Composite Structures 52 (2001) 353±364

www.elsevier.com/locate/compstruct

Dynamic models for low-velocity impact damage of composite

sandwich panels ± Part B: Damage initiation
Michelle S. Hoo Fatt *, Kyong S. Park
Department of Mechanical Engineering, The University of Akron, Akron, OH 44325-3903, USA

Abstract
Equivalent single and multi degree-of-freedom systems are used to predict low-velocity impact damage of composite sandwich
panels by rigid projectiles. The composite sandwich panels are symmetric and consist of orthotropic laminate facesheets and a core
with constant crushing resistance. The transient deformation response of the sandwich panels subjected to impact were predicted in a
previous paper, and analytical solutions for the impact force and velocity at damage initiation in sandwich panels are presented in this
second paper. Several damage initiation modes are considered, including tensile and shear fracture of the top facesheet, core shear
failure, and tensile failure of back facesheet. The impact failure modes are similar to static indentation failure modes, but inertial
resistance and high strain rate material properties of the facesheets and core in¯uence impact damage loads. Predicted damage
initiation loads and impact velocities compare well with experimental results. Ó 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Equivalent dynamic systems; Damage initiation; Shear and tensile fracture; Core shear failure

1. Introduction observed in single skin laminated composites, but the

core in the sandwich panel causes facesheet damage to
Equivalent single and multi degree-of-freedom sys- become more localized under the projectile. In a recent
tems were used to determine the transient deformation study, Roach et al. [5] found that single skin laminates
of composite sandwich panels in a companion paper [1]. absorbed more energy and were more likely to prevent
Damage initiation for the same sandwich panels are projectile penetration than foam-backed laminates
proposed in this paper. Several experimental studies to resting on a solid foundation. Experiments also show
characterize impact damage initiation in sandwich that blunt projectiles induce shear failure around the
panels have been conducted in recent years [2±7]. circumference of a plug in thick laminate facesheets,
C-scans, X-rays, thermography, and micrography reveal while hemispherical and cylindro-conical projectiles in-
damages that include matrix cracking, ®ber failure and duce radial cracks that petal as the projectile penetrates
delaminations in facesheet, core crushing and shear into thin laminate facesheets [3]. Foam or honeycomb
failure, and debonding and facesheets and core. These cores crush and eventually shear to form a conical plug
recent studies suggest that the mode of damage initia- under the projectile.
tion depends on support conditions, projectile nose- Analytical solutions for the static indentation load
shape as well as the geometric and material properties of and impact velocity to initiate damage will be presented
the facesheet and core. for composite sandwich panels that are (a) rigidly sup-
Rigidly supported panels undergo only top facesheet ported (bottom facesheet ®xed), (b) two-sided clamped
fracture under the projectile and core shear failure, while (clamped wide beam), (c) simply supported, (c) four-
simply supported and clamped panels may undergo sided clamped. The composite sandwich panels are sym-
tensile failure of the bottom or back facesheet in addi- metric and consist of orthotropic laminate facesheets and
tion to top facesheet fracture and core shear failure. a core with constant crushing resistance. Except in the
Impact damage of the top facesheet is similar to that case of the rigidly supported sandwich panel, both local
and global deformations occur. Local deformation con-
sists of top facesheet indentation and core crushing.
*
Corresponding author. Tel.: +1-330-972-6308; fax: +1-330-972-
Global deformation consists of bending and shear of the
6027. entire panel. Rigidly supported sandwich panels have no
E-mail address: [email protected] (M.S. Hoo Fatt). global deformation but only local indentation.

0263-8223/01/$ - see front matter Ó 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 2 6 3 - 8 2 2 3 ( 0 1 ) 0 0 0 4 5 - 9
354 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

Notation q static crushing strength

qd dynamic crushing strength
a length of panel Qd total dynamic core crushing strength
Aij laminate extensional sti€ness matrix Qij laminate sti€ness matrix
Aijd dynamic laminate extensional sti€ness R blunt projectile radius
matrix Re e€ective radius of hemispherical-nose
b width of panel projectile
C1 static membrane sti€ness of laminate V0 projectile velocity
C1d dynamic membrane sti€ness of w0 slope of local indentation under
laminate projectile
d length of damage w0cr slope of local indentation at tensile
D1 static bending sti€ness of laminate failure
D1d dynamic bending sti€ness of facesheet w 0 †f slope of local indentation at core shear
Dij laminate bending sti€ness matrix failure
Dijd dynamic laminate bending sti€ness x; y in-plane coordinates of sandwich panel
matrix d top facesheet deformation
Dsij sandwich bending sti€ness matrix df facesheet indentation at damage
Dsijd dynamic sandwich bending sti€ness initiation
matrix d_ initial velocity of top facesheet
Eij laminate or core sti€ness e strain
Fmax maximum impact force e_ strain rate
Ff impact force at damage initiation ecr static tensile strain for facesheet failure
Gij laminate or core shear modulus ecrd dynamic tensile fracture strain
h facesheet thickness c transverse shear strain in core
hk ply thickness cf transverse shear fracture strain in core
H core thickness jc curvature at center of sandwich panel
Kc constraint factor for core crushing jij curvature of sandwich panel
Kg global sti€ness of clamped panel s13 static transverse shear failure stress
Kgd dynamic global sti€ness of clamped of facesheet
panel s13d dynamic transverse shear failure stress
Kld dynamic local sti€ness of top facesheet of facesheet
(linearized) sf static transverse shear failure stress
M0 projectile mass of core
Mij laminate bending moments sfd dynamic transverse shear failure stress
Mmax bending moments at center of of core
p
sandwich panel x ˆ Kl Kg = Kl ‡ Kg †M0
Ncr membrane force at tensile failure frequency of vibration due to impact
Nij laminate in-plane forces n extent of local indentation
P indentation force n:f extent of local indentation at damage
Pf indentation force at damage initiation † ˆ ddt † time derivative
Pl equivalent nonlinear spring response
for top facesheet deformation

In a previous paper [1], a solution for the local in- the projectile mass and an e€ective mass of the deformed
dentation of the top facesheet was found by considering composite sandwich. Results from a static indentation
a laminate resting on a rigid-plastic foundation, while a analysis were used to ®nd the e€ective spring force and
solution for the global deformation was found by ne- dashpot resistance. The spring and dashpot forces were
glecting core crushing in the composite sandwich. The adjusted with high strain rate dependent material
principle of minimum potential energy was used to ap- properties of the facesheet and core. Analytical solutions
proximate the load-indentation response. The sandwich were given for the transient indentation of the top
panel was then modeled as a discrete system of lumped facesheet and the maximum contact force between the
masses, springs and dashpots in order to derive closed- projectile and the top facesheet. In this paper, damage
form solution for the impact response of the composite initiation criteria of the same panels will be proposed.
sandwich panels. Lumped masses were used to represent Analytical solutions to predict impact damage initiation
 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364 355

will be derived based on the above discrete dynamic ter facesheets of composite sandwich panels undergoing
model. Analytical predictions of both quasi-static and impact by blunt cylinders. Hemispherical nose projec-
impact damage will be compared to test results [3,6,8]. tiles with large diameters can also induce circumferential
cracks and plugs. The large diameter hemispherical nose
projectile acts like a blunt projectile with an e€ective
2. Modes damage initiation radius.

Consider a composite sandwich panel of dimensions 2.3. Facesheet thickness

a  b, with laminated facesheets of thickness h and core
thickness H. The composite sandwich can be (a) rigidly When facesheet are thin compared to the lateral ex-
supported, (b) two-sided clamped, (c) simply supported tent of the panel, facesheet de¯ections tend to be large
or (d) fully clamped around all four sides. The panel is (many times the facesheet thickness) and the high in-
subjected to either static indentation or low-velocity plane tensile forces cause tensile cracking (Mode I). A
impact at the origin by a rigid, blunt cylinder of radius R blunt or large diameter hemispherical-nose projectile
and length L. Damage initiation in the composite can also cause plugging in thin facesheets if the radial
sandwich depends primarily on the panel support con- tensile strain reaches the tensile rupture strain of the
ditions, projectile nose-shape and facesheet thickness. facesheet. When the facesheet is thick, facesheet de¯ec-
The e€ects of each of these are ®rst explained below. tions are small (less than the facesheet thickness) and
Then damage initiation criteria are proposed. transverse shear forces in regions that surround the
projectile can be very high and cause transverse shear
2.1. Panel support conditions cracking (Mode II).
Following the above discussion, one can characterize
Damage is initiated only in the top facesheet of rig- damage initiation by three modes: top facesheet failure,
idly supported panels. However, damage can be initiated core shear failure, bottom facesheet failure. These
in either the top or bottom facesheet of clamped or modes are explained below.
simply supported panels. Global panel deformation can
lead to higher tensile forces in the bottom facesheet than 2.3.1. Top facesheet failure
in the top facesheet, which is also undergoing indenta- Cracking and perforation of cross-ply laminates by
tion. Wen et al. [3] showed experimentally that tensile rigid projectiles resembles that in metal plates. For in-
fracture of the bottom facesheet of clamped composite stance, plugging (circumferential cracking) in woven
sandwich panels consisting of GPR facesheets and very glass/polyester facesheets has been observed when it
dense foam cores, occurred before shear fracture of top undergoes impact by blunt projectiles [3], while cross-
facesheet when the panel was subjected static indenta- hair fractures (radial cracking or petalling) have been
tion by a blunt cylinder. The dense core had a very high observed in carbon ®ber facesheets subjected to impact
crushing resistance, thereby preventing local indenta- by hemispherical-nose projectiles [8]. The similarity in
tion. A large global deformation then resulted in high fracture modes suggests that failure criteria that were
tensile strains in the bottom facesheet. used in metals could be applied to the laminated top
facesheet. Two failure criteria are proposed for top
2.2. Projectile nose-shape facesheet damage initiation in the following sections.

Hemispherical-nose shape and cylindro-conical pro- 2.3.1.1. Shear failure. Shear failure of top facesheet oc-
jectiles usually produce radial cracks emanating from curs when the shear stress in the facesheet is equal to its
the point of contact between the projectile and the shear fracture stress s13 . It is likely to occur when the
facesheet. Williamson and Lagace [8] described cross- facesheets are relatively thick and do not undergo very
hair fractures in the [0/90] graphite epoxy facesheets of large indentation so that membrane forces are not well
composites sandwich panels undergoing impact damage developed. A free-body diagram at the point of shear
by hemispherical-nose projectile. These cracks result in failure is shown in Fig. 1. The indentation load at shear
petal formation on the backside of the facesheet as the failure Pf is thus given by
projectile penetrates and perforates the panel. The mode
Pf ˆ 2pRhs13 ‡ Kc pR2 q; 1†
of fracture is tensile (Mode I) in the radial crack. On the
other hand, blunt projectiles produce circumferential where Kc is a constraint factor, resulting from the fact
cracks and eventually, plug formation under the pro- that the core under the indenter is constrained by the
jectile nose. Material right under the projectile nose surrounding material. Eq. (1) was ®rst proposed by Wen
becomes compressed and a fracture zone develops et al. [3]. The constraining e€ect gives an average in-
around the periphery of the projectile. Wen et al. [3] dentation stress that is greater than the uniaxial
described disc shaped cracks in the woven glass/polyes- compressive strength of the core. Theoretically, the
356 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

Fig. 1. Balance of forces on blunt cylinder during shear failure.

constraint factor should increase with decreasing ind-

enter or projectile radius and approach unity when the Fig. 2. Balance of forces during tensile failure: (a) radial cracks and (b)
indenter or projectile radius is in®nite. Reddy et al. [9] circumferential crack.
carried out simple compression and indenter penetration
tests on foams and found that 1:7 < Kc < 2:5. Here we The free-body diagram allows one to estimate the
take an average value of Kc ˆ 2: damage initiation load. The vertical components of the
The same shear failure criterion can be used for im- membrane tensile forces per unit crack length are Ncr w0cr .
pact damage by replacing the failure load with the The area of the crushed core under the hemispherical-
maximum impact force at damage initiation and using nose cylinder is pR2e , where Re is the e€ective projectile
dynamic material properties, i.e., radius.
The damage initiation load for the circumferential
Ff ˆ 2pRs13d h ‡ Kc pR2 qd ; 2† crack is given by
where s13d is the dynamic transverse shear strength of Pf ˆ 2pRe Ncr w0cr ‡ Kc pqR2e : 3†
the facesheet and qd is the dynamic crushing strength.
Nemes et al. [10] performed high strain rate experiments If the radial crack length is small, one can assume that
using a punch shear version of the split Hopkinson Bar Ncr w0cr is roughly uniform. For a total damage length d,
and showed that the transverse shear strength increases the total failure load is given by
with strain rate in graphite/epoxy laminate. Pf ˆ dNcr w0cr ‡ Kc pqR2e : 4†

2.3.1.2. Tensile failure. Tensile failure occurs when the If the top facesheet is modeled as a membrane and in-
strains are equal to the tensile fracture strain ecr or when plane deformations are neglected, the facesheet strains
the corresponding membrane forces reach the mem- are given by
brane fracture force Ncr . Tensile failure is likely to occur 1
2
when the facesheets are thin and de¯ections are large so e ˆ w0 : 5†
2
that high tensile forces are developed in the facesheet. A Setting this equal to the tensile fracture strain gives an
free-body diagram of the forces on the hemispherical- expression for the slope at tensile failure
nose projectile at the point of tensile failure is shown for p
circumferential and radial cracks in Fig. 2(a) and (b), w0cr ˆ 2ecr : 6†
respectively. Circumferential cracks develop under
Furthermore, the tensile membrane forces at failure are
hemispherical-nose projectiles that have a large radius of
approximately
curvature. A plug is expelled as in the case of a blunt
projectile. A nose with a smaller radius of curvature is Ncr  A11 ecr : 7†
more likely to cause radial cracks, as in the case of Substituting Eqs. (6) and (7) into Eq. (3) gives
cylindro-conical projectiles. For instance, cross-hair p
fracture patterns in orthotropic laminates, i.e., those Pf ˆ 2pRe A11 ecr 2ecr ‡ Kc pqR2e 8†
consisting of [0/90] plies, are often observed under as the damage initiation load due to circumferential
hemispherical-nose indenters. Such radial cracks would cracking. Likewise, substituting Eqs. (6) and (7) into Eq.
petal as the projectile penetrates into the facesheet. (4) gives
 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364 357

p
Pf ˆ dA11 ecr 2ecr ‡ Kc pqR2e 9† df
cf ˆ ; 13†
nf Re
as the damage initiation load due to radial cracking.
A dynamic tensile failure criterion for impact damage where df and nf are the de¯ection and deformation ex-
is expressed in terms of the maximum impact force at tent at core shear failure.
damage initiation, a dynamic extensional sti€ness A11d , Equilibrium relationships between d and n as well as
and a dynamic tensile fracture strain ecrd . The last two P and d were derived for a local indentation of the
quantities are dynamic materials properties that can be facesheet into a core with constant crushing resistance
obtained from experiments. The critical impact force at (rigid-plastic) in Ref. [1]. The core should be exhibiting
tensile failure is constant core crushing before shear failure. These
p equilibrium relationships can be used to predict df and
Ff ˆ 2pRe A11d ecrd 2ecrd ‡ Kc pqd R2e 10† nf and the load at damage initiation Pf . At core shear
failure
for circumferential cracks and 8
p
128q
< 255D R nf †4 ; df < h;
df ˆ h q n R†4 i1=3
1
Ff ˆ dA11d ecrd 2ecrd ‡ Kc pqd R2e 11† 14†
: f
; df > h;
9C1
for radial cracks. Eqs. (8) and (10) can also be applied to
blunt projectiles. ( 32 p
2D1 qdf ‡ pqR2 ; df < h;
Pf ˆ 8p 
15
C1 q 32
15†
2.3.2. Core shear failure 3
df ‡ pqR2 ; df > h;
Damage initiation of the sandwich panel can also where
result from sudden core shear failure while the face-
16 384
sheets are still intact. Core shear failure for a rigidly D1 ˆ 7D11 ‡ 7D22 ‡ 4D12 ‡ 8D66 †;
supported panel is shown in Fig. 3. As the honeycomb 11 025
 
crushes under the top facesheet, the transverse shearing A11 ‡ A22 † 2A12 ‡ 4A66 †
C1 ˆ 8 ‡ :
strain in the honeycomb surrounding the indenter ap- 45 49
proach a critical shear fracture for the honeycomb cf . Dij is the laminate bending sti€ness and Cij is the lami-
This transverse shear fracture strain can be calculated nate membrane sti€ness. The ®rst expressions in Eqs.
from the shear strength as follows (14) and (15) are for a plate on a rigid-plastic founda-
sf tion, while the second expressions are for a membrane
cf ˆ ; 12†
G13 on a rigid-plastic foundation.
If local indentations are described by a plate on a
where G13 is the transverse shear modulus of the hon- rigid-plastic foundation, the damage load is
eycomb. When the actual transverse shearing strain in "
the honeycomb is equal to cf , core shear failure occurs. 1=3 #1=2
32 p 255D1 4
Experiments [3] show that the core shears into a trun- Pf ˆ 2D1 q c ‡ pqR2 : 16†
15 128q f
cated cone as shown in Fig. 3.
The actual shearing strain c in the honeycomb can be If local indentations are found by using a membrane on
approximated by the shear angle. Therefore, a rigid-plastic foundation, then the damage load is given
by
p  3=2
8 C1 q 9C1 c4f
Pf ˆ ‡ pqR2 : 17†
3 q
The corresponding impact failure loads are
" 1=3 #1=2
32 p 255D1d 4
Ff ˆ 2D1d qd c ‡ pqd R2 18†
15 128qd fd

and
p  3=2
8 C1d qd 9C1d c4fd
Ff ˆ ‡ pqd R2 ; 19†
3 qd
where D1d is the dynamic bending sti€ness, C1d is the
dynamic membrane sti€ness, and cfd is the dynamic
shear fracture strain. The dynamic bending and mem-
Fig. 3. Core shear failure. brane sti€ness are given by the same expressions for the
358 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

static bending and membrane sti€ness, but they are

4pecrd Ds11d ‡ Ds12d
calculated from laminate bending and membrane sti€- Ff ˆ H
2a


; 26†
ness matrices that have been adjusted for high strain 2
‡ h 2 ln pR ‡ 0:669
rate. Both qd and cfd must be obtained from experi- where Ds11 and Ds12 are the dynamic bending sti€ness of
ments. the sandwich.

2.3.3. Bottom facesheet failure

Bottom facesheet failure can occur in clamped panels 3. Impact damage initiation criteria
if the core crushing resistance is high enough to resist
local deformation or top facesheet indentation. The When the maximum impact force is equal to the
entire panel deforms in a shear-bending mode and fail- above critical impact forces, damage initiates. Expres-
ure would take place when the maximum tensile strains sions for the maximum impact force were given in the
in the bottom facesheet are at the critical fracture strain. previous paper [1]. They are repeated here convenience.
Maximum tensile strain occurs right under the projectile
or middle of the plate, where the bending moment is a (i) Rigidly supported panels. The following nonlinear
maximum. di€erential is solved numerically
The bending moments in a rectangular sandwich
M0 d ‡ Pl d† ‡ Qd ˆ 0; 27†
plate Mxx and Myy are given by
where M0 is the projectile mass, Pl d† is the equivalent
Mxx ˆ Ds11 jxx ‡ Ds12 jyy 20†
spring force and Qd ˆ pR2 qd is the dynamic core
and crushing resistance. The ordinary di€erential equation is
Myy ˆ Ds12 jxx ‡ Ds22 jyy ; 21† solved with initial conditions d_ 0† ˆ V0 and d 0† ˆ 0,
where V0 is the initial velocity of the projectile.The
where jxx and jyy are the curvatures in the plate with maximum impact force occurs at tmax when d_ ˆ 0, and is
respect to the x and y directions and Dsij is the bending given by
sti€ness of the sandwich.
At the plate center, the bending moment is a maxi- Fmax ˆ M0 d tmax †: 28†
mum. Therefore, the curvature at the plate center jc is (ii) Simply supported and clamped panels. The equivalent
given by nonlinear spring force is ®rst linearized
Mmax Pl d†  Kld d; 29†
jc ˆ ; 22†
Ds22‡ Ds12 †
where Kld is the linearized dynamic local spring con-
where Mmax is the maximum bending moment. Timo- stant. Then, a closed-form expression for the maximum
shenko and Woinowski-Krieger [11] give Mmax in terms impact force can be derived as
of the load P as  
    M0 Kgd ‡ Kld †x Q2 x2
P 2a Fmax ˆ q d_20 Kld ‡ d ; 30†
Mmax ˆ 2 ln ‡ 0:669 : 23† Kgd Kld
Qd x† ‡ d_0 Kld †
2 2
4p pR
Therefore, the curvature at the plate center can be ex- where
pressed in terms of the load, bending sti€ness and plate s
geometry. Kld Kgd V0 Kgd
xˆ and d_0 ˆ
By neglecting in-plane deformation, one can directly Kld ‡ Kgd †M0 Kgd ‡ Kld †
relate the curvatures to the maximum bending strain as
follows and Kgd is the dynamic global spring constant. Setting
equal Ff to Fmax in the above equations allows one to
H
emax ˆ jc : 24† calculate a critical impact velocity for impact damage.
2
If the bottom facesheet fails when emax ˆ ecr , then a
failure criterion in terms of the load at which failure 4. Comparison with experimental results
would occur, can be given by substituting expression in
Eqs. (22) and (23) into Eq. (24) The analytical solutions are applied to test data from


4pecr Ds11 ‡ Ds12 three independent studies [3,6,8]. Composite sandwich
Pf ˆ H
2a


: 25† panels with AS4/3501-6 Carbon/Epoxy facesheets and
2
‡ h 2 ln pR ‡ 0:669
Nomex honeycomb core and hemispherical nose-shape
Bottom facesheet damage under impact load occur when indenters/projectiles were used in the ®rst two experi-
the maximum impact force is given by mental studies, while E-glass WR cloth reinforced
 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364 359

laminate (GRP) facesheets and Divinycell H130 foam facesheet damage before top facesheet damage or core
core was used in the last experimental study. shear failure. Cross-hair fractures were observed in the
top facesheets. The cross-hair patterns are typical of an
4.1. Rigidly supported and two-sided clamped composite orthotropic laminate that has failed due to high in-plane
sandwich panels with AS4/3501-6 carbon/epoxy face- tensile strains. Radial cracks run parallel to the ®ber
sheets and Nomex honeycomb core direction, [0/90] in these tests. Tensile failure was the
likely mode of damage because of the hemispherical-nose
Williamson and Lagace [8] performed a series of shape of the tup and the large facesheet de¯ections or
static indentation and low-velocity impact tests on rig- membrane forces that are induced in the relatively thin
idly supported and two-sided clamped composite sand- top facesheet. Cylindro-conical tups will also produce
wich panels made with AS4/3501-6 carbon/epoxy the same cross-hair fractures, which will eventually lead
facesheet with Nomex honeycomb cores. The indenter to petaling when the projectile perforates the facesheet.
and projectile were hemispherical-nose tups or cylinders
made from case-hardened steel (rigid compared to the 4.1.1. Static damage initiation
sandwich panel). Material and geometric properties of Damage is a local phenomenon that depends only on
the facesheet, core and projectile are given in Table 1. the local indentation of the top facesheet. Therefore,
Material properties for the woven [0/90] plies could not rigidly supported and two-sided clamped sandwich
be found and the laminate facesheet was assumed to panels made of the same materials should fail at the
consist of separate 0° and 90° plies. The actual sti€ness same load. In predicting either static and impact dam-
and strength properties of the laminate facesheets would age, an e€ective radius of 0.4R for the hemispherical-
be slightly di€erent than those shown in Table 1. nose cylinder was used.
Williamson and Lagace found that damage was ini- The indentation force at damage initiation based on
tiated by top facesheet failure in all the tests. Core shear the tensile failure was predicted for sandwiches with three
failure would occur only after the indenter/projectile di€erent facesheets thickness using Eq. (9). The analyti-
penetrated through the top facesheet. Furthermore, the cal predictions are compared to experimental results in
tensile strain in the bottom facesheet of the two-sided Fig. 4. The failure load depends on the size of the damage
clamped panels was not high enough to cause bottom length, which is also shown in the diagram. Except for the
sandwich with [0/90] facesheets, the analytical predic-
Table 1 tions for the damage initiation load using Eq. (9) are
Material properties of the rigidly supported and two-sided clamped within 10% of tests results. Notice that the damage load
C/E composite sandwich panels and indenter/projectile for sandwiches with approximately the same damage
Facesheet Hercules AW193-PW prepreg consisting of length are the same for the rigidly supported and two-
AS4 ®bers in a 3501-6 matrix sided clamped cases. Therefore, top facesheet damage
a  a ˆ 102  102 mm2 : square panel
initiation is indeed a local phenomenon.
dimensions
a  b ˆ 203  89 mm2 : wide beam
dimensions 4.1.2. Impact damage initiation
hk ˆ 0:175 mm: ply thickness In order to ®nd the dynamic strength of the facesheet
qf ˆ 1617:3 kg/m3 : mass density and core, strain rates in the facesheet and core had to be
Ply Sti€ness found. The maximum value of the strain rate in the
E11 ˆ 142 GPa: longitudinal sti€ness facesheet during impact damage initiation was approx-
E22 ˆ 9:8 GPa: transverse sti€ness imated by dividing the tensile failure strain by the time it
G12 ˆ 7:1 GPa: in-plane shear modulus
takes to reach maximum impact force, tmax . The strain
m12 ˆ 0:3: Poisson's ratio
rates for the low velocity impact tests were computed
Ply Strength using static sti€ness and tensile failure strain and are
cr ˆ 0:0112: static tensile failure strain
found to be no greater than e_ ˆ 4 s 1 . It was dicult to
s13 ˆ 97 MPa: out-of-plane shear strength
®nd tensile test data of carbon/epoxy laminates at this
Core HRH 10 1/8±3.0 Nomex honeycomb particular strain rate. However, Harding et al. [12]
(Ciba-Geigy)
found that the in-plane sti€ness of woven carbon/epoxy
qc ˆ 48 kg/m3 : density
d ˆ 3:2 mm: cell diameter laminates increased only by 17%, while the failure strain
H ˆ 25:4 mm: core thickness is about the same at rates of about 290 per s. Therefore,
q ˆ 1:389 MPa: crushing resistance it would appear that at these low strain rates, the in-
Indenter/ Hemispherical-nose cylinder made of plane sti€ness of the woven carbon/epoxy laminate
projectile case-hardened steel facesheets were about the same as the static values. It
D ˆ 12:7, 25.4 and 38.1 mm: diameter was assumed that dynamic material properties for AS4/
L ˆ 660 mm: length 3501-6 carbon/epoxy laminate are equivalent to those
M0 ˆ 1:53, 1.61 and 1.69 kg: masses
for static indentation as follows:
360 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

Fig. 4. Failure load vs. facesheet thickness for rigidly supported and two-sided clamped C/E composite sandwich indented by 25.4-mm diameter
hemispherical-nose cylinder.

C1d ˆ C1 , dynamic membrane sti€ness of carbon/ supported composite sandwich panels with AS4/3501-6
epoxy, carbon/epoxy facesheet with Nomex honeycomb cores.
Aijd ˆ Aij , dynamic extensional sti€ness of carbon/ The indenter and projectile were hemispherical-nose
epoxy, tups or cylinders made from case-hardened steel. The
ecrd ˆ ecr , dynamic tensile fracture strain of carbon/ panels were simply supported. The material and geo-
epoxy. metric properties of the facesheet, core and projectile are
given in Table 2.
Goldsmith et al. [13] found that the dynamic crushing
Herup and Palazotto de®ned impact damage initia-
strength of Nomex honeycomb was always 10% greater
tion as the ®rst drop in the quasi-static load or the dy-
than the static crushing strength in low velocity impacts
namic contact load between the projectile and the top
and independent impact velocity. Therefore,
facesheet. C-scans revealed that the sandwich underwent
qd ˆ 1:1q, dynamic crushing strength of Nomex multiple delaminations and matrix cracking in the top
honeycomb. facesheet as well as core crushing and shear fracture.
Critical velocities at damage initiation for sandwiches None of the panels exhibited bottom facesheet failure.
with the three di€erent facesheet thickness were calcu- The ®rst load drop corresponded to core shear failure.
lated from Eqs. (11) and (30) for the two-sided This is in contrast to the experiments by Williamson and
clamped panel. Unlike static damage initiation, the Lagace [8], where damage initiation was due to top
maximum impact force and critical impact velocity at facesheet fracture.
damage depend on the boundary conditions. The an-
alytical solutions are shown together with experimental 4.2.1. Static damage initiation
results in Fig. 5. Except for the [0/90] facesheets, the The static load at the instant of core shear failure was
predicted failure loads were about 20% lower than test calculated from Eq. (16) for several composite sandwich
results. panels with di€erent facesheet thickness. The transverse
shear fracture strain for HRH-10-1/8-4.0 Nomex hon-
4.2. Simply-supported composite sandwich panels with eycomb cf was calculated from Ref. [14] as 2.9%. The
AS4/3501-6 carbon/epoxy facesheets and Nomex honey- analytical predictions of the failure load are compared
comb core to the experimental results in Fig. 6. The analytical
predictions are within 25% of the tests results. Eq. (16)
Herup and Palzotto [2] conducted a series of static seems to be most accurate for the thicker facesheets.
indentation and low-velocity impact tests on simply This is because the shear fracture zone becomes more
 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364 361

Fig. 5. Failure velocity vs. facesheet thickness for two-sided clamped C/E composite sandwich impacted by 25.4-mm diameter hemispherical-nose
cylinder.

localized in sandwiches with thicker facesheets and the 4.2.2. Impact damage initiation
approximation of the shear angle in Eq. (13) is better for The strain rates that were calculated for the AS4/
more localized deformations. 3501-6 facesheets and Nomex honeycomb were similar
to those in the experiments by Williamson and Lagace
[8]. Therefore, one can also assume the following
dynamic material properties:
Table 2
Material properties of simply supported C/E composite sandwich D1d ˆ D1 , dynamic bending sti€ness of carbon/
panels and indenter/projectile epoxy,
Facesheet AS4/3501-6 graphite±epoxy Dijd ˆ Dij , dynamic ¯exural sti€ness of carbon/
a  a ˆ 178  178 mm2 : square panel epoxy,
dimensions
hk ˆ 0:0635 mm: ply thickness
qf ˆ 1632 kg/m3 : mass density

Ply Sti€ness
E11 ˆ 144:8 GPa: longitudinal sti€ness
E22 ˆ 9:7 GPa: transverse sti€ness
G12 ˆ 7:1 GPa: in-plane shear modulus
m12 ˆ 0:3: Poisson's ratio

Ply Strength
cr ˆ 0:0145: static tensile failure strain
s13 ˆ 120:7 MPa: out-of-plane shear
strength
Core HRH 10 1/8±4.0 Nomex honeycomb
(Ciba-Geigy)
qc ˆ 64 kg/m3 : density
d ˆ 3:2 mm: cell diameter
H ˆ 12:7 mm: core thickness
q ˆ 3:83 MPa: crushing resistance
Indenter/projectile Hemispherical-nose cylinder
(Case-hardened steel)
Fig. 6. Static failure load vs. facesheet thickness for simply supported
D ˆ 25:4 mm: diameter
C/E composite sandwich indented by 25.4-mm diameter hemispherical-
M0 ˆ 3:48 kg: mass
nose cylinder.
362 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

qd ˆ 1:1q, dynamic crushing strength of Nomex

honeycomb,
cfd ˆ 1:3cf , dynamic shear fracture strain of Nomex
honeycomb.
The 30% increase in the dynamic shear fracture strain of
Nomex honeycomb was assumed to be the same as that
in aluminum honeycomb [15].
The maximum impact load at the instant of core
shear failure was calculated from Eq. (18) for several
composite sandwich panels with di€erent facesheet
thickness. The analytical predictions of the failure load
are compared to the experimental results in Fig. 7. The
analytical predictions are again within 25% of the tests
results and Eq. (18) appear to be most accurate for the
thicker facesheets.
The variation of both static and impact failure loads
with facesheet thickness are plotted in Fig. 8. In all cases Fig. 8. Comparison of static and impact failure load for simply sup-
ported C/E composite sandwich impacted by 25.4-mm diameter
the impact damage load is higher than the static failure hemispherical-nose cylinder.
load. This is primarily due to the fact that the crushing
and shear strength of the core increase with strain rates. clamped composite sandwich panels made of E-glass
Material properties of the C/E facesheet are almost WR cloth reinforced laminate (GRP) facesheets and
unchanged. Herup and Palazotto [2] noted that the Divinycell H130 foam cores. The GRP composite
di€erences between the static and impact failure loads sandwich panels were clamped along four edges, but in-
were slightly larger for sandwiches with the thicker plane were not constrained. The material properties for
facesheets. They attributed this tendency to the increase the facesheet and core are given in Table 3. As in the
of inertia associated with a thicker facesheet. The failure
load predicted by Eq. (18) does not include the inertial
Table 3
resistance of the facesheet because it was assumed to be
Material properties of the clamped GRP sandwich panel and indenter/
negligible compared to the mass of the projectile. projectile
Facesheet E-glass/polyester laminate
4.3. Composite sandwich panels with E-glass WR cloth laminate a  a ˆ 200  200 mm2
reinforced facesheets and divinycell H130 foam core 850  850 mm2 : square panel dimensions
hk ˆ 0:29 mm: ply thickness
qf ˆ 1650 kg/m3 : mass density
Wen et al. [3] conducted a series of static indentation,
impact and perforation tests with blunt cylinders and Ply Sti€ness
E11 ˆ 24:4 GPa: longitudinal sti€ness
E22 ˆ 6:87 GPa: transverse sti€ness
G12 ˆ 2:89 GPa: in-plane shear modulus
m12 ˆ 0:32: Poisson's ratio

Ply Strength
cr ˆ 0:021: static tensile failure strain
s13 ˆ 45 MPa: out-of-plane shear strength
Core Divinycell H130 foam core (Barracuda)
qc ˆ 130 kg/m3 : mass density
H ˆ 25:4 mm: core thickness
Sti€ness
Ec ˆ 175 MPa: Young's modulus
Gc ˆ 50 MPa: shear modulus
mc ˆ 0:3: Poisson's ratio
Strength
q ˆ 2:5 MPa: crushing resistance
sc ˆ 2 MPa: out-of-plane shear strength
Indenter/projectile Blunt cylinder made of case-hardened
steel
Fig. 7. Impact failure load vs. facesheet thickness for simply supported
D ˆ 10:5±50 mm: diameter
C/E composite sandwich impacted by 25.4-mm diameter hemispheri-
M0 ˆ 17:9 g±42.5 kg: mass
cal-nose cylinder.
 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364 363

case of the C/E laminate, all the material properties for

the woven [0/90] plies could not be found and the lam-
inate facesheet was assumed to consist of separate 0°
and 90° plies.
Test results for damage initiation due to low velocity
impact were not available, but damage initiation loads
for the panels undergoing static indentation was plen-
tiful. Damage initiation was attributed to shear failure
in the top facesheet by the blunt cylinder/projectile.
Shear failure was the likely mode of damage initiation
because the facesheets were relatively thick. When the
facesheet is thick, local indentations are small and no
appreciable membrane force can develop in the face-
sheet. Furthermore, core material right under a blunt
cylinder is compressed and a localized shear zone de-
velops around the compressed region. Core material Fig. 10. Failure load vs. facesheet thickness for rigidly supported and
being compressed under the blunt cylinder is also con- four-sided clamped GRP composite sandwich indented by 50-mm
strained by the adjacent honeycomb cells. The interac- diameter blunt cylinder.
tion of the blunt projectile and facesheet thus makes
shear failure a likely mode of damage. mental data well in Fig. 10. The analytical predictions
Shear failure is a local phenomenon and should be are within 20% of experimental results. Test results for
independent of panel size and boundary conditions. As rigidly supported panels were compared to the clamped
described in Eq. (1), the shear failure load should in- panel results in Fig. 10. The shear failure loads for both
crease parabolically with indenter radius and increase support conditions are approximately the same. There-
linearly with facesheet thickness. Failure loads based on fore, the shear failure load was independent of boundary
shear fracture of the top facesheet were calculated from conditions.
Eq. (1). The analytical results are compared to test re- In an e€ort to induce tensile failure of bottom face-
sults for varying projectile radius and facesheet thick- sheet due to global bending, Wen et al. [3] used denser
ness in Figs. 9 and 10, respectively. The test results PVC foam core, HCP100 with a nominal density of 400
shown in Fig. 9 are for three di€erent panel sizes, but kg/m3 , in the composite sandwich with [0/90]6 or 3.25
despite this, the parabolic function that describes how mm thick E-glass/polyester facesheets. The panel was
the failure load varies with indenter radius ®ts the ex- indented by the 50-mm diameter blunt cylinder and
perimental data rather well. The analytical predictions failed by tensile failure of bottom facesheet at a load of
are within 10% of experimental results. Therefore, the 57 kN. The bottom facesheet tensile failure load calcu-
shear failure load was independent of panel size. Fur- lated from Eq. (25) was equal to 69.5 kN, which about
thermore, the linear function describing how the failure 22% higher than the experimental results.
load varies with facesheet thickness also ®ts the experi-

5. Conclusion

Equivalent single and multi degree-of-freedom sys-

tems were used to predict the transient deformations of
composite sandwich panels under impact loading in a
previous paper, and analytical solutions for low-velocity
impact damage of the same composite sandwich panels
were given in this paper. The composite sandwich panels
were symmetric and consisted of orthotropic laminate
facesheets and a core with constant crushing resistance.
The initial mode of impact damage depended on the
panel support conditions, projectile nose-shape and
geometric and material properties of facesheet and core.
Particular damage initiation modes that were discussed
in this paper include fracture of the top facesheet, core
shear failure, as well as tensile failure of back facesheet.
Fig. 9. Failure load vs. indenter radius for four-sided clamped GRP Fracture patterns in the top facesheet resembled some of
composite sandwich with [0/90]6 laminates indented by blunt cylinder. those in metals, such as shear plugging by blunt
364 M.S. Hoo Fatt, K.S. Park / Composite Structures 52 (2001) 353±364

projectiles and petalling by hemispherical-nose projec- [5] Roach AM, Jones N, Evans KE. The penetration energy of
tiles. The impact failure modes were similar to static sandwich panel elements under static and dynamic loading. Part
II. Compos Struct 1998;42:135±52.
indentation failure modes, but material properties of the [6] Herup EJ, Palazotto AN. low-velocity impact damage initiation in
facesheets and core had to be adjusted for the in¯uence graphite/epoxy/Nomex honeycomb-sandwich plates. Compos Sci
of high strain rate. Predicted damage initiation loads Technol 1997;57:1581±98.
and impact velocities compared well to experimental [7] Ferri R, Sankar BV. Static indentation and low-velocity impact
results. tests on sandwich plates. In: Proceedings of the 1997 ASME
International Mechanical Engineering Congress and Exposition,
vol. 55, 1997; Dallas. p. 485±90.
[8] Williamson JE, Lagace PA. Response mechanism in the impact of
Acknowledgements graphite/epoxy honeycomb sandwich panels. In: Proceedings of
the Eighth ASC Technical Conference, Cleveland, OH; 1993.
This work was supported by the Ohio Aerospace p. 287±97.
[9] Reddy TY, Soden PD, Reid SR, Sadighi M. Impact response of
Institute under the 1999 OAI Core Collaborative thick composite laminates and sandwich panels. Collaborative
Research program. The research was made possible Research Program on the Cost-E€ective Use of Fibre-Reinforced
through collaboration with GE Aircraft Engines, Rolls- Composite O€shore, Marintech Research, Phase 1, Report No.
Royce Allison, BFGoodrich Aerospace and the NASA CP04, 1991.
Glenn Research Center. The authors wish to thank Dr. [10] Nemes JA, Eskandari H, Rakitch L. E€ect of laminate parameters
on penetration of graphite/epoxy composites. Int J Impact Eng
P.A. Lagace, Dr. A.N. Palazotto, and Dr. H.M. Wen 1998;21(1/2):97±112.
for the use of their experimental results. [11] Timoshenko SP, Woinowsky-Krieger S. Theory of plates and
shells. 2nd ed. New York: McGraw-Hill; 1961.
[12] Harding J, Li YL, Saka K, Taylor MEC. Characterization of the
References impact strength of woven carbon ®bre/epoxy laminates. In:
Proceedings of the International Conference on Mechanical
[1] Hoo Fatt MS, Park KS. Dynamic models for low-velocity impact Properties of Materials at High Rates of Strain, Oxford; 1989.
damage of composite sandwich panels ± Part A, Deformation. p. 403±10.
Compos. Struct. 2001;52(3±4):335±51. [13] Goldsmith W, Sackman JL. An experimental study of energy
[2] Palazotto AN, Gummadi LNB, Vaidya UK, Herup EJ. Low- absorption in impact on sandwich plates. Int J Impact Eng
velocity impact damage characteristics of Z-®ber sandwich panels 1991;12(2):241±62.
± an experimental study. Compos Struct 1999;43:275±88. [14] Hexcel Cooperation. Mechanical Properties of Hexcel Honey-
[3] Wen HW, Reddy TY, Reid SR, Soden PD. Indentation, comb Materials. TBS 120, Dublin, CA; 1988.
penetration and perforation of composite laminates and sandwich [15] Goldsmith W, Louie DL. Axial perforation of aluminum
panels under quasi-static and projectile loading. Key Eng Mater honeycombs by projectiles. Int J Solids Struct 1995;32(8/9):
1998;141±143:501±52. 1017±46.
[4] Roach AM, Jones N, Evans KE. The penetration energy of
sandwich panel elements under static and dynamic loading. Part I.
Compos Struct 1998;42:119±34.