You are on page 1of 5

Damping of rotating beams with particle dampers: Discrete element method analysis

D. N. J. Els

Citation: AIP Conference Proceedings 1542, 867 (2013); doi: 10.1063/1.4812069


View online: http://dx.doi.org/10.1063/1.4812069
View Table of Contents: http://aip.scitation.org/toc/apc/1542/1
Published by the American Institute of Physics
Damping of Rotating Beams with Particle Dampers:
Discrete Element Method Analysis
D.N.J Els
Dept Mech & Mechatron Eng, Stellenbosch University, South Africa.

Abstract. The performance of particle dampers (PDs) under centrifugal loads was investigated. A test bench consisting of a
rotating cantilever beam with a particle damper at the tip was developed (D. N. J. Els, AIAA Journal 49, 22282238 (2011)).
Equal mass containers with different depths, lled with a range of uniform-sized steel ball bearings, were used as particle
dampers. The experiments were duplicated numerically with a discrete element method (DEM) model, calibrated against the
experimental data. The DEM model of the rotating beam with a PD at the tip captured the performance of the PD very well
over a wide range of tests with different congurations and rotation velocities.
Keywords: Particles, DEM, Vibration, Damping, Centrifugal Forces.
PACS: 45.70.-n

INTRODUCTION were determined in [2]. The equivalent beam and con-


tainer mass me , stiffness ke and damping coefcient ce in
Particle dampers (PDs) are composed of a container terms of the rotation velocity (unit in s1 ) are
lled with one or more particles (metals, ceramics, etc.).
They function by dissipating energy through inelastic im- me 7.587 16107 4 + 1.863 02105 3
pacts and friction between the particles and the walls, + 2.228 30104 2 + 4.858 02104
and between the particles themselves. Particle dampers
+ 0.094 296 [kg] (1)
are simple and inexpensive devices. The range of appli-
cations of PDs is vast. Simonian [1] gives an in-depth ke 0.480 77 3 + 6.2839 2 7.2071
overview of applications in diverse disciplines such as + 1046.1 [N/m] (2)
aerospace, ground transportation and high performance 7 5
sporting equipment industry. ce 9.173510 + 3.226310
4 3

The main focus of this paper is to determine analyti- 1.8864104 2 + 3.1662104


cally with the discrete element method (DEM) the per- + 0.0180 [Ns/m] (3)
formance of a PD under centrifugal loads. It is based on
the experimental work by Els [2] on a rotating cantilever The body accelerations on the particles inside the
beam with a PD at the tip (see Fig. 1). The tip of the damper due to gravitational and centrifugal forces are
beam were remotely activated with a cam. After activa-
tion the beam were free to vibrate. The vibration decay gx = 4 2 R 2 , gy = g0 = 9.81 m/s2 (4)
and rotation velocity were measured.
For the experiment four damper containers with iden- with R = 0.4 m the damper rotation radius.
tical masses, but with different cavity heights were used. The damper equation of motion is
The dampers were lled with three sets of different sizes
of steel ball bearings. The number of balls were selected ke ce Fy
yd = yd yd + + gy (5)
to give the same total tip mass for all the tests. me me me
with Fy the total sum of all the contact forces in the
y-direction between the particles and the container. E-
PD SIMULATION MODEL quation (5) was implemented by adjusting the damper
container velocity after each system integration step. The
The beam-damper system is simplied as a single de-
increment in damper container velocity is
gree of freedom (SDOF) mass-spring-damper system at-
tached to a cylindrical container lled with spherical par- yd = yd t (6)
ticles as shown Fig. 2. The cylinder dimensions are listed
in Tab. 1 and particle properties in Tab. 2. with t the current integration time step. The initial
The equivalent SDOF properties of the rotating beam displacement is yd0 = 6.2 mm at time t = 0.

Powders and Grains 2013


AIP Conf. Proc. 1542, 867-870 (2013); doi: 10.1063/1.4812069
2013 AIP Publishing LLC 978-0-7354-1166-1/$30.00

867





 
   





  

 

FIGURE 1. Schematic layout of test beam assembly (dimensions in mm)

gx gx gx BASIC DATA ANALYSIS


gy gy gy gy

The equation of motion for the free decay of an under-


me
me damped system, 0 < < 1, as given by Meirovitch [6]
me
yd  
y(t) = A e n t cos 1 2 nt (8)
0
yd0
ke ce ke ce ke ce The period T of the damped motion is
(1) (2) (3) (4) 2
T=  (9)
FIGURE 2. Setup and simulation procedure n 1 2

Consider the i-th peak of the vibration


 displacement at
TABLE 1. Container parameters time ti at the local maximum, cos( 1 2 nti ) = 1,
Container Height Radius then
hd [mm] rd [mm]
Ai = y(ti ) = A e n ti (10)
A 37.0 6.25
B 31.0 6.25 The log ratio between two successive peaks is the
C 25.3 6.25
D 21.8 6.25 logarithmic decrement

Ai y(ti ) 2
ln = ln = = (11)
Ai+1 y(ti + T ) 12
DEM CONTACT PARAMETERS
Equation (10) can be written in logarithmic format
The specic software used for the simulations is the com-
mercial DEM program PFC3D version 3.0 [3]. The con- ln y(ti ) = ln Ai nti = ln Ai ti (12)
tact models employed were the nonlinear Hertz normal
contact model for normal forces FNk 3/2 , viscoelastic with the linear slope of the exponential decay function
damping FNc 1/4 , shear friction and Mindlin non- in log space. The decrement in peak values is
linear shear stiffness FS (FNk +FNc )1/2 with the con-
ln y (ti ) ln y (ti + T ) = ln Ai /Ai+1 = T. (13)
tact overlap between particles.
For the viscoelastic model damping model in PFC3D It is clear from (11) and (13) that = T = n T ,
the critical damping ratio in terms of the coefcient of resulting in the undamped natural frequency
restitution for identical particles is [4]

1  n = 4 2 /T 2 2 . (14)
5/3 (7)

Through interpolation the points t = i can be found
where the signal in Fig. 3(a) crosses zero, y(i ) = 0.
The instantaneous period of the signal at time i is then
dened as

Ti = i+1 i1 , i = 1, 2, . . . (15)

868
TABLE 2. Hertz contact particle parameters

Particle Particle Modulus of Poisons Particle Number


radii rigiditya ratioa densityb of particles
r [mm] G [GPa] [] [kg/m3 ]
2 mm 1.0 73.1 0.305 7779 207
3 mm 1.5 73.1 0.305 7676 62
4 mm 2.0 73.1 0.305 7734 26
NOTE: Standard stainless steel ball bearings.
a Properties from Budynas and Nisbett [5].
b Calculated from measured mass

The instantaneous vibration frequency at time i is then 6


given by (a)

Tip displacement, y [mm]


 4

n (i ) = 2 fn (i ) = 4 2 /Ti2 2 . (16) 2
0
The result of the application of (16) on the example is
2
shown in Fig. 3(c).
The viscous damping coefcient for each zone can 4

then be calculated from 6


101

= /n (17) AI
(b) Peaks

Amplitudes, A [mm]
Exponential ts
with n the average frequency in zones I, II or III. I Intersections
The undamped natural frequency of the SDOF mass-
spring system for the PD is II
AII
ke ke /me AIII
III
n2 = = (18)
me + m p 1 + rm
100
18
with me the effective mass of the beam and damper (c)
container, m p the mass of the particles, rm =m p /me and
Frequency, fn [Hz]

17
the fraction of the particles contributing towards the
total mass of the system in zone I, II or III. If it is 16 fnI fnII fnIII
assumed that the stiffness of the spring and the effective
mass are constant, then for the frequency to change, 15 fn -instantaneous
the contribution of the particle towards the mass of the fn -average
vibrating system must be variable. In zone III there is no 14
particle damping and the assumption is made that the full
Time, t [s]
mass of particles is part of the total mass of the vibrating
system (III = 1). FIGURE 3. Example of data analysis for container A with
4 mm particles and rotation velocity of = 116 min1 .
n2III 1 + I r m 1  n2III 
= , I = (1+rm )1 (19)
n2I 1 + rm rm n2I
TABLE 3. DEM friction and damping parameters
The range of the effective mass factor I is 0 I 1.
When I 0 it means that the system is highly excited Particle Particle
particle wall
and that the particles contribute very little towards the
Parameter contactsa contactsb
system mass. If I =1 then the particles move as a solid
unit together with the damper container. Coef. of restitution 0.95 0.95
Damping factor 0.015 0.021
Shear damping factor s 0.015 0.021
Friction coefcient 0.3 0.5
DEM SIMULATION a Stainless steel on stainless steel contacts.
b Stainless steel on aluminium contacts.
The DEM model was calibrated by simulating all the ex-
perimental data points in [2] and performing identical

869
Container A with 2 mm balls Container A with 3 mm balls Container A with 4 mm balls
8

y-Peak acceleration, [ ]
AI 2n /g0 AI 2n /g0 AI 2n /g0
6

2 AIII 2n /g0 Exp


AIII 2n /g0 DEM
0
0.03
Damping coecient, [ ]

I Exp
II Exp
0.02 I DEM
II DEM

0.01

III III III


0
1
Eective mass factor, I [ ]

0.8

0.6

0.4

0.2 I Exp
I DEM
0
0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30
Radial acceleration, R [ ] Radial acceleration, R [ ] Radial acceleration, R [ ]

FIGURE 4. Test results for container A

data analysis on the output data. The damping and the CONCLUSIONS
friction coefcients were iterated until the results corre-
sponded with experimental results. It was found that the The DEM model of the rotating beam with a PD at the
damping had only a minor inuence and that the friction tip captured the performance of the PD very well over
was the main driver in the performance of the PD. a wide range of tests with different congurations and
The nal parameters are given in Tab. 3. The friction rotation velocities. It can be used to extrapolate the ex-
coefcient = 0.3 for the friction between the particles perimental results for more detailed investigations of the
and the coefcient of restitution of 95 % correspond well PD performance and for a more in-depth investigation
with the experimental values obtained by Wong et al. [7]. of phenomena occurring when PDs are under centrifugal
A comparison between the experimental and DEM loads.
results is shown in Fig. 4 with acceleration factors

= A 2 /g0 , R = 4 2 R 2 /g0 (20) REFERENCES


From the data analysis, it can be seen that there are 1. S. S. Simonian, Particle damping applications, in
two zones of damping, one with a high and one with Proceedings of the 45th AIAA/ASME/ASCE/AHS/ASC
a low damping factor. These damping zones depend on Structures, Structural Dynamics & Materials Conference,
the ratio between the peak vibration acceleration and the AIAA 2004-1906, Palm Springs, CA, USA, 2004.
centrifugal loading. Each zone has a limit in terms of the 2. D. N. J. Els, AIAA Journal 49, 22282238 (2011).
3. PFC3D 3.0, PFC3D 3.0: Users Guide, Itasca Consulting
centrifugal loading beyond which the PD cannot function Group (2003).
if the vibration amplitude is xed. In the high damping 4. D. N. J. Els, The Effectiveness of Particle Dampers under
zone, it was found that the excitation state of the particles Centrifugal Loads, Phd, Stellenbosch University (2009).
was high enough for the system vibration frequency to 5. R. G. Budynas, and J. K. Nisbett, Shigleys Mechanical
change. In the low damping zone, there is only limited Engineering Design, McGraw-Hill, Boston, 2008, 8 edn.
motion between the particles. 6. L. Meirovitch, Fundamentals of Vibrations, McGraw-Hill,
New York, 2001.
7. C. X. Wong, M. C. Daniel, and J. A. Rongong, Journal of
Sound and Vibration 319, 91118 (2009).

870

You might also like