Beruflich Dokumente
Kultur Dokumente
1735
ABSTRACT: In the contribution the first eigenmode of transversal vibrations of a clamped-clamped slender beam with an
electromagnetic actuator, fixed at some distance from the beam at its midpoint is analysed. The beam is excited by harmonic
force acting at its midpoint. The actuator consists of a pot type permanent magnet with an extra winding short-circuited by an
external resistance. The mathematical model of a magnetic force is derived in a simplified, linearised form. Its influence on the
beam transversal vibrations is analysed analytically and by numerical simulation. The extent of natural frequency variation and
of the additional damping is assessed and optimised. Based on the developed theory and experimental data, from a similar
experiment the advantages of this actuating principle for vibration control are exploited.
KEY WORDS: Beam theory; Vibration control; Electromagnetic actuator; Harmonic excitation; Coupled problem; Numerical
simulation.
1
INTRODUCTION
centre of a coil. As the yoke vibrates, the air gap width d(t)
varies with time. The change of d(t) is responsible for the
change of the air gap reluctance and consequently the primary
magnetic flux time variations, which induce alternating
voltage u(t) in the coil. The induced voltage forces a current
i(t) flowing in the closed electrical circuit. The direction of i(t)
is such that the generated magnetic field in the coil opposes
the primary field of the PM (Lenzs law). The alternating
dynamic magnetic force due to the induced current acts both
as a damping and a spring force. The current i(t) is dissipated
in the shunt resistance R (including also the coil resistance).
The combination of the static magnetic force of the PM and
that one of the induced current i(t) influences the structure
stiffness and thus the natural frequency. Moreover, the
induced current i(t) intensity influences the extent of damping.
Both effects are non-linear, because of the 1/d(t)2 character of
the magnetic force expression, which is dependent on the
square of the sum of the current i(t) and the PM
magnetisation.
1736
When the centre of the beam is oscillating, the air gap width
d(t) between the yoke and the core is periodically varying.
According to Faradays law of induction, in the coil of
winding cross-section SW and number of turns N, alternating
voltage u(t) is generated, due to time variable magnetic
induction (magnetic field density) B(t) in the air gap:
u (t ) = -
d
dB (t )
= - NS W
,
dt
dt
(1)
FF
FF (t )
,
=
lg
R
l
lPM
1
M
C
+
+
S C m rFe m 0 m 0 m rPM m 0
where the contour integral over the closed curve C follows the
middle magnetic flux line. Hence:
FF (t ) = M + Ni (t ) = H PM lPM + H C lC + H g lg (t ) ,
B ( t ) - Br
B( t )
B( t )
lPM +
lC +
lg (t ) = M + Ni(t ) .
m 0 m rPM
m 0 m rC
m0
(2)
(3)
B0 = m 0 ( H PM + M ) .
(5)
Hence:
M = m rPM ( - H c ' ) =
Br
.
m0
(6)
(10)
(4)
where:
(9)
where lC, lg, and lPM are in turn the length of the flux line in
the core/armature of relative permeability rFe, of the air gap
and of the PM, with material relative permeability rPM,
respectively. F(t) is the magnetomotive force in the magnetic
circuit, which has a steady state component, due to the PM,
and a time varying component, due to the yoke vibrations.
Assuming that the tangential magnetic induction component
at the interface between the yoke surface and the air gap is
negligible (no fringing effects) and the normal component is
homogenously distributed over the whole core surface of area
SC, then:
(t ) = BN (t )S C = B (t )S C .
(8)
and
m 0 [M + Ni (t )]
m M [1 + i (t ) N M ]
= 0
, (11a)
2d 0 [(1 + d ) + e (t )] 2d 0 [(1 + d ) + e (t )]
FF (t ) = M + Ni (t ) ,
(11b)
where:
d0 is the equilibrium distance between yoke and the PM
w(t) is the dynamic displacement of the yoke from the
equilibrium position d0: w(t) = d(t) d0 in the positive
(upwards) direction (Figure 3)
(t) is the relative dynamic displacement: (t) = w(t)/d0,
positive in the upward direction, further referred to as the
relative air gap width (a non-dimensional quantity).
It will be assumed that |w(t)| << d0, i.e. | (t)| << 1.
Eq. (11a) represents the general form of magnetic induction
in the air gap in the time domain. If the shunt circuit is not
connected, no current is flowing and the last term in
numerator is zero. There is no oscillating magnetic field and
so the magnetomotive force of Eq. (11b) is constant.
According to Lenzs law, the current flowing through the
closed electrical circuit with resistance R (in general external
impedance Zext) has such direction, that its force action is
1737
NS W dB (t )
.
R
dt
(12)
Hence:
F F = M + Ni ( t ) = M -
N 2 S W dB (t )
.
R
dt
(13)
1 l3
,
z m = mg N
192 E Y I b
(14)
B0 =
(15a)
4.2
(15b)
THE
MAGNETIC
FIELD
IN
The time dependence of the magnetic field in the air gap plays
a decisive role in the electro-magnetic system description.
4.1
(17)
m0 M
= const ( d 0 ) ,
2d 0 (1 + d )
(18)
1 2
B0 S C .
m0
N 2 S W dB (t )
M
R
dt
mS
FM (t ) = 0 2C
.
4d 0
[(1 + d ) + e (t )]2
(16)
Static analysis
Dynamic analysis
m 0 N 2 S w d 2 B(t )
dB (t ) B de B di B
=
+
=
e& , (19)
dt
e dt i dt e
2d 0 R(1 + d ) dt 2
where in the second term e in the denominator was neglected
because |e(t)| << 1. From Eq. (11a) follows for B/e:
B
m M [1 + i (t ) N M ]
m0 M
= - 0
2
2 .
e
(
)
2d 0 1 + d
2d 0 (1 + d )
(20)
m 0 N 2 S W 1 dB(t )
B0
=e& .
+
(1 + d )
dt
2d 0 (1 + d ) R
(21)
1738
Let us assume, that the relative air gap width e(t) changes
harmonically with the excitation angular frequency wa, i.e.
e(t) = e0sin( wat+je) = e0Re{exp[j( wat+je)]}, where je is the
phase angle in respect to the excitation force and j = - 1 .
Then the Eq. (21) can be transformed into frequency domain,
using the Fourier transform and neglecting all transient
phenomena (electro-magnetic and mechanic):
B0 ~ jw
~ jw L
B e jw a 0 + 1 = Ee ,
(1 + d )
R
a
or
~
B=-
B0
~
E.
jw a L0
(1 + d )
+ 1
R
(22)
(23)
~
~
cS kS ~ jw t Fe jw t
FM
2
w
+
j
w
+
E
e
=
,
a
a
m m
md 0 md 0
(26)
~
~
Fe jw t FMO
~
,
+ 2 jx S ]Ee jw t =
ks d 0
ks d 0
(27)
and
[1 -
where:
(24)
(25)
m 0 S C FF
m 0 S C FF
2
2 2
3 e (t ) .
4d 0 (1 + d )
2 d 0 (1 + d )
2
FMO (e ) =
(28)
Next, let now convert the magnetic force into the frequency
domain. Since a square of the magnetomotive force FF
according to the numerator of Eq. (15b) has to be performed,
it is a good to assume multi-component magnetic force, while
each component would oscillate (i.e. rotate in the complex
plain) with a specific multiple of the excitation angular
frequency wa and own phase in respect to acting excitation
~
force Fe jw t . Hence:
~2
~2 ~
Q
mS
FF
m 0 S C FF Ee jw t
~
~
FMO = FMn e jnw t = 0 2C
,
(29)
2
2
3
4d 0 (1 + d )
2d 0 (1 + d )
n =0
a
1739
(F~ )
= M
( ),
2j M w a N 2 S W ~ w a2 N 4 S W2 ~
B B
R
R2
F
k w 0 L0 ~
2
(
)
1
k
2
j
x
E
=
. (35)
+
+
S 2 R + jw L
kS d 0
a 0
(30)
jw a m 0 S C S W MN 2 B0
m0 S C M 2 ~
~
E E . (31)
2
3
2d 0 (1 + d ) ( jw a L0 + R )
2d 02 (1 + d )3
x E (w a , R) =
k
w 0 L0
.
2 R + jw a L0
(36)
(32)
~
FM1 =
jw a L0
~
~
k M* 1 E - k M* 1 E .
jw a L0 + R
The factor k*M1 has dimension Newton, thus the ratio k*M1/d0
has the same dimension as the beam stiffness kS, i.e. N/m. It is
possible to define new quantity magnetic field stiffness:
kM =
m0 S C M 2
k
, and the ratio k = M .
2d 03 (1 + d ) 3
kS
(34)
k 1 - j ~
F
2
E
=
, (37d)
1 - k - + 2 j xS +
E
2 1 + 2
kSd 0
1740
S
~
k
k
k
EN
1
2
2,
(38)
k R
k (R R* - j )
.
+ j =
2
2 R*
2 (R R* ) + 2
-1
x E ( , R) =
(39)
-1
B0
R R
~
+ j E .
(1 + d ) R* R*
(40)
-1
~
~
~
R
FM = d 0 k M j
+ j E - d 0 k M E .
R*
(41)
DISCUSION
10
1741
CONCLUSION