You are on page 1of 8

Composite Structures 46 (1999) 163170

Modeling of vibration damping in composite structures


Ahmed Maher *, Fawkia Ramadan, Mohamed Ferra
Production and Mechanical Design Department, Faculty of Engineering, Minuya University, Shebin El-Kom, Egypt

Abstract
An improved dynamical model for vibration damping in composite structures is introduced to investigate the stacking sequence
and the degree of anisotropy as a function of the vibration modes. Extensive investigation has been carried out from the tting of
modal measurements with lowest residual errors to establish quasi-uniform mass damping models in terms of normal coordinates
system. The analysis of the obtained results proves not only the eciency of the proposed model but also its applicability in any wide
range of frequency spectrum of composites. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Composite; Uniform mass damping; Modelling; Code no.; Model parameters; Eigen-value problem

1. Introduction
Composite damping or energy dissipation property of
vibrating composite structures, refers to a complex
physical dynamic nature that is amenable to rheological
modal analysis. In a broad class of composite structures,
the distinguishing characteristic of the damping mechanism is its strong dependence on the eigenfrequencies
such that it exhibits little damping at high frequency
level [1,2].
In contrast to the dynamic nature of isotropic materials, a further complication arises in composite domain
due to the mutual eects of various parameters, such as
code number, degree of isotropism (volume fraction),
boundary conditions as well as the vibrating mode
number on the damping and stiness distributions. As an
example, the decreasing of volume fraction of the ber
enhances energy dissipation by increasing the loss associated with matrix composite [24]. It might be expected
that the natural frequencies of vibrating composite
structures and in sequence the damping capacity, can be
altered by changing the layer's orientations and stacking
sequence [5,6], so that the damping nature as a function
of frequencies of composites should be further studied.
At the present time, it is still dicult to determine
accurately the modal characteristics of composite
structures (particularly damping parameters) by an analytical approach. The experimental conrmation prediction is therefore at very least desirable and can be
*

Corresponding author.

used to form analytically the mathematical model. In


turn it can be used to more clearly understand the eect
of parameters controlling the dynamics of composite
states.
Recently, a mathematical model representing the
damping capacity of the composite was established [1].
Based on the student distribution approximation of the
measured values of damping in the fundamental mode,
the modal relationships between the fundamental frequencies and the damping factors were developed in
equivalence to the uniform mass damping of isotropic
structures.
In the present work, an attempt has been made to
improve the convergence characteristics of the model
within a wide range of frequencies for dierent code
numbers at two levels of volume fraction.
Basically, a weight factor (a) has been introduced for
correlating and updating the mathematical model to the
experimental data throughout the utilization of the
curve tting response function. This has resulted in
generalized quasi-rectangular hyperbolic relationships
between the loss factors and the natural frequencies with
the condence level at 99.5%. These results permit the
uncoupling of simultaneous equations of motion of
composite structures with the lowest residual errors.
In the experimental work, cantilever composite
beams made from ber reinforced plastic FRP are
considered as the object of the study for their simplicity
and for extensive applications. Various specimens made
from three plies, Fig. 1, are utilized for two levels of
volume fraction (a) a weakly composite 15% and (b) an
average composite 45%.

0263-8223/99/$ see front matter 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 2 3 ( 9 9 ) 0 0 0 5 1 - 3

164

A. Maher et al. / Composite Structures 46 (1999) 163170

Fig. 1. 3-layered beam model.

In order to evaluate accurately the inuences of code


number on the damping capacities and natural frequencies, twelve specimens of unidirectional cross-ply
and angle-ply laminate have been fabricated in the laboratory by using hand lay up technique.
Numerically the rst four natural frequencies at the
two levels of volume fraction are computed by the use of
the modied formula MFM [7] and listed in the second
column of Tables 1 and 2.
For the sake of verication, the experimental results
of the natural frequencies and the loss factors for the

rst four natural modes are listed in the third and fourth
columns in these tables, respectively.
To highlight the unpredictable nature of the damping
parameters, various curves representing mutual relationships of modal parameters were plotted in Figs. 3
and 4.
The close agreement of the results of the proposed
mathematical and experimental models proves the ecient applicability of the proposed models for deeply
understanding the dynamic nature of vibrating damping
composite structures.

Table 1
The numerical and experimental modal parameters of the rst four
modes of the xed-free GRP beam of volume fraction 15%

Table 2
The numerical and experimental modal parameters of the rst four
modes of the xed-free GRP beam of volume fraction 45%
Frequency,
x (rad/s)

Frequency,
xd (rad/s)

Damping,
g (%)

13.334
4.370
2.552
2.080

[0/0/0]a
1
2
3
4

299.004
1873.816
5246.742
10281.522

295.310
1859.823
5152.212
10040.530

10.106
3.632
2.256
1.564

185.354
1244.071
3606.548
7200.530

13.898
4.546
2.614
2.094

[0/30/0]b
1
2
3
4

296.372
1857.310
5196.766
10190.975

289.027
1847.256
4988.849
10027.964

10.870
3.996
2.330
1.630

208.936
1307.506
3661.049
7174.191

179.071
1237.788
3581.416
6754.424

14.912
5.076
2.632
2.280

[0/45/0]c
1
2
3
4

294.926
1848.281
5175.234
10141.394

282.743
1859.823
4951.150
10015.397

11.112
4.222
2.348
1.662

[0/90/0]d
1
2
3
4

208.482
1306.526
3658.309
7168.825

175.929
1231.504
3568.849
6572.212

16.072
5.612
2.728
2.486

[0/90/0]d
1
2
3
4

294.631
1846.433
5170.069
10131.272

276.460
1822.124
4913.451
9952.566

11.932
4.396
2.430
1.704

[45/45/0]e
1
2
3
4

171.160
1072.640
3003.425
5885.510

163.363
1043.009
2978.230
5811.946

20.000
6.024
3.798
2.648

[45/45/0]e
1
2
3
4

230.643
1445.403
4047.176
7930.844

213.628
1432.566
4021.239
7929.380

15.294
5.702
2.812
2.298

23.684
8.196
4.534
3.086

[45/0/45]f
1
2
3
4

141.529
886.960
2483.511
4866.691

138.230
841.947
2475.575
4863.185

20.910
7.462
3.808
2.842

Frequency,
x (rad/s)

Frequency,
xd (rad/s)

Damping,
g (%)

[0/0/0]a
1
2
3
4

211.109
1322.981
3704.384
7259.115

188.496
1294.336
3694.513
7250.796

[0/30/0]b
1
2
3
4

209.544
1313.186
3676.958
7205.362

[0/45/0]c
1
2
3
4

Mode no., i

[45\0\45]f
1
2
3
4

123.615
774.666
2169.094
4250.556

119.381
766.549
2161.416
4234.867

gi 2.248738345 (xi )0:53724390 ; condence level 99.4803%.


gi 2.463872500 (xi )0:5463537063 ; condence level 99.5838%.
c
gi 2.743609311 (xi )0:55319680819 ; condence level 99.3578%.
d
gi 3.011008077 (xi )0:5545636963 ; condence level 99. 864%.
e
gi 3.56531542 (xi )0:5694874495 ; condence level 99.6619%.
f
gi 3.797399516 (xi )0:5762650855 ; condence level 99.9998%.

Mode no., i

gi 1.973894157 (xi )0:5243924613 ; condence level 99.9324%.


gi 2.288586882 (xi )0:5362827918 ; condence level 99.9956%.
c
gi 2.400855191 (xi )0:5394284987 ; condence level 99.9872%.
d
gi 2.765085107 (xi )0:5522194691 ; condence level 99.9926%.
e
gi 3.097594465 (xi )0:5536083162 ; condence level 99.6279%.
f
gi 3.591001961 (xi )0:5741157503 ; condence level 99.8897%.

A. Maher et al. / Composite Structures 46 (1999) 163170

2. Problem statement
The governing equations of free damped vibration of
a laminated structural model of n degrees of freedom
may be expressed as
mq cq_ kq 0;
1
where m, c, k are the n  n mass damping and stiness
matrices, respectively and q is the corresponding n  1
displacement vector.
In the absence of damping, Eq. (1) become
m
q kq 0;
substitute the trial solution
q V eixt
in the last equation and premultiplying the resulting
equation by m1 we have
EV VP;
the natural frequencies x and the corresponding mode
shapes V are obtained by solving the eigenvalue problem
EV VP;
2
where E m1 ; k is the n  n inverse dynamic matrix.
V is the n  n orthogonal classical modal matrix
formed from the eigenvectors Vr and P is the n  n diagonal frequency matrix containing the n squared frequencies (x2r ) for r 1; 2; . . . ; n.
Having obtained the complete set of n eigenpairs, the
orthogonality properties of mass and stiness matrices
are then expressed as

M VT mV dmT
r mmr c Mr


K VT mV dmT
4
r kmr c kr ;
where M and K are the n  n diagonal mass and stiness
matrices.
Considering the invariance of the modal matrix under
raising the matrices E and P to any rational positive
number a, the eigenvalue problem (2) may be recast as

DV V^;
5
where D Ea and ^ Pa .
Here a is the control weight factor to be chosen for
minimizing the weighted residual. A special case, at
which amax 1, was considered for the uniform mass
damping matrix of the composite in equivalence to the
damping of isotropic structure presented in Ref. [1].
3. Mathematical model of vibration damping of composite
structure
It was mentioned in Ref. [1] that the hyperbolic relations between loss factors (g) and natural frequencies
(x) of the composite plate, vibrating at the rst mode for
dierent boundary conditions, provide a more reliable
prediction throughout the utilization of the uniform
mass damping scheme. These relations were established

165

by utilizing the student distribution approximation with


condence level at 95%.
The demand for more accurate modeling of composite structures for various code numbers within a wide
range of frequency spectrum requires a modication of
hyperbolic relations by introducing the proper weight
factor (ai ) and the damping constant (ai) for ith code
number.
The complication for selecting the proper current
values of the damping parameters is a result of the
mutual eects of the code number, volume fraction,
aspect ratio, types of xation and natural mode number
[810].
For establishing a proper equivalent mathematical
model let us start by successive premultiplication of
Eq. (5) by a positive integer j a j followed by premultiplication of both sides of the result equation by
VT m and we have:
VT mDj V M^j ;
j

6
j

where D E ; ^ P .
From denition, the right-hand side of the previous
equation is a diagonal matrix. It follows that mDj satises the orthogonality conditions. Let a damping matrix be represented as a linear combination of the
compound matrix given by:
X
aj mDj :
7
c
j0

The proposed form (7) satises the orthogonality conditions such that
cm1 k km1 c:
With regard to Eqs. (6) and (7) this diagonalized damping matrix is then given by
X
aj M^j :
8
C Cr VT cV
Here the damping coecient of the rth mode is expressed as
X
aj  Mr kjr ;
9
Cr mTr cmr
j0

x2a
r .

where kr
In view of the hyperbolic relationship between the
loss factor g and the natural frequency derived in Ref.
[1], the rth loss factor can be modied to be
X
aj xa2j1 :
10
gr
j0

In the composite domain the computation of damping constants requires high computational eort compared with the isotropic one. For simplicity, the series
given by Eqs. (7), (9) and (10) have been truncated respectively to the forms:
c a0 m;

11

C r a0 M r ;

12

166

A. Maher et al. / Composite Structures 46 (1999) 163170

Fig. 2. Schematic block diagram of the measuring circuit.

gr a0 xa
r :

13

In view of Eq. (13), the quasi-hyperbolic relation can be


expressed as
gxa a0 :

14

Here, the damping constant a0 increases as the volume fraction decreases and as the code number leading
to a low sti composite structure.
For the sake of graphical linearization, the previous
equation can be transformed into the following logarithmic form
ln g a ln x ln a0 :

0 < a < a ;
a 1 < a0 < a 2 ;
0 < g < 1:
It is obvious that the rate of change (slope) depends
on the degree of isotropism at which a 6 tan1 45 . Also

15

The isotropic state can be considered as a limiting


case at which the weight factor reaches its maximum
value here as
a tan d 1;
where d represents the slope of the logarithm relation of
g x curves (Fig. 5), consequently Eq. (15) will be recast as
ln g ln x ln a :

16

In Eq. (16) the two limiting isotropic states arise here


as
1. a a1 represents a full ber domain at which vf
100%:
2. a a2 represents a fully matrix domain at which
vf 0:0%:
A set of family curves representing composite domains of various degrees of isotropism in the physical
state can be then bounded as follows:

Fig. 3. Quasi-hyperbolic relation between the damping loss factor, g


and the natural frequency, x for the xed-free GRP beam of volume
fraction 15%.

A. Maher et al. / Composite Structures 46 (1999) 163170

167

The steady-state response in the rth mode is then


expressed as
 
F0
Yr
 dr  sin Xt a:
18
Kr
Here the magnication factor in the rth mode is given by
2

2 1=2

dr 1 Zr2 gr  Zr
where Zr X=xr :
dr 1=gr

19

for a < 1;

1=gr

dr
for a 1;
where gr gr x1a
r :
4. Experimental model of vibration damping of composite
structural beam FRP

Fig. 4. Quasi-hyperbolic relation between the damping loss factor, g


and the natural frequency, x for the xed-free GRP beam of volume
fraction 45%.

the damping constants for each specimen increase as the


volume of the fraction decreases.
The validity of the orthogonality condition permits
the decoupling of natural modes. Consequently, the
equation of motion in the rth modes is expressed as
Mr Y Cr Y_ r Kr Yr Fr :

17

A verication of the equivalent mathematical model


of the damping distribution in relation to the frequency
spectrum of the composite structure in the light of the
tested data is represented. The frequency response tests
were performed on cantilever composite beams made
from ber reinforced plastic (FRP) by utilizing fast
fourier transform dual channel analyzer in conjunction
with the computer as shown in Fig. 2.
A typical specimen FRP composite beam of dimension 210  20  3 mm3 ) made of three plies with 1 mm
thickness for each ply is shown in Fig. 1. Two composite
levels were selected for each code number. These are
specimens with low ber volume fraction Vf 15% and
the average ber volume fraction Vf 45%.
To study the eect of lamina orientation and stacking
sequence on the modal parameters, six code numbers of

Fig. 5. Logarithmic relation between damping loss factor, g and natural frequency, x of xed-free GRP beam of volume fraction 15%.

168

A. Maher et al. / Composite Structures 46 (1999) 163170

Fig. 6. Logarithmic relation between damping loss factor, g and natural frequency, x of xed-free GRP beam of volume fraction 45%.

the specimen were fabricated and stated as (0/0/0), (0/30/


0), (0/45/0), (0/90/0), (45/45/0) and (45/0/45) for each
volume fraction.
Within the frequency range (8000 Hz), the frequency
response and half power tests were performed for the
measurements of the rst four loss factors and the corresponding eigenfrequencies and listed in the third and
fourth columns of Tables 1 and 2. For the sake of verication, the rst four natural frequencies listed in the
second column of the tables were computed by modifying the developed formula [7] and the form:
  12
k2i
D
;
xi
2pL2 qc  t
where D is the condensed bending stiness modulus of
composite posit beam given by
1

D D BA B;
where [D] is the bending stiness matrix of ber reinforced laminated plate, [B] the stretchingbending coupling matrix of a ber reinforced two-dimensional
laminate, [A] the extensional stiness matrix of a ber
reinforced two-dimensional laminate and t the thickness
of composite laminate.

With the utilization of the least square technique on


the measured values of the loss factors and natural frequencies at the rst four natural modes, the nearly hyperbolic curve ttings are plotted with the condence
level at 99.5% as shown in Figs. 3 and 4.
To study the eect of the ply orientation on the
damping capacity, we have rst considered the eect of
changes of inner ply orientations and second the eects
of changes of outer ply orientation compared with the
changes of inner ones. A glance at Figs. 3 and 4 indicates that the damping capacity increases monotonically, (natural frequencies decrease), with the increase of
angle of orientation of the inner layer.
With regard to the experimental results listed in Tables 1 and 2, it is obvious that the changes of outer
orientations have signicant eects on the damping,
(and stiness), of the specimens compared with the
changes of the inner orientation. As an example, the
experimental results of the loss factors for (0/0/0), (0/45/
0) and (45/0/45) show that the loss factor increases by
16% due to the increase of the inner layer (by 45 ), while
it increases by 87% and 105% due to the increase of the
two outer layers (by 45 ) for Vf 45% and 15%, respectively.

Table 3
Damping parameters of the six code numbers of xed-free GRP beam for the two volume fractions (Vf )
Code no.

[0/0/0]

[0/30/0]

[0/45/0]

[0/90/0]

[45/45/0]

[45/0/45]

Vf 15%

a
a

2.248738345
0.53724390

2.4638725
0.5463537063

2.743609311
0.55319680819

3.011008077
0.5545636963

3.56531542
0.5694874495

3.797399516
0.5762650855

Vf 45%

a
a

1.973894157
0.5243924613

2.288586882
0.5362827918

2.400855191
0.5394284987

2.765085107
0.5522194691

3.097594465
0.5536083162

3.591001961
0.5741157503

A. Maher et al. / Composite Structures 46 (1999) 163170

In Figs. 5 and 6 the logarithmic forms of quasi- hyperbolic relations for all cases are plotted in linear
forms. It is noticed that the slopes assigned by the

169

weights factor are dependent mainly on the degree of the


isotropical state while the damping constant depends
mainly on the exibility of the specimens.

Fig. 7. Quasi-linear relation between the damping constant, a and the stacking sequence, s for the xed-free GRP beam of volume fraction 15% and
45%.

Fig. 8. Quasi-linear relation between the damping exponent, a and the stacking sequence, s for the xed-free GRP beam of volume fraction 15% and
45%.

170

A. Maher et al. / Composite Structures 46 (1999) 163170

The interrelations of the weight parameters (a; a0 )


and the code numbers for each level of volume fraction
are listed in a monotonic manner in Table 3 and plotted
in Figs. 7 and 8. It is found that the weight parameters
increase linearly with the increase of ply-orientation of
either inner or outer layers.

ence on the modal parameters compared with the other


angle orientations.
(5) The obtained values of the weight factors may be
considered as indicators of the degree of isotropity of
the composite structures.
References

5. Conclusion
In the present work, the modeling of the damping
distribution in vibrating composite structures is established with the lowest residual errors. The tted results
of the measured values indicate the following.
(1) There is an existence of generalized quasi-rectangular hyperbolic relationships between the loss factors
and natural frequencies of composite structures and can
be found at any selected range of frequency spectrum.
(2) The developed quasi-uniform mass damping matrix by utilizing the proper weight factors permits the
utilization of normal coordinate systems for uncoupling
equations of motion of composite structure with the
high condence level at 99.5%.
(3) The angle orientations of the outer laminate have
signicant eects on the modal parameters of the composite compared with the inner laminate.
(4) The angle of orientation 45 , at which the shearing
parameter reaches highest values has the highest inu-

[1] Maher A. Evaluative eigenanalysis of composite structures. J


Comp Struct 1994;29:19195.
[2] Vinson JR. TSU-Weichou, Composite material and Their use in
Structures. Barking: Applied Science Publishers, 1975.
[3] Jones RM. Mechanics of Composite Materials. Scripta Book
Company, 1975.
[4] Schwartz MM. Composite Material Hand Book. Published, 1976.
[5] Maher A, El-Soaly, Ghoneam S, Hamada A. Eigenanalysis of
ber reinforced composite plates. J Comp Struct 1993;25:52128.
[6] Nadella S, Rao MD. Damping of composite structures using
embedded viscoelastic layers. In: Michigan Technological University (USA) Proceeding of the 13th International Modal
analysis Conference. 1995;2:23339.
[7] Blevins RD. Formula for Natural frequency and Mode shape.
Litton Educational publishers, 1979.
[8] Plunket R. Damping mechanisms in ber-reinforced laminates
mechanics of composite materials. In: Recent Advances, Proceedings. 1982.
[9] Bert CW. Composite Materials: A survey of damping capacity of
ber-reinforced composite. In: Damping Applications for vibration controled. Torvik PJ, ASME, AMD38 , 1980.
[10] Gibson RF, Plunket R. Dynamic, stiness and damping of berreinforced composite materials. Shock and Vib Digest 1977;9(2).