Composite Structures 125 (2015) 8187

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Criteria for skin rupture and core shear cracking induced by impact on
sandwich panels
Robin Olsson a,, Tim B. Block b,1
a
Swerea SICOMP, Box 104, SE-431 22 Mlndal, Sweden
b
Faserinstitut Bremen e.V., Geb. IW 3, Am Biologischen Garten 2, D-28359 Bremen, Germany

a r t i c l e i n f o a b s t r a c t

Article history:
Available online 30 January 2015
Keywords:
Material: carbon bre, fabrics/textiles
Property: impact behaviour
Analysis: analytical modelling
Core shear cracking induced by impact on sandwich panels is a critical failure mode causing severe loss of
structural performance. This paper reviews previous experimental and theoretical work in the area and
derives improved closed form expressions for initiation of skin rupture and core shear cracking during
impact on sandwich panels with foam cores. The criterion for skin rupture is also applicable to laminates
without a core. It is shown that the skin rupture load limits the achievable core shear load, and that core
shear cracking can be prevented by selecting a core thickness above a certain threshold value. The criteria
are successfully validated by comparison with experimental results for a range of thicknesses of skins and
cores in panels with carbon/epoxy skins and a Rohacell foam core. The criterion for skin rupture is also
validated for plain laminates.
2015 Elsevier Ltd. All rights reserved.

1. Introduction
Sandwich panels with laminated composite skins (face sheets)
offer very high bending stiffness per unit weight and are therefore
attractive for use in aircraft structures, but concerns remain over
damage tolerance issues [1,2]. A general review on the effects of
impact on sandwich panels, and a comparison of damage types
in various skin and core materials was provided in [3]. A more
recent review focused on impact response models and damage pre-
diction was provided in [4]. Studies of beam impact are common in
the literature, but will not be considered here due to differences in
response and damage growth of beams and the wide panels used
in most structures. The differences result from the fact that the
transverse shear force is constant in a beam, but inversely propor-
tional to the distance to the loading point in a plate.
Impact damage in sandwich panels may involve core crushing,
skin delamination and penetration, core shear cracking and core-
skin debonding, Fig. 1. Typically core crushing and delamination
occur at low impact energies, while core shear cracking and skin
penetration occur for signicantly higher energies. For large
impact energies both the upper skin and core will be penetrated
so that the impactor pushes on the lower skin and peels it off the
Corresponding author. Tel.: +46 31 706 63 51.
E-mail address: robin.olsson@swerea.se (R. Olsson).
1
Present address: Nordex Energy GmbH, Langenhorner Chaussee 600, 22419
Hamburg, Germany.
core, Fig. 1(d) [5,6], until eventually even the lower skin is
penetrated [5]. Core shear cracking induced by impact on large
naval panels with GFRP skins and PVC foam cores was reported
in [7], while another study found no such cracks in similar panels
[2]. This transition has been highlighted in a number of studies
which demonstrated that changes in the sandwich panel skin
properties or core thickness resulted in appearance or suppression
of impact induced shear cracks, [811].
Most aircraft structures have used honeycomb cores, which
offer high shear stiffness, but have been found to cause problems
with water entrapment. This problem is avoided with foam cores,
which also offer simplied manufacturing, but a major concern is
the risk for core shear cracking or skin debonding, e.g. due to
impact.
Leijten et al. [10] analysed the impact resistance and compres-
sive residual strength of CFRP foam core sandwich panel with
Rohacell PMI foams for application in primary aircraft structures.
They describe a signicant increase in planar damage area during
non-destructive inspection (NDI) for a particular sandwich con-
guration with a relatively thin foam core but otherwise concen-
trate on skin debonding due to core crushing below the point of
impact. Destructive sectioning of this particular sandwich con-
guration revealed core shear cracks that emerge from the local
core crushing area and extend in a circular manner conically
through the foam core. These shear cracks result in a reduced ex-
ural stiffness of the affected specimens compared to the reference
conguration. Block [11] investigated in a recent publication the

http://dx.doi.org/10.1016/j.compstruct.2015.01.028
0263-8223/ 2015 Elsevier Ltd. All rights reserved.
82 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187
low velocity impact response of different CFRP foam core sandwich
panels with Rohacell PMI foams and focussed on the transition
from the damage mode skin rupture to core shear failure. His nd-
ings support the evidence from Leijten et al. [10] that sandwich
congurations with thin foam cores are more likely to fail in core
shear cracking. The experimental results are used in this paper as
reference and will be described later in more detail. Honeycomb
cores are less vulnerable to shear cracking as through the thickness
crack growth has to rupture the cell walls. Shear loads on honey-
comb cores instead typically lead to shear buckling of the cell
walls. An extensive investigation of Nomex honeycomb core sand-
wich panels with CFRP skins by Raju et al. [12] revealed no core
shear cracking as failure mode. It is however reported that with
decreasing core thickness and increasing diameter of the indenter,
the planar damage area in the sandwich increases, too.
Most impacts of concern for sandwich panels involve core
crushing and large skin deections. Several analytical models for
sandwich panel indentation with core crushing have been present-
ed [1316]. The model by Olsson and McManus [13] considered
the skin as a quasi-isotropic plate or membrane on an elasticplas-
tic foundation. Elastic indentation and shear deformations were
included in the model and large skin deections were considered
by an approximate large deection plate theory or by Rayleigh
Ritz method for a pure membrane with an assumed mode shape.
Orthotropic skins were considered by a rescaling to an equivalent
quasi-isotropic skin. Trk and Hoo Fatt [14] used a RayleighRitz
method with an assumed mode shape to consider a pure
orthotropic membrane on a rigidly plastic foundation. Hoo Fatt
and Park [15] used the same method to consider small deections
of an orthotropic plate on an elastic plastic foundation. Koissin
et al. [16] used a more general formulation to consider both the
2D problem and the axisymmetric problem of a quasi-isotropic
plate on an elasticplastic foundation. Large skin deections were
considered in an approximate way by considering the average
membrane strain in the deected region. Elastic indentation and
shear deformation of the skin were not included in the models
[1416].
Analytical predictions of the impact response have been
obtained by combining the indentation models with models for
global panel response, e.g. [17,18]. In addition numerous impact
response models based on the nite element method have been
Fig. 1. Impact damage involving (a) core crushing, (b) core shear cracking (c) skin
rupture/penetration and (d) core penetration.
presented, e.g. [8,19]. Closed form criteria for initiation of various
damage modes in sandwich panels were presented in [17,18] and
have been reviewed in [20].
Local core crushing and skin damage are a relatively benign
damage types resulting in stable dent growth into the core and
moderate strength reductions while core-skin debonding causes
a severe reduction in the strength and stiffness of sandwich panels
[2,9]. The current study is focused on core shear cracks appearing
prior to skin penetration, as such cracks are difcult to detect
and are prone to initiate catastrophic core-skin debonding. In the
current article we derive closed form criteria for the onset loads
of skin rupture (penetration) and core shear cracking, and show
that skin rupture prevents further shear loading of the core. The
article is an expanded version of a previous conference paper
[21], and includes a signicantly extended review and discussion
of previous work in the eld, as well as an improved expression
for skin rupture. The criterion for core shear cracking is primarily
relevant for foam cores. The criterion for skin penetration is not
only applicable to sandwich panels with various skin materials,
but also to monolithic laminates.
2. Theory
Impact and static indentation of sandwich panels with compos-
ite laminate skins result in a sequence of damage events. The local
elastic skin deection is typically followed by core crushing, skin
delamination and eventually skin rupture. Foam cores may also
experience core shear cracking. A typical load-indentation curve
for a panel without core shear cracking is shown in Fig. 2.
Core crushing involves initial elastic buckling of the core cell
walls, followed by cracking or plastic buckling of the cell walls. Cel-
lular core materials have been found to crush under a more or less
constant compressive stress, pcr, up to the point of densication
(>80% strain) where the cell walls are entirely collapsed and the
core modulus approaches the modulus of the solid cell wall mate-
rial [22].
The following theory is based on consideration of vertical equi-
librium for a hemisphere of radius R indenting a sandwich panel
with a facing skin of thickness hf, and a core of thickness hc being
crushed within a radius a, as illustrated in Fig. 3.
The effective (axisymmetric) plate stiffness in bending of the
skin is given by Df [18]
q
Df  Df 11 Df 22 g 1=2
q 1
where g Df 12 2Df 66 Df 11 Df 22
8
7
6
Skin rupture
Load [kN]
5
4
3
2
Delaminaon
1
Core crushing
0
0 2 4 6 8
Indentation [mm]
Fig. 2. Typical load-indentation curve for a sandwich panel without shear cracking.
 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187 83

Thus, the membrane is in a state of uniform plane stress, where

the stressstrain relation is given by:
r0 e0 Er =1  mr 6

where Er
and m are the average Youngs modulus and Poissons

r
ratio of the skin. These properties are obtained by evaluating the
average values of a ply rotated a complete circle, and are identical
to the corresponding properties for a quasi-isotropic laminate. The
membrane strain at the edge of the contact area is given by the rela-
Fig. 3. Forces acting on a sandwich panel with a central crush zone of radius a. tion between membrane strains and slope:

e0 h20 =2 7
Here Dj are the elements of the bending stiffness matrix of the skin,
Vertical equilibrium for the spherical cap in Fig. 4 gives:
as given by laminate theory. We now dene effective moduli Q f of
 
the skin (subscript f) and Q c of the core (subscript c): F 2pRh0  r0 hf h0 pcr pRh0 2 pRhf h20 2r0 pcr R=hf 8
(
2
3 Ec =1  m
c foam Combination with Eqs. (6) and (7) gives the expression:
Q f 12Df =hf Q c 2
Ecz honeycomb 
F 2pRhf e0 2e0 Er =1  mr pcr R=hf 9
where z refers to the thickness direction [18]. The elastic skin
The resulting skin rupture load Fr is obtained by evaluating the load
deection under a concentrated load F is modelled as a plate on
for a strain e0 equal to the ultimate tensile failure strain e1t:
an elastic foundation, where the foundation stiffness is inuenced
by the stiffness and thickness of the skin and core. The core may F r 4pRhf Er e21t 1 w=1  mr
be considered semi-innite when the core thickness hc exceeds a
10
where w 1  mr pcr R=2Er e1t hf
certain maximum value hc max [18]:
 1=3 Inspection of Eq. (10) reveals that the factor w can be neglected in
32 4   most cases of practical interest, i.e. except for extremely blunt
hc max hf Q f =Q cz 3
27 3 impactors or very dense cores. Furthermore, by setting w  0 the
The load for onset of core crushing, Fcr, is given by the theory for criterion is directly applicable to plain laminates.
centrally loaded plates on an elastic foundation, and has been Impact and indentation of sandwich panels typically results in a
derived previously [18]: central zone with core crushing. The radius a of the crush zone is
q obtained from equilibrium between the contact load F, the core
3
F cr 4pcr Q f hf hc =4:14Q cz for hc 6 hc max crush reaction and the skin membrane load N at the edge of the
p h i2=3 4 crush zone, Fig. 3:
3
F cr 3 3 pcr Q f hf =6Q cz for hc > hc max
pcr pa2 F  2paNh 11
which have been rewritten using the rst relation in Eq. (2). The The squared radius may be written on the following dimensionless
load for onset of skin delamination, Fd1, has been derived previously form:
[18], and may be written:
q 2 pcr pa2 =F 1  2paNh=F
a 12
3
F d1 p 8Q f h GIIc =9 5
where 2paNh/F is a reduction factor accounting for the load carried
where GIIc is the interlaminar toughness in mode II. The delamina- by skin membrane action during indentation. Solutions for the
tion threshold load is the same as for monolithic laminates, since indentation of a sandwich skin on a crushing core, and relations
the core support is negligible for an innitesimally small delamina- between a and F were presented in [13].
tion, but increases somewhat for subsequent delamination growth, Classical sandwich plate theory assumes a uniform shear stress
due to increasing core support. over the core thickness. The core shear stress increases proportion-
Skin delamination occurs at an early stage and causes a more or ally to the radius in the region with constant crush stress, and
less complete loss of skin bending stiffness, while membrane decreases outside this region due to decreasing contact stresses
action is maintained. Hence skin rupture may be predicted by between skin and core. Hence, from Fig. 3 the peak core stress sc
considering the equilibrium of a membrane wrapped around a is given by the following relation:
hemispherical impactor. In the contact region the delaminated skin r
p pa2 pcr a 1 Fa 2 pcr
obtains the shape of a spherical cap, with a resulting uniform sc cr 13
contact pressure, Fig. 4. 2pahc 2hc hc 4p
where the relations in Eq. (12) have been used. The load is limited
F by the skin rupture load Fr. Hence the maximum shear stress sc max
in the core is given by:
q
0 h f 0 h f sc max 1 wF r a 2r pcr =4p=hc
q
0 0 1 wRhf Er a 2r pcr =1  mr  e1t =hc 14

where a r is the dimensionless crush radius at the rupture load Fr.

pcr
The quantity w was dened in Eq. (10) and is usually close to zero.
R 0
The results in [13] show that a 2 6 0:8 for the case of a membrane
Fig. 4. Geometry of a membrane wrapping around a hemispherical impactor/ skin, and hence a value a 2r  0:8 may be used for conservative
indentor. design.
84 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187

3. Experiments 350
250
Sandwich panels were made from 75 kg/m3 Rohacell PMI RIST
foam cores covered with skins of HTS carbon non-crimp fabric
position of
(NCF) impregnated by RTM6 epoxy resin. The skins have a stacking impact
sequence of [(45/0/45)s]n and a ply thickness of 0.125 mm lead-
ing to total laminate thicknesses of 0.75 mm, 1.5 mm, 2.25 mm

300

400
and 3.0 mm. The core thickness varies between 6.5 mm and
35.5 mm. The complete test matrix with all congurations is
shown in Table 1. For each sandwich conguration three
specimens were manufactured and tested with variable impact
energies.
Impact tests were performed by the Fraunhofer Institute for
Fig. 5. Sandwich plates and picture frame used for specimen xture during impact
Mechanics of Materials in Halle, Germany. A drop weight impact
testing.
testing machine was used with a 3.1 kg impactor and a hemi-
spherical tup of radius R = 12.7 mm. This impactor is considered
more critical compared to smaller types, as it increases the loading Peak impact force vs. Impact energy
on the weaker core material due to shear bending.
t_skin = 0.75mm t_skin = 1.5mm t_skin = 2.25mm t_skin = 3.0mm
For the impact tests the sandwich panels were cut to a size of
350  400 mm and xed inside a rigid picture frame with a 12

peak impact force [kN]

250  300 mm window as shown in Fig. 5. Impacts were applied
10
to the centre of the specimen with impact energies varying from
12 J to 90 J depending on the strength of the specimen. The size 8
of the test specimens was selected large in order to reduce edge 6
effects and keep the thickness of the test specimens by one order
of magnitude smaller than its in-plane dimensions representative 4
for shell structures. 2
Impactor load and displacement were recorded during the
experiments and skin and core damage inspected afterwards using 0
air-coupled ultrasonics, and in some cases also fractography of 0 20 40 60 80 100
sections cut from panels. The load and displacement curves thus Impact energy [J]
characterise the impact process itself while the damage measure-
Fig. 6. Peak impact force vs. impact energy. Filled symbol = observed core shear
ments characterise the damaged specimen.
cracking, Empty symbol = local core damage.
Fig. 6 displays the peak impact load observed during testing.
The results are grouped into two categories. The shape of the sym-
bol is selected depending on the skin thickness. Filled symbols
mark specimens with observed core shear cracking while empty
Planar dam age diameter vs. Impact energy
symbols represent specimens where there was only local core t_skin = 0.75mm t_ skin = 1.5mm t_skin = 2.25mm t_skin = 3.0mm
damage occurring. 400
planar damage diameter [mm]

It can be observed that the peak impact force is constant for all
specimens of the same skin thickness and impact energies above a
300
certain threshold energy. From investigation of damaged speci-
mens this threshold energy can be related to skin rupture which
coincides with the peak force observed during impact testing. 200
Consequently core shear cracking as a damage mode has if at
all only little inuence. 100
Fig. 7 shows the planar damage diameter observed in the dam-
aged test specimens. The results show only total damage area and 0
do not distinguish between skin and core damage. Core damage
0 20 40 60 80 100
was observed either in the form of local skin debonding due to core
crushing below the point of impact or in the form of additional core Impact energy [J]
shear cracking. In any case core damage was observed larger than
Fig. 7. Planar damage diameter vs. impact energy. Filled symbol = observed core
skin damage and thus describes the total damage area fully. shear cracking, Empty symbol = local core damage.
When looking at the graph two trends become obvious as the
results arrange depending on the observed damage mode.
Specimens with core shear cracking (lled symbols) are subject
to signicantly larger damage sizes at lower impact energies than
Table 1
specimens with a local core damage observed. Here a maximum
Test matrix of impact tests. damage size of 50 mm diameter was observed compared to a max-
imum damage size of about 300 mm for specimens with core shear
Skin thickness hf (mm) Core thickness, hc (mm)
cracking, which is close to the total size of the specimen.
6.5 10.0 16.3 25.7 35.5 Only one particular specimen did not t into either of the two
0.75    trends. The specimen has a very thin skin of only 0.75 mm, a
1.50      16.3 mm thick core and was impacted with a relatively high energy
2.25   
of 35 J. The measured damage diameter was 100 mm and was con-
3.00   
rmed by sectioning. This also revealed that the impactor had
 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187 85

14

12

Skin rupture load [kN]

10

8
Fig. 8. 50 J impact damage in a specimen with a thick core: Core crushing with
skin rupture. Cutting scratches have been retouched for clarity. 6
Theory 2.0%
4 Theory 1.8%
Theory 1.6%
2
Experiment
0
0 1 2 3 4
Skin thickness [mm]

Fig. 10. Comparison between predicted skin rupture load and experimentally
Fig. 9. 50 J impact damage in a specimen with a thin core: Core crushing extended observed peak load.
by a foam core shear crack. Cutting scratches have been retouched for clarity.

punctured the front skin and crushed almost through the entire 6
core. As the impactor approached the rear skin, it created a local
debonding of the rear skin which was captured by the NDI and thus 5

Shear stress [MPa]

explains the large damage size. The damage mode however is core
crushing combined with skin rupture. 4
Destructive sectioning was also performed on selected speci-
3
mens in order to conrm the available NDI results. Fig. 8 shows
the damage mode local core crushing with skin rupture while
2 3.00 mm
Fig. 9 shows foam core shear cracking. The core thickness and 2.25 mm
Core shear
impact energies differ however as the specimen in Fig. 8 has got strength 1.50 mm
1
a 25.7 mm thick core and was impacted with 90 J energy while 0.75 mm
the thinner specimen in Fig. 9 was subject to a 50 J impact and 0
has got only 16.3 mm core thickness. Local bre pull out in the 0 10 20 30 40
damaged skin could not be avoided during sectioning which result- Core thickness [mm]
ed in local scratches in the foam core. For clarity these scratches
have been retouched in the current pictures. The highlighted shear Fig. 11. Predicted core shear stress and observed core shear cracking (trian-
gle = observed core cracking, circle = no observed shear cracking).
cracks are somewhat disguised by the ductile nature of cell wall
failure in the foam core.
closely corresponds to predicted stresses exceeding the shear
4. Comparison between theory and experiments strength of the foam core material and that core shear cracking
can be prevented by exceeding a certain critical core thickness.
Theoretical predictions were based on the following elastic To demonstrate the applicability of the penetration criterion to
properties for the individual UD-plies of the NCF material: plain laminates Eq. (10) was compared with experimental data for
E1 = 130 GPa, E2 = 9 GPa, G12 = 4.5 GPa and m12 = 0.26. The value of carbon/epoxy laminates. This included quasi-isotropic T300/8502
E1 reported in [8] has been slightly reduced to better reect the laminates [26] tested with a R = 12.7 mm impactor, and HTA7/
stiffness reduction caused by bre crimp in NCF composites. 6376 laminates [27,28] with quasi-isotropic, cross-ply and
The resulting average (quasi-isotropic) skin properties are: orthotropic 0/45 layup tested with a R = 7.5 mm impactor. For
Er = 50 GPa, mr = 0.307. The tensile failure strain was assumed to T300/5208 the following data were assumed: Er = 52 GPa,
be e1t = 1.8%, based on the failure strain for bre bundles reported
by the bre manufacturer [23]. This value is signicantly higher
than the tensile failure strain for conventional coupons of the
8
NCF bulk material, but is thought to be more representative of
the small material volumes (a few mm3) strained under the impac-
Penetration load [kN]

tor. The increasing strength for small material volumes is a well

6
known result of Weibulls statistical theory of strength [24].
The following properties were assumed for the 75 kg/m3 RIST T300 QI
foam: pcr = 1.7 MPa, sU = 1.3 MPa [25]. Furthermore the dimension- 4 T300 QI exp
2r 0:8 [13]. HTA7 QI
less squared crush radius was assumed to be a
HTA7 CP
Fig. 10 gives a comparison between predicted and experimen- HTA7 OR
tally observed load for onset of skin rupture, which coincides with 2 HTA7 QI exp
the peak load during impact. To illustrate the effect of skin failure HTA7 CP exp
strain, predictions for a failure strain of 1.6% (obtained in conven- HTA7 OR exp
tional tests on bulk material) and 2.0% have been included for 0
0 1 2 3 4 5
comparison.
Fig. 11 presents curves for predicted shear stress, and symbols Laminate thickness [mm]
representing corresponding experimental observations of core Fig. 12. Predicted and experimentally observed penetration load in various plain
shear cracking. It is evident that the appearance of shear cracks composite laminates (QI = quasi-isotropic, CP = cross-ply, OR = orthotropic layup).
86 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187
mr = 0.3, e1t = 1.2% [29]. Material data is available in more recent
references but [29] is assumed to be more representative for the
quality and properties at the time of testing. The tensile failure
strain reported in [29] has been increased by 20% to represent
the failure strain in small material volumes, in line with the previ-
ous arguments for the sandwich skins. For HTA7/6376 the follow-
ing data were assumed: Er = 54.3 GPa, mr = 0.31, e1t = 1.52% [30].
Fig. 12 demonstrates that predictions and experimental data are
in good agreement, although the data for T300/5208 are somewhat
unreliable. The layup independence of the penetration load is fur-
ther supported by data in [31], where penetration of 3 mm thick
[h2/02]s laminates occurred at Fr  2.3 kN more or less indepen-
dently of h. A reduction was, however, noticed for h = 90, most

likely due to splitting of the resulting 904 plies.
5. Discussion and conclusions
In the current work criteria for skin rupture (penetration) and
core shear cracking of CFRP foam core sandwich panels during
low velocity blunt impact have been presented and compared with
experimental results. Good agreement between experiments and
the damage criteria has been reached.
The proposed expression for the skin rupture load is layup inde-
pendent, as it only depends on the average in-plane modulus of the
skin. This seems to be supported by the results in [5], which provided
the same peak (penetration) load for cross-ply, angle-ply and quasi-
isotropic layups. The only exception was unidirectional layups,
where failure is controlled by splitting rather than by bre failure.
The skin rupture criterion in [17] contained an undetermined
effective contact radius Re, which was assumed to be slightly
smaller than the impactor tup radius. The current explicit criterion
is obtained by use of the geometrical relation Re = Rh0, evident in
Fig. 4. Insertion of the same relation in the expression given in
[17] yields a criterion similar to the current skin rupture criterion,
differing merely by the factor 1/(1 + m).
The current criterion for core shear cracking states that the cri-
tical load is proportional to the square of the core shear strength,
Eq. (13), while the criteria in [17] state a power of 2/3 for the skin
modelled as plate, and a power of 6 for the skin modelled as a
membrane. The reasons for these differences are not fully clear,
but may be either due to differences in assumed skin deformation
shapes, or due to the fact that [17] considered a sandwich panel on
a rigid foundation. A critical load similar to Eq. (13) was quoted in
[20] and attributed to [32]. Private communication [33] has, how-
ever, revealed that the expression was actually derived by the
authors of [20] by assuming a bare core and combining equations
in [32]. Unfortunately no details of the derivation were given in
[20]. The current expression in Eq. (13) includes an additional
2 (0.8), which accounts for the load fraction taken by
factor a
the skin.
Impact resistance is an important part of damage tolerance
which is required for primary load bearing structures in aircraft.
Here impact resistance is the ability of a structure to withstand
impact loads with minimum damage. In contrast to this damage
tolerance is the ability of a structure to sustain loads without catas-
trophic failure despite undetected structural damage. It should be
recognised that increasing the impact resistance of a structure may
sometimes reduce the damage tolerance performance.
For monolithic CFRP structures impact resistance is typically
determined by measuring damage caused by relevant impact
threats. This is then correlated with the appropriate NDI technique,
typically visual inspection, to conrm whether the damage is
detectable or not. From this correlation the maximum non-de-
tectable damage has to be used for damage tolerance analysis
and tested experimentally, e.g. by compression after impact.
The impact resistance of CFRP sandwich structures is normally
determined the same way. Two types of damage modes have been
observed in impact tests on foam core sandwich structures. The
rst damage mode core crushing combined with skin rupture
creates limited damage sizes in the core and tends to have good
visibility. The second damage mode core crushing extended by
core shear cracks creates signicantly larger damage sizes and
tends to have less visibility.
From a damage tolerance point of view core crushing and skin
rupture can be treated similarly to monolithic CFRP. Core shear
cracking, however, is more detrimental to the damage tolerance
performance as it creates larger damage at less visibility. Avoid-
ance of foam core shear cracking could thus lead to signicant
improvement of the damage tolerance of sandwich structures.
By applying the criteria derived in this paper, e.g. during pre-
liminary design, the damage mode of sandwich structures subject
to a relevant impact threat can be estimated and suitable design
changes can be made to avoid unwanted failure modes. It is noted
that prevention of skin rupture and core shear cracking may be
partly conicting aims. Skin rupture is prevented by increasing
the stiffness (modulus and/or thickness) or tensile failure strain
of the skin. These modications are, however, detrimental for core
shear cracking.
Core shear cracking is prevented by reducing the core shear fail-
ure index sc max /sU, where sU is the shear strength of the core.
Inspection of Eq. (14) reveals that this may be done in several
ways. Reduction of the skin failure strain or modulus is rarely an
option as this reduces the mechanical performance of the panel.
Core modication involves increasing the density or thickness of
the core. Both have a similar weight penalty, but it is evident that
an increased core thickness is more weight efcient, as the crush
stress pcr and shear strength sU of foams both are approximately
proportional to the core density.
It has to be kept in mind, however, that the presented failure
criteria are limited to simple geometries and load cases and do
not account for material non-linearity such as e.g. strain rates,
plasticity of the foam core or residual stresses dating back to
manufacturing of the sandwich structure.
Acknowledgements
The authors thank the CTC Stade GmbH for manufacturing of
the test specimens as well as Martin Rinker and Marianne John
of the Fraunhofer Institute for Mechanics of Materials in Halle,
Germany for impact testing of the specimens. The impact tests
were funded by the German Federal Ministry for Economic Affairs
and Energy through the Federal Aeronautical Research Program IV
(LuFo IV-2) within the project LoKosT. The authors gratefully
acknowledge this support.
References
[1] Zuardy MI, Herrmann AS. An advanced centre box of a vertical tail plane with a
side panel from CFRP foam-core sandwich structure. CEAS Aeronaut J
2011;2(14):25369.
[2] Mouritz AP, Thomson RS. Compression, exure and shear properties of a
sandwich composite containing defects. Compos Struct 1999;44(4):26378.
[3] Abrate S. Localized impact on sandwich structures with laminated facings.
Appl Mech Rev 1997;50(2):6982.
[4] Chai GB, Zhu S. A review of low-velocity impact on sandwich structures. Proc
Inst Mech Eng Part L J Mater Design Appl 2011;225(4):20730.
[5] Mohmmed R, Zhang F, Sun B, Gu B. Static and low-velocity impact on
mechanical behaviours of foam sandwiched composites with different ply
angles face sheets. J Compos Mater 2014;48(10):117388.
[6] Ferri R, Sankar BV. A comparative study on the impact resistance of composite
laminates and sandwich panels. J Reinf Thermopl Compos Mater
1997;10(4):30415.
[7] Wiese M, Echtermeyer A, Hayman B. Evaluation of oblique impact damage on
sandwich panels with PVC and balsa core materials. In: Proc fourth int conf on
sandwich construction. Stockholm, Sweden: EMAS Publ; 1998. p. 80718.
 R. Olsson, T.B. Block / Composite Structures 125 (2015) 8187
[8] Block TB, Brauner C, Zuardy MI, Herrmann AS. Advanced numerical
investigation of the impact behaviour of CFRP foam core sandwich
structures. In: Proc third ECCOMAS thematic conf on the mechanical
response of composites. Hannover, Germany; 2011.
[9] Ishai O, Hiel C. Damage tolerance of a composite sandwich with interleaved
foam core. J Compos Techn Res 1992;14(3):15568.
[10] Leijten J, Bersee HEN, Bergsma OK, Beukers A. Experimental study of the low-
velocity impact behaviour of primary sandwich structures in aircraft. Compos
Part A 2009;40(2):16475.
[11] Block TB. Analysis of the mechanical response of impact loaded composite
sandwich structures with focus on foam core shear failure [Doctoral thesis].
University of Bremen, Logos Verlag Berlin, Berlin, Germany; 2014.
[12] Raju KS, Smith BL, Tomblin JS, Liew KH, Guarddon JC. Impact damage
resistance and tolerance of honeycomb core sandwich panels. J Compos
Mater 2008;42(4):385412.
[13] Olsson R, McManus HL. Improved theory for contact indentation of sandwich
panels. AIAA J 1996;34(6):123844.
[14] Trk MH, Hoo Fatt MS. Localized damage response of composite sandwich
plates. Compos Part B 1999;30(2):15765.
[15] Hoo Fatt MS, Park KS. Dynamic models for low-velocity impact damage of
composite sandwich panels Part A: Deformation. Compos Struct 2001;52(3
4):33551.
[16] Koissin V, Shipsha A, Rizov V. The inelastic quasi-static response of sandwich
structures to local loading. Compos Struct 2004;64(2):12938.
[17] Hoo Fatt MS, Park KS. Dynamic models for low-velocity impact damage of
composite sandwich panels Part B: Damage initiation. Compos Struct
2001;52(34):35364.
[18] Olsson R. Engineering method for prediction of impact response and damage
in sandwich panels. J Sandwich Struct Mater 2002;4(1):8395.
[19] Feng D, Aymerich F. Damage prediction in composite sandwich subjected to
low-velocity impact. Compos Part A 2013;52:1222.
[20] Zhu S, Chai GB. Damage and failure mode maps of composite sandwich panel
subjected to quasi-static indentation and low-velocity impact. Compos Struct
2011;101:20414.
87
[21] Olsson R, Block TB. Criteria for skin rupture and core shear cracking during
impact on sandwich panels. In: 19th Int conf on composite materials.
Montreal, Canada; 2013.
[22] Gibson LJ, Ashby MF. Cellular solids. 2nd ed. Cambridge: Cambridge University
Press; 1999.
[23] HTS 5631 800/1600tex 12/24K t0/Z10 data sheet. Document PLS 017 Rev E 01/
29/07. Toho Tenax America Inc.; 2007.
[24] OBrien TK, Salpekar SA. Scale effects on the transverse tensile strength of
graphite epoxy composites. In: Composite materials and design (Eleventh
volume), ASTM STP1206, Ed: Camponeschi ET, Jr. ASTM, Philadelphia; 1993. p.
2352 [also NASA TM 107637].
[25] Product information: ROHACELL RIST. January 2011 edition. Evonik Industries.
Darmstadt, Germany.
[26] Elber W. The effect of matrix and ber properties on impact resistance. Tough
Composite Materials, vol. 2334. NASA Conference Publication; 1984. p.
99121.
[27] Olsson R. A review of impact experiments at FFA during 1986 to 1998. FFA TN
1999-08. Bromma: The Aeron Res Inst of Sweden, Bromma; 1999.
[28] Olsson R. Large mass impact tests on rectangular composite laminates with
various layups. FFA TN 2000-01. The Aeron Res Inst of Sweden, Bromma;
2000.
[29] Garber DP. Tensile stress-strain behavior of graphite/epoxy laminates. NASA
Contractor Report 3592. NASA; 1982.
[30] Olsson R, Iwarsson J, Melin LG, Sjgren A, Solti J. Experiments and analysis
of laminates with articial damage. Compos Sci Technol 2003;63(2):
199209.
[31] Lagace PA, Williamson JE, Tsang PHW, Wolf E, Thomas S. A preliminary
proposition for a test method to measure impact resistance. J Reinf Plast
Compos 1993;12(5):584601.
[32] Hoo Fatt MS, Sirivolu D. Impact perforation of composite sandwich materials.
In: 16th Int conf on composite materials. Kyoto, Japan; 2007.
[33] Zhu S. Private communication. 19 October 2014.