The Forced Vibration of A Three-Layer, Damped Sandwich Beam With Arbitrary Boundary Conditions
The Forced Vibration of A Three-Layer, Damped Sandwich Beam With Arbitrary Boundary Conditions
The Forced Vibration of A Three-Layer, Damped Sandwich Beam With Arbitrary Boundary Conditions
D. J. MEAD
1. INTRODUCTION
The theory of the flexural vibration of damped sandwich beams and plates has been considered
by a number of authors in recent years. Kerwin [l] analysed the three-layer system with the
damping layer sandwiched between two face-plates. He considered travelling sinusoidal
waves of transverse displacement in the beams, and derived an expression for an effective,
complex, flexural stiffness for the beam section. Mead [2] analysed the forced flexural
vibrations of a simply supported plate which vibrated in sinusoidal standing waves. He
derived an expression for the flexural stiffness which was basically the same as Kerwin’s
though more general. The standing waves, or “modes”, were found to be uncoupled, even
when damping was admitted in the core. The solution of the general forced vibration problem
was therefore quite simple and straightforward when these uncoupled modes were used in a
modal-type analysis. Freudenthal and Bieniek [3] found the same uncoupled property to hold
for the sinusoidal modes of a sandwich cylinder.
When a sandwich beam is not simply supported, but is supported in some other way
(e.g. clamped, or free at the ends, etc.) the natural flexural vibrations are no longer in pure
sinusoidal modes. Moreover, the natural modes of such a beam with an undamped core
t On study leave from the Institute of Engineering Mechanics of the Slovak Academy of Sciences,
Bratislava, Czechoslovakia.
11 163
164 D. J. MEAD AND S. MARKUS
become coupled as soon as core damping is introduced [4]. The calculation of the forced
vibrations of these beams therefore presents, apparently, a more difficult problem.
Di Taranto [5] has studied damped sandwich beams with arbitrary boundary conditions.
He derived the sixth-order differential equation of motion for the beam (in terms of longi-
tudinal, and not transverse, displacements), and assigned a complex shear modulus to the
sandwiched core. Furthermore, he assigned a complex form to the square of the vibration
frequency [i.e. o* = wf(l + iq)] and attributed the equations to free vibration. Since free
vibrations of damped systems are necessarily decaying, the complex stiffness notation is
strictly inadmissible. (See reference 6 for a full discussion on this point.) Di Taranto proceeds
in a later paper [7] to show how natural frequencies, loss factors and modes can be found
from the differential equation, in conjunction with six boundary conditions for the beam.
It is the purpose of this paper to show that the natural frequencies and modes discussed
by Di Taranto constitute, in fact, a special class of resonant frequencies and forced modes of
vibration of the sandwich beam. None of Di Taranto’s important conclusions is invalidated
(in particular, that the relationship between the modal loss factor of a beam and the corres-
ponding modal frequency is independent of the beam boundary conditions), but the usefulness
of Di Taranto’s modes is shown to lie in the fact that they are completely uncoupled. This
greatly simplifies the general forced vibration problem. The modes can only exist in the
presence of an external transverse loading which is proportional at all points along the beam
to the local inertia loading. At the resonant frequency (Di Taranto’s “natural” frequency),
the external loading is iq times the inertia loading (7)= modal loss factor).
These important properties are proved by considering the sixth-order differential equation
of motion in terms of the transverse displacement, W.This equation is first derived, and various
boundary conditions are also expressed in terms of W.The equation in terms of w permits a
clearer physical understanding of the problem, especially when a transverse, external loading
is applied to the beam. The sixth-order differential equation is derived from a pair of simpler,
simultaneous equations which are necessary for the proving of the orthogonal (uncoupled)
properties of pairs of the forced modes of vibration.
By considering the geometry of the displacement pattern ul, u3 and &v/8x of the face-plates
[see Figure l(b)] it can easily be shown that
au h‘ + hJaw 241 - 243
az=--+
2hz ax hl ’
so that
aw u1 - u3
d
'=&ax+ hz ’ (1)
where
d = hz + +(h, + h,).
In this, it has been assumed that the shear strain is constant across the depth of the core.
This follows from the assumption of zero longitudinal direct stress in the core, which implies
constant shear stress across the depth. The shear stress in the core is therefore given by
(2)
Now the total shear force on the face XX consists of three components [see Figure l(c)] :
(a) a shear force, Si, associated with the flexural rigidity, D,, of the top face-plate, i.e.
where D1 = El h:/12;
@I a shear force, S3, associated with the flexural rigidity, D3, of the bottom face-plate, i.e.
S cw
3=D 3ax3
where D3 = E,h:/12;
w a shear force, S2, associated with the core shear stress 7; this shear stress must be
considered to be acting uniformly between the mid-planes of the face-plates, and not
just over the depth of the core; actually, it is constant over the depth of the core, and
varies linearly to zero across the thickness of the face-plates; this is equivalent, of
course, to a uniform shear stress between the mid-planes of the face-plates; hence
S, = -rd (3c)
where the minus sign must be included since the positive shear force direction is
opposite to the positive shear stress direction.
The total shear force, S, on the section is therefore given by
1
.
Now the transverse loading,p, on the beam is related to S byp = aS/&. Hence, by differen-
tiating equation (4) putting D, + D3 = D,, and re-arranging, we obtain
p= D a4w Gd2a2w
h2&f($q. (5)
fax4
Let the nett longitudinal forces in each of the face-plates be denoted by Pi and P3. These
forces have their lines of action in the mid-planes of the face-plates and are related to the
longitudinal displacements by
au,;
P, = E, h, ax P3=E3h3z.
au, (6)
166 D. J. MEAD AND S. MARKUS
Since there can be no resultant longitudinal force on the whole section, P, = -P3, or
au au,
E h l=-E3h3-. (7)
l ‘ax ax
Furthermore, considering the physical nature of the system and its displacements, it is
obvious that we can also write
E, h, u1 = E3 h3 u3. (8)
Equation (5) can therefore be written in the alternative form
Undef tected
Overal I moment
ond shear force
X u3
c
Cd)
Face
a.-.- - Ps + 6Ps
p3
plate
I
6X
l-4
Figure 1. Longitudinal forces on lower face-plate element. (a) Dimensions and co-ordinate
system. (b) Displacement system. (c) Forces and moments acting on a section. (d) Longitudinal
forces on lower face-plate element.
VIBRATION OF DAMPED SANDWICH BEAMS 167
In subsequent equations, it is convenient to introduce the symbols
and
(11)
Using these in equation (9a), we may re-arrange it into the form
(9b)
Next, consider the longitudinal equilibrium of a lengthwise element, 8x, of the lower face-
plate [see Figure l(d)]. It is evident that
6Pj = --78x
or
ap3_
ax (12)
Substitute into this the expressions for T and P3 from equations (2) and (6) and also use equa-
tion (8) to eliminate ul. After re-arrangement, equation (12) then becomes
a2 u3
--gu,=-gY&dg (13)
ax2 3 3
Equations (9b) and (13) constitute the simplest pair of differential equations relating the
displacements w and u3 to the applied loading p. On eliminating u3 from this pair, a single
sixth-order differential equation in w is obtained, viz.
a6 w
--g(l+ Y)$=$ ;2;-gp . (14)
ax6 f( 1
(m is the mass per unit length of the whole three-layer section). Substituting this into equation
(14) we obtain the differential equation for the forced vibratory motion of the beam:
Since the motion is now harmonic, it is legitimate to admit hysteretic damping into the core
by putting G = G’(1 + i/I). Evidently, we may now also write
ri;.g(W;-gW,)+T,,(Wf-g(1 + Y) Wk)=Pz(Wi-gW,,)e’“’
t f
or
j: +T W,“‘-g(l+ Y)Wf,
” n( (m/D,)(W: (21)
- gW,) ) =peior’
Now T,, is a function only of o and t, and not of x. The coefficient of T,, in this equation
must therefore be independent of x, and equal to a complex constant, a + ib, say. Then
(26)
which form Di Taranto also considered. The h,,,‘s are the six complex roots of the complex
characteristic equation
Di Taranto expressed the boundary conditions in terms of & (= ui), but they may equally
well be expressed in terms of W. Six conditions must be sought for any particular beam.
Before all these expressions can be written down, we need to have expressions for the face-
plate forces, P, and P,, the bending moment M, and the shear force, S, in terms of W andp
only.
Equations (5), (6) and (7) can readily be manipulated to show that
. (28)
170 D. J. MEAD AND S. MARKUS
The total bending moment, M, acting on the section can be split into three components
analogous to those of the shear force.
(a) and (b) moments MI and AI3 associated with the flexural stiffness D1 and Ds of the
top and bottom face-plates, i.e.
(2% b)
(c) moment M2 associated with the equal and opposite forces, PI and P,, which act along
the mid-planes of the face-plates, i.e.
. (29~)
(30)
When the beam is vibrating in one of the damped normal modes of vibration, p in the
above expressions should be replaced by 1~x1 + iq,) W,, and w by W,. Possible boundary
conditions for the modes of sandwich beams are then as follows.
(a) Transverse displacement prevented at an end
w”lend= 0. (32)
(b) Rotation prevented at an end
[a W,/axi,,, = 0. (33)
(c) Bending moment, A4 = 0 at an end
Equation (17) yields
(f) Face-plate displacements u1 and u3 at an end are restrained by a rigid rivet through the
whole sandwich plate
This means shear strain in the core is prevented and S2 = 0. Hence from equation (4)
we have
1
a3 w
S= Dt[ jg cnd’
VIBRATION OF DAMPED SANDWICH BEAMS 171
But S is also given by equation (31). Equating these two expressions, we find
[
IV;-gwy; -m*(l
f
+i$)W,:
1
end
=o. (37)
(g) Face-plate displacements, uI and u3, at an end are restrained by aflexible rivet through the
whole sandwich plate
The rivet is sheared by the equal and opposite forces Pi and P3 at the end. Let the shear
strain in the rivet be given by
where k, defines an effective rivet shear stiffness. This value of y is also the shear strain in the
adjacent core at the end. From this, it can be shown that the following boundary condition
holds :
(38)
Other boundary conditions can be developed, as required, e.g. for an end having a con-
centrated tip mass, or resting on a flexible support, etc.
(39a)
and
- (39b)
-g’~~~-w:(~)(~“L-g’(~“+~))=o. (Mb)
Now although equation (40a) may have six roots for &, only four of these can satisfy
equation (40b). (This will only be possible for certain discrete values of w, and I”.) The general
solution to equation (24) can then only have four terms, and so will be of the form
The roots of equation (40b) consist of a pair of purely imaginary numbers (&A,) together
with a pair of purely real numbers (&). This follows from w”,and 7” being real and positive.
Equation (41) can therefore be written in the familiar form
W,-Alcos&x+A2sin)r,x+Ajcosh&x+A,sinhh,x (42)
172 D. J. MEAD AND S. MARKUS
and since w, is required to be real, the A’s must be entirely real. There are therefore just four
unknown real coefficients in equation (42), which must be adjusted to satisfy the six complex
boundary conditions of the most general type of sandwich beam. Clearly, therefore, this is
impossible, and the real solution for W,, is not a general possibility.
However, it may easily be shown that if a beam is simply supported at its ends and the
face-plate displacements are unrestrained at both ends [i.e. boundary conditions (32), (34),
(36) apply] then the mode
W= A,sin&x
is a possible solution. This is an entirely real function. On the other hand, if the ends of the
same beam are riveted [boundary conditions (37) applies], conditions (32), (34) and (37)
can no longer be satisfied by real W functions.
It is possible to find other specific boundary conditions which permit a real mode to exist
but we find these to represent elastically restrained ends which either incorporate damping
or at which an externally applied shear force or moment acts in quadrature with the inertia
forces (i.e. in phase with the externally applied loading). Even them if one real damped normal
mode can be found, the other modes are found to be complex. Such systems are so unrealistic
that we conclude that entirely real damped normal modes of a sandwich beam will never exist
in practice unless both ends are simply supported and unrestrained in the sense of condition
4(e) [equation (36)].
s
0
mW,, W,,,dx=O (n#m) (43)
then the solution of the general forced vibration problem can be shown to reduce to a very
simple form (see next section). This relationship cannot be proved from equation (24) alone
as this problem is not self-adjoint. Nevertheless, the non-self-adjoint property does not
necessarily exclude the possibility of the above orthogonality. Consider, instead, therefore,
the pair of equations (9b) and (13) from which equation (24) was derived, and put
p = w,2(1+ ir],)mW,.
The pair of equations for the nth mode become
w~-gYw:+gxu;=w,2,
(1+ bJmW ”
f
and
u;-gu”=-x
gy w;
(45)
where X = E3hj d/D, and U, is the longitudinal displacement function corresponding to the
transverse displacement function W,,.
Multiply equation (44) by W,,,and equation (45) by u,,, and integrate them over the length
of the beam. Likewise take the corresponding equations for W,,, and u,,,, multiply them by
W,, and u, and integrate. Subtracting one set of integrated equations from the other,and
using possible boundary conditions for Wand U, leads to equation (43) being proved.
Physical considerations can lead to further development of equation (43). From it we
deduce that an external loading proportional to m W,,,cannot excite the mode W,. This means
VIBRATION OF DAMPED SANDWICH BEAMS 173
that the loading m W,,, cannot do any work when it moves through a virtual displacement
6 W,,. Since m W,,, and W,, have both real and imaginary components, and the virtual work
must be zero at any instant of time for any phase-relationship between m W,,,and W, we can
deduce that
I
s
0
mRe( W,) Re( W,) dx = 0 Wa)
or
I m(W, W: f
0
W,* WJdx = 0, (47)
Suppose now that the sandwich beam is excited by the general harmonic loading
QW = j, km Wm. (49)
Now write the solution of the general vibration problem in the form
Equation (16), together with equations (48), (49) and (51), now yields
WI’-g(l+ Y)W:-Fw’(w:-gw.))=&zk_(W.-pW.).
f r n
174 D. J. MEAD AND S. MARKUS
From the previous section, we deduce that this equation must be true for each mode
separately, i.e.
Wi’-g(l+ Y)*~-m~(w;-gwn)=~~(w..-gw”
f )
f n
where the functions, W,,, are generated from equation (24). Combining equations (24) and
(52), we find
B,,
= 4,
m[w,2(1 + iyn) - w2]
(53)
where k, (=k,, + ik,,) is fully determined by equation (50). Thus, the forced vibration at any
point x along the beam is
The form of this is identical to that obtained for any elastic system in terms of the undamped
normal modes, provided the modes are not coupled by the damping. The normal modes of
an undamped sandwich beam become coupled when damping is introduced into the core,
unless the beam is simply supported and so vibrates in sine modes. If the undamped normal
modes were used in a modal analysis of the forced vibration of the damped beam, the resulting
expression for the response would be much more complicated than equation (54). The above
analysis shows how the forced vibration problem is simplified by the use of the complex
damped normal modes.
Had we used W,* to develop equation (50), and used the orthogonal relationship
I W,,, W,*dx=O
0
[derivable from equation (47)], then the final expression for w(x, t) would be
The real term rni 1W,21d x in the denominator may be regarded as the real “generalized
mass” of the nth complex normal mode W,.
8. CONCLUSIONS
The forced vibrations of damped sandwich beams may be conveniently analysed by using
a special class of forced, uncoupled and complex modes of vibration. These modes exist for
beams with any boundary conditions. They can only exist on their own in the presence of
an externally applied harmonic transverse loading which is proportional to the local inertia
loading of the mode. At the resonant frequency of the mode the external loading is in quadra-
ture with the inertia loading, and its magnitude is equal to the inertia loading times the modal
loss factor.
VIBRATTON OF DAMPED SANDWICH BEAMS 175
These modes, resonant frequencies and loss factors are identifiable with those discussed by
Di Taranto, who attributed them to free vibration. It would appear that such forced modes
exist for any elastic system with linear damping which may be distributed in any form through-
out the system.
REFERENCES
1. E. M. RERWIN, JR 1959 J. acoust. Sot. Am. 31, 7. Damping of flexural waves by a constrained
v&o-elastic layer.
2. D. J. MEAD 1962 University of Southampton Report No. AASU 160. The double skin damping
configuration. (Subsequently published as Chapter V of USAF Report No. AFML-TR-65284).
3. A. M. FREUDENTHALand M. P. BIENIEK 1960 USAF Report No. WADD Tech. Rep. 60-307.
Forced vibrations of sandwich structures.
4. D. FRANCE 1968 M.Sc. Dissertation, University of Southampton (unpublished). A study of the
vibration of a damped sandwich layer in non-sinusoidal modes.
5. R. A. DI TARANTO1965 J. appl. Mech. 32. (Tram ASMESeries E 87,881.) Theory of the vibratory
bending for elastic and viscoelastic layered finite length beams.
6. S. NEUMARK1962 ARC(London) R& M No. 3269. Concept of complex stiffness applied to problems
of oscillations with viscous and hysteretic damping.
7. R. A. DI TmNro and W. BLASINGAME 1967 J. Engng Znd.(Trans. ASME633.) Composite damping
of vibrating sandwich beams.