Impact Damage and Energy-absorbing Characteristics and Residual In-plane
Compressive Strength of Honeycomb Sandwich Panels
G. Zhou and M.D. Hill†
*
Department of Aeronautical and Automotive Engineering, Loughborough University,
Loughborough, Leicestershire, LE11 3TU, UK
†
Composites Research Centre, GKN Aerospace Engineering Services, Isle of Wight
PO32 6LR, UK
ABSTRACT: An experimental study of the in-plane compressive behaviour of both aluminium
and nomex composite sandwich panels with 8 ply carbon/epoxy skins was conducted. All
sandwich panels were impact-damaged with a range of impact energies from 1 J to 55 J.
Dominant damage mechanisms were found to be core crushing, skin delamination and fracture
with the former two absorbing most of the impact energy. While the intact panels failed in
region close to one loaded end, all the impact-damaged nomex panels failed around the midsection region. Two thirds of the aluminium panels also failed in the mid-section region and one
third failed in the loaded end region. The presence of the core played a unique role in in-plane
compression with a substantial stabilising support to the skins, which counteracted the
deleterious effect of impact damage. The in-plane compressive behaviour has shown the
combined effects of impact damage and the core in a complex manner.
KEY WORDS: sandwich panel, honeycomb, impact damage, CAI strength, impact damage
tolerance.
INTRODUCTION
Composite sandwich constructions are widely used in aerospace structures due to their light
weight, high strength-to-weight and stiffness-to-weight ratios, good buckling resistance, good
energy-absorbing capacity and design versatility. A major concern with the mechanical
performance of these sandwich structures is their susceptibility to localised manufacturing
defect and/or impact damage; the latter could be caused by tool dropping, hail stones, or
runway debris. As a result, a multitude of damage mechanisms such as skin-core debonding,
core crushing and/or shear failure in addition to skin delamination and fibre fracture could
occur. To maintain the aforementioned advantages in the case of any of these impact events or
manufacturing defect, both mechanical properties from skins and core and geometric properties
must be tailored in a design analysis such that the sandwich structures are damage-tolerant. This
requires a thorough understanding of how the damage mechanisms and their energy-absorbing
characteristics affect a reduction of the in-plane compressive strength (commonly known as
compression-after-impact (CAI) strength) of the damaged sandwich structures and of their
damage tolerance performance.
A great deal of research has been carried out to investigate CAI strength in damaged sandwich
structures as reviewed in [1-3], and thus our overview of the past research will be restricted to
those for aerospace applications only. Since a standard test method for CAI strengths of
sandwich panels, equivalent to the CAI standard ASTM-D 7137/D 7137M for monolithic
laminate plates [4], is yet to be developed, over the years very different approaches of
evaluating CAI strengths have been employed, involving 4-point beam bending [5-6], wide
coupon compression [7], wide column compression [8-16] and panel in-plane compression [17
Author to whom any correspondence should be addressed. G.Zhou@Lboro.ac.uk.
1
20]. Clearly most of research works adopted the wide column compression approach, in which
the unloaded edges of sandwich specimens were left free. This approach was relatively simple
and often produced reasonably well defined failure mechanisms in the test specimens. The
employment of the panel in-plane compression approach, of present interest, was relatively
recent. With the unloaded edges of sandwich specimens being simply supported, their
compressive behaviour was much more complex. Although the compressive failure
mechanisms induced by these two approaches might look similar in some way, the respective
contributions of the skins and core to the reduction of CAI strengths could be very different. At
present, there is little understanding of these differences, which is vitally important to the
assessment of impact damage tolerance. McGowan and Ambur [17] investigated impact
damage using an impactor of 12.7 mm diameter and CAI strength of graphite/epoxy skinned
Korex core sandwich panels with the dimensions of 254 mm by 127 mm. They found that the
global compressive response of the panels was not affected by the presence of impact damage
with an area of less than 8 mm in diameter. However, the residual compressive loads measured
showed interestingly a steady reduction with the increase of impact energy. Tomblin and his
associates [18-20] conducted a series of impact and CAI tests using various sandwich
configurations. They found that an impactor of 76.2 mm diameter resulted in lower CAI
strengths than that of 25.4 mm diameter at the given energy level. They also observed that the
sequence of failure modes was dependent on the skin stiffness and the through-the-thickness
normal properties of the core.
The present work reports results of an investigation, using the panel in-plane compression
approach, into the in-plane compressive behaviour of intact and impact-damaged composite
sandwich panels with both aluminium and nomex honeycombs. In early reports [21-22],
damage mechanisms induced in both aluminium and nomex honeycomb sandwich panels were
identified; the effects of skin thickness, core density and material, indenter nose shape, panel
diameter and support condition on the damage characteristics were studied. The energyabsorbing characteristics of the identified damage mechanisms were examined. This paper
reports the results of a systematic investigation of how those characteristics affect the in-plane
compressive behaviour of impact-damaged composite honeycomb sandwich panels.
SANDWICH MATERIALS AND PANEL MANUFACTURE
Composite skins were made of unidirectional carbon/epoxy T700/LTM45 prepreg with a
nominal ply thickness of 0.128 mm. A symmetric cross-ply lay-up of (0/90)2s was selected due
primarily to strength considerations. Although four skin thicknesses, each with 4 plies, 8 plies,
12 plies, or 16 plies, have been made, the only 8 ply skins were used in fabricating in-plane
compression panels. Two types of honeycomb cores used were 5052 aluminium with a density
of 70 kg/m3 (4.4-3/16-15) and nomex with a density of 64 kg/m3 (HRH-3/16-4.0). The core
depth of both aluminium and nomex honeycombs was 12.7 mm and a nominal panel thickness
was 14.7 mm. The nomex honeycomb was dried in an oven overnight at 65°C before being
bonded to the skins to remove any moisture absorbed from the atmosphere. Adhesive VTA260
was selected for interfacial bonding.
Skin laminates of 300 × 200 mm were laid up and cured first in an autoclave at 60C under a
pressure of 0.62 MPa (90 psi) for 18 hours. To aid skin-core adhesion, the cured skin laminates
were degreased before bonding. The designated 0° direction of carbon fibres within the skins
was aligned with the ribbon direction of honeycomb core. For aluminium panels, each skin was
separately bonded to the core in an oven at 80C for 5 hours under a pressure of 0.1 MPa (15
psi). For nomex panels, skins were bonded to the core in an autoclave under the same
conditions. Because of the condensation of nomex cells (a couple of cells around the panel
edges), the pieces of nomex core used were slightly larger than the skins so that the condensed
2
cells could be trimmed off after bonding. Each sandwich panel was then cut into two nominal
200 mm × 150 mm specimens with the longer side aligned with the direction of compressive
loading. Back-to-back strain gauges were bonded on the panel surfaces at selected locations in
both the longitudinal and transverse directions (shown in Figure 2(a)) to monitor mean (or
membrane) and panel bending strains. These strain data allowed both the local and global
behaviour of the panels to be examined.
EXPERIMENTAL PROCEDURES
Drop-weight Low-velocity Impact Tests
Low-velocity impact tests were carried out on a purpose-built instrumented drop-weight impact
rig in Figure 1(a) by using a hemispherical (HS) impactor of 20 mm diameter with a 1.5 kg
mass. Desired impact energies were regulated by adjusting drop height and they ranged from 1
J to 55 J in this investigation. Each rectangular carbon/epoxy plate of 200 mm by 150 mm with
a circular testing area of 100 mm in diameter was clamped by using a clamping device, as
illustrated in Figure 1(b). A pair of photodiodes fixed over a known distance and two timers
were used to record respective times that were used to calculate both impact and rebound
velocities. This allowed absorbed energies to be calculated directly. For high impact energies,
rebound velocities could be slightly underestimated due to the substantial panel deflections,
thereby overestimating the absorbed energies. Both impact force and strain gauge responses
could be recorded by a Microlink 4000 data acquisition system with a sampling rate of 50 s.
At each selected impact energy level, three impact tests were conducted with one being
diametrically cut up for examination of damage mechanisms and the remaining two being
reserved for subsequent in-plane compression test. The absorbed energy could also be estimated
via the closed area under load-displacement curves. While surface dent was measured after each
impact test, crushed core depth was measured over the cut cross section. All impact test results
along with the extents of crushed core and skin delaminations are summarised in Table 1.
In-plane Compression Test
As part of specimen preparations for compression test, the core at the panel ends intended for
compressive load application was removed to a depth of about 5 mm (slightly more than one
cell size). Epoxy end pots were cast between the two skins to prevent an end-brooming failure
and after curing the two potted ends were machined to parallel. In each in-plane compression
test, a panel was placed in a purpose-built support jig, as illustrated in Figure 2(b), such that the
L or ribbon direction of the honeycomb core was aligned with the loading direction. The jig
provided simple support along the unloaded edges, which were free to move in the width
direction during loading. Quasi-static load was applied to the panel at the machined ends via a
universal testing machine at speed of 1 mm/min. Although the loaded ends were not clamped,
they were effectively close to the clamped condition but without surface clamping pressures.
Load, strain and cross-head displacement in all tests were recorded through a data acquisition
system at a sampling rate of 0.5 Hz. All tested panels were cut up for a study of damage
mechanisms. All in-plane compression results for both intact and impact-damaged panels are
summarised in Table 2. To aid an understanding of the in-plane compressive behaviour of
sandwich panels, a small number of 16 ply monolithic cross-ply panels were also compressiontested.
DAMAGE CHARACTERISTICS IN SANDWICH PANELS
As an assessment of the compressive strength reduction of impacted sandwich panels required
quantitative information on damage states with varying severity induced over the range of
3
impact energies, the initiation and propagation of those induced damage mechanisms must be
characterised in terms of impact energy, absorbed energy and damage extent. As the strain rate
sensitivity of the aluminium honeycomb induced between 1 m/s and 6 m/s impact velocity was
considered to be very small, carbon/epoxy skin and nomex core were generally strain-rate
insensitive within this impact velocity range. As a result, this characterisation effort was
significantly facilitated in combination with much more cost-effective quasi-static bending and
indentation tests and with systematic microscopic inspections of cut cross sections in [21-22].
Unless expressly stated, the damage characteristics as discussed below were restricted to those
associated with sandwich panels with the 8 ply skins, either impacted by the HS impactor or
loaded by the HS indenter.
For the aluminium sandwich panels, the initial damage threshold that was indicated by either a
load drop or a significant loss of the slope of a responsive curve occurred at the average force of
0.51±0.03 kN (measured quasi-statically). Such indication could be seen in Figure 3 from two
sample load-displacement curves from both quasi-static and impact testing. Under the impact
condition, the initial damage threshold was exceeded even at the lowest impact energy of 1.43
J. As the stabilised compressive and crush strengths of the core from manufacturer are usually
obtained using a flat-ended (FE) indenter, they could not be used directly to check to see if the
threshold was associated with core crushing here. Nevertheless, further diagnostic tests in
conjunction with a systematic cross-sectioning of tested specimens revealed that this incipient
damage was due to a combination of core crushing and small delaminations in the loaded (or
top) skin in the shape of a cone towards the skin-core interface. These two damage mechanisms
did not seem to be related. The estimated contact pressures of the HS indenter/impactor ranged
from 2.54 MPa at the apex to 2.11 MPa on the annular region of about 5 mm radius [23] and
they were greater than the core crush strength of 1.71 MPa but significantly less than the
stabilised compressive strength of 4.1 MPa. This discrepancy could be attributed to the fact that
the core crushing in the sandwich panels in bending induced by the HS indenter involved
substantial honeycomb cell rotations, tearing and oblique crushing in addition to normal plastic
folding [21], which was the primary crushing mechanism in the measurement of the crush
strength. As impact energy was increased, the damage mechanisms in the sandwich panels were
characterised by the continued core crushing and by propagation of the top skin delaminations.
A cross-sectional view of impact-damaged aluminium panel (Al 13J 2) is shown in Figure 4. At
about 25 J, top skin fracture occurred and the extent of damage in both the skin and core was
very significant as shown in Figure 5. The extent of crushed core was the same as the extent of
skin delamination at the low impact energies (see Figure 8(a-b)). The gap between the two
started to grow when impact energy was greater than 12 J and became greater for the given
higher impact energy. The surface dent depth (in Table 1) was also greater with the greater
impact energy, as expected. There was no local skin-core interface debonding found in all tests
before the ultimate load. The maximum depth of the crushed cells at an impact energy of 25 J
reached only about the middle of the core thicknesses and the bottom skin remained intact on
all occasions.
The damage characteristics of the nomex panels were more or less the same as the aluminium
ones as a cross-sectional view of impact-damaged nomex panel (Nom 16J 2) shows in Figure 6,
though nomex cells were fractured with a lesser degree of cell folding and showed a substantial
spring-back upon unloading. Furthermore, the initial flexural rigidity of the nomex sandwich
panels (within the elastic range) was significantly lower than those of the aluminium ones due
to the fact that the shear properties of the nomex cores were significantly lower. In addition, the
local indentation in the nomex panels was found to be very dominant. This is interesting in
considering the fact that the impacted nomex panels exhibited much less surface indent with the
maximum depth of crushed cells reaching not even the middle of the core thickness from the
23-J impact test, as can be seen in Figure 7.
4
ENERGY ABSORPTION CHARACTERISTICS IN SANDWICH PANELS
The energy absorption characteristics of both aluminium and nomex sandwich panels were
established respectively over three regions shown in Figures 8(a-b) and 9(a-b) in terms of
damage extent and absorbed energy. In Figure 8(a-b), both extents of skin delamination and
crushed core exhibit distinctive transitions at around 5 J and 23 J, respectively, thereby defining
those three regions. In Region 1, the minimum impact energy required to induce skin
delamination and core crushing can be extrapolated to be around 0.75 J for the current sandwich
material systems. The subsequent trends for these two damage mechanisms seem nonlinear,
irrespective of the type of core materials. The magnitudes of both extents at any given impact
energy are roughly the same and this feature continues up to the middle of Region 2, which
approximately corresponded to about 12 J. These characteristics seem to suggest that the
respective propagations of core crushing and skin delamination were collectively responsible
for absorbing most of the impact energies. Nevertheless, among the two damage mechanisms,
impacted skin delaminations and core crushing, the latter was believed to be the major damage
mechanism that absorbed impact energy significantly more than the former. As the interlaminar
shear (ILS) strength of cross-ply skins in the present sandwich panels is around 62 MPa [22]
and the core crush strengths are 4.02 MPa and 3.93 MPa, respectively, it was thus much easier
for the impact energy to be dissipated throughout the continued core crushing in the impact
direction than driving the local ILS stress level up to above the ILS strength in the annular
region, which was substantially away from the impactor in the direction perpendicular to the
impact direction.
From the middle of Region 2 onwards, the respective extents of skin delamination and crushed
core start separating, with the former being substantially less than the latter. These diverging
trends have continued into Region 3 when fibre fracture in the impacted skin started occurring.
While the diverging trends of both damage mechanisms in Region 2 are linear for the
aluminium sandwich panels, the core crushing extent for the nomex sandwich panels seems to
escalate with an increase in impact energy. This observation seems to suggest that the nomex
honeycomb absorbed impact energy through the extra extension of core crushing in the plane of
the sandwich rather than through the continuous cell fracturing in the loading direction. In
Region 3, while both extents start decreasing for the aluminium sandwich panels and thus
indicate localised penetration of the impactor into the sandwich panels, they seem to level off
for the nomex sandwich panels. It is noted that the gap between the two extents for the nomex
panels is much wider due likely to the lack of the plastic folding capability of nomex in addition
to its relatively low flexural rigidity associated with low through-the-thickness shear properties.
The maximum crushed core depth at an impact energy of 25 J reached only the middle of the
core thickness in both types of panels.
In Figure 9(a-b), it is interesting to note that the rate of energy absorption in the damaged panels
can be seen to be linear and remain constant throughout Regions 1 and 2 for both types of
panels. This constant level of energy absorption is about 67% for the aluminium sandwich
panels and about 59% for the nomex sandwich panels. This strongly suggests that the greater
percentage of energy absorption occurred in the through-the-thickness direction in Region 2.
This trend continues up to either an impact energy of about 23 J or Region 3. This finding is in
accordance with the early observations about the nonlinear trends and divergence of the two
damage extents. At the beginning of Region 3, fibre fracture of the impacted skin set in and the
amount of energy absorption went up to over 94%, irrespective of the type of core materials.
This just confirms that this final similar level of energy absorption was attributed to multiple
ply fractures in the impacted skin, which were identical for both types of panels.
5
In Figure 10, the surface dent depths produced by impact in both aluminium and nomex
sandwich panels are plotted against impact energy. The BVID criterion from a perspective of
maintenance inspections requires that an 8 J impact produces a dent depth of no more than 0.5
mm, which is regarded as ‘barely visible’ under normal lighting condition from a distance of
about 1.5 m. In the figure, the aluminium sandwich data indicate the BVID energy threshold of
only about 3 J. In other words, the 8 J impact would produce a surface dent depth of around 1.3
mm in the current panels with 8 ply skins and 70 kg/m3 core. On the contrary, the nomex
sandwich panels seem to meet the BVID criterion with ease with the BVID energy threshold of
being about 25 J. This is of course deceptive, as the extents of core crushing and skin
delamination in both types of sandwich panels were similar up to a level of 10 J. Thus such low
dent depths in the nomex sandwich panels were due primarily to the nomex core’s capability of
storing part of the impact energy elastically, which were released during unloading in the form
of spring-back.
RESIDUAL IN-PLANE COMPRESSIVE BEHAVIOUR
In-plane Compressive Behaviour
In addition to the damage characteristics as discussed above, two other factors were considered
also to contribute to the complexity of the in-plane compressive behaviour of the damaged
sandwich panels. One was that the two skins during in-plane compression were locally
stabilised and supported, to some extent, by the core in the through-the-thickness direction.
Unless it was very extensive with respect to panel width, the impact damage alone may not
have been sufficient to weaken the local stability of the skins in the mid-section region. As a
result, the stabilising and supporting effect of the core could significantly reduce the effect of
the impact damage. This could particularly be the case if the shear rigidity and compressive
strength of the core was relatively high for the given flexural rigidity of the skins and if the
impact damage was relatively small. Thus, much of the in-plane compressive characteristics
established from monolithic composite panels [24-25] could not necessarily be the same as
those from the sandwich panels. The other was that the presence of the impact damage resulted
in through-the-thickness local asymmetry with respect to the compression loading and
supporting conditions. Therefore, the in-plane compressive strength reduction of the damaged
sandwich panels was attributed not only to the effect of impact damage but also to asymmetry
and the relative magnitude of the shear rigidity and compressive strength of the core.
For the intact sandwich panels with no damage and thereby with no local asymmetry, the first
aforementioned factor could have much greater influence over the compressive behaviour.
Consequently, the likelihood of failure at one of panel ends increased significantly. From the
compressive responses of the intact aluminium sandwich panels as examined in [26], a tilting of
one panel end could be seen at the early stage of loading via the bending strain response of the
far-field strain gauges (SG1) as shown in Figure 11. As the local strain gauges (SG2) on the
mid-section exhibited very small bending strain with large mean strain in the figure, these
responses provided no indication of local buckling and thus the panel failed in the region close
to the loaded end, as shown in Figure 12. The compressive response of the intact nomex panels
was very similar in terms of both strain response and failure characteristic.
Once impact damaged, about two thirds of the aluminium sandwich panels failed around the
mid-section region, as one example from specimen ‘Al 13J 2’ shows in Figure 13, and all the
impact damaged nomex sandwich panels failed in the same way. The remaining third of the
impacted aluminium sandwich panels failed in a similar way to the control panels with the
failure being close to one loaded end. One of such panels’ strain responses are shown in Figure
14, which can be seen to be similar to those of the intact panel in Figure 11. Moreover, this
6
phenomenon did not appear to be affected by the severity of the impact damage, as the impact
energies applied to those four panels spread from 3.7 J to 25 J. In the last case with the impact
energy of 25 J, the respective extents of delamination in the impacted skin and core crushing
were close to a half of the panel width. A further examination of the aluminium sandwich
panels revealed that there was little difference in CAI strength between those with the midsection failure and those with the end failure. Nevertheless, the majority of the impact damaged
nomex sandwich panels not only failed around the mid-section region but showed also classic
strain reversal, as one exhibits in Figure 15, which immediately triggered catastrophic failure.
As the aluminium core had substantially greater through-the-thickness shear and compressive
properties than the nomex, these differences increased the probability of the aluminium
sandwich panels failing at one of panels ends for the given impact damage.
Contribution of Honeycomb Core
The early discussion of the in-plane compression test results clearly suggests that the presence
of the core between the two skins during in-plane compression played a complicated role in the
compressive behaviour of the sandwich panels, irrespective of whether or not they were impact
damaged. The intact sandwich panels had the average panel width-to-thickness ratio of 10 and
the significant longitudinal flexural rigidity. With the core providing a stabilising support to the
skins in the through-the-thickness direction through both transverse shear and normal
compression properties, thus both skins of the panels in the mid-section region were constrained
from deforming and a tendency of local buckling in the mid-section region was very small.
Moreover, with the two skins being apart, the effective longitudinal load transmission was
limited. As a result, all the intact panels failed at the location of being close to one of the panel
ends.
For the impact damaged panels, in some cases, their in-plane compressive resistance was not
weakened sufficiently by even the significant impact damage in terms of damage extent (e.g.
‘Al 9.5J 2’ and ‘Al 25J 2’) so that those panels did not fail respectively in the impact damaged
(or mid-section) region. In those cases, the impact damaged skin was able to provide a level of
the compression resistance similar to the intact skin as in the case of the intact sandwich panels.
Although the imbalance of local bending moments induced by the shift of the mid-plane
towards the intact skin in the mid-section region could facilitate local buckling, the stabilising
effect of the core was greater so that the impact damage in the impacted skin could neither
propagate transversely nor facilitate local buckling.
For the remaining impacted damaged panels which failed in the mid-section region, it was their
stabilised compressive strengths that affected the final failure characteristics of the compressive
behaviour. As the adhesive’s tensile strength of 5.5 MPa was substantially greater than the
stabilised compressive strength of 4.2 MPa for aluminium (and 4.0 MPa for nomex), the
undamaged (distal) skin could have only bent inwards, thereby crushing the core, as the
photographs in Figure 13 show that the zone of the core in the aluminium panels crushed during
in-plane compression was close to the distal skin. As for the damaged skin in the dished shape,
whether it compressed the already crushed core further (see Figure 13) or buckled outward as
shown in Figure 16 depended on, among others, a state of local interfacial bonding, dent depth
and surface curvature. Although both inward and outward local skin buckling were observed,
the latter could be in favour if local debond around the curved region occurred during impact.
This seems to suggest that the denser core and/or core with greater transverse shear stiffness
provides the sandwich panels with stronger compressive resistance, which seems in agreement
with other results in [27]. The only additional feature of the nomex sandwich panels was that
the spring-back of the fractured cells led to a lesser degree of local dishing in the impacted area
and thereby the less dent depths (see Figures 6 and 7), as the fractured cells were close to the
7
mid-plane of the panels. Nevertheless, this lesser local curvature did not seem to affect the
aforementioned characteristics.
IMPACT DAMAGE TOLERANCE ASSESSMENT
The impact damage tolerance of both aluminium and nomex panels can be assessed by plotting
the variation of either CAI strengths or residual compressive (mean) strains against either
damage extent (in the width direction) or impact energy in Figures 17-19. The actual reduction
of their CAI strengths is seen more clearly when CAI strengths were normalised with the
baseline values of the sandwich panels in Figure 18. Although a damage area could also be used
as a damage measure [24-25], the damage extent is an equally useful indicator as demonstrated
here.
In Figure 17, once the panels were impact-damaged, there seems an immediate but moderate
reduction in compressive strength. From 3.7 J to 13 J on the impact energy axis, however, the
variation of CAI strengths is limited, as a state of damage in these panels was dominated by a
steady growth of core crushing and skin delaminations. Although the growth of the extents
exceeded over 50% of the panel width, they seemed insufficient to trigger local buckling and/or
sideway propagation of skin delaminations due likely to the aforementioned contribution of the
core for stabilising the mid-section region. This seems to suggest that much of the effect of
impact damage was ‘cancelled out’ so that it could not lead to a further reduction of CAI
strengths. At the BVID threshold of 8 J, the reduction of CAI strengths was unfortunately
substantial. When the impact energy level reached over 20 J, a further reduction of CAI
strengths became noticeable because of the fracture of the impacted skins. At highest impact
energies (about 26 J for the aluminium sandwich panels and 28 J for the nomex sandwich
panels), the reduction of compressive strength was about 50% for both sandwich panels. In
addition, it is noted that the different failure locations among the CAI tested panels are not
distinguishable over the values of CAI strength. Nevertheless, this does imply that the relative
magnitude of the shear rigidity of the aluminium honeycomb was at the critical state such that a
reduction of the shear properties could lead to the consistent mid-section failure, or vice versa.
The normalised CAI strengths results are presented in Figure 18 with compressive strength
retention factor plotted against damage extent. The data trends are very similar to those
observed from Figure 17. The residual compressive mean strains obtained from the far-field
strain gauge locations can also be used in the impact damage tolerance assessment in Figure 19,
with the advantage for the fact that allowable strains are often used in a composite structure
design. Although the overall trends here are similar to those in Figure 17, the values of strain
loss seem to be slightly worse than CAI strengths.
Although the trends of data in the form of either CAI strengths or residual compressive strains
are similar when plotted against impact energy in Figure 17, the fact that a 75% increase in the
extent of skin delaminations did not lead to a further reduction in CAI strengths seem to
indicate another possibility. That is, the impact energy may not be an effective damage
measure, as the majority of the impact energies were actually absorbed by a crushing of the
honeycomb cores, while the skins that absorbed less energy provided the major compression
resistance. Unlike in damaged monolithic laminates, both impact energy and energy absorption
were intimately related to the creation of damage or fracture surfaces, which in turn weakened
the laminates in subsequent in-plane compression, so that their variation corresponded to the
variation of CAI strengths. Therefore, they have proven to be effective damage measures along
with damage area [24-25]. However, sandwich panels had a role-sharing multi-functioning
characteristic and posed a significant challenge. Unless the through-the-thickness shear and
normal properties of the core were low, they could significantly affect the in-plane compressive
8
behaviour of the sandwich panels as in the present case. Thus, a further investigation over the
role of the core in the in-plane compression in the future is essential to the impact damage
tolerance assessment of sandwich panels.
CONCLUSIONS
Both aluminium and nomex composite sandwich panels were impact damaged with energies
ranging from 1 J to 55 J. Dominant damage mechanisms were found to be core crushing, skin
delamination and skin fracture with the former two absorbing the most impact energies. As the
initial damage occurred at relatively low load, energy absorption prior to the skin fracture was
dominated by cell crushing in both in-plane and through-the-thickness directions in the initial
region and primarily in the latter direction in the extent growth region. Different core materials
with a similar density made little difference on either the damage or energy-absorbing
characteristics.
Both intact and impact damaged composite sandwich panels were tested in in-plane
compression. They failed in three different modes because the presence of the core affected the
way in which the panels failed. The presence of the core provided the sandwich panels with
high flexural rigidity, enhanced shear rigidity and normal compressive strength such that the
likelihood of local buckling around the mid-section region in these panels was substantially
reduced through local stabilisation and support. Because of this, the intact panels failed in a
region being close to one loaded end due to limited load transmission. Two thirds of the impact
damaged aluminium panels along with all the damaged nomex panels failed around the midsection region due to the significant effect of the impact damage. However, one third of the
impact damaged aluminium panels failed in the loaded end region similar to the intact panels.
This was found because the presence of the core seemed to have counteracted the deleterious
effect of the impact damage in such a way that local buckling or sideway propagation of the
impact damage in those panels was constrained.
REFERENCES
1. Abrate, S. (1997). Localized impact on sandwich structures with laminated facings, Applied
Mechanics Review, 50: 69-82.
2. Tomblin, J.S., Lacy, T., Smith, B., Hooper, S., Vizzini, A. and Lee, S. (1999). Review of
damage tolerance for composite sandwich airframe structures, Report DOT/FAA/AR-99/49,
USA.
3. Hayman, B. (2007). Approaches to damage assessment and damage tolerance for FRP
sandwich structures, J. of Sandwich Structures and Materials, 9: 571-596.
4. ––. (2005). Standard test method for compressive strength properties of damaged polymer
matrix composite plates, ASTM-D 7137/D 7137M – 05.
5. Gottesman, T., Bass, M. and Samuel, A. (1987). Criticality of impact damage in composite
sandwich structures, In: Proc. of 6th ICCM and 2nd ECCM, London, pp. 3.27-3.35.
6. Charles, J.P. and Guedra-Degeorges, D. (1991). Impact damage tolerance of helicopter
sandwich structures, In: Proc. of 23rd Int. SAMPE Technical Conference, New York.
7. Akay, M. and Hanna, R. (1990). A comparison of honeycomb-core abd foam-core carbonfibre/epoxy sandwich panels, Composites, 21: 325-331.
8. Kassapoglou, C., Jonas, P.J. and Abbott, R. (1988). Compressive strength of composite
sandwich panels after impact damage: an experimental and analytical study, J. of
Composites Technology and Research, 10: 65-73.
9. Palm, T.E. (1991). Impact resistance and residual compression strength of composite
sandwich panels, In: Proc. of 8th ICCM, Honolulu, pp. 3-G-1-3-G-14.
9
10. Tsang, P.H.W. and Lagace, P.A. (1994). Failure mechanisms of impact-damaged sandwich
panels under axial compression, AIAA-94-1396, pp. 745-754.
11. Kassapoglou, C. (1996). Compression strength of composite sandwich structures after
barely visible impact damage, J. of Composites Technology and Research, 18: 274-284.
12. McGowan, D.M. and Ambur, D.R. (1997). Damage-tolerance characteristics of composite
fuselage sandwich structures with thick facesheets, NASA TM 110303, USA.
13. Cvitkovich, M.K. and Jackson, W.C. (1999). Compressive failure mechanisms in composite
sandwich structures, J. of the American Helicopter Society, Oct.: 260-268.
14. Moody, R.C., Harris, J.C. and Vizzini, A.J. (1999). Curvature effect on the damage
tolerance iof impacted damaged composite sandwich panels, In: Proc. of 12th ICCM, Paris.
15. Lagace, P.A. and Mamorini, L. (2000). Factors in the compressive strength of composite
sandwich panels with thin facesheets, J. of Sandwich Structures and Materials, 2: 315-330.
16. Moody, R.C., Harris, J.C. and Vizzini, A.J. (2002). Scaling and curvature on the damage
tolerance iof impacted composite sandwich panels, J. of Sandwich Structures and Materials,
4: 71-82.
17. McGowan, D.M. and Ambur, D.R. (1998). Damage characteristics and residual strength of
composite sandwich panels impacted with and without a compression loading, AIAA-981783, pp. 713-723.
18. Tomblin, J.S., Raju, K.S., Liew, J. and Smith, B.L. (2001). Impact damage characterization
and damage tolerance of composite sandwich airframe structures, Report DOT/FAA/AR00/44, USA.
19. Tomblin, J.S., Raju, K.S., Liew, J. and Smith, B.L. (2002). Impact damage characterization
and damage tolerance of composite sandwich airframe structures - Phase II, Report
DOT/FAA/AR-02/80, USA.
20. Tomblin, J.S., Raju, K.S. and Arosteguy, G. (2004). Damage resistance and tolerance of
composite sandwich panels - scaling effects, Report DOT/FAA/AR-03/75, USA.
21. Zhou, G., Hill, M.D., Loughlan, J. and Hookham, N. (2006). Damage characteristics of
composite honeycomb sandwich panels in bending under quasi-static loading, J. of
Sandwich Structures and Materials, 8: 55-90.
22. Zhou, G., Hill, M.D. and Hookham, N. (2007). Investigation of parameters governing the
damage and energy-absorbing characteristics of honeycomb sandwich panels, J. of
Sandwich Structures and Materials, 9: 309-342.
23. Zhou, G. (1996). Static behaviour and damage of thick composite laminates, Composite
Structures, 36: 13-22.
24. Zhou, G. and Rivera, L. (2005). Investigation for the reduction of in-plane compressive
strength in preconditioned thin composite panels, J. of Composite Materials, 39: 391-422.
25. Zhou, G. and Rivera, L. (2007). Investigation for the reduction of in-plane compressive
strength in preconditioned thick composite panels, J. of Composite Materials, 41: 19611994.
26. Zhou, G. and Hill, M.D. (2007). Impact damage and residual compressive strength of
honeycomb sandwich panels, In: Proc. of 16th ICCM, Kyoto.
27. Nokkentved, A., Lundsgaard-Larsen, C. and Berggreen, C. (2005). Non-uniform
compressive strength of debonded sandwich panels - I. Experimental investigation, J. of
Sandwich Structures and Materials, 7: 461-482.
10
Figure Caption
Figure 1. (a) Instrumented drop-weight impact test rig and (b) experimental set-up of
a clamped panel for bending test
Figure 2. (a) Specimen dimensions and strain gauge locations and (b) experimental set-up of
a panel in compression with unloaded edge support rig
Figure 3. Load-displacement curves from 8 ply skinned aluminium honeycomb
sandwich panels
Figure 4. An aluminium sandwich panel impacted at 13 J
Figure 5. An aluminium sandwich panel impacted at 25 J
Figure 6. A nomex sandwich panel impacted at 16 J
Figure 7. A nomex sandwich panel impacted at 23 J
Figure 8. Extent of delamination and crushed core in (a) aluminium honeycomb sandwich
panels and (b) nomex honeycomb sandwich panels
Figure 9. Absorbed energy of (a) aluminium honeycomb sandwich panels and
(b) nomex honeycomb sandwich panels
Figure 10. Variation of surface dent depths as a function of impact energy in sandwich panels
Figure 11. Mean and bending strain responses of an intact aluminium sandwich panel
Figure 12. Photographs of an intact aluminium sandwich panel after compression test
Figure 13. Photographs of an impact-damaged aluminium sandwich panel after CAI test
Figure 14. Mean and bending strain responses of aluminium sandwich panel ‘Al 13J 2’
impacted at 13 J
Figure 15. Mean and bending strain responses of nomex sandwich panel ‘Nom 16J 1’
impacted at 16 J
Figure 16. Photographs of an impact-damaged aluminium sandwich panel after
CAI test showing crushed core around the mid-section region
Figure 17. Variation of CAI strength with IKE for (a) aluminium honeycomb sandwich panels
and (b) nomex honeycomb sandwich panels
Figure 18. Variation of compressive strength retention factor with damage length for
(a) aluminium honeycomb sandwich panels and (b) nomex honeycomb
sandwich panels
Figure 19. Variation of residual compressive strain with damage length for (a) aluminium
honeycomb sandwich panels and (b) nomex honeycomb sandwich panels
11
Table Caption
Table 1. Summary of impact tests on honeycomb sandwich panels
Table 2. Summary of CAI tests on honeycomb sandwich panels
12
Table 1. Summary of impact tests on honeycomb sandwich panels
Panel ID
IKE
Al 1.4J
Al 3.7J 1
Al 3.7J 2
Al 3.7J 3
Al 3.7J 4
Al 3.7J 5
Al 5J 1
Al 5J 2
Al 5J 3
Al 9.5J 1
Al 9.5J 2
Al 9.5J 3
Al 13J 1
Al 13J 2
Al 13J 3
Al 16J
Al 20.6J
Al 25J 1
Al 25J 2
Al 25J 3
Al 30.6J
Nom 1.5J
Nom 3.0J 1
Nom 3.0J 2
Nom 5J 1
Nom 5J 2
Nom 9J 1
Nom 9J 2
Nom 9J 3
Nom 16J 1
Nom 16J 2
Nom 20J
Nom 23J
Nom 25J 1
Nom 25J 2
Nom 28J
Nom 30J
J
1.43
2.60
3.66
3.62
3.69
3.69
5.20
5.40
5.24
9.54
9.51
9.06
13.68
13.76
13.18
16.24
20.63
24.34
25.54
25.18
30.57
1.54
3.04
3.41
5.13
5.28
9.23
9.29
9.43
16.29
16.37
20.15
22.93
25.45
25.80
28.20
30.03
Absorbed Percentage
impact
energy
energy
absorption
J
0.63
0.44
1.75
0.67
2.44
0.67
2.45
0.68
2.54
0.69
2.39
0.65
3.49
0.67
3.55
0.66
3.59
0.69
6.11
0.64
6.60
0.69
6.20
0.68
9.19
0.67
9.22
0.67
8.92
0.68
11.24
0.69
14.01
0.68
24.24
1.00
24.15
0.95
24.32
0.97
30.49
1.00
0.7
0.45
1.65
0.54
1.83
0.54
2.83
0.55
2.74
0.52
5.45
0.59
5.04
0.54
4.95
0.52
10.03
0.62
9.64
0.59
12.12
0.60
12.67
0.55
24.54
0.96
14.76
0.57
26.54
0.94
29.24
0.97
13
Delamination
extent in the
impacted skin
mm
23
30
42
51
53
55
69
66
62
17
31
30
35
42
45
59
61
62
53
58
66
Crushed
core
extent
mm
23
32
39
50
57
67
75
74
65
22
33
29
33
38
59
64
77
98
101
100
101
Surface
dent
depth
mm
0.20
0.60
0.75
0.60
1.20
1.00
1.50
1.30
2.00
2.00
2.00
2.00
2.50
3.50
4.50
2.00
2.50
0
0
0
0
0
0
0
0
0
0
0
0.25
0.50
0.50
2.00
1.50
Table 2. Summary of CAI tests on honeycomb sandwich panels
CSRF1 Longitudinal Longitudinal Transverse Failure
Panel ID
delamination core damage damage location2
extent
mm
J
kN
MPa
MPa
mm
mm
Al control 1
0
92.00
270.6
0.93
0
0
0
LE
Al control 2
0
111.59 322.0
1.10
0
0
0
LE
Al 3.7J 1
2.60 71.53
207.0
0.71
29
31
150
MS
Al 3.7J 2
3.66 67.11
190.7
0.65
LE
Al 3.7J 3
3.62 72.94
219.0
0.75
42
33
150
MS
Al 5J 1
5.20 75.50
221.8
0.76
38
38
38
FE
Al 5J 2
5.40 67.58
195.3
0.67
37
38
150
MS
Al 9.5J 1
9.54 84.45
243.9
0.83
53
62
150
MS
Al 9.5J 2
9.51 68.32
190.3
0.65
50
54
52
LE
Al 13J 1
13.68 71.54
203.8
0.70
54
70
150
MS
Al 13J 2
13.76 73.00
215.2
0.74
55
67
150
MS
Al 25J 1
24.34 63.16
187.6
0.64
50
100
150
MS
Al 25J 2
25.54 51.00
150.4
0.51
67
85
75
LE
Nom con. 1
0
97.65
283.9
0.97
0
0
0
LE
Nom con. 2
0
102.69 293.1
1.00
0
0
0
LE
Nom 5J
5.13 55.87
162. 4
0.56
35
34
463
MS
Nom 9J
9.23 71.60
205.6
0.70
60
63
150
MS
Nom 16J
16.29 65.01
187.0
0.64
63
93
150
MS
Nom 28J
28.20 46.53
134.1
0.46
75
102
150
MS
1
CSRF – compressive strength retention factor
2
LE, MS and FE denote loaded end, mid-section and far end, respectively
3
Fracture point above cross sectioned plane
IKE Failure
CAI
Load strength
14
(a)
Indenter
Specimen
Steel bolts
Clamping
plates
Steel plate to
raise specimen
25.4mm thick
steel base plate
Box support
Mand machine base
LVDT
(b)
Figure 1. (a) Instrumented drop-weight impact test rig and (b) experimental set-up of a
clamped panel for bending test
15
33.33mm
Loading end
region
SG 1
Epoxy end-pot
Mid-section
SG 2
SG 3
Mid-section
region
45mm
100mm
150mm
Far end region
~15mm
(a)
Unloaded simple
support
Compression load
Specimen
Adjustable bolt
Epoxy
end-pot
(b)
Figure 2. (a) Specimen dimensions and strain gauge locations and (b) experimental set-up of a
panel in compression with unloaded edge support rig
16
3
Quasi-static test
2.5
Impact test at 1.43J
Load (kN)
2
1.5
1
0.5
0
0
0.5
1.5
1
2
2.5
3
Displacement (mm)
Figure 3. Load-displacement curves from 8 ply skinned aluminium honeycomb sandwich
panels
17
Figure 4. An aluminium sandwich panel impacted at 13 J
Figure 5. An aluminium sandwich panel impacted at 25 J
Figure 6. A nomex sandwich panel impacted at 16 J
Figure 7. A nomex sandwich panel impacted at 23 J
18
120
8 ply skins with 70kg/m3 aluminium honeycomb core
Damage length (mm) .
100
Core crushing length
80
60
40
Delamination length
20
3
2
1
0
0
5
10
15
20
25
30
35
30
35
IKE (J)
(a)
120
3
8 ply skins with 64kg/m nomex honeycomb core
Damage length (mm) .
100
80
Core crushing length
60
40
Delamination length
20
1
3
2
0
0
5
10
15
20
25
IKE (J)
(b)
Figure 8. Extent of delamination and crushed core in (a) aluminium honeycomb sandwich
panels and (b) nomex honeycomb sandwich panels
19
35
100% Absorption
3
8 ply skins with 70kg/m
aluminium honeycomb core
.
30
Absorbed energy (J)
25
20
15
10
1
5
3
2
0
0
5
10
15
20
25
30
35
30
35
IKE (J)
(a)
35
100% Absorption
3
8 ply skins with 64kg/m
nomex honeycomb core
.
30
Absorbed energy (J)
25
20
15
10
1
5
3
2
0
0
5
10
15
20
25
IKE (J)
(b)
Figure 9. Absorbed energy of (a) aluminium honeycomb sandwich panels and (b) nomex
honeycomb sandwich panels
20
Surface dent depth (mm) .
5
4
8 ply skins with 70kg/m3 aluminium core
3
2
3
8 ply skins with 64kg/m
nomex core
1
BVID threshold
0
0
5
10
15
20
25
30
35
IKE (J)
Figure 10. Variation of surface dent depths as a function of impact energy in sandwich panels
120
SG2 0°
- mean
SG1 - bending
SG1 - mean
100
SG2 90°
- bending
Compressive load (kN)
SG2 90°
- mean
SG2 0°
- bending
80
60
40
SG1
SG2
20
0
-8000
-6000
-4000
-2000
0
2000
Strain (με)
Figure 11. Mean and bending strain responses of an intact aluminium sandwich panel
21
Figure 12. Photographs of an intact aluminium sandwich panel after compression test
Figure 13. Photographs of an impact-damaged aluminium sandwich panel after CAI test
22
120
Compressive load (kN)
SG2 90°
100
- bending
SG2 0°
- mean
SG1
- mean
SG2 90°
- mean
80
60
SG2 0°
- bending
SG1
40
SG2
SG1
- bending
20
0
-8000
-6000
-4000
-2000
0
2000
Microstrain (με)
Figure 14. Mean and bending strain responses of aluminium sandwich panel ‘Al 13J 2’
impacted at 13 J
120
Compressive load (kN)
SG2 0°
100
- bending
SG2 0°
- mean
SG1 mean
SG2 90° bending
80
60
SG2 90°
- mean
SG1
SG1 bending
40
SG2
20
0
-8000
-6000
-4000
-2000
0
2000
Microstrain (με)
Figure 15. Mean and bending strain responses of nomex sandwich panel ‘Nom 16J 1’ impacted
at 16 J
23
Figure 16. Photographs of an impact-damaged aluminium sandwich panel after CAI test
showing crushed core around the mid-section region
24
Residual compressive strength (MPa)
350
300
250
200
150
100
50
0
0
5
10
15
20
25
30
20
25
30
IKE (J)
(a)
Residual compressive strength (MPa)
350
300
250
200
150
100
50
0
0
5
10
15
IKE (J)
(b)
Figure 17. Variation of CAI strength with IKE for (a) aluminium honeycomb sandwich panels
and (b) nomex honeycomb sandwich panels
25
Compressive strength retention factor .
1.2
1
0.8
0.6
0.4
0.2
Aluminium sandwich panel - end failure
Aluminium sandwich panel - central failure
0
0
10
20
30
40
50
60
70
50
60
70
Damage length (mm)
(a)
Compressive strength retention factor .
1.2
1
0.8
0.6
0.4
0.2
0
0
10
20
30
40
Damage length (mm)
(b)
Figure 18. Variation of compressive strength retention factor with damage length for
(a) aluminium honeycomb sandwich panels and (b) nomex honeycomb sandwich panels
26
Residual compressive strain (με)
-6000
-5000
-4000
-3000
-2000
-1000
Aluminium sandwich panel - end failure
Aluminium sandwich panel - central failure
0
0
10
20
30
40
50
60
70
50
60
70
Damage length (mm)
(a)
Residual compressive strain (με)
-6000
-5000
-4000
-3000
-2000
-1000
0
0
10
20
30
40
Damage length (mm)
(b)
Figure 19. Variation of residual compressive strain with damage length for (a) aluminium
honeycomb sandwich panels and (b) nomex honeycomb sandwich panels
27