ARTICLE IN PRESS

International Journal of Impact Engineering 34 (2007) 522–543

www.elsevier.com/locate/ijimpeng

Response of composite sandwich panels with transversely

flexible core to low-velocity transverse impact:
A new dynamic model
K. Malekzadeha, M.R. Khalilib, R.K. Mittalc,
a
Department of Mechanical Engineering, Malek Ashtar University of Technology, 4th Kilameter, Makhsous RD, 1387763681 Tehran, Iran
b
Department of Mechanical Engineering, K.N. Toosi University of Technology, Vafadar East Ave, 4th Tehran pars SQ., Tehran, Iran
c
Department of Applied Mechanics, I.I.T. New Delhi 110016, India
Received 20 January 2005; accepted 16 October 2005
Available online 20 December 2005

Abstract

A new computational procedure based on improved higher order sandwich plate theory (IHSAPT) and two models
representing contact behavior between the impactor and the panel are adopted to study the low velocity impact
phenomenon of sandwich panels comprising of a transversely flexible core and laminated composite face-sheets. The
interaction between the impactor and the panel is modeled with the help of a new system having three-degrees-of-freedom,
consisting of spring–mass–damper–dashpot (SMDD) or spring–mass–damper (SMD). The effects of transverse flexibility
of the core, and structural damping are considered. The present analysis yields analytic functions describing the history of
contact force as well as the deflections of the impactor and the panel in the transverse direction. In order to determine all
components of the displacements, stresses and strains in the face-sheets and the core, a numerical procedure based on
improved higher order sandwich plate theory (IHSAPT) and Galerkin’s method is employed for modeling the layered
sandwich panel (without the impactor), while the analytic force function developed on the basis of SMDD or SMD model,
can be used for the contact force between the impactor and the panel. The contact force is considered to be distributed
uniformly over a contact patch whose size depends on the magnitude of the impact load as well as the elastic properties and
geometry of the impactor. Various boundary conditions for the sandwich panel have also been considered. Finally, the
numerical results of the analysis have been compared either with the available experimental results or with some theoretical
results.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Sandwich panel; Core flexibility; Low-velocity transverse impact; Higher-order theory; Structural damping

1. Introduction

Sandwich structures comprising of fiber-reinforced composite face sheets bonded to a lightweight core
material are finding increasing use in a wide range of load-bearing applications. During service, these panels

Corresponding author. Tel.: +91 11 2659 1219; fax: +91 11 2658 1119.
E-mail address: rkmittal44@rediffmail.com (R.K. Mittal).

0734-743X/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijimpeng.2005.10.002
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 523

Nomenclature

a; b length and width, respectively, of the sandwich panel

C ef effective viscous damping coefficient of the panel
½C e  matrix of viscous damping
EP effective elastic modulus of the sandwich panel
EI effective elastic modulus of the impactor
F static indentation force
F cr critical crushing load of the core
F̄ dash dashpot force representing the dynamic crushing resistance of the core
F c ðtÞ; F c ðtÞ
contact forces based on the Hertzian and linearized Hertzian contact laws, respectively
ht ; hb ; hc ; h
thicknesses of the top and the bottom face-sheets, the core and the panel, respectively
kF stiffness of the core modeled as elastic foundation
K c ; K c coefficients of the Hertzian and linearized Hertzian contact laws, respectively
K face ; K core ; K sand effective stiffnesses of the impacted face-sheet, the core and the panel at the impact
location
[K] stiffness matrix of sandwich panel
MI mass of impactor
M xx ; M yy ; M xy shear and bending moments per unit length of the edge
M̄ sand effective mass of the sandwich panel
M̄ face effective mass of the impacted face-sheet of the panel
[M] mass matrix of the sandwich panel
N xx ; N yy ; N xy in-plane forces per unit length of the edge
P exponent in the Hertzian contact law
q; qd static and dynamic crushing strengths of the core
qi ðx; y; tÞ
contact loads over the (top or bottom) impacted face-sheet (i ¼ t; b)
qmn ðtÞ Fourier coefficients of the impact force distribution
Qx ; Qy shear forces in face-sheets, per unit edge length
[Q] vector of the impact forces
R contact radius
RI impactor radius
t2  t1 time interval of analysis
Te kinetic energy
u; v length and width of the assumed rectangular contact area
uc ; vc ; wc displacement components of the core
u0i ; v0i ; w0i
displacement components of the face-sheets, (i ¼ t; b)
u€ c ; v€c ; w€ c acceleration components of the core
u€ 0i ; v€0i ; w€ 0i
acceleration components of the face-sheets, (i ¼ t; b)
Ue internal potential energy
Ve external energy
V0 transverse velocity of the impactor at contact
W ðx; yÞ mode shape corresponding to the fundamental natural frequency of the impacted face-sheet
W nc structural damped energy
ðx0 ; y0 Þ point of impact on the face-sheet
zt ; zb ; zc normal coordinates in the mid-plane of the top and the bottom face-sheets and the core
 ARTICLE IN PRESS
524 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

Greek letters

aðtÞ contact indentation

amax maximum contact indentation
bi terms representing parts of the viscous damping effects
d variational operator
dT deflection under the impactor in static load-indentation test
D0 ðtÞ; D1 ðtÞ; D2 ðtÞ transverse displacement functions in discrete dynamic systems (SMD and SMDD
models)
Zst effective structural loss factor of the panel
nP effective Poisson’s ratio of the sandwich panel
nI effective Poisson’s ratio of the impactor
rt ; rb ; rc material densities of the face-sheets and the core
sczz normal stress in the core
tcxz ; tcyz shear stresses in the core
o11 fundamental natural frequency (rad/s) of the panel
cx ; cy rotation of the normal section of the face-sheet along y and x axes

may encounter low-velocity impacts caused by runway stones, hails, tool drop, tire blowout debris, etc.
Although extensive research has been devoted to the impact behavior of composite laminates in general [1–9],
the work on sandwich structures is somewhat limited. In this context, the work of Ambur and Cruz [10] may
be mentioned in which a local–global analysis was carried out to determine the contact force and displacement
of the panel. In deriving closed-form solution for the impact response of the composite sandwich panels, the
composite structures have sometimes been modeled as a discrete dynamic system with equivalent masses,
springs and damper. For instance, Shivakumar et al. [11] presented a two-degrees-of-freedom model that
consisted of four springs for bending, shear, membrane and contact rigidities for predicting the impact
response of a circular plate. They calculated on the basis of this model, the contact force and contact duration
for low-velocity impact on circular laminates. Other two-degrees-of-freedom models for impact response of
composite plates include those by Sjoblom et al. [12] and Lal [13]. Caprino et al. [14] used a single degree-of-
freedom system to analyze drop weight impact tests on glass/polyester sandwich panels. Anderson [15]
described an investigation using single degree-of-freedom model for large mass impact on composite sandwich
laminates. The stiffness parameters of the model were derived from the results of a three-dimensional quasi-
static contact analysis of a rigid sphere indenting a multi-layered sandwich laminate. Gong and Lam [16] used
a spring–mass model having two degrees-of-freedom in order to determine the history of contact force
produced during impact. They included structural damping also in their model.
As noted above, much effort has been made to analyze composite plates subjected to low-velocity impact
using a discrete spring–mass dynamic system. However, none of those studies considered the effects of
transverse flexibility and structural damping of the core of sandwich panels on their transient response. Thus,
an analytical procedure that includes the transverse flexibility and structural damping of the core material and
gives accurate results for impact behavior of sandwich panels has not yet been dealt with.
Complete impact analysis of sandwich panels having a core of elastic foam has been reported by Nemes and
Simmonds [17] and Lee et al. [18]. Davies [19] carried out experimental study of impact on different sandwich
panels with laminated face-sheets subjected to low velocity impact. Recently, a complete review of the subject
of impact on sandwich structures was carried out by Abrate [20]. Olsson and McManus [21] set up an
analytical model for the indentation of sandwich panels, while Hoo Fatt and Park [22] obtained analytic
solutions for the transient deformation response of sandwich panels. Olsson [23] also gave an engineering
method for predicting the impact response and damage in sandwich panels. Various approaches have been
proposed for the analysis of impact response of sandwich panels. The classical method decouples the local and
global responses and ignores any interaction between the two. The first-order shear deformation theory
(FSDT) and higher order shear deformation theory (HSDT) do not consider the flexibility of the core in
transverse direction, and the interaction between face-sheets and the soft flexible core is neglected [24,25].
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 525

Most recently, the higher-order sandwich plate theory (HSAPT) has been used [26–29]. For a soft core, the
vertical flexibility of the core must be taken into account since this flexibility of the core influences the stress
and displacement fields in the face sheets. The equations describing the static behavior of sandwich panels with
a soft core have been obtained in Ref. [28] by applying the principle of virtual work. In the present
spring–mass–damper model, equivalent stiffnesses of the impacted top face-sheet and the sandwich panel have
been obtained from the static analysis of sandwich panel based on an improved higher-order sandwich plate
theory (IHSAPT) as used in [28].
Recently, Malekzadeh et al. [30,31] introduced FSDT for face-sheets and a non-linear variation of vertical
acceleration in the core, instead of linear acceleration variation as is commonly assumed. With this
modification, solutions were obtained for damped and undamped free vibrations of simply supported
sandwich panels based on improved higher-order sandwich plate theory (IHSAPT). Another important step in
the solution of the impact problem is the contact law which gives the relationship between the impact force
and the indentation of the target surface. For isotropic homogeneous linear elastic bodies, the use of Hertz’s
contact law is well established [1–6] when the indentation is much smaller than the plate thickness. But for
sandwich panels, the face sheets are stiff and often anisotropic while the core is flexible as compared to the face
sheets. Therefore, the Hertz’s indentation law is not applicable in the present case because of this
inhomogeneity.
Spring–mass models are used extensively to analyze the dynamics of impact. An analytical procedure that
includes the transverse flexibility and structural damping of the core of sandwich panels has not yet been dealt
with. In this paper, a new three-degrees-of-freedom (TDOF) springs–masses–damper (SMD) model is
proposed to predict the contact force history for composite sandwich panels with transversely flexible core. In
order to determine all components of the displacements, stresses and strains in the face-sheets and the core, a
commercial FEM software or a specially developed numerical program like ours which is based on Galerkin’s
method, can be employed only for modeling the layered sandwich panel (without the impactor), while the
force function presented in this paper can be used to determine the contact force between the impactor and the
panel. In this paper, the full dynamic effect which includes the horizontal vibrations as well as the rotary
inertia, in addition to the normal mass inertia of the core and face-sheets is considered.

2. Formulation of the problem

The rectangular sandwich flat panel studied in this paper is composed of two FRP composite laminated and
symmetric face sheets and a core, as shown in Fig. 1 where coordinates and sign conventions are also shown.
The assumptions used in the present analysis are those encountered in linear elastic small deformation
theories. The face sheets are considered as ordinary thin or thick plates. The stacking sequences of laminate in
top and bottom face sheets are assumed to be symmetric. The core behavior follows the assumption which has

Fig. 1. Sandwich composite panel with laminated face sheets. Panel coordinates and panel dimensions are also shown. t, b, and c mean the
top and the bottom face-sheets and the core, respectively.
 ARTICLE IN PRESS
526 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

been adopted by many researchers for the honeycomb type of core [28]. The core has shear resistance but is
free of in-plane normal and shear stresses. This assumption is practically valid for a foam core, since elastic
modulus and flexural rigidity are, respectively, about three and two orders smaller than those of the face
sheets. The core is assumed to behave in a linear elastic manner with small deformations, although its height
may change and its transverse plane takes a nonlinear shape after deformation. The top and bottom face
sheets and the core are assumed to be perfectly bonded, i.e. there is no relative displacement between the core
and the adjacent face sheets at the interface. The impactor is assumed to have a spherical/hemispherical shape
and is made of an elastic material with high stiffness in comparison with the transverse stiffness of panel. The
friction between panel and impactor is assumed to be negligible. In this work, the composite sandwich panel
can be simply supported (SSSS) or fully clamped (CCCC) around all four edges of top and bottom face sheets.
Alternatively, the boundary conditions of the top and bottom face-sheets may be prescribed separately. For
example, the four edges of the bottom sheet may be simply supported while the four edges of the top sheet are
clamped (CTBS). The effects of secondary contact loadings are assumed to be negligible. Therefore, the
contact force acts only over the impacted surface of the panel during the first contact. Various stages of the
problem formulation are discussed below.

2.1. The contact forces

The contact loads qi ði ¼ t; bÞ are assumed to be represented by a double Fourier series expansion and are
separable into functions of time and position as follows:
1 X
X 1
qi ðx; y; tÞ ¼ qmn ðtÞ sinðam xÞ sinðbn yÞ. (1)
m¼1 n¼1

The Fourier coefficients qmn(t), depend on the nature of the load distribution, for example, if the load is
assumed to be concentrated at the impact point (x0, y0), then
4F c ðtÞ
qmn ðtÞ ¼ sinðam x0 Þ sinðbn y0 Þ, (2)
ab
where am ¼ mp=a and bn ¼ np=b.
A concentrated load generally results in a singularity in stress and bending moment distributions at the
point of the load application. Instead, the actual load distribution obtained in contact problems may be used.
However, in order to simulate actual contact conditions, one may note the following.
If the indentation due to impact load is much smaller than the plate thickness, Hertz’s contact theory [32]
may be used for isotropic homogeneous media. For an anisotropic medium Sveklo’s contact theory [33] may
be used. Since this theory has many similarities with the Hertz’s theory, it will be referred to as Hertz–Sveklo
theory in the following.
According to Hertz’s theory, the contact patch between the contacting sphere and the target medium is
circular and the load distribution is hemispherical. The radius of the circular patch, R(t), depends on the
contact load in the following manner:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3 F c ðtÞ:RI
RðtÞ ¼ 1:109: . (3)
EI

On the other hand, for an orthotropic target medium, such as an aligned fiber-reinforced panel, Hertz–Sveklo
theory predicts an elliptical contact patch, whose semi-major and semi-minor axes are also influenced
by the magnitude of the contact load. These axes are proportional to ½F c ðtÞ1=3 , as in Eq. (3). A detailed
analysis of carbon fiber-reinforced epoxy plates subjected to impact by steel spheres was carried
out by Frischbier [34] using Hertz–Sveklo theory. It was shown that for these plates, the contact patch
was elliptical of low eccentricity. The ratio of semi-minor to semi-major axes was nearly 0.88, unless the
fiber volume fraction is very high (470%). In view of this observation, the contact patch may be
approximated as circular.
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 527

Since Hertz–Sveklo theory also predicts a parabolic load distribution, the following contact load distribution
may be used in the present case.
 1=2
3F c ðtÞ ðx  x0 Þ2 þ ðy  y0 Þ2
qi ðx; y; tÞ ¼ 1 . (4)
2pR2 R2
Unfortunately, this impact load distribution [Eq. (4)] does not result in closed form expressions for the Fourier
coefficients. Instead, numerical integration has to be resorted to due to the non-linearity of the integral. For a
given F c ðtÞ the integral can be solved numerically with the help of commercial software and the Fourier
coefficients qmn ðtÞ obtained for various values of m and n. Since the impact force itself is generally unknown for
the impact problem and many iterations are needed, as will be shown later, for its determination, it is desirable
to have an explicit expression for the Fourier coefficients. Therefore, a further simplification of the contact
load distribution was sought.
One possibility, used initially by Dobyns [35] and later used often, is to assume a small rectangular patch
with a uniform contact load distribution over it. This assumption simplifies the analysis very much and the
Fourier coefficients are obtained as given below
16F c ðtÞ
qmn ðtÞ ¼ sinðam x0 Þ sinðbn y0 Þ sinðam u=2Þ sinðbn v=2Þ, (5)
p2 mnuv
where u and v are the dimensions of the rectangular patch. Since pffiffiffi this patch is an approximation of the circular
patch, as discussed above, it has been assumed that u ¼ v ¼ pR. This ensures that the rectangular patch has
the same area as the circular patch. Therefore, in place of Eq. (5), the following equation was used for Fourier
coefficients:
 pffiffiffi   pffiffiffi 
16F c ðtÞ pR pR
qmn ðtÞ ¼ 3 2
sinða x
m 0 Þ sinðb y
n 0 Þ sin am sin bn , (6)
p mnR 2 2
where R and F c ðtÞ are related through Eq. (3).
A comparison of qmn ðtÞ based on Eq. (6) with those obtained for a circular patch using numerical integration
showed a close agreement between the two. This can be explained in view of the fact that the size of the contact
patch is much smaller than the size of the plate. Therefore, in further analysis, Eq. (6) was adopted.

2.2. The improved high order sandwich plate theory (IHSAPT)

These equations were derived by Frostig [28], using the principle of virtual work and have been used by
present authors with some modifications for the dynamic analysis of sandwich plates [30]. They have been
further modified by incorporating FSDT for face sheets and assuming a nonlinear variation of vertical
acceleration of the core instead of the usual assumption of linear variations. The full dynamic effects (i.e.
besides the mass inertia of the core, both the horizontal vibrations and rotary inertia of the core and face-
sheets are also included) are considered in the analysis. The governing equations and the boundary conditions
are derived using Hamilton’s principle which requires that
Z t2
d ½U e þ V e þ W nc  T e  dt ¼ 0. (7)
t1

By carrying out the variational procedure, the equations of motion are obtained and are given below
(see Ref. [30])
N txx;x þ N txy;y þ tcxz ðzc ¼ 0Þ ¼ I 0t u€ 0t þ b1 ,
N bxx;x þ N bxy;y  tcxz ðzc ¼ cÞ ¼ I 0b u€ 0b þ b2 ,
Qtx;x þ Qty;y þ qt þ sczz ðzc ¼ 0Þ ¼ I 0t w€ 0t þ b3 ,
Qbx;x þ Qby;y þ qb  sczz ðzc ¼ cÞ ¼ I 0b w€ 0b þ b4 ,
 ARTICLE IN PRESS
528 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

N tyy;y þ N txy;x þ tcyz ðzc ¼ 0Þ ¼ I 0t v€0t þ b5 ,

N byy;y þ N bxy;x  tcyz ðzc ¼ cÞ ¼ I 0b v€0b þ b6 ,
M txx;x þ M txy;y  Qtx þ tcxz ðzc ¼ 0Þ:ðht =2Þ ¼ I 2t c€ xt þ b7 ,
M bxx;x þ M bxy;y  Qbx þ tcxz ðzc ¼ cÞ:ðhb =2Þ ¼ I 2b c€ xb þ b8 ,
M t þ M t  Qt þ tc ðzc ¼ 0Þ:ðht =2Þ ¼ I 2t c € þb ,
xy;x yy;y y yz yt 9

M bxy;x þ M byy;y  Qby þ tcyz ðzc € þb ,

¼ cÞ:ðhb =2Þ ¼ I 2b cyb 10

tcxz;z ¼ rc u€ c ; tcyz;z ¼ rc v€c ; tcxz;x þ tcyz;y þ sczz;z ¼ rc w€ c . ð8Þ

The boundary conditions are obtained in Appendix A. Further, I jt and I jb are the moments of inertia for the
upper and lower face sheets. tcxz, tcyz (zc ¼ 0, c) and sczz ðzc ¼ 0; cÞ are the shear and the vertical normal stresses
at the upper and the lower interfaces between the core and the face sheets, respectively, Qjx and Qjy are
distributed shear forces per unit length of the edge (j ¼ t, b) in the x- and y-directions, respectively. Using the
last thee equations of the set of Eqs. (8), constitutive law of the core material and the continuity conditions for
the top and the bottom face-sheets, the analytical relations for the normal stress as well as the horizontal and
vertical displacements in the core were derived [30].
These relations are given in Appendix B. bi (i ¼ 1–10) are the terms representing parts of equivalent viscous
damping effects and are given in Appendix C.
The impact solution for a rectangular panel with simply supported boundary conditions at the top and
bottom face sheets is assumed to be in the following form:
2 3 2 3
u0j ðx; y; tÞ u0jmn ðtÞ: cosðam xÞ: sinðbn yÞ
6 7 6 v0jmn ðtÞ: sinðam xÞ: cosðb yÞ 7
6 v0j ðx; y; tÞ 7 6 n 7
6 7 6 7
6 w0j ðx; y; tÞ 7 6 w ðtÞ: sinða xÞ: sinðb yÞ 7
6 7 X 1 X
1 6 0jmn m n 7
6 7 6 7
6 cxj ðx; y; tÞ 7 ¼ 6 A0jmn ðtÞ: cosðam xÞ: sinðbn yÞ 7. (9)
6 7
6 c ðx; y; tÞ 7 m¼1 n¼1 6 B ðtÞ: sinða xÞ: cosðb yÞ 7
6
7
6 yj 7 6 0jmn m n 7
6 c 7 6T 7
6 txz ðx; y; tÞ 7 4 cxmn ðtÞ: cosða m xÞ: sinðb yÞ 5
4 5 n
c
tyz ðx; y; tÞ T cymn ðtÞ: sinðam xÞ: cosðbn yÞ

The above double Fourier series functions satisfy the boundary conditions of panel, i.e. simply supported on
all edges. However, when all edges are clamped, the functions cosðam xÞ and cosðbn yÞ in series expansions of cxj
and cyj , respectively, are replaced by sinðam xÞ and sinðbn yÞ.
In Eq. (9) u0jmn ; v0jmn ; w0jmn ; A0jmn ; B0jmn ; T cxmn and T cymn are time dependent unknown Fourier coefficients,
m and n are, respectively, the half wave numbers in x and y directions and j ¼ t, b, where t and b mean the top
and the bottom face sheets. The dynamic equations of motion in terms of deformation and rotation
components and shear stresses in the core are derived by using the field equations alongwith the constitutive
relations and the governing equations (8). Then by applying the Galerkin method, the governing equations are
reduced to the following system of ordinary differential equations:

½Mf€wg þ ½C e f_wg þ ½Kfwg ¼ fQg. (10)

Therefore, the problem of impact on a sandwich panel reduces to the standard structural response equation. In
Eq. (10), [M] is a (10mn)  (10mn) square symmetric mass matrix, [K] is a (10mn)  (10mn) square symmetric
stiffness matrix, [Ce] is a (10mn)  (10mn) square damping matrix and {Q} is a (10mn)  1 vector of impact
forces. The system damping is simulated by proportional viscous damping terms. Eq. (10) can be readily
solved with a suitable numerical integration procedure [30].
The symmetric mass and stiffness coefficients for simply-supported rectangular sandwich panels are listed in
Appendix D. For the case of general dynamic analysis, the vector fwðtÞg½ð10mnÞ;1 contains ten sets of time
dependent unknowns: u0t , v0t , u0b , v0b , w0t , w0b , cx , cy , tcxz and tcyz .
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 529

2.3. Effective contact stiffness

The contact force between impactor and the target has often been assumed to be known. However, in
practice, the contact force is the result of contact deformation between the impactor and the target and should
be evaluated. The effective stiffness Kc can be obtained by the experimental static load-indentation test [22], or
it can also be estimated by the following modified analytical methods.

2.3.1. The modified Hertzian contact stiffness of impacted sandwich panels with elastic flexible core
Note that the solution suggested below is only applicable prior to fiber fracture, core crushing
(plastic deformation) or the penetration of the panel. Let D1 ðtÞ; D2 ðtÞ and D0 ðtÞ represent the transverse
displacements at the load point of the sandwich flat panel in the impacted top face sheet, bottom face
sheet (see Fig. 2) and that of the impactor, respectively, at any time t during impact. The contact
deformation is
aðtÞ ¼ D0 ðtÞ  D1 ðtÞ. (11)
The contact force between the impactor and the sandwich panel during the impact is assumed to be governed
by the nonlinear Hertzian contact law of the form

F c ðtÞ ¼ K c aP , (12)
where Kc and P can be obtained by static indentation tests or they can also be estimated by Hertzian contact
theory [9]. However, for a complete characterization of the contact behavior including the unloading phase,
the contact law must be determined experimentally. As Eq. (12) can be highly nonlinear, seeking an analytic
solution for the contact force might pose a formidable task. The present approach, as in Ref. [16], employs an
effective contact stiffness K c and the assumption of linear relationship between equivalent contact force and
contact deformation. This relationship is given below
F c ðtÞ ¼ K c a ¼ K c ½D0 ðtÞ  D1 ðtÞ. (13)

Fig. 2. Equivalent three-degrees-of-freedom models of the panel and impactor system: (a) Model I: spring–mass–damper; (b) Model II:
spring–mass–damper–dashpot.
 ARTICLE IN PRESS
530 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

The effective stiffness K c can be estimated with the help of the following relationship [16]:
   
P pffiffiffi P þ 1
  2G þ 1 þ pG
pffiffiffi P þ 1 2 2
K c ¼ pG     aP1
max K c , (14)
2 2 P 2 Pþ1
4G þ 1 þ pG
2 2
where GðxÞ is the gamma function. For an impactor of small mass ðM̄ sand =M I 410Þ, amax was estimated by
Gong and Lam [16], as given below
 1=ðPþ1Þ  
M̄ sand :M I P þ 1 1=ðPþ1Þ 2
amax ¼ ðV 0 =K c Þ1=ðPþ1Þ . (15)
M̄ sand þ M I 2
Many experimental studies have shown that P is equal to 1.5 for laminated composite targets. Also, Hertzian
indentation ðP ¼ 1:5Þ dominates at small loads and for thick face-sheets and stiff cores. For most cases of
interest in sandwich structures, the Hertzian indentation is negligible and the combined action of softening due
to core crushing and stiffening due to membrane/bending effects in face-sheets results in a nearly linear load-
indentation relationship, implying that P  1. In Hertzian indentation [1], the contact stiffness Kc can be
estimated using

1=2 1 1  n2I 1  n2P

K c ¼ ð43ÞERI ; ¼ þ . (16)
E EI Ep
Because the elastic moduli in z direction of all laminas are identical, with some simplifications, i.e. quasi-
isotropic assumption and by neglecting the effect of in-plane moduli, EP and nP can be calculated. An initial
estimate can be obtained using the rule-of-mixtures in z direction of the panel [31].
The modified contact law with a proper reduction in the elastic constant of the core provides reasonable
prediction for low-speed impact behavior of composite sandwich structures [31]. It has been found that the
proposed contact law can predict the measured contact forces and the contact duration for most cases,
especially for thick or stiff flat sandwich panels.

2.3.2. The contact stiffness of impacted sandwich panels assuming rigid plastic flexible core in contact zone
Many analytical methods for determining the local deformation assume a Hertzian contact [17–20].
However, the local deformation considered here causes transverse deflections of the entire top face sheet and
core crushing, so that the Hertzian contact laws are inappropriate for finding the local indentation response.
Hoo Fatt and Park [22] divided the whole impact indentation process into three stages and applied three
different mathematical models to the corresponding stages. These stages are: (I) plate on an elastic foundation;
(II) plate on a rigid-plastic foundation; (III) Membrane on a rigid-plastic foundation. The load-indentation
response has been obtained for the stage (II) by using the principle of minimum potential energy and is given
as follows [22]:
32 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
F ¼ pffiffiffiffiffiffiffiffi 2D1 dT q þ pqR2I , (17)
255
where
16 384
D1 ¼ ð7D11 þ 7D22 þ 4D12 þ 8D66 Þ. (18)
11 025
Dij (i, j ¼ 1; 2; 6) are coefficients of bending stiffness matrix of the panel. D1 is the bending stiffness of the
orthotropic face sheet alone. The first nonlinear term of Eq. (17) is the resistance due to the face sheet bending
and crushing of honeycomb/foam outside the contact area of indentor, while the second term is due
to crushing of core under the indentor (see Ref. [22]). The effective stiffness K c can be obtained by linearizing
Eq. (17).
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 531

2.4. The new proposed procedure for low-velocity impact analysis

In this study the equivalent three-degrees-of-freedom (TDOF) spring–mass–damper–dashpot (SMDD) and

spring–mass–damper (SMD) models as well as a new procedure are proposed and used to predict the low-
velocity impact response of a composite sandwich panel having a core that can be either transversely stiff or
flexible (see Figs. 2a and b). In Figs. 2a and b, the core is assumed to be elastic or rigid plastic (in contact
zone), respectively. When the indentation is very small and core crushing is elastic (i.e. transverse strain is less
than the elastic limit eec ), the impact response is found by considering the model I (Fig. 2a). As the face sheet
indentation becomes larger and core crushing is rigid plastic (in the contact zone), impact response is found
using the model II (Fig. 2b).
In Figs. 2a and b the boundary conditions of the two face-sheets are assumed independently. Thus, the
equivalent stiffness of the impacted face-sheet K face is connected to the fixed reference frame. After obtaining
the contact force history analytically, using SMD or SMDD models, a commercial FEM software or other
specially developed numerical program like ours (based on Fourier series and Galerkin’s methods) can be
employed only for modeling the layered sandwich panel (without the impactor) in order to determine all
components of displacement, strain and stress in face sheets as well as the core. The analytic force function
presented in this paper can be implemented to handle the contact force between the impactor and the panel.
Therefore, the problem of impact on the sandwich structures can be simplified to solving a standard structural
response equation for a known impact loading.

2.4.1. Low-velocity impact response of sandwich panel with elastic flexible core (SMD model)
The equations of motion of the three-degree-of-freedom system (model I—Fig. 2a) are as follows:
MID € 0 þ K  ðD0  D1 Þ ¼ 0,
c
€ 1 þ K face D1 þ K core ðD1  D2 Þ þ K  ðD1  D0 Þ ¼ 0,
M̄ face D c
M̄ sand D€ 2 þ K sand D2 þ K core ðD2  D1 Þ þ C ef D
_ 2 ¼ 0. ð19Þ
The effective structural loss factor for a two-layer composite beam has been proposed by Gong and Lam [16]
and their procedure has been modified for the calculation of the effective structural loss factor Zst for a
three-layer sandwich composite sheet. Then the following effective viscous damping coefficient C ef can be
obtained [16]:
K sand
C ef ¼ Zst . (20)
o11
Local deflection prior to core crushing at the critical crush load F cr is modeled using small deflection theory
for a plate on an elastic foundation, which is given in Ref. [23]. The effective compressive stiffness of elastic
flexible core was given as follows [23]:
pffiffiffiffiffiffiffiffiffiffiffi
K core ¼ 8 kF Df (21)
where Df is the effective stiffness of the impacted face sheet given as follows:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Df ¼ D11 D22 ðg þ 1Þ=2 (22)
and
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
g ¼ ðD12 þ 2D66 Þ= D11 D22 . (23)
The elastic region of the core is modeled as a Winkler foundation, i.e. without the out-of-plane shear stiffness.
Such a core is in a uniaxial stress state with stresses and displacements related through the elastic foundation
stiffness [23].
For a core thickness less than
   1=3
32 4Qf 12D
hc max  hf where Qf ¼ 3 f (24)
27 3E c hf
 ARTICLE IN PRESS
532 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

the deviation from a uniaxial stress state is small [23]. In Eq. (24) hf is the thickness of the impacted face sheet.
The foundation stiffness in thick cores is not uniquely defined, since displacements and stresses cannot be
simultaneously matched to the proper three-dimensional solution. The available solution for isotropic foam
cores may also be applied for honeycombs, since they have a similar relation between the average out-of plane
shear modulus and Young’s modulus [23]. For the case of thick cores, Eq. (25) below gives the foundation
stiffness kF which results in a good agreement for deflections while for thin cores, Eq. (26) gives good
agreement for stresses and deflections
Ec
kF ¼ where h̄c ¼ hc =1:38 for hc phc max and h̄c ¼ 2hc max for hc 4hc max , (25)
h̄c
 2
27
h̄c ¼ 2hc max for hc 4hc max . (26)
64
The above classification of h̄c is considered in the calculation of kF and kcore , respectively.
The equivalent stiffness of impacted face sheets for impact at center or off-center can be calculated from
static analysis of laminated composite face sheets with arbitrary boundary conditions by using the analytical
method [25] based on the first shear deformation theory (FSDT). In the special case of central impact, the
equivalent stiffness of face sheets can be obtained as follows:
K face ¼ o211 M̄ face , (27)
where o11 is the fundamental natural frequency parameter of face sheets that can be calculated from free
vibration analysis of laminated composite plates [25]. The equivalent stiffness of the impacted sandwich panel
with soft flexible core with an arbitrary location of impact point can be calculated from static/free vibration
analysis of laminated composite sandwich panels with simply supported or fully clamped boundary conditions
based on improved higher-order sandwich plate theory (IHSAPT) [30]. In special case of impact at center, the
equivalent stiffness of sandwich panel can be obtained as follows:
K sand ¼ o211 M̄ sand , (28)
where o11 is the fundamental natural frequency parameter of sandwich panel that can be calculated from the
free vibration analysis of sandwich composite panel [30]. The effective mass of impacted face sheet can be
determined from Ref. [16]
Z Z
M̄ face ¼ rt ht W 2 ðx; yÞ dAP (29)
AP

in which dAP is the differential surface area of the target face sheet. The system of ordinary differential
equation (19) can be solved using the following initial conditions:
D0 ðt ¼ 0Þ ¼ 0; D1 ðt ¼ 0Þ ¼ 0; D2 ðt ¼ 0Þ ¼ 0,
D _ 1 ðt ¼ 0Þ ¼ 0; D
_ 0 ðt ¼ 0Þ ¼ V 0 ; D _ 2 ðt ¼ 0Þ ¼ 0. ð30Þ
By applying the equivalent damping concept due to Gong and Lam [16], the eigenvalue equation can be
obtained. Therefore,
  
ðM I M̄ face M̄ sand Þo6  ½K gbc M̄ face M I þ K gcc M̄ sand M I þ K c M̄ face M̄ sand o4
 
þ ½M I ðK gcc K gbc  K 2core Þ þ K c ðK gbc M̄ face þ K gcc M̄ sand Þ  K c 2 M̄ sand Þo2
þ ½K c 2 K gbc  K c ðK gcc K gbc  K 2core Þ ¼ 0, ð31Þ
where

K gcc ¼ K face þ K core þ K c ,


K gbc ¼ K sand þ K core ; M̄ sand ¼ M̄ sand ð1  Zst Þ. ð32Þ
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 533

Eq. (31) can be solved analytically as follows:

 
pffiffiffiffiffiffiffiffi B3
½o2 1;2;3 ¼ 2 L cosðy þ 2npÞ  ; n ¼ 0; 1; 2,
3B1
pffiffiffiffiffiffi pffiffiffiffiffiffi pffiffiffiffiffiffi
3B1 B5  B23 9B1 B3 B5  27B21 3 B7  2B33
L¼ ; cosð3yÞ ¼ p ffiffiffiffiffiffiffiffiffi
ffi , ð33Þ
9B21 54B31 L3
where

B1 ¼ ðM I M̄ face M̄ sand Þ,
 
B3 ¼ ½K gbc M̄ face M I þ K gcc M̄ sand M I þ K c M̄ face M̄ sand ,
 
B5 ¼ ½M I ðK gcc K gbc  K 2core Þ þ K c ðK gbc M̄ face þ K gcc M̄ sand Þ  K c 2 M̄ sand Þ,
B7 ¼ ½K c 2 K gbc  K c ðK gcc K gbc  K 2core Þ. ð34Þ
Then, the analytic functions of dynamic deflections and impact force can be obtained as follows:
D0 ¼ c1 f10 sinðo1 tÞ þ c2 f20 sinðo2 tÞ þ c3 f30 sinðo3 tÞ, (35a)

D1 ¼ c1 sinðo1 tÞ þ c2 sinðo2 tÞ þ c3 sinðo3 tÞ, (35b)

D2 ¼ c1 f12 sinðo1 tÞ þ c2 f22 sinðo2 tÞ þ c3 f32 sinðo3 tÞ, (35c)

F c ðtÞ ¼ K c ½c1 ðf10  1Þ sinðo1 tÞ þ c2 ðf20  1Þ sinðo2 tÞ þ c3 ðf30  1Þ sinðo3 tÞ, (35d)
where
K c K core
fi0 ¼ ; fi2 ¼  ; i ¼ 1; 2; 3, (36)
K c  M I o2i K gbc  M̄ sand o2i

V 0 ðf22  f32 Þ
c1 ¼ ,
o1 ½ðf20  f10 Þðf12  f32 Þ  ðf22  f12 Þðf10  f30 Þ
V 0 ðf12  f32 Þ
c2 ¼ ,
o2 ½ðf20  f10 Þðf12  f32 Þ  ðf22  f12 Þðf10  f30 Þ
V 0 ðf12  f22 Þ
c3 ¼ . ð37Þ
o3 ½ðf20  f10 Þðf12  f32 Þ  ðf22  f12 Þðf10  f30 Þ
The general solution procedure proposed above can be used for three classes of impact problems based on
Olsson’s classification [23]. (1) Large mass impacts ðM I =M̄ face 42Þ, (2) small mass impacts ðM I =M̄ face o0:2Þ,
and (3) medium mass impacts ð0:2oM I =M̄ face o2Þ.

2.4.2. Low-velocity impact of sandwich panel with rigid-plastic flexible core in contact zone (SMDD-model)
The quasi-static force-indentation response of a rigidly supported panel can be described in terms of a
nonlinear spring force and a constant force dashpot. The response of this model is given by Eq. (17). The
effective stiffness K c can be obtained by linearizing Eq. (17). For impact loading, additional resistance will be
produced by the inertia of the impactor and deforming face sheets and the core. The inertia of the deformed
core is negligible when compared to that of the face sheets and the impactor [22]. For dynamic analysis, D1
and q in Eq. (17) are replaced by D1d and qd, respectively, where these quantities are the dynamic counterparts
of D1 and q. The dynamic bending stiffness is given by the same expression as for the static bending stiffness,
but it is calculated from a laminate bending stiffness matrix that has been adjusted for high strain rate tests
[22]. The dynamic crushing resistance qd is also measured from tests on core materials. The constant force
dashpot represents the dynamic crushing resistance of the core and is given by
F̄ dash ¼ pR2 qd . (38)
 ARTICLE IN PRESS
534 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

The equations of motion of the SMDD system (model II) are as follows:
MID € 0 þ K  ðD0  D1 Þ ¼ 0,
c
€ 1 þ ðK face þ K  ÞD1  K  D0 þ F̄ dash ¼ 0,
M̄ face D c c
M̄ sand D€ 2 þ K sand D2  F̄ dash þ C ef D_ 2 ¼ 0. ð39Þ
The dynamic response of the impacted panel is described by the system of equations (39). The third equation is
decoupled and independent, but the first two equations are coupled. The first two equations of the set of
Eqs. (39) can be rewritten as follows:
 
MI €
D1 ¼ D0 þ D0 , (40)
K c
___

ðM̄ face M I ÞD0 þ ½ðK face þ K c ÞM I þ M face :K c D€ 0 þ ðK face K c ÞD0 þ K c F̄ dash ¼ 0. (41)
Using Eqs. (30) and (40), the above stated fourth-order ordinary differential equation can be solved
analytically with the following initial conditions:
V0 
___

D0 ð0Þ ¼ 0; _
Dð0Þ ¼ V 0; € 0 ð0Þ ¼ 0;
D D0 ð0Þ ¼  K . (42)
MI c
Therefore, the impactor deflection function is obtained analytically as follows:
F̄ dash
D0 ðtÞ ¼ c1 ea1 t þ c2 ea2 t þ c3 ea3 t þ c4 ea4 t  , (43)
K face
where ai ði ¼ 1; 2; 3; 4Þ are the analytical roots of following polynomial equation:
ðM̄ face M I Þa4 þ ½ðK face þ K c ÞM I þ M̄ face :K c a2 þ ðK face K c Þ ¼ 0 (44)
and the coefficients ci ði ¼ 1; 2; 3; 4Þ can be obtained using Eqs. (42)–(44). Then, the general contact force
function can be obtained as follows:
€ 0 ðtÞ ¼ M I ðc1 a2 ea1 t þ c2 a2 ea2 t þ c3 a2 ea3 t þ c4 a2 ea4 t Þ.
F c ðtÞ ¼ M I D (45)
1 2 3 4

The above general procedure can be used for the solution of low-velocity impact case involving an impactor of
arbitrary mass. Also, for the special case of large mass impact, the fourth-order ordinary differential equation
(41) can be reduced and simplified to a second-order ordinary differential equation yielding a solution in the
form of trigonometric functions.
Then, the analytic time dependent response functions for deflections and impact force for large-mass impact
can be obtained as follows:
V0 F̄ dash
D0 ðtÞ ¼ sinðotÞ þ ½cosðotÞ  1, (46a)
o K face
    
V 0 M IV 0o F̄ dash M I o2 F̄ dash
D1 ðtÞ ¼  sinðotÞ þ 1 cosðotÞ  , (46b)
o K c K face K c K face

F̄ dash
D2 ðtÞ ¼ ½1  cosðōtÞ, (46c)
K sand
 
 € F̄ dash o
F c ðtÞ ¼ M I D0 ðtÞ ¼ M I o V 0 sinðotÞ þ cosðotÞ , (46d)
K face
where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffi
K c K face K sand
o¼ ; ō ¼  . (47)
M I ðK face þ K c Þ M̄ sand
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 535

The proposed spring–mass–damper–dashpot (SMDD) and spring–mass–damper (SMD) models facilitate the
general formulation of the analytic impact force functions for various categories of impact (i.e. small-mass
impact, large-mass impact and medium-mass impact conditions), when the flexible core deforms according to
elastic or rigid plastic behaviors. Based on the impact state, the derived force function (see Eqs. (35d), (45) and
(46d)) is incorporated in the dynamic system of equations of the panel (see Eq. (8)).

2.5. Verification of results and discussions

The characteristics of the proposed models in predicting the transient impact response of orthotropic,
laminated composite sandwich panels are evaluated. In order to validate the present formulation, the
results obtained from the present method are compared with those already reported by other authors
in literature.

2.5.1. Numerical accuracy of the approach for an isotropic plate subjected to low-velocity impact
To verify the present analysis, the results obtained for the case of an isotropic plate subjected to low velocity
impact are compared with those reported by others [7,9]. Consider an isotropic rectangular steel plate, 0.2 m
long, 0.2 m wide and 0.008 m thick, subjected to an impact at its center by a 20 mm diameter steel ball striking
with a velocity of 1 m/s.
The material properties of the plate and the impactor are: E ¼ 200 GPa; r ¼ 7971 kg=m3 , n ¼ 0:3 and the
loss factor Z ¼ 0:02. The contact force histories calculated by Wu et al. [7], Gong et al. [9], and the present
analysis are shown in Fig. 3. It is observed that the force–time curve obtained from present three-degrees-of-
freedom spring–mass–damper model (SMD-Model) and previous impact force functions are almost the same.

2.5.2. Numerical accuracy of the solution procedure for sandwich panels subjected to low-velocity impact
In this section, three examples are considered. The results are compared with other analytical and
experimental results for sandwich panels with edge support to verify the accuracy of the procedure.

2.5.2.1. Example 1: Dynamic response of sandwich panel subjected to low-velocity large-mass impact. The
panel is simply supported and made of AS4/3501-6 carbon/epoxy face sheets and Nomex honeycomb core
(HRH 10 1/8-4.0) with crushing strength q ¼ 3:83 MPa. The local indentation before damage initiation was
about 0.7 mm which is small compared to the face sheet thickness which is approximately 2 mm.
The value of critical strain eec at which the Nomex core begins to exhibit nonlinear elastic behavior is
0.02 mm/mm [22]. Therefore, the 12.7 mm thick core should deform more than 0.24 mm before it can be
modeled as plastic. In impact analysis, the dynamic crushing strength of Nomex honeycomb qd is equal to 1.1q
[22]. The material and geometrical properties of impactor are as follows:
Geometry: Hemispherical-nosed cylinder (case-hardened steel): V 0 ¼ 1:42 m=s; d ¼ 25:4 mm; M I ¼ 3:48 kg.

Fig. 3. Contact force history of steel panel impacted by steel ball.

 ARTICLE IN PRESS
536 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

Table 1
Material properties of sandwich panels

Examples Example 1: Ref. [22] Example 2: Ref. [23] Example 3: Ref. [10]

Properties Face Core Face Core Face Core

E11 (GPa) 144.8 0.0804 12.5 0.092 153 0.000689

E22 (GPa) 9.7 0.0804 12.5 0.092 8.96 0.000689
E33 (GPa) 9.7 1.005 8.0 0.092 8.96 0.4756
G12 (GPa) 7.1 0.0322 4.0 0.0325 5.1 0.000275
G13 (GPa) 7.1 0.1206 4.0 0.0325 3.79 0.2754
G23 (GPa) 3.76 0.0758 4.0 0.0325 3.3 0.0965
n12 0.3 0.25 0.25 0.41 0.29 0.25
n13 0.3 0.02 0.5 0.41 0.29 0.03
n33 0.3 0.02 0.5 0.41 0.29 0.03
r ðkg=m3 Þ 1632 64 1630 80 1650 88.14
hc (mm) — 12.7 — 22 — 9.525
Ply thickness
(mm) 0.0635 — 1.1 — 0.3 —
a (mm) 178 178 800 800 241.3 241.3
b (mm) 178 178 800 800 114.3 114.3
Material AS4/3501-6 HRH 10 1/8-4 Glass-Polyester H80/PVC IM7/5260 Glass-Phenolic

Fig. 4. Contact force histories of the impacted composite sandwich panel and the effects of various boundary conditions on contact force
history.

The material properties of the core and the face sheets of example 1 are shown in Table 1. The stacking
sequences of 32 plies in the top and the bottom face sheets are [0/90]8s. The contact force history is compared
with the analytical and experimental results in Fig. 4. Also, the effects of boundary conditions on contact force
history are shown in Fig. 4. There is a negligible difference in the phase and magnitude of contact force results
for a fully clamped and simply supported boundary conditions. Further, the contact force result for a
sandwich panel with top and bottom face sheets clamped on all edges (CCCC) is the same as the contact force
result for a sandwich panel with fully clamped boundary in top face sheet and simply supported in bottom face
sheet (CTSB). Therefore, the effects of the boundary conditions of bottom face sheet on contact force history
are negligible. These observations indicate that the impact response of a sandwich panel is a very localized
phenomenon as compared to that of a monolithic plate. This may be explained by the fact that most of the
impact energy is taken up by the flexible core during its deformation in the region close to the impact point.
For the clamped case, contact force magnitude is a little larger, and duration is smaller than for the simply
supported case. Contact force history for simply supported panel obtained from the present method is in good
agreement with that obtained from the experiments [22]. The theoretical predictions based on SMDD model,
shown in Fig. 4, indicate that the largest error in the value of maximum contact force obtained from the
present analysis vis-a-vis the measured experimental values [22] is about 4.76% which corresponds to the
impactor velocity, V 0 ¼ 1:42 m=s. A possible reason for this small discrepancy may be the non-inclusion of the
effects of loading rate in case of both the honeycomb constitutive model as well as the linear elastic model used
for the face sheets and the bond layer.
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 537

2.5.2.2. Example 2: Dynamic response of simply supported composite sandwich panel subjected to low-velocity
small-mass impact. In this example, the numerical results are compared with the results of impact
experiments on sandwich panels with edge supports [19]. This example has been selected as explicit
measurement of deflections on the back face and is available for comparison with the results of present work.
The details of the sandwich construction are presented in the column labeled as example 2 of Table 1. The
sandwich panel was hit at 7.67 m/s by a 2.5 kg impactor with 12.5 mm tup radius. The impactor/panel mass
ratio of 0.44 puts the problem in the category of small mass impact response. The ply thickness is 1.1 mm and
thickness of each face-sheet is 2.2 mm (hb ¼ ht ¼ 2:2 mm).
For the Galerkin’s method based on Fourier series, the optimum number of terms for summation is
determined by the required accuracy and computational efficiency. Accuracy obviously increases with more
terms, though not very significantly beyond some finite number. Practical considerations, like the computation
time, limit the accuracy further. Fig. 5 shows the convergence of the solution with increasing number of terms
of the Fourier series. It was observed that the difference in deflection response for m, n415 (225 terms) is very
small. The size of time step ðDtÞ is generally chosen to be a fraction of the period corresponding to the highest
frequency of the vibrating structure to ensure stability of the numerical solution process. Dt is commonly taken
to be equal to 0:1T k , where T k ¼ 2p=ok is the time period corresponding to the circular frequency ok , and k is
the mode number of the highest frequency. A comparison of the contact force and deflection results obtained
by the present method with analytical and experimental results obtained from [19,23] are, respectively, shown

Fig. 5. Effect of varying the number of terms of the Fourier series included in the solution for transverse deflection of the panel.

Fig. 6. Comparsion of the results of the present analysis with the analytical and experimental results: (a) contact force; (b) transverse
deflection of the bottom face sheet.
 ARTICLE IN PRESS
538 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

in Figs. 6a and b. The contact force and deflection histories obtained from the present method are in fairly
good agreement with those obtained in Ref. [23]. Any small disagreement between the two sets of results may
be attributed to some uncertainty in the properties of the core material as used in the present analysis. Also,
large deformation of the top face-sheet due to core crushing has not been considered in our analysis.
Further, the difference between our results and theoretical results [23] may be due to the different sets of
assumption, e.g. an infinite plate was considered and the effects of rotary inertia and transverse shear were
neglected in Ref. [23] while we considered a finite plate with rotary inertia and shear effects, etc. Also, it was
noted that the variation of contact area with respect to time consumes a lot of computer time during dynamic
analysis of structures and its effect on the impact response of sandwich panels was so small that it can be
neglected.
Also, in Fig. 6, the effects of structural damping on contact force and central deflection of panel are shown.
The loss factor of core was assumed to be 0.1 and loss factors of face sheets were neglected in computation.
Maximum induced vertical shear stresses in the core tcxz ; tcyz and maximum in-plane shear stresses txy in the top
and the bottom face sheets on the section at y ¼ b=2 and parallel to the x-axis are shown in Fig. 7a. This figure
shows that the shear stresses in the top face sheet txy are larger than the shear stresses in the bottom face sheet
and in the core. Also, Fig. 7b shows the maximum normal stress in the mid-plane of the core on the section at
y ¼ b=2 parallel to the x-axis.

2.5.2.3. Example 3: Dynamic response of simply supported composite sandwich panel subjected to low-velocity
medium-mass impact. The panel considered is simply supported and is fabricated with graphite-bismaleimide
(IM7/5260) face sheets and glass fiber-reinforced phenolic honeycomb core. The stacking sequence of plies in
the top face sheet and the bottom face sheet is ½ð45=0=  45=90=0Þ2 =90s . The details of this sandwich
construction are presented in the column ‘example 3’ of Table 1. The panel is 114.3 mm wide and 241.3 mm
long. Its exposed area was impacted by means of a dropped weight with 12.7 mm diameter hemispherical tup
ðM I =M sand ¼ 1:602Þ at an energy level of 2.7138 J, which is below the damage threshold level to generate
contact force and surface-strain results [10] that are used for validating the present approach. Bottom surface
strain for this sandwich panel are presented in Fig. 8. The strain profile so obtained is compared with the
experimental and analytical results of Ref. [10] in which the face sheet is treated as a plate on an elastic
foundation. The strain results, based on present SMD model (model I) are in good agreement with the
experimental results. The results obtained from SMD model are closer to experimental results than those
obtained from the plate on elastic foundation model. This close agreement of back-surface strain with

Fig. 7. Maximum induced normal and shear stresses in the core and the face sheets along the section y ¼ b=2 and parallel to x-axis: (a)
maximum shear stresses in the mid-plane of the core and the face sheets; (b) maximum normal stress in the core.
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 539

Fig. 8. Variation of strain on the lower surface of the impacted panel at a point opposite to the impact point.

Fig. 9. Central deflections of the top and the bottom face sheets and indentation of impacted panel.

Fig. 10. Normal stress (sczz )-profile in the mid-plane of the core with increasing contact time during impact: (a) along the section at
x ¼ a=2 and parallel to y-axis; (b) along the section at y ¼ b=2 and parallel to x-axis.

experimental results reflects that the through-thickness deformation and the transverse flexibility of the core of
panel which influence the back-surface strain have been appropriately accounted for in the present analysis
based on the improved high order sandwich plate theory (IHSAPT). The central deflection histories of top and
bottom face sheets and the total indentation of the panel are shown in Fig. 9. The figure illustrates that the
 ARTICLE IN PRESS
540 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

compressibility and transverse flexibility of the core cause core indentation and therefore, the Hertzian
indentation of top face sheet is relatively small as compared to the core indentation. The indentation of core is
almost equal to the difference of the top and the bottom transverse deflections that are shown in Fig. 9. Fig. 10
shows the profiles of normal stress sczz in the mid-plane of the core with increasing contact time. Fig. 10a shows
the normal stress on a section at x ¼ a/2 and parallel to y-axis while Fig. 10b shows the normal stress on a
section at y ¼ b/2 and parallel to x-axis. Maximum normal stress is obtained at the impact point, as expected.

3. Conclusions

A new computational method based on the improved higher order sandwich plate theory (IHSAPT) for face
sheets has been introduced to analyze transverse low velocity impact on sandwich panels caused by a spherical
impactor. In this method the transverse flexibility of the core has been taken into account, as is necessary for
foam cores or non-metallic honeycomb cores. For the solution procedure the response functions of the panel
are represented by double Fourier series and then the set of the governing partial differential equations reduces
to a set of ordinary differential equations. The interaction between the panel and the impactor is represented
by one of the two models (SMD or SMDD) of three degrees of freedom, depending on the behavior of the core
under transverse load. The solution procedure has been validated by comparing its results with those obtained
either experimentally or analytically by other researchers. Three examples were considered. In one case, the
predictions of the present theory for the history of the contact force were found to be in close agreement with
experimental measurements of contact force, in spite of some uncertainty about the physical properties of the
core material as used in the analysis.
For the other two examples, the comparison of the results of the present work with those available in the
literature was made and found to be quite good, notwithstanding the difference in assumptions. The examples
considered also showed that the effects of structural damping (of face sheets as well as the core) are very small
in the low-velocity impact analysis, although the present method is capable of accounting for these effects.
Similarly, in the present method the variability of the contact patch size with the change in impact load (as
predicted by Hertz–Sveklo contact theory) was not neglected. However, for the low velocity impact case, this
variability has only a small effect on the response of the sandwich panel as long as the contact patch does not
reduce to a point.
It was also observed that when the core is soft and flexible, the solution of the low-velocity impact problem
is quite insensitive to the type of boundary conditions, which may be prescribed independently on the top and
bottom face sheets. Therefore, in such cases, the impact phenomenon is very much localized. This may be
attributed to the large proportion of the impact energy being absorbed by the core in deformation or crushing.
The present analysis is based on linear elastic or viscoelastic behavior of the face sheets and the core. Hence,
the deformations are assumed to be small. The in-plane stresses in the core are also neglected. However, there
is no restriction on the ratio of the impactor mass and the panel mass, i.e. the analysis can be applied to all
three categories of mass ratio as defined by Olsson.
Computationally, the present method is very economical as compared to other numerical procedures, such
as F.E.M. In other methods, the contact force history has to be obtained as a part of the solution and this
involves a lot of computation effort. However, the F.E.M. gives a detailed information about the distributions
of stresses and strains in the face sheets and the core, separately. This is not so with the present procedure.
Instead, the contact force obtained by the present procedure is used as an input for further analysis of the
panel, like the deflection under the point of impactor and the failure prediction of the core, etc.

Appendix A

The boundary conditions at the edges of upper and lower face sheet are:
i
N ixx ðx ¼ 0 or x ¼ aÞ ¼ aN̄ xx or u0i ðx ¼ 0 or x ¼ aÞ ¼ 0
i
N ixy ðx ¼ 0 or x ¼ aÞ ¼ aN̄ xy or v0i ðx ¼ 0 or x ¼ aÞ ¼ 0
i
N iyy ðy ¼ 0 or y ¼ bÞ ¼ aN̄ yy or v0i ðy ¼ 0 or y ¼ bÞ ¼ 0
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 541

i
N ixy ðy ¼ 0 or y ¼ bÞ ¼ aN̄ xy or u0i ðy ¼ 0 or y ¼ bÞ ¼ 0
M ixx ðx ¼ 0 or x ¼ aÞ ¼ 0 or cxi ðx ¼ 0 or x ¼ aÞ ¼ 0
M iyy ðy ¼ 0 or y ¼ bÞ ¼ 0 or cyi ðy ¼ 0 or y ¼ bÞ ¼ 0
M ixy ðy ¼ 0 or y ¼ bÞ ¼ 0 or cxi ðy ¼ 0 or y ¼ bÞ ¼ 0
M ixy ðx ¼ 0 or x ¼ aÞ ¼ 0 or cyi ðx ¼ 0 or x ¼ aÞ ¼ 0
M ixy atððx ¼ 0 or x ¼ aÞ and ðy ¼ 0 or y ¼ bÞÞ ¼ 0
w0i ððx ¼ 0 or x ¼ aÞ and ðy ¼ 0 or y ¼ bÞÞ ¼ 0
Qiy ðy ¼ 0 or y ¼ bÞ ¼ 0 or w0i ðy ¼ 0 or y ¼ bÞ ¼ 0
Qix ðx ¼ 0 or x ¼ aÞ ¼ 0 or w0i ðx ¼ 0 or x ¼ aÞ ¼ 0
i ¼ t; b
Subscripts and superscripts t and b mean the top and bottom face sheets. The boundary conditions at the
edges of the core at zc ¼ z, read: tcxz (x ¼ 0 or x ¼ a) ¼ 0 or wc (( x ¼ 0 or x ¼ a), z) ¼ 0; tcyz (y ¼ 0 or
y ¼ b) ¼ 0 or wc (( y ¼ 0 or y ¼ b), z) ¼ 0.

Appendix B

Using the results obtained in [31] analytically, the vertical normal stress and the vertical deformations
through the depth of the core are as follows:
ðtcxz;x þ tcyz;y Þ ðw0b  w0t ÞE c
sczz ðx; y; zc Þ ¼  ð2zc  cÞ þ
2 c
  2   2 
zc c zc c
€
þ rc w0b  €
 w0t  zc þ ,
2c 6 2c 3

ðtcxz;x þ tcyz;y Þ zc
wc ðx; y; zc Þ ¼  ðz2c  c:zc Þ þ ðw0b  w0t Þ: þ w0t
2E c c
  3   3 
r z z
þ c w€ 0b c  czc  w€ 0t c  3z2c þ 2czc .
6E c c c
Also, the in-plane displacements of the core in x- and y-directions through the depth of the core are as follows:
tcxz :zc ðtcxz;xx þ tcyz;yx Þð3cz2c  2z3c Þ ðw0b;x  w0t;x Þ:z2c
uc ðx; y; zc Þ ¼    w0t;x :zc
Gc 12E c 2c
     
r z4 z4
þ c w€ 0b;x 2cz2c  c þ w€ 0t;x 4cz2c þ c  4z3c þ ðht =2Þ:cxt þ u0t ,
24E c c c

tcyz :zc ðtcxz;xy þ tcyz;yy Þð3cz2c  2z3c Þ

ðw0b;y  w0t;y Þ:z2c
vc ðx; y; zc Þ ¼    w0t;y :zc
Gc 12E c 2c
     
r z4 z4
þ c w€ 0b;y 2cz2c  c þ w€ 0t;y 4cz2c þ c  4z3c þ ðht =2Þ:cyt þ v0t .
24E c c c

Appendix C

The terms representing parts of equivalent viscous damping effects are as follows:
Ce
b̄c ¼ ðhc =hÞ
ða  bÞ
_ =12 þ ht :c
b1 ¼ b̄c ðu_ 0t =3 þ u_ 0b =6  hb :c _ =6Þ,
xb xt
_ _
b ¼ b̄ ðu_ 0b =3 þ u_ 0t =6  hb :c =6 þ ht :c =12Þ,
2 c xb xt
 ARTICLE IN PRESS
542 K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543

b3 ¼ b̄c ðw_ 0t =3 þ w_ 0b =6Þ,

b4 ¼ b̄c ðw_ 0b =3 þ w_ 0t =6Þ,
_ =12 þ ht :c_ =6Þ,
b5 ¼ b̄c ð_v0t =3 þ v_0b =6  hb :cyb yt

b ¼ b̄ ð_v0b =3 þ v_0t =6  hb :c _ =6 þ ht :c_ =12Þ,

6 c yb yt
_ =24 þ h2 :c
b7 ¼ b̄c ðu_ 0t :ht =6 þ u_ 0b :ht =12  hb :hb :c _ =12Þ,
xb t xt
2 _ _ =24Þ,
b ¼ b̄ ðu_ 0t :hb =12  u_ 0b :hb =6 þ h :c =12  ht :hb :c
8 c b xb xt
_ =24 þ h2 :c
b9 ¼ b̄c ð_v0t :ht =6 þ v_0b :ht =12  hb :hb :c _ =12Þ,
yb t yt

b10 ¼ b̄c ðv_0t :hb =12  v_0b :hb =6 þ hb :cyb =12  ht :hb :c_ yt =24Þ.
2 _

Appendix D

Symmetric stiffness matrix-coefficients in the free vibration eigenvalue equation:

K 11mn ¼ At66 b2n þ At11 a2m ; K 12mn ¼ ðAt12 þ At66 Þam bn ; K 1;11mn ¼ 1,
K 22mn ¼ At22 b2n þ At66 a2m ; K 2;12mn ¼ 1; K 33mn ¼ Ab66 b2n þ Ab11 a2m ;
K 34mn ¼ ðAb12 þ Ab66 Þam bn ; K 3;11mn ¼ 1; K 44mn ¼ Ab22 b2n þ Ab66 a2m ;
K 4;12mn ¼ 1; K 55mn ¼ kAt55 a2m þ kAt44 b2n þ R11 ; K 56mn ¼ R11 ;
K 57mn ¼ kAt55 am ; K 59mn ¼ kAt44 bn ; K 5;11mn ¼ am hc =2;
K 5;12mn ¼ bn hc =2; K 66mn ¼ kAb55 a2m þ kAb44 b2n þ R11 ; K 68mn ¼ kAb55 am ;
K 6;10mn ¼ kAb44 bn ; K 6;11mn ¼ am hc =2; K 6;12mn ¼ bn hc =2;
K 77mn ¼ Dt66 b2n þ Dt11 a2m þ kAt55 ; K 79mn ¼ ðDt12 þ Dt66 Þam bn ; K 7;11mn ¼ ht =2;
K 88mn ¼ Db66 b2n þ Db11 a2m þ kAb55 ; K 8;10mn ¼ ðDb12 þ Db66 Þam bn ; K 8;11mn ¼ hb =2;
K 99mn ¼ Dt66 a2m þ Dt22 b2n þ kAt44 ; K 9;12mn ¼ ht =2; K 10;11mn ¼ hb =2;
K 10;10mn ¼ Db66 a2m þ Db22 b2n þ kAb44 ; K 11;11mn ¼ ðR12 þ R13 a2m Þ; K 11;12mn ¼ ðR13 am bn Þ;
K 12;12mn ¼ ðR12 þ R13 b2n Þ:
Except for the above elements and their symmetric counterparts, the other elements of the complex stiffness
matrix are zero.
Symmetric mass matrix-coefficients:
M 11 ¼ R5 , M 13 ¼ R2 , M 17 ¼ 2R3 , M 18 ¼ R1 , M 22 ¼ R5 , M 24 ¼ R2 , M 29 ¼ 2R3 ,
M 2;10 ¼ R1 , M 33 ¼ R4 , M 37 ¼ R3 , M 38 ¼ 2R1 , M 44 ¼ R4 , M 49 ¼ R3 , M 4;10 ¼ 2R1 ,
M 55 ¼ R5 , M 56 ¼ R2 , M 66 ¼ R4 , M 77 ¼ R9 þ I 2t , M 78 ¼ R7 , M 88 ¼ R10 þ I 2b , M 99 ¼ R9 þ I 2t ,
M 9;10 ¼ R7 , M 10;10 ¼ R10 þ I 2b ,
where
R4 ¼ I 0b þ mc =3; R9 ¼ mc h2t =12; R12 ¼ hc =G c ; R13 ¼ h3c =ð12E c Þ; R14 ¼ mc h2c =ð24E c Þ;
R5 ¼ I 0t þ mc =3; R11 ¼ E c =hc , R7 ¼ mc hb ht =24; R9 ¼ mc h2t =12; R10 ¼ mc h2b =12;
R1 ¼ mc hb =12; R2 ¼ mc =6; R3 ¼ mc ht =12:
Except for the above elements and their symmetric counterparts, the other elements of the mass matrix are
zero.

References

[1] Abrate S. Impact on laminated composite materials. Appl Mech Rev 1991;44:155–89.
[2] Abrate S. Impact on composite structures. Cambridge: Cambridge University Press; 1998.
 ARTICLE IN PRESS
K. Malekzadeh et al. / International Journal of Impact Engineering 34 (2007) 522–543 543

[3] Sun CT, Chattopadhyay S. Dynamic response of anisotropic laminated plates under initial stress to impact of a mass. J Appl Mech
1975;42:693–8.
[4] Khalili MR. Analysis of the dynamic response of large orthotropic elastic plates to transverse impact and its application to fiber
reinforced plates. PhD thesis, Indian Institute of Technology, Delhi, May 1992.
[5] Mittal RK. A simplified analysis of the effects of transverse shear on the response of elastic plates to impact loading. Int J Solids
Struct 1987;23(8):1191–203.
[6] Mittal RK, Khalili MR. Analysis of impact of a moving body on an orthotropic elastic plate. AIAA J 1994;32(4):850–6.
[7] Wu H-YT, Chung F-K. Transient dynamic analysis of laminated composite plates subjected to transverse impact. J Comput Struct
1989;31:453–66.
[8] Shivakumar KN, Elber W, Illg W. Prediction of impact force and duration due to low velocity impact on circular composite
laminates. J Appl Mech 1985;52:674–80.
[9] Gong SW, Toh SL, Shim PW. The elastic response of orthotropic laminated cylindrical shells to low-velocity impact. J Compos Eng
1994;4(2):247–66.
[10] Ambur DR, Cruz JR. Low-speed impact response characteristics of composite sandwich panels. AIAA J 1995;95-1460-CP.
[11] Shivakumar KN, Elber W, Illg W. Prediction of low-velocity impact damage in composite laminates. AIAA J 1984;23(5):442–9.
[12] Sjoblom PO, Hartness JY, Cordell TM. On low-velocity impact testing of composite materials. J Compos Mater 1988;22:30–50.
[13] Lal KM. Residual strength assessment of low-velocity impact damage of graphite epoxy laminates. J Reinf Plast Compos
1983;2:226–38.
[14] Caprino G, Teti R. Impact and post-impact behavior of foam core sandwich structures. J Compos Struct 1994;29:47–55.
[15] Anderson TA. An investigation of SDOF models for large mass impact on sandwich composites. J Compos B 2005;36(2):135–42.
[16] Gong SW, Lam KY. Effects of structural damping and stiffness on impact response of layered structures. AIAA J 2000;38(9):1730–5.
[17] Nemes JA, Simmonds KE. Low-velocity impact response of foam-core sandwich composites. J Compos Mater 1992;26:500–19.
[18] Lee LJ, Huang KY, Fann YJ. Dynamic response of composite sandwich plate impacted by a rigid ball. J Compos Mater 1993;
27:1238–56.
[19] Davies P, Choqueuse D, Pichon A. Impact behaviour of composite sandwich panels to impact loading. In: Impact and dynamic
fracture of polymers and composites, vol. 19. London: Mechanical Engineering Publication; 1995. p. 341–58.
[20] Abrate S. Localized impact on sandwich structures with laminated facing. Appl Mech Rev 1997;50:69–82.
[21] Olsson R, McManus HL. Improved theory for contact indentation of sandwich panels. AIAA J 1996;34:1238–44.
[22] Hoo Fatt MS, Park KS. Dynamic models for low-velocity impact damage of composite sandwich panels—Part A: deformation. J
Compos Struct 2001;52:335–51.
[23] Olsson R. Engineering method for prediction of impact response and damage in sandwich panels. J Sandwich Struct Mater 2002;
4:3–29.
[24] Reddy JN. Mechanics of laminated composite plates. Theory and analysis. Boca Raton, FL: CRC Press; 1997 [chapter 5].
[25] Khalili MR, Malekzadeh K, Mittal RK. A new approach in static and dynamic analysis of composite plates with different boundary
conditions. J Compos Struct 2005;169:149–55.
[26] Frostig Y, Baruch M, Vilnay O, Sheinman I. High-order theory for sandwich beam behavior with transversely flexible core. J Eng
Mech 1992;118(5):1026–43.
[27] Frostig Y, Baruch M. Free vibrations of sandwich beams with a transversely flexible core: a high order approach. J Sound Vib
1994;176(2):195–208.
[28] Frostig Y. Buckling of sandwich plates with a flexible core: high-order theory. Int J Solids Struct 1998;35(3–4):183–204.
[29] Sokolinsky V, Steven R, Frostig Y. Boundary condition effects in free vibrations of higher-order soft sandwich beams. AIAA J
2000;40(6):1220–7.
[30] Malekzadeh K, Khalili MR, Mittal RK. Local and global damped vibrations of sandwich plates with a viscoelastic soft flexible core:
an improved high-order approach. J Sandwich Struct Mater 2005;7(5):431–56.
[31] Malekzadeh K, Khalili MR, Mittal RK. Prediction of low-velocity impact response of composite sandwich panels using new three
degrees-of-freedom model. 13th International conference of mechanical engineering, Esfahan University of Technology, Esfahan,
Iran, Paper code: 24.1418505, 2005.
[32] Timoshenko S, Goodier JN. Theory of elasticity. New York: McGraw-Hill; 1951.
[33] Sveklo VA. Hertz problem of compression of anisotropic bodies. J Appl Math Mech 1974;38(6):1023–7.
[34] Frischbier J. Theorie der Stossblastung orthotroper Platen und ihre experimentelle Ueberpruefung am beispiel einer unidirectional
verstaerkten CFK-Verbundplatte. Report No. 51, Institut fuer Mechanik, Ruhr-Universitaet, Bochum, March 1987.
[35] Dobyns AL. Analysis of simply supported orthotropic plates subject to static and dynamic load. AIAA J 2000;19(5):642–50.