular acceleration under different load classes.

Fig. 13. Diagram of particles with different diameters filled in the lightening hole (single hole).

3.3. Angular acceleration analysis

In order to accurately calculate the motivation on the driven gear under the different load, we need to analyze the
angular acceleration. The acceleration curves of the driven gear are shown in Fig. 11. In general, there is a maximum value of
the acceleration at the startup of the motor. From 0 s to 0.1 s, the acceleration of the driven gear is maximum, and after 0.1 s,
the angular acceleration of the driven gear changes periodically, due to the periodical gear excitation from time-varying
meshing stiffness. To illustrate, when the tooth starts to mesh, the excitation as well as the vibration becomes more
intensive, and then declines as the decrease of the time-varying meshing stiffness. The excitation begins to increase as the
next tooth meshing, and the cycle repeats itself.
With the increase of the load, the average of the driven gear's angular acceleration grows accordingly. The average of the
acceleration is 9.04  105°/s2 at third-class load and 1.04  106°/s2 at seventh-class load.
Fig. 12 shows the values of acceleration under different load class. The acceleration variation obtained from ADAMS can
be transformed as the excitation on the tooth surface, which is defined as the boundary condition of the particle dampers
and will be used in the following discrete element analysis.

4. DEM simulations and discussion

4.1. Motion of particles

Particles are made of stainless steel. Fig. 13 shows the diagram of particles with different size filled in the lightening hole.
Fig. 14 shows the motion of particles with 3 mm in diameter at 600 rpm and first-class load. Particles with high energy
are denoted in red and those with low energy are denoted in blue. As is shown in the figure, apparently being different from
72 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 14. Diagram of particles with different energy.

Fig. 15. Particle motion within a single period of tooth engagement.

Table 4
Parameters of the gear and particles at different loads.

Parameters of the gear Parameters of particles

Me/ kg 2.546 Diameter d Normal coeffi- Tangential coef- Normal damping Tangential damping Coefficient of
Ke/ (kN/m) 8.1792 (mm) cient kn (kN/m) ficient ks (kN/m) coefficient cn coefficient cs restitution e
Ce/ (N s/m) 0.2886 (N U s/m) (N U s/m)

Gear's rota- 600 rpm 1 1,143,764 957,570 3.276837 3.914

tional speed 2 3,396,805 2,843,837 16.20837 19.36
3 4,160,219 3,482,974 32.91907 39.32
Damper' 15 mm 4 4,803,807 4,021,792 54.46046 65.05
diameter 5 5,370,820 4,496,500 80.49767 96.15 0.7
6 5,883,438 4,925,669 110.7377 132.27
7 6,354,839 5,320,330 145.0247 173.22
8 6,793,609 5,687,673 183.1647 218.78
9 7,205,711 6,032,688 225.11 268.87

the steady state, in centrifugal fields, particles are pressed on the inner wall away from the rotation axis. Thus, the energy
dissipation law may not be the same with that in the steady state because the distribution of energy dissipation by inelastic
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 73

d=3mm d=5mm

d=7mm d=9mm

Fig. 16. The energy dissipation trend with the particles' diameter of (a) 3 mm, (b) 5 mm, (c) 7 mm and (d) 9 mm.

Fig. 17. Collision times for each particle size.

collision and friction will change—rolling friction will be the main movement in centrifugal fields (whereas in the steady
state inelastic collision dissipates more energy).
Fig. 15 shows the the motion of particles at different time point within a single period of tooth engagement (at 600 rpm,
third-class load, 3 mm in diameter). At first, due to the centrifugal force, most particles move adhering to the wall. Then,
because of the excitation of the gear, as the speed of the gear changes, particles start to bounce inward. Whereafter, particles
74 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 18. Total energy dissipation for different particle-size damper at varied load.

Table 5
Parameters of the system at different rotational speeds.

Rotational speed/rpm 200 400 600 800 1000 1200

Me/kg 1.9496 2.2594 2.4692 2.5790 2.5890 2.4990

Ke/(kg/m) 4.5942 6.3852 8.1959 8.9909 7.7349 3.3924
Ce/(N s/m) 0.1850 0.2401 0.2896 0.2988 0.2329 0.0572

being excited begin to move toward the other side of the damper and rebound. Finally, those particles bouncing back collide
with other particles. During the process, the energy of the gear system is transferred to particles. The mechanical energy is
converted to heat by the inelastic collision and friction of particles. Hence, the vibration of the gear transmission system is
reduced.

4.2. The influence of particle size on the damping effect at different loads

In order to study the influence of particle diameter on the damping effect at different load, we choose stainless particles
with the diameter of 1 mm, 2 mm, 3 mm, 4 mm, 5 mm, 6 mm, 7 mm, 8 mm, 9 mm. Some parameters of the particles are:
density ρ ¼ 7850 kg=m , Poisson's ratio ν ¼ 0:28, shear modulus G ¼77 GPa, Young's elastic modulus E¼ 206 GPa, filling rate
3

ψ ¼ 60%. The rotational speed of the gear is fixed as 600 rpm and the load exerted on the driven gear is set from first-class
to seventh-class. The calculated parameters of the gear and particles at different loads are shown in Table 4.
Fig. 16 shows the histogram of particle dampers' energy dissipation with different particle diameters at 600 rpm, second-
class load. Though the total energy dissipation is different at each time point, the overall trend is similar. At about 0.0215 s, a
peak appears. Such an energy peak is a result of the excitation from gear's time-varying meshing stiffness. And it levels off
gradually along with the disappearance of the gear excitation. At 600 rpm, the period of a single tooth engagement is
0.0042 s for the gear with 24 teeth. As is discussed in Section 3, the excitation rises and falls cyclically. That is to say, the next
excitation will occur in approximately 0.0257 s, which meets the results of the simulation, as is shown in Fig. 16.
The relationship between collision times and particle size is shown in Fig. 17. The smaller the particle, the less the energy
produced by each collision, but the more the collision times. Therefore, it is hard to determine without further discussion
the optimum particle size which dissipates the most energy and achieve the best damping effect.
Fig. 18 shows the damper's total energy dissipation within a simulation period with different particle diameters at varied
load. From Fig. 18, it can be concluded that the total energy dissipation reaches the maximum when the particle diameter is
about 4–5 mm at whatever load. The energy dissipation grows before 5 mm and falls after 5 mm. The reason for this
phenomenon is just as what we discussed before. Smaller particles collide far more times than bigger ones, but dissipate
much less energy at each collision, thus the total energy dissipation may still be low. However, as the volume of the damper
is fixed, when the diameter reaches to 9 mm, the number of particles filled in the damper is only two (as shown in Fig. 13),
in which case, the collision times are extremely few and the total energy dissipation is also very low. Therefore, the optimum
particle size must be an intermediate value between 1 mm and 9 mm and the simulation results agree well with the
conjecture.
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 75

Fig. 19. Total energy dissipation for different particle-size dampers at varied rotational speed.

Fig. 20. Diagram of test system.

The change of load class has little to do with the damper's total energy dissipation trend. Fig. 18 shows that with the
increasing load class, the energy dissipation is also enlarged. However, the general trend is the same at each load. At first-
class load, the total energy dissipation is the least. At seventh-class load, the total energy dissipation reaches the maximum.
The reason is that as the load class being aggrandized, the excitation of the gear system is intensified. As a result, the
particles' acceleration is enhanced, and the heat converted from collision and friction motion become intensified.

4.3. The influence of particle diameter on the damping effect at different rotational speed

In order to study the damper's damping effect with different particle diameters at varied speed, we set the load as
fourth-class and the rotational speed at 200 rpm, 400 rpm, 600 rpm, 800 rpm, 1000 rpm and 1200 rpm relatively. The filling
rate is also set at 60%. The particle's parameters are the same as that in Table 4. The gear's parameters are shown in Table 5.
76 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 21. Diagram of the devices (a) test devices and (b) the wireless acceleration sensor.

Fig. 22. Diagram of the gears (a) the non-fixed gears and (b) the fixed gear filled with particles.

Fig. 19 shows the damper's total energy dissipation within a single simulation period with different particle diameters at
varied rotational speed.
When the rotational speed is below 500 rpm, with the increase of the speed, the total energy loss of the particle damper
slowly rises. However, when the speed exceeds 600 rpm, the energy dissipation increases rapidly. And after 1200 rpm, the
increase rate declines. The reason for this is the vibration intensifies along with the increase of the rotational speed.
Therefore, the larger the rotational speed, the more vibration energy, and the more energy dissipation by particle dampers.
However, when the speed is relatively high, because of the centrifugal force, the particles adhere to the inner wall of the
enclosure and have difficulty in sliding friction. As a result, the rolling friction becomes the main way of energy con-
sumption. Since the rolling friction coefficient is much less than sliding friction coefficient, the increase rate of the total
energy dissipation goes down. We can still reach the conclusion from the simulation that the optimum diameter is still 4–
5 mm at varied rotational speed (Fig. 24 illustrates this point).

5. Experimental validation

5.1. Setup of the particle damping experiment

The experiment equipment includes four modules: the gear transmission system, the exciting device, the acceleration
sensor, and the data analytical equipment. The gear box's vibration is transmitted to the wireless acceleration sensor. After
the acquisition of the vibration signal, we will analyze the signal data and summarize the influence of different elements on
the damping effect. The equipment is composed of the GCL-100A gear test platform, which is supplied by Kebucheng
Company; the wireless sensors, the signal acquisition unit and the analytical software, which are supplied by Beetech
Company. The diagram of test system is shown in Fig. 20.
The gear bench adopts the test method of mechanical closed energy flow, the energy being returned, which largely
diminishes the power required and contributes to repeatable tests. The layout of the test gear is cantilevered. After
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 77

Fig. 23. Gear's acceleration with and without particle dampers.

Fig. 24. Energy dissipation and damping effect of particles at varied rotational speed.

disassembling the gear box cover, the gear can be easily dismantling and assembling without removing the shaft and
bearings, contributing to the error reduction of the experiment. Diagram of the devices are shown in Fig. 21(a). Because of
the resonance occurs in the process of start and stop, the signal collected may be unstable during this period of time.
Therefore, the signal is collected 15 s after the load and speed getting to be stable. Test five times for each working condition
and take the average as the result. In order to measure the vibration of the running gear, put the 3-direction wireless
acceleration sensor A103 at the bearing cap closest to the test gear. The sensor can collect the acceleration signal of X, Y, Z
direction simultaneously. After signal being collected, the acceleration signal is transmitted to the data-collecting instru-
ment. The diagram of the wireless acceleration sensor is shown at Fig. 21(b).
The gears adopted are spur gears, made of 45Cr. The gear's module is 4.5 mm, tooth number is 24, pressure angle is 20°,
and tooth width is 30 mm. There are 8 identical through-holes, whose diameter is 15 mm, and 4 identical screwed holes,
whose diameter is 4.5 mm on both sides of the gear. After the damping particles are filled, fix the sealing rings on each side
of the gear with screws. Make sure that particles cannot fall out during the test. The particles adopted in the test are made of
stainless steel, with the diameter from 1 mm to 3 mm. Precisely weigh the total weight of the particles in each hole by an
electronic balance, so as to maintain the dynamic balance of the gear while rotating. The diagram of the gear is shown in
Fig. 22(a). The diagram of the fixed gear filled with particles is shown in Fig. 22(b).

5.2. Experimental validation

We get the acceleration of the gear filled with particles under varied load and rotational speed by experiment, which is
expressed as apij . Likewise, the acceleration of the gear without particles is expressed as aeij . Use the factor De to indicate the
damping effect, which is expressed as
R 2 R 2 R 2
axpij dt þ aypij dt þ azpij dt
De ¼ 1  R 2 R 2 R 2
(17)
axeij dt þ ayeij dt þ azeij dt
78 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

Fig. 25. Energy dissipation and damping effect of particles at varied loads.

where i and j represents for load class and rotational speed, respectively. i¼1, 2, 3, 4, 5, 6, 7 and j¼200, 400, 600, 800,
1000, 1200.
Comparison of the gear's acceleration with and without particle dampers is shown in Fig. 23. The dampers do reduce the
acceleration of the gears, which also means the vibration is reduced by dampers. We can also get the information from the
figure that the ratio of the reduced acceleration increases with the increase of rotational speed. At 600 rpm, the dampers
reduce about 20% acceleration whereas at 1200 rpm, the dampers reduce more than 30% acceleration.
To verify the simulated results, we compare the simulated data with the tested data. Fig. 24 shows the comparison of the
two under different rotational speed, 7th-class load. Dashed lines are the tested data indicated in the left axis, and solid lines
are the simulated data indicated in the right axis. The simulation damping effect is based on the value of the particles'
energy dissipation, the greater the energy dissipates, the better the damping effect.
The figure indicates that the simulation and experimental data are closely relevant. So the validity of the proposed
theoretical model is demonstrated. To be specific, when the load is fixed, the damping effect becomes stronger along with
the increase of the rotational speed, and the tested trend also shows the damping effect in experiment grows with the
rotational speed. The reason for such an increase damping effect is a result of the intensification of the particles' motion. The
larger the rotational speed, the more intense the particles' motion, and the stronger the damping effect of the dampers. We
can also come to the conclusion that for the three compared diameters, the best damping effect is achieved by particles with
the diameter of 5 mm, which also proves the validity of the simulation.
Fig. 25 shows comparison of the simulated and tested data under different loads, 600 rpm. The dashed lines are the
experimental data indicated in the left axis, and the solid lines are the simulation results indicated in the right axis. The
figure indicates that the simulated and test data are closely relevant, thus the correctness and validation of the simulation
model is demonstrated. To be specific, firstly, the tendency of the simulated data and the tested data all show that as the
load increases, the dampers' damping effect, that is, the dampers' total energy dissipation in simulation, and the damping
factor in experiment, also aggrandize. Moreover, at fixed load, as the particle diameter increases, the damping effect gets
stronger at first and reaches its maximum at 4 mm in experiment and 5 mm in simulation. After that, as the diameter
increases, the damping effect goes down. Such discrepancy of the optimum diameter value in simulation and experiment
can be explained by the difference of the simulated and tested system. In experiment, there are many other additional
apparatuses besides the gear system such as the shaft, bearings, couplings, so on and so forth, which could change the
characteristics of the system. In simulation, we only simulate the performance of the gear system and no other disturbance
is considered. Thus, it is reasonable such discrepancy exists. Despite the little discrepancy of the optimum diameter value in
simulation and experiment, the tendency of the two is similar and the results all show that the optimum diameter is an
intermediate value between the greatest and smallest diameter value, which corresponds well with theory.

6. Conclusion

Through theoretical calculation, simulation analysis and experiment, the particle damper's damping mechanism has
been obtained. The main conclusions are

(1) The vibration frequency and the excitation frequency derived from the modal analysis of the gear are not relevant, and
therefore no resonance occurs. The inherent frequency of the gear with and without additional containers is different
which is shown in the aberration nephogram.
 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80 79

(2) Gear's first mode frequency changes with the rotational speed, resulting in the change of gear's parameters such as
mass, stiffness coefficient, and damping coefficient. Such parameters can be put as the polynomial of rotational speed.
(3) The gear engagement is analyzed, and the torsional vibration model of spur gear pair is established. The vibration of the
driven gear is defined by its rotational speed and acceleration. By comparison, the variation of the driving gear's
rotational speed and acceleration is far less than that of the driven gear's, which suggests the main source of the
vibration is the driven gear.
(4) The load imposed on the teeth change as the gear teeth engaging in and engaging out. The energy dissipation by
collision and friction is intensified as the growth of the gear's excitation and decreases as the decline of the gear's
excitation. The process repeats itself in a loop.
(5) As the rotational speed increases, the dampers' total energy dissipation increases alike. However, the positive corre-
lation of damping effect and rotational speed is correct only within a limited speed range. When the rotational speed is
very high, the centrifugal effect become very strong, and the rolling friction is in the dominant position instead of the
sliding friction, so that the discipline of energy dissipation of particle dampers has been changed. According to the test,
the damping effect goes down slightly when the rotational speed exceeds a certain value. In addition, the energy
dissipation goes up as the load increases.
(6) The optimum diameter is 4–5 mm under whatever condition. The optimum diameter in experiment is slightly different
from that in simulation. Such discrepancy can be explained by the changed characteristic of the system due to additional
apparatuses in experiment.
(7) Choosing an optimal particle diameter will help to effectively reduce the vibration of gear system for different rotation
speed and loads. The data obtained from the study can provide theoretical guidance for the vibration and noise
reduction of the gear transmission, and can be used as a reference for other similar systems as well.

Acknowledgments

The authors would like to acknowledge financial support from National Natural Science Foundation of China (No.
51205382) and the Fundamental Research Funds for the Central Universities, Xiamen University (No. 20720150094) and
Collaborative Innovation Center of High-End Equipment Manufacturing in Fujian and Xiamen Southern Oceanographic
Center.

References

[1] Y.T. Zhong, L.Q. Wang, W.L. Li, Research on damping technology of gear transmission, Journal of Mechanical Transmission 11 (2011) 17–20.
[2] G. Marunic, Rim stress analysis of thin-rimmed gear, Key Engineering Megye Materials, Advances in Fracture and Damage Mechanics VI (2007) 141–144.
[3] H.X. Liu, S.M. Wang, Y.S. Fan, The influence of radiation on the nonlinear dynamic characteristics of gear transmission, Mechanical Science and
Technology 28 (7) (2009) 965–970.
[4] W.A. Yang, M. Xu, Research on vibration and noise reduction of gear, Journal of Vibration and Shock 3 (1995). 49–52þ 56þ 85.
[5] C.Z. Gu, X.Y. Wu, Research on passive vibration reduction technology of gear, Journal of Mechanical Design and Manufacturing 09 (2010) 74–76.
[6] Y. Duan, Q. Chen, Experimental investigation of damping properties of the particle dampers with softlining, Journal of Vibration Engineering 24 (2)
(2011) 215–220.
[7] R.D. Friend, V.K. Kinra, Particle impact damping, Journal of Sound and Vibration 233 (1) (2000) 93–118.
[8] A. Papalou, S.F. Masri, An experimental investigation of particle dampers under harmonic excitation, Journal of Vibration and Control 4 (4) (1998)
361–379.
[9] C. Saluena, T. Poschel, S.E. Esipov, Dissipative properties of vibrated granular materials, Physical Review E 59 (4) (1999) 4422–4425.
[10] B.M. Shah, J.J. Nudell, K.R. Kao, L.M. Keer, Q. Jane Wang, K. Zhou, Semi-active particle-based damping systems controlled by magnetic fields, Journal of
Sound and Vibration 330 (2011) 182–193.
[11] X.M. Bai, B. Shah, L.M. Keer, Q.J. Wang, R.Q. Snurr, Particle dynamics simulations of a piston-based particle damper, Powder Technology 189 (1) (2009)
115–125.
[12] S.C. Dragomir, M. Sinnott, E.S. Semercigil, Ö.F. Turan, Energy dissipation characteristics of the particle sloshing in a rotating cylinder, Journal of Sound
and Vibration 331 (5) (2012) 963–973.
[13] Z. Lu, S.F. Masri, X.L. Lu, Studies of the performance of particle dampers attached to a two-degrees-of-freedom system under random excitation,
Journal of Vibration and Control 17 (10) (2011) 1454–1471.
[14] R.D. Nayeri, S.F. Masri, J.P. Caffrey, Studies of the performance of multi-unit impact dampers under stochastic excitation, Journal of Vibration and
Acoustics 129 (2) (2007) 239–251.
[15] Z.W. Xia, S.B. Wei, H.B. Wen, X.D. Liu, Research on the application of particle damping technology in the vibration control of the brake drum, Journal of
Vibration Engineering 6 (2014) 893–899.
[16] Z. Lu, X.L. Lv, W.M. Yan, Survey on the technology of particle damping, Journal of Vibration and Shock 7 (2013) 1–7.
[17] W.M. Yan, X.D. Zhang, Y.W. Huang, H.X. He, Based on the technique of particle damping structure vibration control, Journal of Beijing University of
Technology 9 (2012) 1316–1320.11.
[18] W.M. Yan, J. Wang, W.B. Xu, Based on single degree of freedom structure of particle damping mechanism study, China Civil Engineering Journal S1
(2014) 76–82.
[19] D. Guyomar, A. Badel, Nonlinear semi-passive multi-modal vibration damping: an efficient probabilistic approach, Journal of Sound and Vibration 294
(1–2) (2006) 249–268.
[20] K. Li, A.P. Darby, Experiments on the effect of an impact damper on a multiple-degree-of-freedom system, Journal of Vibration and Control 12 (5) (2006)
445–464.
[21] Z. Lu, X.L. Lu, W.M. Yan, A survey of particle damping technology, Journal of Vibration and Shock 32 (7) (2013) 1–7.
[22] H.V. Panossian, Structural damping enhancement via non-obstructive particle damping technique, Journal of Vibration and Acoustics 114 (1) (1992)
101–105.
80 W. Xiao et al. / Journal of Sound and Vibration 366 (2016) 62–80

[23] Z.G. Wan, Y.Y. Zi, H.R. Cao, Z.J. He, S. Wang, Time-varying mesh stiffness algorithm correction and tooth crack dynamic modeling, Journal of Mechanical
Engineering 49 (11) (2013) 153–160.
[24] W.N. Yu, Y.M. Shao, C.K. Mechefske, The effects of spur gear tooth spatial crack propagation on gear mesh stiffness, Engineering Failure Analysis 54
(2015) 103–119.
[25] Z.C. Shen, Application and prospect of particle damping in vibration reduction of spacecraft [J], Chinese Journal of Applied Mechanics 06 (2011). 608–
612þ 673.
[26] H.B. Wen, Z.W. Xia, X.D. Liu, Simulation and testing of the structure and vibration characteristics of the particle damper, Journal of Agricultural
Machinery 08 (2011) 26–29.
[27] J.Z. Qiu, Particle damping technology and flexible strip particle damper, Precision Manufacturing and Automation (2011) 3.
[28] Keiko Onitsha, Novel image restoration method using spherical Nano-particles for atomic force microscopy topography data containing tip-induced
distortions, Journal of the Vacuum Society of Japan 1 (3) (2013) 195–198.
[29] C. Zhang, T.N. Chen, X.P. Wang, Y.G. Li, Discrete element method model and damping performance of bean bag dampers, Journal of Sound and Vibration
33 (3) (2014) 6024–6037.
[30] W.Q. Xiao, W. Li, Experiment for vibration suppression based on particle damping for gear transmission, Advances in Manufacturing Science and
Engineering V (2014) 703–706.
[31] L. Hu, Q.B. Huang, Z.S. Xu, Regression analysis of particle damping, Chinese Mechanical Engineering 19 (23) (2008) 2834–2837.
[32] H. Jiang, Q. Chen, Application of restoring force surface method to particle damping research, Journal of Vibration, Measurement Diagnosis 27 (3) (2007)
228–231.
[33] H.W. Zhou, Q. Chen, S. Lin, An Experimental Investigation on NOPD coupled system in vertical direction, Journal of Vibration and Shock 26 (9) (2007)
124–127.
[34] A.H. Tanrikulu, Application of ANN techniques for estimating modal damping of impact-damped flexible beams, Advances in Engineering Software 40
(10) (2009) 986–990.
[35] P.A. Cundall, O.D.L. Strack, A discrete numerical model for granular assemblies, Geotechnique 29 (1) (1979) 47–65.
[36] Z. Lu, S.F. Masri, X.L. Lu, Parametric studies of the performance of particle damper under harmonic excitation, Journal of Structural Control and Health
Monitoring 18 (1) (2011) 79–98.
[37] K.F. Malone, B.H. Xu, Determination of contact parameters for discrete element method simulations of granular systems, Particuology 6 (6) (2008)
521–528.
[38] M. Saeki, Analytical study of multi-particle damping, Journal of Sound and Vibration 281 (3–5) (2005) 1133–1144.
[39] C. Zhang, T.N. Chen, X.P. Wang, Y.G. Li, Discrete element method model and damping performance of bean bag dampers, Journalof Sound and Vibration
333 (23) (2014) 6024–6037.
[40] L. Hu, Q. Huang, Z.X. Liu, A non-obstructive particle damping model of DEM, International Journal of Mechanics and Materials in Design 4 (1) (2008)
45–51.
[41] Z. Lu, X.L. Lu, S.F. Masri, Studies of the performance of particle damper under dynamic loads, Journal of Sound and Vibration 329 (2010) 5415–5433.
[42] S. Xu, F. Zhu, J. Zhang, An overview of the discrete element method and its coupling algorithms, Journal of Mechanics in Engineering 35 (1) (2013) 8–14.