See discussions, stats, and author profiles for this publication at:

https://www.researchgate.net/publication/229182743

Three-dimensional vibration of
laminated composite plates and
cylindrical panels with arbitrarily
located lateral surfaces point supports

Article in International Journal of Mechanical Sciences · March 1996

DOI: 10.1016/0020-7403(95)00056-9

CITATIONS READS

17 9

2 authors, including:

Jianqiao Ye
Lancaster University
140 PUBLICATIONS 1,339 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Jianqiao Ye
letting you access and read them immediately. Retrieved on: 23 August 2016
 Pergamon lm d '~tech. S('i Vol. 38. No. 3, pp 271 281. 1996
Copyrigh'L i~-, 1995 Elsevier Science Ltd
Printed in Great Brilain. All rights reserved
0020 7403i96 $15.00 + 0.00

0020-7403(95) 00056-9

THREE-DIMENSIONAL VIBRATION OF LAMINATED

COMPOSITE PLATES AND CYLINDRICAL PANELS
WITH ARBITRARILY LOCATED LATERAL SURFACES
POINT SUPPORTS

JIANQIAO YE t and K. P. SOLDATOS *

•Department of Civil Engineering, University of Leeds, Leeds LS2 9JT, U.K. and ~Department of Theoretical
Mechanics, University of Nottingham, Nottingham, NG7 2RD, U.K.

(Received 29 duly 1994; and in rertsed (orm 21 March 1995)

Abstract On the basis of fully three-dimensional elasticit 3, considerations, this paper presents a free
vibration analysis of simply supported, cross-ply laminated plates and cylindrical panels that are
subjected to an arbitrary number of lateral surfaces point supports. The analysis is based on
a recursive approach suitable for the vibration analysis of corresponding unconstrained structural
elements. By means of dynamic equilibrium considerations, the reaction of the point supports is
imposed by using the Lagrange multipliers method. This yields the eigendeterminant of a con-
strained panel by appropriately coupling the response of a suitably large number of natural
vibration modes of the corresponding unconstrained structural element.

1. I N T R O D U C T I O N

Plate and shell panels with interior point supports are widely used by civil or mechanical
engineers in many industrial applications. However, although free vibrations of point
supported plates have extensively been studied (see for instance some recent papers [1 6]
from which past relevant publications can further be found), corresponding shell-type
analyses [7] are very rare in the literature. Moreover, all the afore-mentioned analyses
[1-7] are based on two-dimensional theories.
Dealing in particular with the flat plate configuration, classical plate theory has mostly
been employed in the literature, in spite of the fact that vibrations of not only isotropic [1,4]
but also orthotropic [5] and laminated composite point supported plates [2, 3] have been
studied. To the authors' best knowledge, the only relevant study dealing with shear
deformable point supported plates is the Mindlin-type analysis presented in Ref. [6] which
also deals with homogeneous isotropic plates. Transverse shear deformation effects are
much more pronounced in highly reinforced orthotropic and laminated composite struc-
tural elements. However, vibration studies dealing with shear deformable composite plates
and shells having interior point supports have not yet appeared in the literature. The
corresponding shell analysis presented in Ref. [7] deals with homogeneous isotropic
circular cylindrical shells and is also based on a classical Love-type shell theory.
On the other hand, by averaging all material and mechanical properties through the plate
or shell thickness, two-dimensional theories are essentially inappropriate for solving prob-
lems with point supports that are pinned on the lateral surfaces. Such a pinned lateral
surface support is erroneously considered by a two-dimensional theory as if the correspond-
ing normal to the middle surface line was rigidly fixed. Hence, unless the dynamic behaviour
of extremely thin structural elements is investigated, two-dimensional theories may lead to
considerably inaccurate results.
On the basis of fully three-dimensional elasticity considerations, this paper presents a free
vibration analysis of simply supported, cross-ply laminated plates and cylindrical panels
that are subjected to an arbitrary number of lateral surfaces point supports. The analysis is
based on the recursive vibration solution presented in Refs [8, 9] for corresponding
unconstrained structural elements. Nevertheless, that recursive three-dimensional solution
is here suitably reformulated in a manner that yields a symmetric form of the final 6 x 6
271
M~ 382-3
272 Jianqiao Ye and K. P. Soldatos

dynamic stiffness eigenmatrix. By means of dynamic equilibrium considerations, the reac-

tion of the point supports is then imposed by using the Lagrange multipliers method. This
yields the eigenmatrix of the constrained panel considered by appropriately coupling the
response of a suitably large number of natural vibration modes of the corresponding
unconstrained structural element. In a particular application, numerical results for an
isotropic plate are obtained and compared with corresponding results based on two-
dimensional plate theories. Further results dealing with laminated plates and cylindrical
panels are presented and discussed.

2 REFORMULATION OF A FREE VIBRATION SOLUTION FOR

LAMINATED CYLINDRICAL PANELS

Consider an arbitrarily thick circular cylindrical panel, having a constant thickness h and
an axial length Lx (Fig. 1) and denote with R and L~ the radius and circumferential length of
its middle-surface, respectively. As the radius R approaches infinity (th = 0), the geometrical
configuration of a flat plate is obtained as a particular case. The axial, circumferential and
normal to the middle-surface co-ordinate length parameters are denoted with x, s and z,
respectively, while u, v and w represent the corresponding displacement components. It is
assumed that the cylindrical panel considered is made of an arbitrary number of linearly
elastic orthotropic layers whose material axes of orthotropy coincide with the axes of the
adopted curvilinear co-ordinate system. It is finally assumed that all four of the panel edges
are imposed on the simply supported boundary conditions considered in Ref. [9]. The free
vibration problem considered is described by the following Navier-type differential equa-
tions:
CllU,xx + C6e,(l + z/R) 2u.~ + C 5 5 R - I ( 1 + z / R ) - l u z
+ Cssu,._z + (C12 + (7661(1 + z/R)-lV,x~
+ (C13 + C 5 s ) w ~ + ~C~2 + C55)R-1(1 + z/R)-lw,~, = pu,,,

(C12 + C66)(1 -r- z/R) 1 laxs At_ C 6 6 u , x x

+ C~2(1 + z/R) -2 t:~ -- C44R-2(1 + z / R ) - 2 v
+ C~,~,R l(1 + z/R) Ir.: + C,,4v=: + (C2 2 + C44)R-a(I + z / R ) - l w s
+ (C23 + C44)(1 + z R ) - l w . s : = pt'.n, (1)

(Cx_s + C55)u.x.- + (C~3 - C 1 2 ) R - lu ~, - (C22 + C44)R-1(1 + z / R ) - 2 v , ,

"~- (C23 -~- C4.4}(1 -~- z R ) 1 Us_" q_ C55Wx x + C44(1 + z/R)-2W.ss

+ C33w= - C22R-2(1 + z / R ) - 2 w + C33R-1(1 + z / R ) - l w . z = p w , ,

where t denotes time, p is the material density and the elastic stiffnesses Cij (i,j = 1,2, ..., 6)
are, in general, considered as piece-wise constants through the thickness of such a cross-ply
laminate [10].
The free vibration analysis detailed in Ref. [9] was based on the division of that initial
cylindrical panel into a total number of N coaxial and successive sublayers. Different layers
may have different thicknesses or material properties. It was assumed, however, that the
thickness of each layer approaches zero with N approaching infinity. Upon choosing
a suitably large value of N, such an individual layer becomes sufficiently thin and, as
a result, the through thickness variation of curvature affects negligibly the corresponding
vibration solution. Hence, by replacing the term (1 + z/R) with 1, Eqns (l) are reduced into
a corresponding set of approximate equations with constant coefficients, whose exact
solution may be used for an approximate but sufficiently accurate vibration analysis for
each one of those thin layers. In Ref. [9], however, all such solutions obtained were suitably
connected by means of appropriate continuity conditions, imposed on all fictitious and real
material interfaces. As a result, a recursive but arbitrarily close solution was obtained for the
free vibration problem of the initial, thick, circular cylindrical panel.
 Three-dimensional vibration of laminated composite plates 273

?
"\,

///"

Fig. I Nomenclatureof a circular c_qindrical panel.

In more detail, a displacement model of the form.

q r
u ( x , s , z ; t ) = (dx, s. zl T(tt = ~ ~ U m . [ = l c o s ( m g x / L x ) s i n ( n ~ s / L ~ ) e i~",
m=l n=l

v ( x , s , z : t ) = g(x,s,z}'l'lt~ = ~ l ~ . ( z ) s i n [ m T z x / L ~ ) c o s ( n ~ s / L s ) e i'°', (2a)

m=l n=l

q r

w ( x , s , z ; t ) = ¢,(x.s,z) Tlt) = ~ ~ I4',..Iz)

" sm {mlrx/L~)
' sin
" (nns/L~)
' e '~' ,
m=i n=l

represents a superposition of the first q x r normal modes of vibration and is associated with
transverse stresses of the following form:
q r
zx=(x,s,z;t) = rx:lX, S.z) I'tt) = ~ ~- X,..(z}cos(mnx,/'L~)sin(nrcs/LOe i°'',
m=l n=l

r = ( x , s , z ; t ) = -c=(x,s.:) l[t) = ~.° i S , . . I z t s i n ( m r t x / L . ) c o s ( n r r s / L ~ ) e i°'~, (2b)

m = 1 n: 1

~:(x,s,z;t} = c~:(x.s,:) Tit) = ~ Z,.. zIsin mnx/Lx)sin(nrrs/L~)e TM.

m=l n=l

Upon employing the displacement model (2a) for the/-th of the afore-mentioned real or
fictitious sublayers of the panel considered (j = 1,2 ..... N), the differential eigenvalue
problem (lj can be represented in the following matrix different equation form,
,~~l I-7.(,j i. ;I, = [ G . , . ] ~¢ -~, .?.{,j,) l {f}r = ~,U,U',V,V'.W,W'}, (3)
for any given combination of the half-wave numbers m and n. Here, a prime denotes
ordinary differentiation with respect to z, while the components of the 6 x 6 matrix [G,.,]
are given in the Appendix of Ref. [1 1] and are, in general, dependent on the unknown
natural frequency eJ. Upon imposing appropriate continuity conditions on all real and
fictitious material interfaces and using the recursive formula introduced in Ref. [9], the
following equation is obtained:
[-IN1
/~,,, th [ U l / ,21} = [H,,,] ¢ ~,'( 1, i, h(l
',2)}. (4)
where, t-.,.IvN)~,¢NjV,,,~,~"'~ and ~'-~.v'~,~ _ h~('2)} denote the values of {F.,. (~at the top and bottom
lateral surfaces of the cylindrical panel rcspcctively, and [Hm.] is a frequency dependent
274 Jianq~ao Ye and K. P. Soldatos

transfer matrix [9]. Hence, the final form of the solution of the vibration problem con-
sidered was obtained in Ref. [9] by suitably connecting Eqn (4) with the stress boundary
conditions imposed on the panel lateral surfaces. Independently of the number of the real
and fictitious layers employed, the natural frequencies of vibration were finally obtained as
roots of a transcendental equation which is produced by nullifying the determinant of
a 6 × 6 non-symmetric matrix.
Instead of following the solution procedure detailed in Ref. [9], the exact stiffness matrix
of the problem considered is obtained here on the basis of Eqn (4) and the following matrix
transformation:

k~'~'~
m n )"
= [B,..] #F~lq
~ ran "
- h~tL/2)}, (5)
where.
I ,e'~"l" = [ z , x , s , t . ~. ~l . . . . ~, f"~'}r=[Z,X,S,U,V,W]:= h,,2, (6)

and the components of the 6 × 6 matrices [A,..] and [B,..] are given in the Appendix.
Hence. the column matrices appearing in the left-hand-sides of Eqns (5) give the values of
the stress and displacement components on the top and bottom lateral surfaces of the panel,
respectively, and a combination of Eqns (4) and (5) yields further,
,~F'~"'}m. = [,4,.,,][H,..][B.,,,] ''17'"4,
m.. h'~' 2)} = [/4,,,.] { F,..
(-"') h(1)/2)}. (7)
On the basis of this later equation, the following equations can finally be obtained,
',,t,..'~ = [Q,.,,], F'"'m.,, ',~,..I = [ R , . . ] ~ ,..,, (8)
where.
{d ~,7 = [U(h,~,/2), V(h,.V)2). It, (h ''~ 2), U( - h~'/2), V( - httl/2), W( - hill/2)],
(9)
, ~7 = [ X ( h ~ , a / 2 ) , S l h ~ 2).Z(]?~:~ 2).X( -h(l'/2),S( h~/2),Z( - hiX~/2)],

and, therefore, the column matrices {d.,,, ~

' and ~s~.. ] consist of the values of the displacement
and stress components, respectively, on the panel lateral surfaces. Hence, Eqn (8) yields the
dynamic stiffness equation of the panel considered in the following form:
"~.~m.l = [t¢~,,] [em.] ~ ',d,..', = [Km.] {din.}, 00)
where the 6 ,~ 6 dynamic stiffness mamx [K,..] is symmetric and has frequency-dependent
elements. Due to the zero traction lateral boundary conditions assumed on a freely
vibrating panel without lateral surfaces point supports, its natural vibration frequencies can
be found as roots of a transcendental equation which is produced by nullifying the
determinant of [K,..]. These are obviously identical to frequencies predicted on the basis of
the corresponding 6 x 6 non-symmetric eigenmatrix formulated in Ref. [9].

V I B R A T I C ) N S O[: I A M 1 N A 1 1 1) ( " f l I N D R I C A L PANELS WITH LATERAL

St RI A C E S P O I N T S U P P O R T S

Suppose lhat the first q × r normal modes that appear coupled in Eqns (2a) represent
a vibration mode of a cylindrical panel having a certain number of point supports on its
lateral surfaces. On the basis of the notation adopted in Eqns (9), two vectors that represent
the values of displacements and stresses on those lateral surfaces (z = + h/2) can respective-
ly be formed as follows:

',b' =
¢ +._
,.... [X.,,,(x, s)] t~drnn ~
m : ] rl :

(11)
r

f' ~ v [N...(x,~)]s ,
m :: ] n : 1

where.
'/) ~ [ulh 2),~'(h 2i.~fh 2),u( h 2),(( - h2),~i~( - h.2)],
 Three-dimensional vibration of laminated composite plates 275

t~T,
-,I = [f~:(t 21. f~_. h,2),c}.ih,2)
. f.~:( h 21. f~:( h 2).a:( - h/2)],

[N,,,,,t\. ~1] = Diag [ cos (m~zx-')

,. Lx / sin ,..L~. .sin \ L.. J

× cos
t/7"/'5 ' t
sin
m~x) sin
I.~ / EL, l '
(n~s)
/tl/r x
cos C- sin -L);' sin \ L., }

x sin sin .
" 1-i, JJ
Consider initially, for simplicity, a particular case in which the cylindrical panel con-
sidered is subjected to two rigid point supports thai are located symmetrically with respect
to its middle surface, that is at the points (x*, s*. + h/2) on its lateral surfaces. Since all three
degrees of freedom of a rigid point support are supressed. Eqns (1 1) and (12) yield the
following constraint equations:

/i*
[,%,..l.x . s*~] ,d ...... = 0 ~, (13)
m = 1 n : l

t, L~ '," ~LF!sin\ L~ I '

,. L, ' l_~., sin k L~ }J

It is essential to notice, however, that in the particular case of a single rigid point support (at
z = - h/2, say), the corresponding constraint equations would be represented in the
following form:
r

--~'. ~ [E*.~.~*. s*l] :d ..... = ,'0'.,. (14}

m = 1 n= I

with the 3 × 6 matrix [E*.Ix*.s*)] being identical with the bottom half of the matrix
[N*.(x*, s*)] above. In the case of a single but non-rigid point support, in which only one or
two of the degrees of freedom are suppressed, the corresponding constraint equations can
again be represented in the matrix form (14). with [E,..(x , s*)] being a l x 6 or a 2 x 6 part,
respectively, of its original form.
Consider next the most general case in which, due to a certain number of either rigid or
non-rigid point supports imposed on the panel lateral surfaces, a total number k of
corresponding degrees of freedom has been suppressed. Upon denoting with x* and s* the
vectors consisting the co-ordinates of all point supports involved, all constraint equations
that correspond to Eqns (14) can be assembled in one equation of the form:
q r

,_.v. V s*(] :d.,.: = ,'o,: (15)

m = ] tl=: 1

where [E.,.(x*, s'l] is an appropriate k × 6 matrix.

Dynamic equilibrium considerations suggest that the total energy, V, of the cylindrical
panel considered equals the total work done by the external forces acting on the displace-
ments of the lateral surfaces. Hence, the total energy of the panel can be written as follows:

.,T, ID, ~ 'd... I [Nij]{dij dA,

J" ~ 2 ..4 ~ ~ .... m=ln:l i=lj~]

(16)
276 Jianqiao Ye and K. P. Soldatos

where A represents the area of the lateral surfaces. It should be mentioned that, after the
refined formulation (10) of the vibration problem studied in Ref. [9], all inertia terms
involved in expression [16) have been incorporated into the dynamic stiffness matrix [K,~.].
Applying the Lagrange multipliers method involves minimization of a functional • that
equals V plus a zero energy contribution due to the constraint Eqns (15). In this respect, the
functional to be minimized takes the following form:

(1)= 2
m=tn=1 i=lj=l

+ ~z.;
ln=l

Since the Lagrange multipliers vector, ~2}, represents the dynamic reaction of the point
supports, due to the constraints imposed, its k components 21, 22,..., 2k, have dimensions
of force. Upon performing the surface integration denoted in Eqn (17) yields:

- dr.., + {;.}7 [ E . , . ( x * , s*)] . (18)

8 m=ln=l m=ln=l

Equating to zero the derivative of @ with respect to {d,.. }, yields the stationary values of
as follows:
[Km.]{dm.}+[E.~.]7"'~2*}=O, (m = 1,2 ..... q; n = l , 2 , . . . , r ) , (19)
where {2"I = 41,;.}"L.,L~.
Equations (15) and (19) can be written in the following compact matrix form:
[/(]{O; = 0, (20)
where.
-[K,,] [0] [0] [Ell] T
[o] [K 2] [El ]r {d,,}
' {d,2}

[/(] = [0] [K,,,] [em,] r , {D}---, {dr..} ~.

[

[Kqr] [Eqr]7 {dqr} I

. d
[G,] [E..] [Gr] [0]
(21, 22)
Hence, the natural frequencies of a laminated cylindrical panel having an arbitrary number
of lateral surfaces point supports can be found as roots of a transcendental equation which
is produced by nullifying the determinant of the constrained dynamic stiffness matrix [/(].
It becomes apparent therefore that, upon employing a suitably large number, q x r, of
unconstrained vibration modes, these predictions can be obtained arbitrarily close to the
exact natural frequencies of the constrained cylindrical panel considered•
It is of interest to notice that employing {2*} = {0} is essentially equivalent to neglecting
the last column and the last row of the constrained dynamic stiffness matrix [/(] which then
occurs in a diagonal form. Since in that case it is det [/(] = d e t [ K l l ] d e t [ K ~ z ]
... det [Kq,], Eqn (20) provides the natural frequencies of a corresponding unconstrained
panel. There are cases that some of the frequencies of the unconstrained panel are also
frequencies of the corresponding constrained one (see, e.g. Refs [1, 4, 6])• Such solutions
may occur when all point supports involved were placed on the nodal line(s) of a vibration
mode of the unconstrained structure. Accordingly, any motions around such point supports
are due to the particular modal pattern and not to the point supports themselves which are
therefore considered as not "active" (see, for instance, Ref. [12], p. 236). Such trivial
solutions of Eqn {20) are therefore not discussed in the present study.
 Three-dimensional vibration of laminated composite plates 277

4. N U M E R I C A L EXAMPLES

In the example cases to follow, all point supports involved were assumed as suppressing
the transverse displacement of the lateral surface only and, therefore, as leaving the
corresponding in-plane displacement components unconstrained. For the results based on
the present three-dimensional analysis, two cases were always considered: (i) the panel has
one or more point supports which are located on the bottom lateral surface only; and
(ii) corresponding point supports are located on both lateral surfaces of the panel and are
symmetrically located with respect to the panel middle surface. To obtain satisfactorily
converged numerical results, the first 20 x 20 unconstrained harmonic modes appearing in
Eqn (2a) were used and retained coupled to represent the vibration modes of a correspond-
ing point supported panel.
For the purpose of comparison, the present analysis was initially restricted to cases
dealing with homogeneous isotropic plates (~b = 0) for which numerical results based on the
classical [12] and the Mindlin-type shear deformable [6] plate theory are available in the
literature. In this respect, the lowest value of the natural frequency parameter,

11 = ~o L 2 .v'Ph/D . (23)
of a centrally point-supported isotropic rectangular plate is presented in Table 1, for
different aspect, L , / L , , , and thickness to a typical-side-length, h/Lx, ratios. Here, D repres-
ents the bending rigidity of the plate.
Table 1 shows that for very thin plates, h/Lx = 0.001, frequencies based on the present
three-dimensional analysis are practically independent on whether a single point support is
used or a point support on each one of the lateral surface is employed. As was expected in
this case, two-dimensional theories provide with very good frequency results. The surprising
fact that the classical theory prediction [12] is better than the corresponding one based on
Mindlin theory [6], should be attributed to the "shear-locking" effect which is a well-known
draw-back of certain shear deformable theories and is always connected with very thin
plates and shells. Since, as is also well-known, results based on the classical plate theory are
independent of the plate thickness, Mindlin's theory frequency predictions [6] for thicker
plates (h/L,, = 0.1 and h/Lx = 0.2), were always closer to the frequencies obtained on the
basis of the present analysis. By increasing the plate thickness, however, the difference
between corresponding results based on three- and two-dimensional theories increases
substantially.
It is of interest to further notice that two-dimensional frequency results are closer to
corresponding three-dimensional frequencies obtained when point-supports are attached

Table 1. Lowest natural frequency parameter, f~, of isotropic rectangular plates (q~ = 0, v = 0.3)
with central point supports: (i) bottom surface, (ii) both lateral surfaces

h/L~ Theory L,/L~ = 1 I,~;L~ = 2 L,/L~ = 3

0.001
Present li! 52.346 22.797 15.298
Present 0il 52.349 22.798 15.299
Mindlin [6] 54.120
Classical [ 12] 52.640 22.80 15.40

0.1
Present t0 39.528 21.093 14.919
Present (ii) 42.548 (7.6%) 21.447 11,68%! 14.962 (0.3%)
Mindlin [6] 45.600
Classical [ 12] 52.640 22.80 15.40

0.2
Present Ill 23.811 16482 13.325
Present (ill 27.295 t14.6%1 17,784 17.9%) 13.656 (2.5%)
Mindlin [6] 34.048
Classical [12] 52.640 22.80 15.40
278 Jianqiao Ye and K. P. Soldatos

on both lateral surfaces. This observation justifies an earlier comment (see Introduction)
according to which, unless extremely thin structural elements are studied, two-dimensional
theories are essentially inappropriate in solving problems dealing with plates and shells
having point supports that are pinned on the lateral surfaces. In this respect, it is further
observed that the frequencies obtained when a point support is pinned on the bottom
surface only (higher flexibility) are always lower than the ones obtained when point
supports are pinned on both lateral surfaces (less flexible structure). To illustrate this, the
percentage increase of the values of the natural frequencies when a second point support is
introduced on the plate top surface is also given in parentheses. Apparently, these percent-
age values are increasing as either the thickness to the typical-side-length, h/Lx, or the side
aspect ratio, Ls/Lx, is decreasing. This later case is eventually equivalent to increasing the
alternative thickness of typical-side-length ratio, h/L~.
After the comparisons made for isotropic plates, the analysis was further applied in
connection with vibrations of point-supported laminated plates and circular cylindrical
panels. All symmetric and antisymmetric cross-ply lay-ups considered in what follows are
composed of orthotropic layers having the following material properties:
EL'Er = 10, GLr ET = 0.6, G r r / E r = 0.5, VLT = v71- = 0.25. (24)
For such laminates, Tables 2 4 show the lowest value of the natural frequency parameter,

(-i = ~'JLx \;/p/' ET • (25)

Table 2 shows the lowest natural frequency parameter of two-layered [0 °/90 ° ] laminated
panels having different number (k) of equally spaced point supports along one of the
diagonals. All panels have equal axial and circumferential lengths but the shallowness
angle, q~, varies. In a close connection with the results tabulated in Table 1, the percentage
increase of a frequency value, when corresponding point supports are also introduced on
the top surface, is again given in a parentheses. As was expected, natural frequencies increase
with increasing curvature (that is, with decreasing ~b). Moreover, the percentage values
shown in Table 2 suggest that upon increasing the panel curvature the difference between
cases (i) and (ii) is decreasing. Assuming, therefore, that an increase of the lowest frequency
is a beneficial effect, it is concluded that the panel configuration that takes most benefit by
introducing "type (ii)" point supports on both lateral surfaces is that of the flat plate. In this
respect, it is of interest to notice that the tabulated frequency parameter of a flat plate with
"'type (ii)" point supports is higher than the frequency parameter of a corresponding "type
(i)'" point supported panel having a shallowness angle gb = 0.5.
Similar remarks can be made from the results tabulated in Table 3, where corresponding
four-layered panels having a regular antisymmetric or symmetric lay-up are considered. For
such laminated panels having a central point support pinned either on the bottom or on
both lateral surfaces, lowest frequency parameters are tabulated in Table 3 for various
values of the shallowness angle 4~. It is of further interest to notice in this case, that all
frequencies shown in Table 3 are substantially higher than the corresponding ones
tabulated in the top half (central point supports) of Table 2. This is evidently due to the
considerable decrease in the [0~'/90 ]2 case, or the absence in the [0"/90°]s case, of the

Fable 2. Lowest natural frequency parameter, &, of two-layered [0~/90 '] cylindrical
panels ~h~L~ - 0.1. Lx L~ = 1) with different number, k, of point supports: (i) bottom
surface, (ii) both lateral surfaces

h 1.0 0.5 0.1 0 (platel

I (i) 1.762 1.638 1,487 1.482

2 (ii) 1.822 (3.4%) 1725 (5.3%) 1,650 (10.9%) 1.649 (11.3%)

2 lii 1.681 1.624 1,525 1.521

4 (6) 1773 (5.5%) 1719 15.8%) 1.667 (9.3%) 1.667 (9.6%)
 , , ,,. , ~ 1 '.?Dldtlt~ii ol ia{lltI~dtcd c o l l i p , ) s i t e p l a t e s 279

" I a b l e ~, I . o ~ ; c s l u : t l t : r a l ~ .~,u~r=,, ', p a r a m e t e r , e). ol t o m ia',crcd laminated cylindrical panels

Jl l . , = I ) l . 1, l, ~ .... :] .cai~r~d p o l l l i sdpporls ~ b o t l o m s u r f a c e . (ii) both lateral
~tll'la 2C~.

c, ' ! 0 (pla e)

[o ~ O l
7X9
Oii gtl? !.: ?;, s,}:: t l )°;,i 1806(11.7%}

(i) 7~{ 152~i 1.617

(ii) .874 i a; 0 0 . I 7X9, 0. " i l 1.784 ( 1 0 . 3 % )

[~tbie J. ! {!;~.cs( !lattlr,.tl I r e q u c n c y p a I a m L ' t c ! . ,,J, o1 three-la.'cered

[ } ~10 I~ I p a n e l s ,/_, L , = x 2,¢) = "tlwvhdiltercnl number, k, of
pOIIl! ~uO[!~Ii" Ik! b~HIOIll st.lrfi/cc. Hi} h o I h laleral surfaces

IlO:
(i r at,l< i~ ~)- 0464
,, >~; ~,Xi1< ~ 2 . , " , 1~484t4.0%i

£"~ s
2 - a i!, ~," ,I 599 I 1 7 6 % )

It 2{
11)
tl!' i 'V: 7 ~! i '~ ~,' 229~130.7%1

well-known coupling between bending and extenskm due to lamination (see, e.g. Ref. [10]).
This coupling, which is not present in laminates having a symmetric lay-up, results in
a substantial lowering of natural frequencies. For antisymmetric cross-ply lay-ups, its
strongest effect occurs always for tw, o - p l y l a m i n a t e s ~ S i n c e . however, in four-layered anti-
symmetric cross-ply laminates that coupling i~ considerably diminished, all frequencies
shown in the top half of Table 3 are as high as the corresponding ones given for symmetric
lay-ups. Under these considerations, it may be of interest to further notice that the
percentage increase of the frequency, due to the additional top-surface point support, is
always slightly lower tbr a symmetric than for a corresponding antisymmetric lay-up.
Table 4 shows corresponding lowest frequent) parameters lot certain three-layered
[ 0 ' / 9 0 / 0 ] semi-circular cylindrical panels !q5 :: ;r~. ti>r various values of thickness to
middle-surface radius ratio h.R, and for different nmnber (k) of point supports. The point
supports are equally spaced along the line x - i . , 2 and are again located either (i) on the
bottom or (ii) on both lateral surfaces. In a close agreement with the discussion followed the
results presented in Table 1: the percentage difference between corresponding "type (i)" and
"type (ii)'" frequency ~alues increases by increasing the thickness to middle-surface radius
ratio, h/R. Due to the high curvature value, however, that difference appears practically
negligible even for quite thick panels (h R = 0.1 or 021 having a single central point support
on their lateral surfaces (k = It. Nevertheless, upo!! increasing the number of point sup-
ports, that difference increases substantially. Heace. for k = 3, difference between corres-
ponding "'type {i)'" and "'type 0ii" frequency values be~or, aes as high as 30.7% for relatively
thick panels (h R = 0 2t. ~abile i~ ,:. ~c..m~l be cons~de~:d a,~ negligible (4%) even for very thin
280 lianqlao Ye and K, P. Soldatos

ones (h,R = 0.0l ~. This observation is considered as a further justification of the inappro-
priateness of two-dimensional theories in solving problems dealing with thin-walled struc-
tures having point supports that are pinned on their lateral surfaces.

5 ('ONCLUSIONS

Free ~.ibrat~ons of simply supported, cross-ply laminated plates and cylindrical panels
with arbitrarily located lateral surfaces point supports have been studied in this paper on
the basis of fully three-dimensional elasticity considerations. The analysis was based on
a recursive approach that was previously applied for a three-dimensional vibration analysis
of corresponding unconstrained plates and panels [9]. In the present paper, however, the
transfer matrix part of the approach employed in Ref. [9] was suitably converted into an
equivalent stiffness matrix formulation on which the reaction of the point supports was
further imposed by means of the Lagrange multipliers method.
In the particular case of isotropic rectangular plates, numerical results were obtained and
compared with corresponding results available in the literature on the basis of two-
dimensional plate theories. As far as very thin plates are concerned, a very good agreement
was observed between corresponding three-dimensional elasticity and two-dimensional
plate theory results. For thick or even moderately thick plates, however, substantial
differences were found, especially in cases with point supports pinned on one (bottom)
lateral plane only. These comparisons revealed that applying two-dimensional theories to
plates and shell panels having lateral surface point supports may cause significant errors.
This was not unexpected and is attributed to not only the approximations introduced by
two-dimensional plate theories, but mainly to the different manner they deal with lateral
surface point supports. Apparently, two-dimensional theories provide substantially higher
frequency predictions since they assume that a pinned support on one of the lateral surfaces
prevents the transverse displacement at all points of the corresponding normal to the plate
middle-plane.
Dealing with free vibration of point supported composite panels, further results were
obtained for laminated composite plates and curved panels having either antisymmetric or
symmetric cross-ply lay-ups. As was expected, for a fixed number of point supports, natural
frequencies increase with increasing the panel curvature. Evidently, higher frequencies were
predicted by locating point supports symmetrically with respect to the middle-surface, on
both lateral surfaces, than on one [bottom) lateral surface only. In such cases, the percentage
increase of frequencies due to the additional top-surface point supports was always slightly
lower for a symmetric than for a corresponding antisymmetric cross-ply lay-up. Moreover,
that increase of natural frequencies is more pronounced for flat plates, while it decreases
with increasing the curvature of a cylindrical panel. Nevertheless, that percentage difference
is also dependent on the number of point supports employed, in the sense that it increases
substantially with increasing the number of point supports.

Acknowled~¢eme~t Fh4~ work described in th~s paper was supported by an SERC Research Grant No.
GR, H42680

Rl F E R E N C E S

1 A. V Bapat and S Suryanarayan. Free vibrations of rectangular plates with interior point supports. J. Sound
l'ibr 134. 251 311 11989).
2. Y. Narita and N. lwato, Vibration studies for symmetrically laminated circular and elliptical plates resting on
elastic point supports. Compos. Sci. Tech. 39. 75-88 (1990).
3 K . M . kiew, A hybrid energy approach for vibrational modelling of laminated trapezoidal plates with point
supports. Int. J. Solids Struct. 29, 3087 3097 (1992).
4 L. A Bergman, J K. Hall, G. G. G. kueschen and D. M. McFarland, Dynamic Green's functions for Levy
plates. J. Sound Vibr. 162, 281 310 119931.
5 D. J. Gorman, Free-vibration analysis oi point-supported orthotropic plates. J. Enyng Mech. ASCE 120,
58 74 i1994).
6 K M Liew. Y Xiang and S. Kitipornchm. V~bration of Mindlin plates on point supports using constraint
functions ,/ t'm#n# Mech 4 S ( ' E 120, 499 '~13 11994).
 ]"brcc-dlmenslonal ~ibration of laminated composite plates 281

7. S.-C. Huang and B.-S. Hsu. Modal anal.~sis of a spinning Qhndrical-shell v, llh interior-point or circular line
supports. J. Vibr..4¢ous~ 4 S M E 115. 535 543(1993).
8. K. P. Soldatos and V P. Hadjigeorgiou. Three-dimensional solution of the free vibration problem of
homogeneous isotropic <,lindrical shells and panels. J. Sound Vibr. 137. 369 384 (1990).
9. J. Q. Ye and K. P. Soldmos. Three-dimensional vibrations of laminated cylinders and cylindrical panels with
a symmetric or an antisymmetric cross-ply lay-up. Compos. Enqnc] 4, 429 444/19941.
10. R. M. Jones, Mechanics ,,~ Comp,~site Materials. Hemishere. New York 11975).
I 1. K. P. Soldatos and I. D Ha~kes. An iterative three-dimensional solution of the free vibration problem of
homogeneous orthotropic holl,w, cylinders, in Structural D y n a m k s iedited by W. B. KrS.tzig et al.). Vol. 2. p p
883 889. Balkema. Rotterdam (!991L
12. D. J. Gorman. Free Vihr,mon b m h s 1 ~ ot Rectongular Ptate~ Elsevier. Nev, York (19821.

&PPEND1X

The nonzero elements ol the t~, ~ ¢, mamces [ t j and [BI appearing in Eqns ~5~ are as follows:

,4~ : ( I, = ~( ~, 4~:: (¢~ R~. 4~ =(~5. A35=flCss '

t 4, = t , , 4 , : 1.
IAI)

• - , 5 5" . _.

where :x - mr~ l , . l: ," ,~,.t R. R h2andR~ : R ~ h2aretheradilofthebottomandtopsurfaces

of the panels, respe,~:~ ',