Modal Analysis of Beams - An Experiment Symposium On Dynamic Problems of Mechanics
Modal Analysis of Beams - An Experiment Symposium On Dynamic Problems of Mechanics
osciloscopes, etc... Nevertheless, no doubt the students become more motivated and so
learn much more when they can actually see a real structural vibration problem.
Bearing this in mind, the authors prepared an experiment aiming at teaching modal
analysis of beams for students who have attended an introductory vibration course. This
paper aims to describe it in details the basic theoretical and experimental procedure adopted
in order to awake the students curiosity to the important problem of modal analysis.
2. MODAL ANALYSIS OF A CANTILEVER
An experimental set up was deviced to study the vibration of beams. It was decided to
use a very simple beam arrangement which could suit well experimental facilities restraints.
Accordingly, a cantilever was chosen since it is very easy to fix its clamped support to a
shaker. The analysis is also somewhat simple, as described next.
Consider a cantilever beam in line with the axis x with length l, breadth b, height h,
made of a material with density and elastic modulus E. The moment of inertia in relation
to the axis z is I(x) and the cross-section area is A(x).
Figure 1 presents the beam under analysis with a free body digramm where equilibrium
of transverse force gives
V ( x, t )
2 w( x, t )
dx V ( x, t ) + f ( x, t )dx = A( x)dx
V ( x, t ) +
x
t 2
(1)
dx M ( x, t ) + V ( x, t ) +
dx dx + [ f ( x, t )dx ] = 0
M ( x, t ) +
x
x
2
(2)
such that transverse force, V (x,t), and bending moment, M(x,t), are related by
f(x,t)
f(x,t)
M ( x, t ) +
M ( x, t )
y
A(x)=b(x)h(x)
x
w( x, t ) dx
V ( x, t )
V ( x, t ) +
dx
x
x
x+dx
M ( x, t )
dx
x
V ( x, t )
dx
x
V ( x, t ) =
M ( x, t )
x
(3)
2 w( x, t )
(4)
x 2
From equations (1) to (4) and focusing on a free vibration problem, i.e. f(x,t) = 0, with A(x)
e EI(x) constants, it follows that
2 w( x, t )
t
+ c2
4 w( x, t )
x
= 0,
c=
EI
A
(5)
X ( 4) ( x)
T(t )
=
=2
X ( x)
T (t )
(6)
T(t ) + 2T (t ) = 0
(7)
whose solution is
T (t ) = A sin(t ) + B cos(t )
(8)
The constants A and B are obtained from the initial conditions of the problem.
Equation (6) also gives the spatial behaviour of a beam when solving
X ( 4) ( x) 4 X ( x) = 0,
4 =
2
c2
A 2
EI
(9)
whose solution is
X ( x) = a1sin ( x) + a2 cos( x) + a3 sinh( x) + a4 cosh( x)
(10)
where and three of the constants ai should be determined from the boundary conditions,
which are
(11)
(12)
(13)
which gives,
1 l = 1.87510407
2 l = 4.69409113
(14)
3 l = 7.85475744
The actual cantilever to be tested as described later has the following physical
parameters
l = 0.5 m
E = 210 GPa
I = 2.258 x 10-9 m4
A = 3.002 x 10-4 m2
= 7860 kg/m3
which, when using equation (9), (13) and (14), gives the following natural frequencies
1 = 199.4 rad/s .: f1 = 31.7 Hz
2 = 1249.4 rad/s .: f2 = 198.8 Hz
3 = 3498 rad/s .: f3 = 557 Hz
As for the vibration modes, they become
n
n
sin( nl ) + sinh( nl )
n = 1,2,3...
(15)
whose necessary parameters are all known except the constant a4 which is an arbitrary
constant. These modes of vibration are depicted in Figure 2.
To uncouple the various recorded signals, the equation of motion for the beam can be
written as
M + Cx + Kx = fe j dr t
x
(16)
where M, C and K are the mass, damping and stifness matrix. The solution of this equation
for an excitation f is x(t)=ue jdrt
2
( K dr M + j dr C )u = f
(17)
where u is the displacement vector whose components are associated with the motion in
each degree of freedom, given by
2
u = ( K dr M + j dr C ) 1 f
(18)
u=(dr)f
(19)
or
2
i-
dr )=[Sdiag[
T -1
dr j]S ]
1
1
= S T diag 2
S
2
i dr + i i dr j
(20)
2
2
i =1 ( i dr ) + ( 2 i i i ) j
ui uiT
( dr ) =
(21)
sr ( dr ) =
2
i =1 i
(22)
Now, by assuming that continous system exhibit modes of vibration well separated
apart, the sum in equation (22) will be dominated by term associated with the natural
frequency, so that, for dr=i , equation (22) becomes
| sr ( dr ) |=
| ui uiT |sr
| ( i2 i2 ) 2 i i i j |
| ui uiT |sr
(23)
2
2 i r
or
| ui uiT |sr =| 2 i i2 || H sr ( i ) |
(24)
where |Hsr(i)|=|sr(i)| is the measured magnitude of the transfer function between the
points s and r for the natural frequency i.
Equation (24) gives only the magnitude but not the signal such that it is not possible to
determine the direction of the movement. However, the phase of H(i) can be used to
determine the signal of |uiuiT|sr.
The beam shown in Figure 3 has at its tip an accelerometer connected to a charge amplifier,
whose analogic signal is fed to a data acquisition board.
beam
By processing the data generated by the beam response to the various hammer
excitations, the spectrum in Figure 4 is obtained. It is evident in the figure the peaks in the
amplitude related to the natural frequencies. This is further corroborated by plotting the real
and imaginary part of this spectrum, so to obtain the Nyquist circle in Figure 4b. In Figure
4c, the phase of the transfer function is plotted, allowing the calculation of the motion
direction.
The natural frequencies for the theoretical model were 31.7 Hz, 198.8 Hz and 556.7 Hz,
whereas the ones obtained by the technique here described are 28.5 Hz, 184 Hz e 507 Hz.
Figura 4. FFT, Nyquist circle and phase of the transfer function H().
Figura 5. The first three modes of vibration for the cantilever beam. Green- analytical, redinterpolation of the experimental data, dotted in blue.
The natural frequencies were obtained within an error of no more than 10%. This is
considered to be a reasonable result, when bearing in mind factors like the beam having a
not perfectly rectangular cross-section and, mainly, that it is difficult to assure a fully
clamped condition to the beam.
It is evident in Figure 5 that the first mode of vibration was not accurately described
although higher modes were. Despite this shortcoming, the technique used here gives to the
students a much more attractive way to study vibration and moda analysis. The technique
used is relatively easy to implement and the experimental set up allows the determination of
the natural frequencies also by merely tunning the frequency with a signal generator.
The possibility of showing to the students other techniques and to present in a clear way
basic concepts like natural frequency and mode of vibration make the experiment above
quite useful and attractive.
6. REFERENCES
All the theoretical information given here was extracted from Inman, D. J., 2000, Engineering
vibration, 2a edio, Prentice Hall.
Another important reference in modal analysis is the book by D.J. Ewins, Modal Testing:
Theory and Practice, Research Studies Press, 1984.
We also suggest the book Dimaragonas, A., 1996. Vibration for Engineers, 2a edio,
Prentice Hall, for a general introduction to vibration.