You are on page 1of 7

Use of Comsol Multiphysics to Characterize the Electromagnetic Vibration Absorber

N. Zhou, Y. He, and K. Liu


Department of Mechanical Engineering, Lakehead University
Thunder Bay, ON P7B 5E1

Abstract: This paper reports a multiphysics analysis for an spring kc1 created by the absorber beam, constant spring
electromagnetic vibration absorber (EMVA). The EMVA uses kc 2 formed by the interaction between the PM and the core
an electromagnet, permanent magnet, and aluminum beam to
form a variable spring. By changing the current of the of the electromagnet, and variable spring kv generated by
electromagnet, the stiffness of the EMVA can be adjusted the interaction between the PM and the electromagnet. The
instantaneously. Characterization of the EMVA intends to value kc1 is adjusted by the tension and length of the beam.
study the factors that affect the stiffness of the EMVA. The The value kc 2 is determined by the gap between the PM
problem involves interaction of structural mechanics and
and the core of the electromagnet and the value kv varies
magnetism. Comsol, a commercial software package, is used
to simulate the system. A simplified 2–dimensional model is with the current of the electromagnet.
built using Structural Mechanics module, AC/DC module,
and the Moving Mesh (ALE) application mode. Using the
model, various studies are conducted. The results are
compared with those from a simplified analytical method.

Keywords: Tunable vibration absorber, variable stiffness,


coupling of structural mechanics and magnetics.

1. INTRODUCTION

Tunable vibration absorbers belong to the family of


semi-active control devices which possess the advantages of
both passive and active control devices. Passive vibration
absorbers are an effective means to suppress steady-state
vibration of a primary structure excited by a harmonic Figure 1: The EMVA
excitation. A narrow operation bandwidth is a main
limitation of passive vibration absorbers. Tunable vibration
absorbers have been used to achieve on-line tuning such that
the operation bandwidth can be increased. In general,
adaptability of such devices is achieved by varying either
stiffness [1-4] or damping [5-8]. The electromagnetic
vibration absorber (EMVA) proposed in [9] belongs to the
family of variable stiffness devices. The notable features that
distinguish the EMVA from the previously proposed similar
devices are non-contact, non-mechanical transmission, and
instantaneous adjustment.
Figure 1 shows a schematic drawing of a modified
EMVA. The modification allows the present device to be
adjusted both manually and electrically such that it is suitable Figure 2: Three springs involved in the EMVA
for non-collocated control [10]. The system consists of a
C-shaped electromagnet and a clamped-clamped beam whose The characterization of the EMVA intends to study the
center is embedded with a permanent magnet (PM). The factors that affect its stiffness. An approximate analytical
PM-beam assembly is placed between the poles of the method was adopted in [9]. The study conducted in [10]
electromagnet. Two beam holders and tensioners are new explored how a commercial finite-element based package,
additions to the EMVA. They are used to achieve a large Comsol Multiphysics, can be utilized to characterize such a
adjustment in the stiffness. As shown in Figure 2, the system. Part of the study results was reported in [11]. This
stiffness of the EMVA is composed of three parts: constant paper presents some new results of using Comsol
Multiphysics in the study of the EMVA.

2. MODEL

Comsol Multiphysics is a finite-element based program


for simulating multiphysics and single-physics applications.
Two physics are involved in the EMVA, namely, structural
mechanics and electromagnetism. These two physics are
coupled. Thus Comsol is suitable for tackling such a
problem. Figure 3 shows a simplified 2-dimensional (2-D)
model, built using Structural Mechanics module, AC/DC
module, and the Moving Mesh (ALE) application mode of
Comsol. A 2-D model is used in order to reduce a
computational burden. The electromagnet is replaced by a
pair of PMs referred to as upper PM (R2) and lower PM
(R3), respectively. The upper and lower PMs are fixed in
space. A PM (R4) referred to as middle PM is placed in the
middle of the upper and lower PMs. Two beams (R5, R6) are
attached to the left and right sides of the middle PM,
respectively. The other end of each beam is fixed. The beams
and middle PM assembly will be deformed when it is
subjected to an external force. The beams are made of Figure 3: 2-D coupling model
aluminum with Young’s modulus of 70 GPa. The beam
length is 0.1 m. Two different beam thicknesses, 0.0015 m 3. RESULTS
(thin beam) and 0.002 m (thick beam), are considered. The
size of the PMs is 0.048 × 0.04 m (width × height). The Note that the net magnetic force acting on the middle PM
depth of the PMs and beams is 0.022 m. The size of the behaves in the following way. When the middle PM is
overall free space is 0.248 × 0.363 m (width × height). The located exactly in the middle of the gap between the upper
magnetization of the upper and lower PMs is assumed to and lower PMs, the net magnetic force is zero. When the
have the following constant values: -395,800 A/m, -268,000 polarity of the upper and lower PMs is negative and the
A/m, -140,000 A/m, 140,000 A/m, 268,000 A/m and 395,800 middle PM moves up, the net magnetic force acts downward,
A/m to simulate a current change in the electromagnet. The which corresponds to a positive magnetic stiffness. When the
magnetization of the middle PM is fixed to be 395,800 A/m. polarity of the upper and lower PMs is positive and the
When the magnetization of the upper and lower PMs changes middle PM moves up, the net magnetic force acts upward,
from negative to positive, the polarity of the upper and lower which corresponds to a negative magnetic stiffness.
PMs switches. To find the stiffness of this setup, an external force is
The model addresses a 2-way coupling between the applied in the center (PT1) of the middle PM in the vertical
structural deformation and the magnetic field. The model is direction ( y - direction). Then a simulation is conducted and
built using three application modes: 1. the Plane Stress mode
from the Structural Mechanics Module; 2. the Perpendicular the displacement of the middle PM can be obtained. The
Induction Currents mode, Vector Potential in Quasi-Statics relationship between the applied forces and the
Magnetic from AC/DC Module; and 3. The Moving Mesh displacements of the middle PM can be obtained. From these
(ALE) mode from Multiphysics Module. Comsol solves this relationships, the stiffness can be found.
coupling problem in the following way. First, the Figure 4 shows an arrow plot of the magnetic flux
Perpendicular Induction Current application mode computes density where the magnetization of all PMs is 395,800 A/m,
the magnetic force acting on the middle PM. Then the Plane which corresponds to a negative current case in the
Stress application mode computes the deformation of the electromagnet. Note that in this case, the directions of the
beams under the magnetic force and any external applied magnetic flux density of the three PM align. Figure 5 shows
force. An iteration solution procedure must be used because an arrow plot of the magnetic flux density when the
the beam deformation will change the position of the middle magnetization of the upper and lower PMs is equal to
PM so that the interacting magnetic force alters -395,800 A/m and the magnetization of the middle PM is
correspondingly. The ALE application mode deals with the 395,800 A/m, which corresponds a positive current case in
deformed mesh. The solver is set as Automatic which will the electromagnet. Note that in this case, the directions of the
detect any nonlinearity that may exist in the model. magnetic flux density of the middle PM are opposite to those
of the upper and lower PMs.
50

40

30

20

Applied force (N)


10

-10

M = -395800 A/m
-20
M = -268000 A/m
-30 M = -140000 A/m
M = 140000 A/m
-40 M = 268000 A/m
M = 395800 A/m
-50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3
x 10
Figure 6: The applied force vs. the displacement of the middle PM for the
thin beam case.

Figure 4: Arrow plot of the magnetic density for the magnetization of all 100
PMs is 395,800 A/m and the applied force is 22 N 90
80
70
60
50
40
30
Applied force (N)

20
10
0
-10
-20
-30
M = -395800 A/m
-40
-50 M = -268000 A/m
-60 M = -140000 A/m
-70 M = 140000 A/m
-80 M = 268000 A/m
-90 M = 395800 A/m
-100
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3
x 10
Figure7: The applied force vs. the displacement of the middle PM for the
thick beam case.

Figure 8 illustrates the relationship between the applied


forces and the displacements of the middle PM by varying
the gap distances from 2.5 mm, 5.0 mm to 7.5 mm while the
Figure 5: Arrow plot of the magnetic density for the magnetization of upper magnetization of the middle PM remains 395,800 A/m, the
and lower PMs is -395,800 A/m, the magnetization of the middle PM is
395,800 A/m and the applied force is 22 N
magnetization of the upper and lower PMs remains -395,800
A/m.
Figures 6 and 7 present the relationship between the
applied forces and the displacements of the middle PM by
varying the magnetization of the upper and lower PMs for
the thin beam case and the thick beam case, respectively.
50 4
x 10
5
40 M = -395800 A/m
4.5 M = -268000 A/m
30
M = -140000 A/m
20 4 M = 140000 A/m
M = 268000 A/m
Applied force (N)

10 3.5 M = 395800 A/m

Stiffness (N/m)
0 3

-10 2.5

-20
2
-30
Distance = 2.5mm 1.5
-40 Distance = 5.0mm
Distance = 7.5mm 1
-50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
0.5
Displacement of the middle PM (m) -3
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
x 10
Figure 8: The applied force vs. the displacement of the middle PM for the Displacement of the middle PM (m) -3
x 10
thin beam with the different gap distances. Figure 9: Stiffness vs. displacement for the thin beam case.

From the preceding figures, the following observations x 10


4

9
can be drawn. First, the relationship between the applied M = -395800 A/m
forces and the displacements of the middle PM is not linear. M = -268000 A/m
8
Second, such a nonlinear relationship indicates that the setup M = -140000 A/m
possesses the characteristics of a hardening spring, i.e., the M = 140000 A/m
M = 268000 A/m
stiffness increases with the increase of the spring deflection. 7
M = 395800 A/m
Third, an increase of the magnetization of the upper and
Stiffness (N/m)

lower PMs results in an increase of the displacement of the 6


middle PM, i.e., a decrease of the stiffness. This means that
when the polarity of the upper and lower PMs is the same as 5
that of the middle PM, the magnetic stiffness is negative.
Fourth, a decrease of the magnetization magnitude will 4
increase the stiffness. This means that when the polarity of
the upper and lower PMs is opposite to that of the middle 3
PM, the magnetic stiffness is positive. Fifth, an increase of
the distance in the gap weakens the effect of the magnetic
2
stiffness. -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3
x 10
A localized stiffness ki can be estimated by Figure10: Stiffness vs. displacement for the thick beam case.

4
x 10
F − Fi −1 5
ki = i , i = 2,3,... (1) Distance = 2.5mm
yi − yi −1 Distance = 5.0mm
4.5
Distance = 7.5mm

where Fi is the applied force that causes the middle PM to 4

move to yi . Using equation (1), the localized stiffness


3.5
Stiffness (N/m)

values corresponding to Figures 6 and 7 are shown in Figures


9 and 10, respectively. The localized stiffness values for the 3
different distances in the gap are given in Figure 11. It can be
seen that the stiffness values vary with the middle PM 2.5
position. The stiffness difference for a given displacement
represents the range of stiffness variation when the 2
magnetization of the upper and lower PMs varies from the
maximum to the minimum. 1.5

1
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3
x 10
Figure 11: Stiffness vs. displacement for the different gap distances.
50
4. COMPARISON WITH THE ANALYTIC RESULTS
40
It is worthwhile to compare the Comsol results with those
30
from a simplified analysis. The force acting on the middle
PM can be found by [12] 20

Applied force (N)


10
Fm = M m ( ∫ B1ds − ∫ B2 ds ) (2)
S1 S2
0

where M m is the magnetization of the middle PM, B1 is -10

M = -395800
the magnetic density created by the upper and lower PMs at -20
M = -268000
the upper face S1 of the middle PM and B2 is the -30 M = -140000
magnetic density created by the upper and lower PMs at the M = 140000
-40 M = 268000
lower face S 2 of the middle PM. To find the magnetic M = 395800
density between the upper and lower PMs, firstly, the -50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
magnetic density of one of the PMs on the central axis at Displacement of the middle PM (m) -3
x 10
distance y from the surface of this PM can be found by Figure 12: Analytic relationships between the applied force and displacement
for different magnetizations of PMs
µ0 M lw
B= [sin −1 Figures 13 and 14 shows the applied forces computed
π (l + 4 y )( w 2 + 4 y 2 )
2 2
using three different ways when the magnetization of middle
lw (3) PM is 395,800 A/m and the magnetizations of the upper and
− sin −1
[l 2
][
+ 4( y + h) 2 w 2 + 4( y + h) 2 ] lower PMs are -268,000 A/m and 268,000 A/m, respectively.
The method 1 utilizes the Comsol coupling model directly. In
method 2, the magnetic force is found using an
where µ0 = 4π × 10−7 H/m, M is the magnetization of the upper-middle-lower PM model built by AC/DC module only.
The beam restoring force is computed using a
lower/upper magnet, l = 0.048 m, w = 0.022 m,
beam-middle-PM model built by using Structural Mechanics
h = 0.04 m are the width, depth, and thickness of the PM,
module only. Then the total force is sum of the magnetic
respectively. Then, the total magnetic density can be obtained force and the beam restoring force. In method 3, equation (5)
by superposition. The magnetic density is considered to be is used. It can be seen that the third method fails to predict
uniform over the surfaces S1 and S 2 . Each of the beams the hardening effect of the system.
attached to the middle PM can be considered as a beam fixed
50
at one end and free to translate at the other end. The total
stiffness of the beam is given by 40

30
24 EI
kb = (4)
L3 20
Applied force (N)

10
where I = 6.1875 × 10−12 m4 for the thin beam, E = 70
0
GPa and L = 0.1 m. The applied force Fa needed to force
the middle PM to displace to y can be calculated by -10

-20
Fa = k b y − Fm (5)
-30
Method 1
Figure 12 shows the analytic results corresponding to -40 Method 2
Method 3
those in Figure 6. It is noted that the analytical results are -50
quite different from the Comsol results. First the analytical -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3
results are linear as expected. Second, the discrepancies x 10
Figure 13: Comparison of the applied forces computed by three methods.
between the Comsol results and the analytical results become The magnetization of upper and lower PMs is -268000 A/m.
more significant when M > 0 . In particular, when
M = 395,800 A/m, the stiffness becomes negative. This
indicates the system at the equilibrium point is not stable.
50 20

40
15
30
10
20

Magnetic force (N)


5
Applied force (N)

10

0 0

-10
-5
-20
-10
-30
Method 1 Method 1
-15
-40 Method 2 Method 2
Method 3 Method 3
-50 -20
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Displacement of the middle PM (m) -3 Displacement of the middle PM (m) -3
x 10 x 10
Figure 14: Comparison of the applied forces computed by three methods. Figure 16: Comparison of the magnetic forces computed by three methods.
The magnetization of upper and lower PMs is 268000 A/m In the method 1, the magnetization of upper and lower PMs is 268000 A/m

Figures 15 and 16 compare the magnetic forces obtained Figure 17 compares the restoring forces of the beam
by three different ways. In method 1, the magnetic force is attained by two different ways. The solid line is from a
obtained from the Comsol coupling model directly. In beam-middle-PM model built by using Structural Mechanics
method 2, the magnetic force is from the upper-middle-lower module only. The dash line is obtained using equation (5)
PM model built by AC/DC module only. In method 3, the where Fm = 0 . It can be observed that equation (4) can
magnetic force is computed using equation (2). It can be calculate the stiffness of the beam accurately only when the
observed that method 3 or the analytical method deformation of the beams is very small. Therefore, the
overestimates the magnetic force. It is also interesting to see analytic result will not precisely describe the stiffness of the
that the magnetic forces from both methods 1 and 2 are system when the deflection of the beams reaches a certain
relatively linear, which indicates that the hardening effect is point.
mainly due to the beam nonlinearity.
50
20
Method 1 40
15 Method 2
Method 3 30

10 20
Restoring force (N)

10
Magnetic force (N)

0
0
-10
-5
-20

-10 -30

-40 Structural Mechanics module


-15
Analytic Result
-50
-20 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Displacement of the middle PM (m) -3
x 10
Displacement of the middle PM (m) -3
x 10 Figure 17: Comparison of the restoring force of the beam computed by
Figure 15: Comparison of the magnetic forces computed by three methods. analytic method and structural mechanics module
In the method 1, the magnetization of upper and lower PMs is -268000 A/m
The reason that equation (2) overestimates the magnetic
force is that the magnetic density along the center line is used
everywhere on the surfaces S1 and S 2 . To illustrate this,
Figures 18 and 19 compare the magnetic flux density
distributions that are obtained from an upper-lower PM
model built by AC/DC module only and those based on
equation (3). It can be observed that the magnetic density
from the Comsol model in the x - direction varies
significantly. Even along the center line, i.e., x = 0 , the magnetization will result in change the stiffness. An increase
analytical method underestimates the magnetic density in magnetization will decrease the stiffness. The Comsol
around the center of the gap, i.e., y = 0 while it results follow the general trend of those predicted by the
overestimates the magnetic density for the positions far away analytical method when the deformation of the beams is
from the center. The latter explains why the analytical small and the magnetic stiffness is positive. When the
method overestimates the magnetic force. magnetic stiffness is negative, the results from the analytical
method differ significantly from those from the Comsol
simulation. The study has shown that Comsol offers a great
Comsol Result
0.15
potential in solving mechatronics systems such as the
Analytic Result
EMVA.
0.1
Magnetic Flux Density (T)

0.05 ACKNOWLEDGMENT

0 The financial support provided by the National Science


and Engineering Research Council of Canada is greatly
-0.05
appreciated. The technical assistant provided by Comsol
-0.1
engineers during the course of this study is acknowledged.

-0.15
REFERENCES
-0.2 0.01 0.02
0.03 0.02 0.01 0 -0.010
[1] M. A. Franchek, M. W. Ryan and R. J. Bernhard, “Adaptive Passive
-0.01 -0.02 -0.03 -0.03-0.02
Vibration Control,” Journal of Sound and Vibration, 189, 565-585
y (m) x (m)
(1995).
Figure 18: Comparison of the magnetic flux density when the magnetization [2] K. Nagaya, A. Kurusu, S. Ikai, and Y. Shitani, “Vibration Control of a
of upper and lower PMs is -268000 A/m Structure by Using a Tunable Absorber and an Optimal Vibration
Absorber Under Auto-tuning Control,” Journal of Sound and Vibration,
228, 773-792 (1999).
Comsol Result [3] N. Varadarajan and S. Nagarajaiah, “Response Control of Building with
0.25 Analytic Result Variable Stiffness Tuned Mass Damper Using Empirical Mode
Decomposition and Hilbert Transform Algorithm,” 16th ASCE
0.2 Engineering Mechanics Conference, Seattle, July 2003.
[4] K. Liu, L. Liao, and J. Liu, “Comparison of Two Auto-tuning Methods
Magnetic Flux Density (T)

0.15 for a Variable Stiffness Vibration Absorber,” The Transactions of


Canadian Society for Mechanical Engineering, 29, 81-96 (2005).
0.1 [5] W. N. Pattern, R. L. Sack, and Q. He, “Controlled Semiactive Hydraulic
Vibration Absorber for Bridges,” Journal of Structural Engineering, 122,
0.05 87-192 (1996).
[6] M. D. Symans and M. C. Constantinou, “Seismic Testing of a Building
0
Structure with a Semi-active Fluid Damper Control System,” Earthquake
Engineering and Structural Dynamics, 26,759-777 (1997).
[7] P. Dupont, P. Kasturi, and A. Stokes, “Semi-active Control of Friction
-0.05
Dampers,” Journal of Sound and Vibration, 202, 203-218 (1997).
[8] L. Y. Lu, “Semi-active Modal Control Seismic Structures with Variable
-0.1 0.02
0.03 0.02 0.01 Friction Dampers,” Engineering Structures, 26, 434-454 (2004).
0.01 0
0 -0.01 -0.01 [9] J. Liu and K. Liu, “A Tunable Electromagnetic Vibration Absorber:
-0.02 -0.03 -0.02
-0.03 Characterization and Application,” Journal of Sound and Vibration, 295,
y (m) x (m) 708-724 (2006).
Figure 19: Comparison of the magnetic flux density when the magnetization [10] Y. He, “Non-collocated Control of Vibration of a 3-DOF Structure
of upper and lower PMs is 268000 A/m Using the Electromagnetic Vibration Absorber,” Msc. Thesis, Lakehead
University (2007).
[11] Y. He and K. Liu, “Multiphysics Analysis for the Electromagnetic
Vibration Absorber”, 2007 Comsol User Conference, Oct. 2007, Boston,
5. CONCLUSIONS USA.
[12] D. J. Craik, “Magnetism: Principles and Applications,” Wiley & Sons,
The Comsol multiphysics functionalities have been New York (1995).
employed to study the stiffness of the electromagnetic
vibration absorber. A simplified 2-D model has been used in
the study. The Comsol simulation results have been
compared with those from a simplified analysis. From the
study, several conclusions can be drawn. The system
possesses a nonlinear stiffness that increases with the
displacement of the middle PM. Such a hardening property is
mainly due to the structural stiffness of the beams. Change of

You might also like