Beruflich Dokumente
Kultur Dokumente
257-275, 1995
Copyright © 1995 Eiaevies Science Ltd
Pergamon Printed in Great Britain. All rights reserved
0961-9526/95 $9.50+ .00
0961-9526(94) 00096-4
Abstraet--A set of analytic solutions is presented to predict the dynamic response of simply-
supported, doubly curved, cross-ply laminated shells impacted by a solid striker, The solutions are
based on a higher-order shear deformation theory (HSDT) which accounts for the parabolic
distribution of transverse shear strain through the thickness of a shell and tangential stress-free
boundary conditions on the surface of the shell. An analytic impact force function recently
proposed by the authors is used to predict the contact force between the striker and the shell and
this is incorporated into the solutions. Fundamental frequencies of cross-ply laminated spherical
and cylindrical shells are calculated using the present solutions and the results are compared with
those published by others. In terms of the characterising coefficients, the solutions can be reduced
to versions of the first-order shear deformation theory (FSDT) and the classical shell theory
(CST). A comparison of the solutions based on the various theories (HSDT, FSDT and CST) is
also made.
NOMENCLATURE
Au, B,./, D~, E,i, F,s, H,~ laminate stiffness defined in eqn (4)
F concentrated load
contact force
Fc* equivalent contact force
KI equivalent stiffness of a simply-supported shell
contact stiffness
/ct contact stiffness
Ni, M i , Pi stress resultants defined in eqns (2)
Ql, Q2, S1, S2 stress resultants defined in eqns (2)
R~, R z principal radii of shell curvature
T contact duration
V impact velocity
a, b, h shell arc lengths along the x 1- and xz-directions and shell thickness
k4, k5 shear correction factors
m1 mass of shell
m2 mass of striker
P material constant describing contact stiffness
q. surface load component along radial z-axis
t time
II, O, W displacements along the xm-axis, xz-axis and radial z-axis
Xl , X2, Z orthogonal curvilinear coordinates for shell
xlp, Xzp coordinates of impact point
wz radial displacement of striker
1. INTRODUCTION
(1981) used the first-order shear deformation theory (FSDT) developed by Whitney and
Pagano (1970) to analyze a simply-supported, specially orthotropic plate subjected to
central point impact. Ramkumar and Thakar (1987) used Donnell's equations based on
classical shell theory (CST) to analyse cylindrical thin shells and presented the response of
large radius cylindrical laminated shells subjected to low velocity impact. They assumed
that impact forces vary linearly with time and followed Sun and Dobyns' procedure to
determine the response. Christoforou and Swanson (1990) used the FSDT developed by
Bert and Birman (1988) for cylindrical shells to analyze simply-supported orthotropic
cylindrical shells subjected to impact and presented an anlytic solution to the problem. In
their solution, the deceleration of the impacting mass was used to estimate the impact
force, and local contact deformation between the striker and shell during impact was
neglected. This solution was confined to the response when the impactor is in contact with
the shell. Gong et al. (1994) adopted Christoforou and Swanson's procedure and under-
took an analysis of laminated open cylindrical shells subjected to impact loading. In their
analysis, contact deformation was considered and an analytic function to describe it was
proposed and incoporated into the analysis.
Classical shell theory assumes that straight lines normal to the undeformed middle
surface remain straight and normal to the deformed middle surface. This leads to neglect
of transverse shear strains. It h~s been reported that application of this theory to layered
anisotropic composite shells can result in errors of as much as 30°7o or more in terms
of deflections, stresses and frequencies (Reddy and Liu, 1985). The first-order shear
deformation theory assumes a uniform distribution of transverse shear strain through the
shell thickness, which does not satisfy the condition of zero transverse shear stress at the
top and bottom surfaces of the shell. The higher-order shear deformation theory (HSDT)
accounts not only for transverse shear strains, but also for a parabolic variation of
transverse shear strain through the shell thickness. This paper presented an analysis based
on a higher-order shear deformation theory to predict the dynamic response of simply-
supported, doubly curved, cross-ply laminated shells excited by impact. The higher-order
shear deformation theory developed by Reddy and Liu (1985) is employed in this analysis.
The load, displacements, rotations, as well as strains, are approximated by double Fourier
series expansions that satisfy simply-supported boundary conditions. Fourier series
solutions are assumed to be separable into functions of time and position. By neglecting
in-surface and rotary inertia, the problem is reduced to a second-order ordinary differen-
tial equation with respect to time for the Fourier coefficients for radial deflection.
Solutions for the impact response of the shell can be obtained by invoking a convolution
integral. The contact force between the striker and the shell is calculated using an analytic
function recently presented by the authors (Gong e t al., 1994). This force function is
substituted into the convolution integral and consequently, the integral equation of
motion of the impacted shell can be solvcd analytically.
The present analysis furthers previous work in three areas: first, a comparison among
the solutions based on the three theories--HSDT, FSDT and CST--is made for the
problem of laminated shells impacted by a solid striker; secondly, the analysis is applic-
able to doubly curved shells; and thirdly, the present solutions facilitate a study of the
dynamic response of shells excited by impact for a total time history (0 < t < oo).
2. GOVERNINGEQUATIONS
Consider a doubly curved, cross-ply laminated shell shown in Fig. 1. The shell
coordinates (orthogonal curvilinear coordinates) are denoted by Xl, x2 and z which
correspond, respectively, to two lines defining the curvature of the middle surface (z = 0)
and the normal to it. The principal radii of curvature of the middle surface are denoted
b y R 1 and R 2 . The shell is simply-supported along its four edges of arc lengths a and b in
the x r and x2-directions, respectively. The sign convention for the stress components
using reduced subscripts is also illustrated in Fig. 1. The equations of motion for the
Impact response of laminated shells 259
Z,W
Z x2 % G2
x1
6
- O~i,
0NI
ax~ + ~0N6 + ,to R--~
Qt : ~ / ~ ..{_ ~,/~1 -- `t2/3 0X 1 (la)
(lb)
OXl + ~x2 + `t°-~2 Ox2
OQI OQ2 ]los I OS2"~ -~ ( 02PI 02p2 + 2 02P6 "~ NI N2
Ox----;+ ~ -- `tl~x 1 -I-OX2/ + 0Xl2 + 0~22 OX1 0X2/ R1 R2
-- O~ 0~1 -- OiJ 0~2 `t2l z 7 1/02W
---- `t2 I3~XI + Is-~xI + I3-~X2 + IS-~x 2 -~ \Ox2 OX2j j + 1li~' -- qn (lc)
where Ni, M i, Pi, Q1, Q2, $1, $2 are stress resultants; ill, f12 are slopes generated by
flexure in the x : z and x2-z planes; and I i , I[ are inertia factors defined in Appendix A.
Also,
(QI, s 0 =
k=l f zk
Zk-- 1
astk)(1, Z2) dz (2b)
£=1 f ~* a(k)(1, z 2) dz
Zk-I
(2c)
260 S.W. Gong et al.
L is the number of layers in the shell. Equations (1) are general and can be modified to
describe flat plates, cylindrical shells and spherical shells by assigning appropriate values
to the curvatures 1/R1 and 1/R2. If 20 = O, 2~ = 4/h 2 and 22 = 1, eqns (1) correspond
to the HSDT (Reddy, 1984; Reddy and Liu, 1985). The case where 2o = 1, 21 = 22 = 0
corresponds to the version of the FSDT developed by Whitney and Pagano (1970)
and Dong and Tso (1972). If a value of 20 = 21 = 22 = 0 is used in conjunction with
fll = --OW/OXI and f12 = -aw/Ox2, eqns (1) reduce to the version of the CST defined by
Donnell (1933).
The relationships between the stress resultants and the strains induced are described
by the following equations (Reddy and Liu, 1985):
(Aij, Bij, L)ij, Eij, fij, Hij) = EL I Zk Qij(k) ( l , z , z 2 ,Z3, Z4, Z6) dz (i,j= 1,2,4,5,6)
k= I Zk_1
(4)
Q~k) are reduced stiffness of the kth lamina in the shell coordinate system.
Strains associated with the deflections and rotations are given by Reddy and
Liu (1985):
~, = ~o + z(,<o + zOr5
e2 = : + z ( : + z2r ~'2,
~3 = 0
(5)
~4 = ~0 -I- Z2K 1
aw u
: = B, + ~ - 2o~ x~ = - 2 ~ ~+
av au a#2 afl~
,<°:N+ N ,@x, + ~ + 2 ax-~U
Impact response of laminated shells 261
By substituting eqns (3), in conjunction with eqns (6), into the governing eqns (1), the
equations o f motion o f the shell can be expressed in terms of deflections and rotations
using the H S D T and FSDT:
where [ ] and [ } represent matrices and vectors, respectively, and the differential operators
L o for the HSDT, FSDT and CST are given, respectively, in Appendix A.
3. CLOSED-FORM SOLUTIONS
We consider Navier solutions to eqns (7) and (8) for the shell shown in Fig. 1. The
simply-supported boundary conditions are:
a
at X1 = + - - v = w= N, = M, = & = & = O when the HSDT is used
-2
v= w= N, = M, = #2 = O when the FSDT is used (9a)
b
at x2=--- 2 u=w=N2=M2=P2=Pl=O when the HSDT is used
The loading qn(Xl, X2, X2, t) is assumed representable by a double Fourier series
expansion, separable into functions of time and position as follows:
® ® m 7/x1 177[X 2
q n ( X l , X = , t) = ~ ~ Omn(t) cos cos-- (10)
m=l n=l a b
For a concentrated (impact) load F(t) applied at the point (xtp, x2p), Qmn(t) is given by:
4F(t) nnx2p
Qm~(t) = - ~ cos mnXIP cos m = 1,3,5, n = 1 3,5, (11)
a b "'" ....
and the solutions to eqn (7) that satisfy the boundary conditions (9a) and (9b) are sought
via the following:
00 00 /'/7[X2
w(xl, x2, t) = ~ ~ Wm,(t) COSmrtxl cos (12a)
ra=l n=l a b
co oo
u(x,, x2, t) = ~ ~ Um,(t) sin mlrx, cos -nrtx
- 2 (12b)
m=l n=l a b
~1 ~2 ~3 ~4 ~5~ I Umn(t) 0
~2 ~3 724 ~5 Vmn(t) 0
Win(t) Phfi/m(t) - Qm(t) 03)
7,4 ~5 Xmn(t) 0
sym ~5. Tmn(t) 0
JVm.(t) = 0 (14)
sym T33 J ~. Wren(t) J ph W(t) - Qmn(t)
in the case of the CST. The coefficients T0 used in the HSDT, FSDT and CST are given
in Appendix A.
Equations (13) and (14) can be reduced to a single linear second-order differential
equation:
Wm~(t) = Km~[phf~'m~(t) - Qm~(t)] (15a)
or
Qmn(t)
((Vm.(t) + ¢.02mnWmn(t) = - - (15b)
ph
2 m
1 (16)
(,Omn
Km,,Ph
and
~3T
Kmn = det (17)
W,nn(t) = C°mnP--'
1 -"~~iJot Qmn(r) sin[COm.(t - r)] dr (18)
mTtxip nTtx2p I t
Wren(t) = cos a cos--7 F(r)[sin COmn(t - ~')] dr (19)
OJranml ,JO
where the coefficients Ku, Kv, Kx and Ky are given by Bert and Birman (1988):
Impact response of laminatedshells 263
Ku = 0/1 + °t2Kx + 0/3Ky Kv "~ 0/4 "l- 0/5Kx + 0/6Ky
Kx = A,,/A Ky = Ay/A
a~ = (T~5 + T~50/1 + T250/.)(T45 + T~40/~ + T240/6)
- (T34 + T~40/1 + Tu0/4)(T55 + T~50/~ + Tz50/6)
0~1 ~-~ (T12 T23 - T22 T13)/0/ 0l2 = (TI2 T24 - T22 T14)/0/
Substituting (12a)-(12e), in conjunction with (20), into (5) yields expressions for
strain in the shell:
m=°° ~o [ a 2,z3( _~)]mn mnxl nnxz
eI = ~ ]~ g u + + zK x - - Kx- Wmn(t )cOs cos--
=1 " = 1 ~ 3 --~ a b
(24a)
(~4 = =1 n~= l b a b
E6 = m = ! . E 1 "~Kv u ,-: y +
+ -321z3(mn
k---~-Ky + --ff
nn K x - 2 mnn__~_ff)]
a Wm.(t) sin m~x, sin mnx,
-b (24e)
The expressions for strain using the CST can be deduced from (24a)-(24e) by setting
mn nn
Kx = a and Ky = --.b
By substituting (24a)-(24e) into the stress-strain relations for the kth lamina, the
stress distributions of the shells can be predicted:
[O'i](k ) = [Q(k)]lc, i](k ) (25)
264 S.W. Gong et al.
where Q~) are the transformed stiffness constants of the kth lamina in the shell coordinate
system. In eqn (25) the transverse normal stress is given by:
~(k) e + ~(k)~ + ~(k)~
0"3 = "~13 1 'k~23 2 ~36 ~6
and may cause delamination in composite laminated shells impacted by a striker (Wu and
Chang, 1989); therefore its magnitude and sign should be examined with respect to the
strength of inter-ply adhesion.
4. IMPACT RESPONSE
K 1 = co~1m* (26)
v@m2
Boundaries:
Simply supported
W2
K2 = K (w2-w)
,.--]
l½ K1
W
r b
• K1
\ \ \ \ \ \ \ \ \ \ \\N.
:(b) (c)
Fig. 2. Spring-mass model.
Impact response of laminated shells 265
where K 2 and p are material constants which can be obtained by static indentation tests
(Tan and Sun, 1985); w(t) and w2(t) are the radial displacements of the load point of the
shell and the striker, respectively. By linearizing the Hertzian contact force-deformation
relationship, an equivalent contact force function may be expressed in terms of an
effective stiffness K~ as follows (Gong et al., 1994):
where
091,2 = ~ \ m• + T +--
m2/ %1 4 \ m* m2/ mtm2
C1=
K~ - 0921m2
c2=
K~' - 092m2 (29)
V
AI=
( O l ( C 2 -- C1)
V
A2 =
092(C - C2)
T is the contact duration and the effective stiffness, K~', can be obtained by dynamic
experiments based on the principle of mechanical impedance (Moriwaki, 1978; Wu e t a l . ,
1983), and can also be estimated by Gong et al. (1994):
where F() is the Gamma function, and t~m the maximum contact deformation. For a
striker of small mass (m I > 10mE), t~m can be stimated by (Greszcuk, 1982):
=( m l m 2 ~2/5(5V2~2/5
Om krnl + m2/ \ 4 K 2 / (31)
Wmn(t) = - - 4 K ~ ' cos rnztXlP o nnx2v ~AI(C1 - 1) [091 sin(09mnt) - 09m.(sin 09t t)]
09_ml a c S--b-- (09 _ --2--09m.
A2(C2 - 1)
+ 09~2~ 09mn--72-[092sin(09mnt) - 09ran sin(092 t)]] 0 < t < T (32a)
By combining (32a), (32b), (20), (12a)-(12e), (24a)-(24e) and (25), the displacements,
strains and stresses in the impacted shell can be evaluated.
266 S. W. Gong et al.
In the present analysis, in-surface and rotary inertia are assumed to be negligible to
simplify the solution procedure. In order to verify this assumption, nondimensionalised
fundamental frequencies calculated by eqn (16) are compared with solutions for the free
vibration of shells given by Reddy and Liu (1985). Table 1 shows results for [O/90/0]
cross-ply laminated cylindrical shells, while Table 2 is for [O/90/0] cross-ply laminated
spherical shells. The nondimensionalised fundamental frequency used in Tables 1 and 2
was defined by Reddy and Liu (1985). The material properties are (Reddy and Liu, 1985):
El = 25E,, G,, = G,, = 0.5E,, G,, = 0.2Ez, v12 = v13 = 0.25, p = 1 and the shear
correction factors K~ and K~ are assumed to be 5/6. It is seen from Tables 1 and 2 that the
present analysis neglects in-surface and rotary inertia, and so gives slightly higher values
than Reddy’s solutions for most of the fundamental frequencies. However, the present
predicted frequencies differ from Reddy’s exact solutions by 1% at most. This indicates
that the effects of in-surface and rotary inertia can be assumed to be negligible.
The impact response of a [O/90/0] cross-ply graphite/epoxy spherical shell is studied
using the HSDT, FSDT and CST, respectively. The shell has a radius of 0.5 m, arc
lengths of 0.14 m for the four edges and a thickness of 3.81 mm. The material properties
of the shell are identical for all layers (AS/3501-6 graphite/epoxy) (Sun and Liou, 1989):
El= 142.73GPa,E2= 13.79GPa,G,,=4.64GPa,G,3=4.14GPa,~,,=0.3,~,3=0.28,
p = 1610 kg/m3 and each ply has a thickness of 1.27 mm. The shear correction factors
K~ and K~ are assigned a value of n2/12 introduced by Mindlin (195 1). A blunt-ended
T a b l e 3. C o n v e r g e n c e o f series solution f o r n o r m a l d e f l e c t i o n
cylindrical steel striker with a length o f 25.4 mm, a diameter of 9.525 mm and a mass o f
0.0142 kg is used in this study. The dimensions and mass are the same as that used by
Sun and Liou (1989); accordingly, the values o f p and K2 are taken from the data given
by Sun and Liou (1989) and have values o f 1.5 and 100 M N / m LS, respectively. Impact
velocity is assumed to be 6 m/s. Unless otherwise stated, these geometrical and material
properties, as well as impact conditions, are common to all cases.
Convergence of the present solutions is first investigated. Since all solutions for
deflections and stresses can be expressed as functions of Wren(t), eqn (12a), in conjunction
with eqn (32), is chosen to examine convergence. Table 3 shows the convergence of
eqn (12a) in conjunction with eqn (32) for calculation of the deflection w at x~ = x2 = 0
and t = 120/as. It is observed that the double Fourier series expansion for normal
deflection tends to converge at m = n = 30; the predicted deflection for m, n = 1, 3 ... 30
differs from that for m, n = 1, 3 ... 200 by 4°70 at most. In the present study, 180 terms
(m = n -- 180) were evaluated for deflections and stresses.
The contact force during impact is determined from the impact force function (28)
which is governed by five parameters K~, KJ', m~, m 2 and V. Only the equivalent stiffness
K~ ( = t o ~ m * ) depends on the fundamental frequency of the shell, which in turn is
dictated by the theory employed (HSDT, FSDT and CST). In this analysis, the difference
between K~ values determined from the three theories is small--less than 10070. Figure 3
shows that the resulting contact force during impact, corresponding to HSDT, FSDT and
CST, are identical, confirming the negligible influence of K 1 . In terms of the other param-
eters, contact force magnitude is most sensitive to impact energy, which is governed by the
14
1
HSDT a/b = 1
% 1 F S D T h/a = 0.05
X
oJ
2
I I I I
0
10 20 30 40 50
El/E2
Fig. 3. C o n t a c t f o r c e b e t w e e n striker a n d spherical shell d u r i n g i m p a c t , V = 6 m / s .
268 S . W . Gong et al.
1.5
LU
ot r
,9
oo
1.0
0.5
0.0
0
=•j 100 200
Time (i~s)
HSDT
FSDT
CST
t
300
,
400
Fig. 4. Variation of maximum central deflection with E=/E 2 for spherical shell.
mass and velocity of the striker, m 2 and V, while the contact duration is most sensitive to
contact stiffness between the striker and the shell, K~'. The effects of material anisotropy
and the thickness-to-lateral dimension ratio (h/a or h/b) on the central deflection of the
impacted shell are examined using the HSDT, FSDT and CST, respectively. A nondimen-
sionalised central deflection ~, = Wma~h3E2/Fmaxab is used to eliminate the influence of
actual shell dimensions. This nondimensional quantity is based on the fact that the
maximum central deflection of an isotropic plate subjected to a static central load F, is
proportional to Fsab/Eh 3 (Timoshenko et al., 1959). Figure 4 illustrates the nondimen-
sionalised maximum central deflection of the spherical shell during impact for various
ratios of moduli E t / E 2 . Deflections calculated using the HSDT and FSDT are practically
identical, whereas the deflection calculated using the CST differs significantly. As the
ratio E~/E 2 increases, the difference between the deflections predicted by the HSDT
(or FSDT) and the CST increases. Clearly the larger the ratio E J E 2 , the more anisotropic
the laminated shell is. According to Whitney and Pagano (1970), severity of shear
deformation effects depends on material anisotropy of the layers. As the ratio E I / E 2
increases, the transverse shear deformation becomes greater. The CST does not account
for this effect on the deflections; as a result, it predicts a smaller deflection than the
HSDT and FSDT. Severity of shear deformation effects also depends on the ratio h/a
(thickness to arc length along the side of the open shell), since the transverse shear
stresses, obtained from equilibrium conditions, are of an order h/a times the bending
stresses (Reddy and Liu, 1985). Figure 5 depicts the variation of the nondimensionalised
maximum central deflection with the value of h/a for the spherical shell. The deflections
calculated using the HSDT and FSDT increase non-linearly with the ratio h/a, the
increase becoming steeper with h/a; whereas, deflections calculated using the CST remain
almost unchanged. The fact that deflections vary with h/a because transverse shear
deformation increases is reflected in the results derived using the HSDT and FSDT.
40 - - HSDT ,,,,-'"" 1
........ FSDT
~o 3o ..... CST
x
m~ ~_~=20
E
• 10
0 •
0.2 0.2 0.3 0.4
h/a
-0.25 -0.25
-0.50
- .0 -0.5 0.0 0.5 1.0 -0.50
°.
.0 -0.5 0.0 0.5 1.0
o,/lo, I G2/l~2maxI
(a) (b)
Fig. 6. Normal stress distribution through the shell thickness at impact point (x~ = 0, x 2 = 0),
120ps after impact.
However, the CST neglects the effects of transverse shear deformation and therefore
underestimates the deflections. It is also observed from Fig. 5 that as h / a increases, the
deflection calculated using the FSDT increases more steeply than that calculated using
the HSDT. The difference between deflections predicted by the H S D T and FSDT exceeds
10070 as h / a becomes larger than 0.2. This is explained by the fact that the FSDT under-
predicts the fundamental natural frequencies as the ratio h / a increases (Reddy and Liu,
1985), which results in an overestimation o f deflections for shells with larger h / a ratios.
From Figs 3-5, it is seen that all three theories agree in their prediction o f contact force;
both H S D T and FSDT produce similar results for deflection when the ratio h / a is less
than 0.2; and usefulness o f the CST is restricted to rough estimation of deflection for
quasi-isotropic (E~/E2 = 1) shells with very small h / a ratio ( < 0.05).
Figure 6 shows the distribution o f normal stresses tr~ and 02 through the thickness at
the impact centre (Xl = 0, x z = 0) at t = 120gs after commencement o f impact, when the
impact force reaches its peak value (see Fig. 3). Since the reduced stiffness Q,~k)is different
for each layer of the laminated shell, there are stress discontinuities at the interface o f
the laminae; also, the stress gradients in adjoining laminae are different. The variation
of normal stresses through the shell thickness is almost linear; the magnitudes of the
compressive stresses above the mid-surface. This is a result of the superposition o f
bending and compressive membrane stresses. Since the HSDT incorporates third-order
terms of the thickness variable z into the expressions for normal stresses, a small differ-
ence (less than 10°70 for the maximum stress) between results derived using the H S D T and
the other two theories is observed. However, all three theories yield similar descriptions
o f the normal stress distribution through the shell thickness. Hence, instead of employing
the more complex HSDT, both the simpler FSDT and CST may be used as a compromise
between accuracy and ease of analysis to predict normal stresses. Figure 7 shows the
distribution of transverse shear stresses a4 and (75 through the thickness at x~ = 5 mm,
x2 = 5 mm, 120gs after commencement of impact. The point examined (x~ = 5 mm,
x2 = 5 mm) is chosen because maximum values of a4 and (r5 occur near the edge of
the impact area (abour 5 mm from impact centre x~ = 0, x 2 = 0) and the impactor is
a 9.5 mm diameter cylindrical striker. The HSDT produces a parabolic distribution o f
transverse shear stresses through each ply o f the shell, with the stresses vanishing at
the top and bottom surface. The FSDT yields a uniform distribution (as expected) of
transverse shear stresses through each ply. (Note that this does not satisfy tangential
stress-free boundary conditions on the surfaces o f the shell.) The maximum stress
predicted by the FSDT is 20070 less than that predicted by the HSDT. Transverse shear
stresses predicted by the FSDT differ significantly from those predicted by the H S D T
because the FSDT assumes a constant transverse deformation through the thickness.
It is therefore obvious that transverse shear stresses (r4 and tr 5 computed using the H S D T
are more realistic and accurate.
270 S.W. Gong et al.
Figures 8 and 9 show the variation of normal stresses with time at the point of impact
(x 1 = 0, x2 = 0) while Figs 10 and 11 depict the transverse shear stress response at
xl = 5 ram, x2 = 5 mm. It can be seen that the stress-time curves obtained using the
HSDT and FSDT resemble each other closely, although the maximum stress during
impact predicted using the FSDT is 20°70 smaller than that predicted using the HSDT,
as evident also in Fig. 7. The normal and transverse shear stress histories in Figs 8-11
are examined in relation to the contact force in Fig. 3. During contact (0 < t < T),
normal and shear stresses vary according to the contact force and increase during loading
0.50 - - HDST
--HSDT
0.50
FSDT
0.25
0.25
0.00 t-
"~ 0.00
.,
-0.25
-0.25
• J
-0.50
1.2 -0.50 I
0.0 0.2 0.4 0.6 0,8 1.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
/l%maxl
a5/l~smaxl
(a)
(b)
Fig. 7. Transverse shear stress distribution through the shell thickness at x~ = 5 ram, x 2 = 5 ram,
120 # s after impact.
250
20O - - HSDT
• ~
150
(M 100
o- 5o
o
v 0
6 -5o
i i i t i
-100
0 100 200 300 400 500 600
Time (l,s)
Fig. 8. Variation of normal stress a~ at impact point (x~ = 0, x 2 = 0) with time, V = 6 m/s.
60
50 - - HSDT
£L 40
v
30
20
o 10
0
C
-10
-20 i I i I , I , I , i ,
Fig. 9. Variation o f normal stress az at impact point (x~ = O, x 2 = O) with time, V = 6 m/s.
I m p a c t r e s p o n s e o f l a m i n a t e d shells 271
10
--- HSDT
8 . . . . . . FSDT
~Z 2
-2 , I i I I i I i I i
10
-~ HSDT
8
~" ~ , _ ' ~ FSD1
12. 6
-2 i i i i i
(0 < t < T / 2 ) and decrease thereafter ( T / 2 < t < T ) . Maximum values occur at t = T / 2 ,
when the contact force reaches its peak value. This indicates that during contact
(0 < t < T) the stresses are governed by the contact force. After separation of the striker
from the shell, the stresses decay rapidly as the shell undergoes free vibration. The largest
normal stress induced after loss o f contact (t > T) is about one-third the maximum value
during contact (0 < t < T), while the largest shear stress following loss o f contact is only
about one-seventh the maximum values induced during contact. Hence, the maximum
stresses generated occur during contact. This result shows that the contact force history
and its relation to stress magnitudes must be monitored to determine the occurrence of
maximum stresses in an impacted shell.
6. C O N C L U S I O N S
A set of analytic solutions has been presented to predict the displacement, strain
and stress responses of simply-supported, cross-ply laminated shells o f double curvature
excited by impact. In terms of the characterising coefficients, the solutions can be
classified as versions of HSDT, FSDT and CST. Results show that the HSDT and FSDT
yield similar predictions for shell deflection when the ratio h / a is less than 0.2. Use of CST
produces acceptable estimations of deflection only for quasi-isotropic shells ( E I / E 2 = l)
with a very small h / a ratio ( < 0.05). The HSDT provides a more reasonable description
o f stresses in an impacted shell because it accounts for the parabolic variation of trans-
verse shear strain through the thickness. However, the FSDT and CST can be employed
in place o f the more complex HSDT to predict normal stresses and their use corresponds
to a compromise between accuracy and ease of analysis. Results also show that the normal
and transverse shear stress response to impact are governed by the magnitude of the
contact force during contact between the striker and the shell; maximum values of these
stresses occur approximately half-way through the contact duration.
272 S. W. Gong et al.
REFERENCES
Bert, C. W. and Birman, V. (1988). Parametric instability of thick, orthotropic, circular cylindrical shells. Acta
Mech. 71, 67-76.
Christoforou, A. P. and Swanson, S. R. (1990). Analysis of simply-supported orthotropic cylindrical shells
subject to lateral impact loads. A S M E Trans. J. AppL Mech. 57, 376-382.
Dobyns, A. L. (1981). Analysis of simply-supported orthotropic plates subject to static and dynamic loads.
A I A A J. 19, 642-650.
Dong, S. B. and Tso, F. K. W. (1972). On a laminated orthotropic shell theory including transverse shear
deformation. A S M E Trans. J. Appl. Mech. 39, 1091-1096.
Donneil, L. H. (1933). Stability of thin walled tubes under torsion. N.A.C.A. Report. 479.
Gong, S. W., Toh, S. L. and Shim, V. P. W. (1994). The elastic response of orthotropic laminated cylindrical
shells to low velocity impact. Compos. Engng 4(2), 247-266.
Greszczuk, L. B. (1982). Impact Dynamics. John Wiley, New York.
Mindlin, R. D. (1951). Influence of rotary inertia and shear on flexural motions of isotropic, elastic plate.
A S M E Trans. J. Appl. Mech. 18, 31-38.
Moriwaki, T. (1978). Optimizing dynamic force in shock excitation testing. SME, Dearborn, MI, pp. 427-434.
Ramkumar, R. L. and Thakar, Y. R. (1987). Dynamic response of curved laminated plates subjected to low
velocity impact. A S M E Trans. J. Engng Mater. Tech. 109, 67-71.
Reddy, J. N. (1984). A simple higher-order theory for laminated composite plates. A S M E Trans. J. Appl. Mech.
51, 745-752.
Reddy, J. N. and Liu, C. F. (1985). A higher-order shear deformation theory of laminated elastic shells. Int. J.
Engng Sci. 23(3), 319-330.
Sun, C. T. and Chattopadhyay, S. (1975). Dynamic response of anisotropic laminated plates under initial stress
due to impact of a mass. A S M E Trans. J. Appl. Mech. 42, 693-698.
Sun, C. T. and Liou, W. J. (1989). Investigation of laminated composite plates under impact dynamic loading
using a three-dimensional hybrid stress finite element method. Comput. Struct. 33, 879-884.
Tan, T. M. and Sun, C. T. (1985). Use of statical indentation laws in the impact analysis of laminated composite
plates. A S M E Trans J. AppL Mech. 52, 6-12.
Timoshenko, S. et al. (1959). Theory o f Plates and Shells. McGraw-Hill, New York.
Whitney, J. M. and Pagano, N. J. (1970). Shear deformation in heterogeneous anisotropic plates. A S M E Trans.
J. AppL Mech. 37, 1031-1026.
Wu, H.-Y. T. and Chang, F.-K. (1989). Transient dynamic analysis of laminated composite plates subjected to
transverse impact. Comput. Struct. 31, 453-466.
Wu, P. J. et al. (1983). Investigation of characteristics of the hammer tip and the force impulse signal. J. Shock
and Vibration (China) 6, 58-69.
APPENDIX A
(1) The inertia factors ~ and ~' are defined by (Reddy and Liu, 1985):
~ =I1 + 2 ~ ° I
R, 2
- 23"0 I
11' = It + - -
Rz 2
~ = I2 + ~l°13 - ~ 1 4 - /l, I 5
3"oI A, I A1 1
= 12 + R2 3 - - ' ~ 4-- 3R2 5
+ 3R, 5
(Al)
2A'1 )'~I
~=I~- T ~+~-
~7
( I , , 12, I 3 , 1 4 , 1 5 , / 7 ) = ~
Lf k
k=l , Zz_ i
ptk)(l,Z, z2,Z 3,Z4,Z 6)dz
where A0 = 1 and A~ = 4/h 2 for the HSDT, and 20 = ~.~ = 0 for the FSDT and CST.
Impact response of laminated shells 273
(2) Differential operators L o for the HSDT and FSDT in eqn (7) are:
032 032 _ d2
LII = All ~X12 + A~bx'-"~2- Ii-.
~
0=
02 A1 ( E l 2 + 02
L~5 = LsI = (Blz + B66)~X 1 ~3x2 3 E66) 0X1 0X2
02 032 _ d2
L ~ = A ~ 0--~1
~ + A=~ a--~~ - 1,'d-~
2X2 04 _ (
- T(H12 + 2H~)~ A, m 5 5 -- ,~.1F55 2Etl "~tZ~ °32
3R 1 3Rz/IOxZt
_ d2
+ 2A,D55 - A~F55 - 14-~
~2 [ 2~.1 A~ 1 02
L.~ = L~. = ~D,~ + n~6) o - - T ~~ - [ T W , ~ + ~.) - ~(.,~ + ,%) Ox, o~2
~2 ~2 ~1
L55=D66~x21+D22_~x _]c4A44_T(2F66 )tLIH ~ 2 ~t2,F22 ,~,IH ~ 2
3 W ox~, T ~V
_ d2
+ 221D44 - A~F44 - 14d-~
COE5:3-0
274 S . W . Gong et al.
a' o /e',
L,3 =-L3I =-J~lla'~Xl3 (Bl2 + 2 J ~ 6 6 ) ~ +
R 1 + R2//{9X1+ 12.~xit.~_.~
kCAll )
82 82 d2
(A3)
0' O' //1,2 A22~ 0 0 /d2\
Lzs = -Ls2 = -B22"ff~X~ - (B,2 + 2Bef~)~ + t-~l + R2/'~-Z2x2 + I2~x2 t ' ~ )
a4 {94 ~4
L,, = - 0 , , ~xl4 - 2(0,2 + 2066) ~ - 022 {9"-'4~2
(3) The coefficients T0 for the HSDT and FSDT in eqn (13) are:
Tll = - A l l M2 - A66N2
'1-1
715 = TSI = -(BI2 + B66)MN + "~-(EI2 + E66)MN
T22 = - A ~ M 2 - A22 N2
_(AI + ÷
T23 = T32 = \R, + R2/ 3
+ ~1(2D55--~IF552Eli
( 3 2E12~M2
3R2/#I -4-~l 2D44- ~lF44 2E122E22~N23R
I 3R2/]
3.2 Al --E'2~M
+ 9 (El2 + 2H66)MN2 + T ( E l l + -- AI(2D55 -- ~l F55)M
\Rl Rz/
2
+ ~ ( H , 2 + 2H~)M N + -~
2 ~I (El2 + E22~N - At(2D44 -- AIF44)N
~# \RI R2/
Impact response of laminated shells 275
[ 221 2~ ]
T45 = 7"54 = - ( D , z + D66)MN + L--T- (F12 + F66) - --~-(Hi2 +/-/66 )• M N
(A4)
The coefficients T~j for the CST are given as:
Til = - A l l ME - A66 N2
T12 = T21 = -(AI2 + A66)MN
_
+ +
B22"~ _
//All + A22 +
2 A12 "~
\R, ,%/ R2/ \R~ R~ R,,%/
where k 4 and k s are share correction factors for the FSDT and hence k 4 = k 5 = l for the HSDT M = m n / a
and N = n n / b .