51st AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Conference<BR>18th
12 - 15 April 2010, Orlando, Florida
AIAA 2010-2770
2010 AIAA SDM Student Symposium
Barely Visible Impact Damage Evaluation of Composite
Sandwich Structures
Abhendra K. Singh,1 Barry D. Davidson2 and David P. Eisenberg3
Department of Mechanical and Aerospace Engineering, Syracuse University, Syracuse, NY 13244
Michael W. Czabaj1 and Alan T. Zehnder4
Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, NY 14853
Abstract
The nature of quasi-static indentation damage is studied in aluminum honeycomb core sandwich panels
with eight ply, quasi-isotropic, graphite/epoxy face sheets. Parameters that are varied include the core
thickness, core density, face sheet layup and indentor diameter. The majority of induced damage is in the
vicinity of the barely visible threshold. The permanent dent in the panel is found to always be larger than the
contact area of the indentor, and specimens with denser cores exhibit smaller dent diameters for a given dent
depth. Regardless of specimen layup, delaminations occur essentially only at the 3rd, 5th, 6th and 7th interfaces.
Stiffer cores, either in terms of a higher density or, for those cores considered, a smaller thickness, result in
more face sheet delamination for a given indentation event. Regardless of the core, larger delaminations
occur in face sheets that contain only 90° angle changes between adjacent plies in comparison to those that
contain only 45° angle changes. Thus, when delamination area is considered as a damage resistance metric, a
low density core with a [45/90/-45/0]s face sheet will provide the best results of those geometries considered.
However, if dent area or dent depth is chosen as the damage metric, this geometry will provide the least
damage resistance, and a high density core with a [±45/0/90]s face sheet is best. These and similar results are
discussed in the context of choosing the most damage resistant structural configuration and how this
translates to damage tolerance.
I.
Introduction
S
ANDWICH composites are stiff, light-weight structures that offer high resistance to compressive stresses and
bending moments. However, one major problem with these types of structures is impact damage. Relatively low
velocity impacts can cause core crushing, face sheet debonding, delaminations and/or matrix cracking.1-4 This
internal damage can be extensive and can have negative effects on structural performance, even if the external
damage is non-visible or barely visible.5,6 Thus, many studies have focused on barely visible impact damage
(BVID), as this is indicative of the maximum amount of damage that may exist undetected in practical structures.
The goal is that, through understanding the nature of the damage at impact energies in the vicinity of BVID, better
models may be developed that predict the effect of this damage on strength, stiffness and/or life.7
The objective of this study was to experimentally determine the effects of core thickness, core density, face sheet
stacking sequence and indentor diameter on the nature and extent of low velocity impact damage. The primary focus
is on damage states in the vicinity of BVID. Sandwich composites with three types of aluminum honeycomb cores
and four different graphite/epoxy face sheet layups are considered. This is part of a larger effort where this
information is used to develop and refine models for post-impact compressive strength, with the overall goal of
developing a clear mechanistic and parametric understanding of the effects of different variables on impact damage
and compression-after-impact response.7
1
2
3
4
Graduate student member, AIAA
Laura J. and L. Douglas Meredith Professor, Associate Fellow, AIAA
Undergraduate student member, AIAA
Professor and Associate Director for Undergraduate Programs
1
American Institute of Aeronautics and Astronautics
Copyright © 2010 by Abhendra K. Singh. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
II. Damage Creation
All “impact damage” in this work was created through quasi-static indentation (QSI). That is, it has been shown
that the nature and distribution of damage in composite sandwich structures is essentially identical for damage
induced via low-velocity impact versus that induced via QSI.8-11 QSI testing also allows for more consistent and
controllable damage levels; for example, as based on the externally evident dent depth. Details on the QSI procedure
are presented subsequently.
III. Materials, Manufacturing and Test Matrix
All sandwich panels used eight ply quasi-isotropic IM7/8552 graphite/epoxy face sheets and 5052 aluminum
honeycomb core. A total of seven different panel geometries were considered. These were comprised of three
different cores and four different face sheet layups. Each of the three cores used a 3.175 mm cell size. Core type
“C1” had a thickness of 25.4 mm and a density of 49.7 kg/m3. Core type “C2” had a thickness of 16.5 mm and a
density of 49.7 kg/m3, and core “C3” had a thickness of 25.4 mm and a density of 72.1 kg/m3. Thus, with equivalent
face sheets, a comparison of sandwich laminates with C1 vs C2 cores indicates the effect of core thickness, and C1
vs C3 indicates the effects of core density. The four different face sheet combinations that were used are referred to
as Q1, Q2, Q3 and Q4. The corresponding layups were [45/0/-45/90]s, [45/-45/0/90]s, [-45/45/90/0]s and [45/90/45/0]s, respectively. These were intended to show the effects of 45° vs 90° angle changes, as well as the effects of
the locations of the 0° and 90° plies.
In addition to the seven geometric permutations described above, QSI testing was performed with both 25.4 mm
and 76.2 mm diameter indentors. A few exploratory tests were also performed using a 12.7 mm diameter indentor,
and it was found that the post-QSI state of damage and the resulting residual compressive strength was essentially
the same as for specimens that were indented using the 25.4 mm diameter indentors. Thus, to provide more test
replicates at each condition, we chose to concentrate on testing with the 25.4 mm and 76.2 mm diameter indentors.
All specimens were fabricated in the Syracuse University Composite Materials Laboratory (SU-CML). Initially,
panels were manufactured that were 381 mm square. These panels were fabricated in a single autoclave cure cycle
using prepreg tape for the face sheets and 3M AF-555 adhesive for the face sheet to core bond. The ribbon direction
of the core was always placed to coincide with the 0° direction for the face sheet layup. Prior to placing the face
sheets in place, 3M EC 3524 void filling compound was used to pot the core in a manner that, after cutting the panel
into specimens and inducing the damage, an edgewise compressive load could be introduced through the potted
region of the core to obtain compression after impact (CAI) strengths.7 A schematic of a panel after manufacturing
and trimming the edges is shown in Figure 1. Each “line” of potting compound was originally 25.4 mm wide and
spanned the full width of the panel in the 90° direction. After trimming the edges, the potted region that remained
was approximately 12.7 mm wide. The panel was also subsequently cut through the center strip of potting
compound, such that the all edges that would be loaded had a potted region of approximately the same width.
Figure 1. Typical panel dimensions after manufacturing and edge trimming (all dimensions in mm).
2
American Institute of Aeronautics and Astronautics
Figure 1 can also be used to understand how panels were cut into specimens. This figure presents the “nominal
configuration,” which corresponded to each panel being cut into four “full-sized” specimens that were 177.8 mm
long (0° direction and direction of CAI loading) and 152.4 mm wide (90° direction), and two specimens that were
177.8 mm long x 50.8 mm wide for edgewise compression testing12, which was used to obtain undamaged
compressive strengths.7 However, as will be discussed subsequently, for the first several panels manufactured,
“small QSI specimens” were needed to determine the QSI load and indentation levels that would correspond to
BVID. In these cases, instead of cutting the lower right quadrant of the panel into one full sized and one edgewise
compression specimen, this was cut into one specimen that was approximately 100 mm square, two that were
approximately 75 mm square, and two that were approximately 50 mm square. These were used for “QSI evaluation
tests” using the 76.2 mm, 25.4 mm and 12.7 mm diameter indentors, respectively. In addition, in some panels, a 12.7
µm thick x 150 mm long x 25 mm wide teflon insert was placed within the plate in the location corresponding to the
lower right edgewise compression specimen. These produced face sheet flexural test specimens that were used to
calibrate the material properties used in the CAI finite element model.7
Table 1 presents the test matrix listing all full-sized specimens that were used for the QSI and damage evaluation
studies that are described in this work. The first column presents the specimen type, where Q1-Q4 corresponds to
the face sheet layups and C1-C3 to the core types described previously. The second and third columns show the
number of specimens tested using the 25.4 and 76.2 mm diameter indentors, respectively. The final column presents
the number of 381 mm square panels that were manufactured to create these specimens. Note that more than one
panel was manufactured for the first four geometries that are listed. In these cases, at least two specimens that were
tested at a given indentor diameter were taken from the same panel, and at least one specimen was taken from a
different panel. In this way the effects of possible inter- and intra-panel variations in the specimens could be
assessed.
Table 1. Number of specimens tested and panels manufactured.
Panel
Type
Q1-C1
Q1-C2
Q1-C3
Q2-C1
Q3-C1
Q4-C2
Q4-C3
TOTAL
IV.
Indentor Diameter
25.4 mm
3
3
3
5
1
2
2
19
76.2 mm
2
3
3
6
2
1
1
18
Panels
Manufactured
3
2
2
3
1
1
1
13
Quasi Static Indentation (QSI) Testing
A servo-hydraulic MTS machine was used for QSI testing. Figure 2 shows the test set-up. The center of the
specimen was first marked using two diagonal lines on the top face sheet. Next, the sandwich specimen was placed
on a circular rigid substrate and aligned so that the center of the indentor coincided with the center of the specimen.
Two spring clamps were then used to hold the specimen in place. All QSI tests were performed under displacement
control with a loading and unloading ramp rate of 0.05 mm/s. The majority of the specimens tested using a given
indentor diameter were tested to the same load level. For the 25.4 mm diameter indentor, the BVID load threshold
was chosen to be 1300 N. This corresponded to a maximum permanent dent depth of approximately 0.5 mm. For the
76.2 mm diameter indentor, the BVID load threshold was chosen to be 2800 N, which corresponded to a maximum
permanent dent depth of approximately 1.0 mm. These load levels were determined using the small QSI specimens
from the first few manufactured panels. That is, it was first ascertained that the small QSI specimens were
sufficiently large that their load-deformation responses were essentially the same as those observed in the full sized
panels. Small QSI specimens were then indented to varying loads and indentation levels in order to choose QSI load
and indentation levels that corresponded to BVID. This determination was made qualitatively by consensus of the
research team. However, as would be expected, not all panel types responded the same. For example, for a given
3
American Institute of Aeronautics and Astronautics
load level, less permanent deformation would be observed in panels with the high density C3 core than in those with
the C1 or C2 cores. Thus, some specimens were tested to different loads in order to obtain similar dent depths, i.e.,
so that the results from different panels could be compared at the same dent depth as well as at the same QSI load
level. A limited number of specimens were also tested to higher load levels than the BVID threshold in order to
provide additional insights into trends and mechanisms.
Figure 2. QSI test set-up, 76.2 mm diameter indentor.
Figure 3 shows the load versus displacement plots for the specimens from the various panel types. These results
are from the 25.4 mm diameter indentor, and the trends and relative differences in behaviors for specimens loaded
using the 76.2 mm diameter indentor were essentially the same. The general shape of the curves is similar for all
specimen types. There is an initial region of elastic deformation. Inelastic effects show up as a knee or bend in the
curve, which indicates the onset of core crushing and potentially other inelastic processes, such as delamination,
matrix cracking and/or fiber failures. Beyond this, the response up to the maximum load is somewhat linear, but
shows the combined effects of both elastic and inelastic deformation. The unloading line is highly nonlinear and
evidences permanent deformations in the vicinity of the target value of 0.5 mm. The slopes in both the initial elastic
region and secondary region between the knee and the maximum load point are primarily controlled by the core,
with the high density C3 core providing the stiffest response. For the same face sheet, the specimens with the C2
core are slightly stiffer than the C1 core. Although they have the same density, the smaller thickness of the C2 core
gives it less ability to redistribute the load and provides an effectively stiffer structural response than C1. There is
also a secondary effect of face sheet layup. Considering specimens with the C1 core, in comparison to Q1,
specimens with the Q2 and Q3 face sheets show a longer range of nonlinear behavior before the slope stabilizes
between the knee in the curve and the maximum load point. As will be shown subsequently, more delamination was
observed in the Q2-C1 and Q3-C1 than the Q1-C1 specimens, and it is likely that this is why the former specimen
types show this extended nonlinear region. As the load level increases, the overall displacement at a given load
becomes increasingly similar for all face sheet types. Along with the previous observation, this likely indicates that
delamination onset occurs relatively early, and the response at higher load and displacement levels is dictated wholly
by the core.
4
American Institute of Aeronautics and Astronautics
Figure 3. QSI load versus displacement plots for full-sized specimens, 25.4 mm diameter indentor.
V.
Evaluation of Indentation Profiles
The permanent dent left in all specimens was mapped using ultrasonic inspection. To the authors’ knowledge,
this is the first time that this approach has been utilized for impact damaged laminates. In order to develop the
method, a metal specimen with a known step size was initially manufactured and used to assess and refine the
technique. The SU-CML c-scan unit, which employs a 500 MHz transient waveform digitizer, was utilized in pulseecho mode for the process. The c-scan’s software was configured to capture the time of flight (TOF) between the
transducer’s pulse and the first wave reflection from the specimen’s surface, and this result was used along with the
wave speed in water to determine the depth of the step. This was relatively straightforward for the stepped metallic
specimen, so the initial development work primarily went into defining which signal peak corresponded to the onset
of the pulse and the return of the echo, as well as the evaluation of various transducers between 10 and 75 MHz. It
was found that a 50 MHz transducer provided the most accurate results. This study was then repeated with the small
specimens that had been QSI tested, and the results for the dent depth versus location profiles from the ultrasonic
evaluations were compared to those obtained by a mechanical surface profile measurement system. This consisted of
a dial gage rigidly mounted above a precision sliding x-y scale, thereby allowing accurate depth measurements to be
obtained at discrete points. In all cases, the 50 MHz transducer was again found to be the most accurate and
provided results that corresponded quite closely to those obtained by the mechanical measurement system.
Following the above initial verification of method, the ultrasonic approach was utilized for all full sized
specimens, with just a few additional verifications performed using the mechanical system. In these cases, however,
an extra step was required due to the fact that the specimens generally did not sit perfectly flat in the c-scan tank. In
order to apply a “corrective rotation” to the data, a few different methods were evaluated. This was done because
small differences in the data sampled by any approach made different methods produce slightly different results. The
methods considered consisted of approaches using two perpendicular lines to correct for the specimen’s orientation
versus those using a plane for correction. After assessing the various methods on several specimens, the former
approach was adopted. Here, dented specimens were first scanned and the software used to obtain the TOF to the
first wave reflection from the panel’s surface. This produced an array of times corresponding to the different x-y
locations on the specimen’s surface, and the wave speed of sound in water was used to convert these to z-direction
distances from the transducer. In the first data manipulation, the (x,y) location of the point that corresponded to the
maximum dent depth was located. A plot was then made of the line that went through this point in the x-direction,
i.e., a plot was made showing the z-direction distances versus x along a line that went through the center of the dent.
5
American Institute of Aeronautics and Astronautics
An analogous line was then plotted of the distances along the y-direction line that went through the center of the
dent. The slopes of these two lines were determined using the values outside of the dented region. Using these two
slopes, the entire data set was then rotated such that the two lines would be horizontal. Next, a mean displacement
was obtained from those data points remote from the dent, and this mean value was subtracted from all data. After
this final data manipulation, the full data set was evaluated to determine the maximum dent depth and other
parameters of interest.
Typical indentation profiles that resulted from the above process are shown in Figures 4a and 4b. Note that the
planar scales of these plots are contracted in comparison to their depth, and the physical dents do not appear nearly
as steep as they do in the figures. Both figures present results for the Q2-C1 specimen configuration. Figure 4a
presents the results for a specimen that was loaded to a force of 1300 N with a 25.4 mm diameter indentor and which
had a maximum dent depth of 0.46 mm. Figure 4b presents the results for a specimen that was loaded to a force of
2800 N with a 76.2 mm diameter indentor and had a maximum dent depth of 0.91 mm. In addition to dent depth, the
data also allowed ready determination of dent diameters and areas. To this end, the general procedure was to first
plot the x-z and y-z surface profiles for the x- and y-direction lines that went through the center of the dent. Next,
the dent width in each direction was determined and, if there was a slight difference, an average was taken. There
was a certain amount of subjectivity in the definition of diameters, i.e., in defining the exact point where the dent
ended and the flat region outside of the dent began. Thus, to make the diameter assessments, all x-z and y-z surface
profiles from all specimens were plotted using identical expanded scales, and the determination of dent diameters for
all specimens was made at the same time. In this manner, it was possible to ensure that a consistent definition was
used for all specimens. In all cases, it was found that the dents were essentially round, i.e., there was very little
difference in the x- and y-direction diameters. In Figure 4a, the dent diameter is 34.3 mm, and in Figure 4b it is 57.2
mm. Additional results from this process are presented subsequently.
(a)
(b)
Figure 4. Typical dent profiles using (a) 25.4 mm and (b) 76.2 mm diameter indentors.
VI.
Internal Damage Evaluations
Both pulse-echo and through-transmission c-scan methods for making internal damage evaluations were
evaluated. Perhaps due to the large overall panel thickness, the through-transmission approach produced relatively
poor results. As such, all internal damage evaluations were performed using the SU-CML c-scan unit with the same
50 MHz pulse-echo transducer that was used for the surface indentation profile evaluations. The small wave-length
of this transducer allowed resolution of relatively near-surface damage and, due to the small thickness of the face
sheets, did not appreciably affect the ability to resolve delaminations that were close to the core to face sheet
bond.13,14
For the undamaged regions of the face sheet, the TOF between the front and back surface reflections was found
to be approximately 0.6 µs. Thus, for the 8 ply face sheets considered, a reflection received 0.075 µs after the front
surface reflection will correspond to one from the first interface, a reflection after 0.15 µs corresponds to the second
6
American Institute of Aeronautics and Astronautics
interface, and so on. The front surface reflection was generally so long as to make resolution of any delaminations at
the first interface nearly impossible, and reflections from the 7th interface were a bit problematic due to the loss of
energy, the proximity to the back surface, and the “shielding” provided by the shallower delaminations, but
reflections from the remaining interfaces were quite easy to resolve with our system. For specimens indented with
the 25.4 mm diameter indentor, the full extent of delamination damage could be obtained. However, this was not the
case for the specimens indented with the 76.2 mm diameter indentor. Here, the slope of the specimen’s external
surface was such that the reflected signal in the regions of the dent near its perimeter could not adequately be
captured. Thus, for specimens indented with the 76.2 mm diameter indentor, accurate quantitative information about
the length of the delaminations at each interface could not be obtained. Nevertheless, the information from the
interior portions of the dent were sufficient to ascertain that the delaminations in these specimens were larger than
those in the specimens indented with the 25.4 mm diameter indentor, and that the pattern of the delaminations were
essentially the same. Thus, in what follows, quantitative assessments on the effect of the delaminations will be
restricted to those specimens indented with the 25.4 mm diameter indentor, and only qualitative assessments will be
made for specimens indented with the 76.2 mm diameter indentor.
In order to initially validate the non-destructive evaluation (NDE) results, as well as to determine the amount of
shielded damage, the small QSI specimens were first c-scanned. Destructive evaluations were then performed and
compared to the NDE results. In the destructive evaluations, the small QSI specimens were cut in the four ply angle
directions of 0, 90, 45 and -45 degrees. The cuts were made through the dent center using a thin diamond blade. Due
to the large number of cuts that were made, we did not polish the specimens, but rather viewed them with an optical
microscope “as is.” This likely somewhat affected our ability to resolve all delaminations but, as will be shown, is
not believed to affect the overall interpretations. Thus, after cutting, sections were imaged using a “scanning optical
microscope”; that is, an optical microscope that has the ability to take several images across a large planar area, auto
focus at each image, and then appropriately reconstruct the grid of images to correspond to the specimen being
evaluated. These images were then inspected for delaminations and matrix cracks, and corresponding “2D damage
maps” were constructed. A typical sequence is shown in Figure 5, which presents the NDE image of a Q1-C1 small
QSI specimen, the corresponding destructive evaluation (DE) from a 0° direction cut, and the 2D damage map. This
Q1-C1 specimen was QSI tested using a 25.4 mm diameter indentor to 1300 N and had a dent depth of 0.53 mm.
Delaminations that appear in the photomicrograph from the DE have been enhanced with white lines to aid in
viewing at the scale of the image. The damage map at bottom is a scale drawing of the image and was made for our
own ease-of-use, i.e., so that the actual micrographs would not need to continually be referred to. They are intended
to show all salient details. Also included in the figure is a c-scan scale bar showing the TOF and corresponding color
scales. Along the cut line, the NDE and DE images show good correlation for the center delamination at interface 3
(blue) and for the delaminations extending to the left and right of it at the 5th (green) and 6th (yellow) interfaces.
Information on the shielded delaminations, i.e., those beneath the nearest surface delaminations that are captured by
the c-scan, is not obtained. It can also be seen that the core produces a certain amount of fiber waviness in the
innermost plies, which somewhat complicates the NDE interpretations at interfaces 6 and 7. The DE images also
showed matrix cracks between interfaces, something which obviously could not have been captured nondestructively. These matrix cracks span outwards from the dent center through various plies towards the core. No
core to face sheet debonding was seen in the destructive evaluations. Finally, note that all the delaminations in this
image are at interfaces 3, 5, 6 and 7. In a few specimens, very small delaminations – significantly smaller than those
at the other interfaces – were observed at interfaces 1 and 2. No specimens showed midplane delaminations.
2D damage maps like the one described above were constructed for each of the cuts (0, 90, 45 and -45 degrees)
from each of the small QSI specimens. In order to better understand the nature of the delaminations at shielded
interfaces, information from these four 2D damage maps were assembled into 3D damage maps. To this end, the
distance of each delamination’s starting and ending points were measured from a common datum and then
assembled together for each interface. A typical result is presented in Figure 6, which presents the 3D damage map
of the Q1-C1 specimen shown in Figure 5. In this 3D damage map, the solid lines present the delamination
information obtained at each interface from the section cuts. The lightly blue shaded regions represent overlays of
the delamination information revealed by the NDE image for that particular interface; in this case, as taken from the
c-scan of Figure 5. It may be observed that these comparisons and those of Figure 5 provide strong corroboration of
the NDE results, and this was true in all specimens evaluated.
The dotted lines in Figure 6 utilize the combination of the NDE and DE results to make a conservative
estimation of the delaminated area at each interface. This was done primarily to better understand the nature of the
shielded damage. For example, based on the NDE and DE images, it is likely that the delamination in interface 5 is
continuous and takes a shape that is something like that shown, or perhaps somewhat larger. The delamination at
interface 6 might be larger than that which is drawn. It may be continuous and perhaps elliptical or lemniscate
7
American Institute of Aeronautics and Astronautics
shaped. There is insufficient information to make these conclusions. However, it is likely that the delaminated area
is at least as large as that which is enclosed by the dotted lines. A similar statement can be made for interface 7. By
making these damage maps for multiple specimens of each configuration, it was possible to extract some general
trends. These are described below. First, however, it was necessary to establish that the damage in the small QSI
specimens was representative of that in the full sized specimens. To this end, c-scan results from all small QSI
specimens of a given type were compared to those of the large QSI specimens of that same type. No significant
differences were observed. In addition, comparison of the load versus displacement plots of small and large QSI
specimens showed essentially no differences. Based on these observations, it was concluded that the 3D damage
maps and conclusions based on their correlation to the NDE images for the small QSI specimens applied equally to
the full sized specimens.
Figure 5. C-Scan, destructively obtained cross-sectional view, and corresponding 2D damage map.
8
American Institute of Aeronautics and Astronautics
Figure 6. 3D damage map of the Q1-C1 specimen of Figure 5.
Figure 7 presents typical c-scans of the four different face sheet layups from specimens that were indented with
the 25.4 mm diameter indentor. The color scale and ply angle convention used here is the same as that introduced in
Figure 5. As previously indicated, delaminations were observed only at interfaces 3, 5, 6 and 7. Also, delaminations
at interface 7, depicted in red, are often difficult to distinguish from the back surface reflection that is obtained in the
regions where no delaminations exist. Note that for layups Q1 and Q4, good information was obtained for interfaces
3, 5 and 6, with limited information at interface 7 due to shielding effect. For Q2, the results are a bit worse due to
the shielding of near-surface delaminations, and the scans for Q3 indicated only those delaminations at interfaces 3
and 5.
By combining results such as those shown in Figure 7 with the 3D damage maps, it was possible to make a
variety of observations across and between face sheet and core types. For all panel geometries, if it is assumed that
all delaminations are continuous through the shielded region, then all delaminations are oblong, with the direction of
the major axis of each delamination primarily controlled by the direction of its back surface ply. This is similar to
what has been observed in low velocity impacts of non-sandwich laminates.13 The delaminations at interface 3 were
always lemniscates. For layups Q2 and Q3, their major axis is aligned with their back surface ply angle, whereas for
Q1 and Q4, the two portions of the lemniscates are slightly offset, giving the major axis a slight tilt towards the
positive 45° direction. Each layup has one other delamination at 0 or 90 degrees, which occurs at either interface 5
or 6, depending on the layup. These delaminations are generally continuous through the shielded region and are
either elliptical or lemniscate shaped. As in the case of interface 3, the interface 5 and 6 delaminations are aligned
with the back surface ply angle for Q2 and Q3, but have a slight tilt towards the positive 45° direction for Q1 and
Q4. For all panel types, these are the widest delaminations of all that occur; for layups Q1, Q2 and Q4, they are also
the longest.
Layups Q1 and Q4 have similar delamination patterns. They both have tilted, lemniscate shaped delaminations at
interface 3 and a somewhat longer and wider trapezoidal shaped delamination at interface 5. The largest
delaminations occur at interface 6 in these face sheet types and are either elliptical or lemniscate shaped and are
slightly tilted with respect the back surface ply angle. The delaminations at interface 7 appeared to be smaller than
those at interface 5, and often were discontinuous throughout the shielded region. For Q2, the delamination at
interface 3 was similar in shape to Q1 and Q4, but somewhat wider, without any obvious tilt of the major axis. The
largest delamination was oriented at 0° at interface 5 and was similar in size to the delaminations at interface 6 in Q1
and Q4. The 45° delamination at interface 6 was smaller than the 45° delamination at interface 5 in Q1, and the
destructive results indicated that this delamination may have not always been continuous through the shielded
region. The delamination at interface 7 was similar to Q1 in terms of size, shape and potential discontinuity.We had
9
American Institute of Aeronautics and Astronautics
limited results for Q3, but the NDE and DE data suggest that it had the longest and widest delamination at interface
3 out of all layups. The delamination at interface 5 was wider but similar in length or perhaps a bit shorter than that
at interface 3. The delaminations at interfaces 6 and 7 were similar to those seen in Q2 face sheet type.
As will be shown subsequently, the size of delamination distribution also differs between different core types.
For the same face sheet, the delaminations are larger in panels with the C2 and C3 cores than in those with C1. The
0 and 90 degree delaminations are thinner and longer in C2 core type than in C1 and C3, where in the latter two they
are similar sized.
10 mm
10 mm
Q1: [45/0/-45/90]s
Q2: [45/-45/0/90]s
10 mm
10 mm
Q3: [-45/45/90/0]s
Q4: [45/90/-45/0]s
Figure 7. Typical c-scan scan results for the four different face sheet types, 25.4 mm diameter QSI.
VII.
Damage Resistance
In this section, the preceding methods are applied to evaluate the damage resistance of the various panel
geometries. The damage event may be thought of as a particular QSI force and indentor, in which case the damage
metrics of dent depth, dent diameter and delamination area are considered. Alternatively, imparting a dent of a given
depth could be taken as the damage event, in which case the latter two metrics would be of interest.
Figure 8 presents the average QSI force versus average dent depth for all panels tested. As indicated in the
legend, different shaped symbols are used to represent the different face sheet types, and different colors are used to
represent the different cores. Filled symbols depict specimens that were loaded to the BVID force level, while open
symbols depict specimens that were loaded to a higher force, for example, to match dent depths of the different
panels or to elucidate trends. To obtain the average QSI values in the figures, specimens of the same type were first
grouped. These were then further divided into groups that experienced the same force within 1-2 N. The average
force and dent depth were then determined for each group. These average results are presented in order to illustrate
general trends. Analogous results for average QSI force versus average dent diameter are presented in Figure 9.
10
American Institute of Aeronautics and Astronautics
Figure 8. Results for average QSI force versus average dent depth.
Figure 9. Results for average QSI force versus average dent diameter.
Considering the specimens with the Q1 face sheet, Figures 8 and 9 show that, for a given QSI load and indentor,
the largest dent depths and areas are observed in those that have the C1 core, the next largest in C2 core specimens,
and the smallest dents in C3. This corresponds to the QSI load vs displacement results presented previously.
Considering specimens with the C1 core, at the BVID force level, specimens with the Q1 face sheet show larger
dent depths and diameters than those with Q2. Interestingly, this trend is reversed for the higher force level
specimens indented with the 25.4 mm diameter indenter. Both the C2 and C3 core results indicate that specimens
with the Q4 face sheet show larger dent depths and diameters than those with Q1. Thus if dent depth or dent
diameter were the damage metric, the average results show that the C3 core provides the most damage resistant
panels and C1 the least, and that the Q2 face sheet is more damage resistant than Q1 which provides better resistance
than Q4. Consistent results were not obtained for Q3, and no conclusions can be drawn about this layup.
Figure 10 presents results that are similar to those in 8 and 9, but here all specimens tested are presented. This
allows the scatter in results to be observed. The same symbol definitions are utilized as in the previous two figures.
11
American Institute of Aeronautics and Astronautics
Although not explicitly identified in the figure, no difference in scatter was observed between specimens cut from
the same panel versus those cut from different panels. Interestingly, note that when the force levels are increased so
that the dent depth in the C3 core specimens agrees with those in the other specimen types, the dent diameter for C3
remains smaller. This is further illustrated in Figure 11, which shows the ratio of dent diameter to dent depth for the
various panel types. As before, filled symbols indicate specimens at the BVID loads, whereas open symbols indicate
those at higher loads. The ratio of dent diameter to dent depth is essentially independent of the load level, and only
slightly dependent on the indentor diameter. Specimens with the C3 core clearly show a lower diameter for a given
depth. However, although the C2 core was shown to be stiffer than C1, the ratio in Figure 11 for specimens from the
two cores and with the same face sheet is essentially the same. This shows that the ratio of dent diameter to dent
depth is a function of core density, and that denser cores will show smaller dent diameters for a given depth.
Figure 10. Dent diameter versus dent depth for all specimen types.
Figure 11. Ratio of dent diameter to dent depth by specimen type.
12
American Institute of Aeronautics and Astronautics
Figure 12 presents the ratio of the dent diameter to the local chord length of the indentor for all specimen types.
The local chord length, c, is as defined in Figure 13 and was computed for each specimen based on the indentor
radius, r, that was used and the permanent dent depth, d, that was obtained using the relation
c = 2 d ( 2r − d )
(1)
Figure 12 shows that, regardless of indentor, the permanent dent was always larger than the region of the indentor
that was contacting the specimen. That is, bending of the face sheet outside of the contact region contributes to core
crushing. The difference between the dent diameter and the chord length increases with increasing QSI load and
displacement, i.e, as the face sheet undergoes more bending. As would be expected, the ratio of Figure 12 is always
greater for the smaller diameter indentor. As a result of the larger density and associated smaller dent diameter, the
ratio for specimens with the C3 core is the smallest, whereas the ratio for the other core types are essentially the
same for a given indentor diameter.
Figure 12. Ratio of dent diameter to local chord length of the indentor.
Figure 13. Sketch illustrating local chord length definition.
Figure 14 considers the case where planar delamination area is chosen as the damage metric. Here, planar
delamination area refers to the overall area of delamination that is obtained from the NDE images, without regard to
the delamination’s depth. The area of shielded delaminations is not included. To compute this value, the major and
13
American Institute of Aeronautics and Astronautics
minor axes of the boundary of all delaminations obtained by NDE were first determined. The average of these two
values was then defined as the equivalent circular diameter, and the area of this equivalent circle was computed. As
previously described, this was only possible for specimens indented with the 25.4 mm diameter indentor. Thus,
Figure 14 presents QSI force versus planar delamination area for only these specimens. Considering specimens with
the C1 core, if damage resistance is based on QSI force versus planar delaminated area, then specimens with the Q1
face sheet are more resistant than those with Q2 which are more resistant than Q3. The C2 results indicate Q4 is
more resistant than Q1, but the C3 results indicate the reverse. Either way, it is clear that less delamination occurs in
Q1 and Q4 than in Q2 and Q3. This likely relates to the fact that Q1 and Q4 only have 45° angle changes between
adjacent plies. This results in a lower thermal and mechanical mismatch, and therefore lower interlaminar shear
stresses during QSI, than what occurs in Q2 and Q3, which have 90° angle changes between adjacent plies. This
result agrees with that reported in Ref. 15.
The effect of core on the delaminated area may be observed by comparing specimens with the same face sheet.
Specimens with the C1 core show lesser amounts of delamination than those with the C2 or C3 cores. Since the
damage due to low velocity impact and QSI has been shown to be essentially the same, it is perhaps easiest to think
of this via an energy argument. Also, recall the load versus displacement plots, where it was shown that the majority
of the delamination occurs near the first inelastic bend in the curve. Since the C2 and C3 cores are stiffer than C1,
less energy is dissipated early on through core crushing, and more energy goes into delamination in the former than
the latter case. There is no clear trend between C2 and C3: with the Q1 face sheet, specimens with the C2 core show
a greater amount of delamination, but the reverse is true with the Q4 face sheet.
Figure 14. QSI force vs delamination area, 25.4 mm diameter QSI.
Figure 15 illustrates the situation where the damage event is considered to be indentation to a certain depth, and
the damage metric is again planar delamination area. This figure illustrates the same behaviors as those described
above with respect to the effect of QSI load. As would be expected, Figures 14 and 15 show that the delaminated
area generally increases with increasing QSI force and with increasing dent depth although, particularly in the cases
of Q2-C1 and Q1-C2, there is a reasonable amount of scatter in both parameters. Overall, if delamination area is the
damage metric, then either the Q1 or Q4 face sheet along with the C1 core will provide the most resistant result.
This is in contrast to the case where dent depth or dent area was the metric, where the Q2 face sheet combined with
the C3 core provided the best damage resistance, and Q4-C1 was the worst. This is better illustrated in Figure 16,
which shows the ratio of the planar area of delamination to the dent area as a function of panel type. Since dent area
and dent depth were shown to follow the same general trend, conclusions from this figure can also be extended to
delaminated area versus dent depth. Here, Q1-C1 is observed to have a small amount of delamination for a given
dent area, i.e., it shows a large dent area, but a small amount of delamination. Specimen types Q1-C3 and Q4-C3
show the reverse: a small dent area, but a large amount of delamination. Presenting the data in the format of Figure
16 also appears to clear up some apparent inconsistencies pointed out previously. Specifically, it indicates that, for a
14
American Institute of Aeronautics and Astronautics
given face sheet, the C2 core type provides intermediate results to the C1 and C3 cores. It also shows that for a given
core type and dent depth or area, the Q4 face sheet will show slightly less delaminated area than Q1.
Figure 15. Dent depth versus delamination area, 25.4 mm diameter QSI.
Figure 16. Ratio of planar area of delamination to dent area, 25.4 mm QSI diameter.
15
American Institute of Aeronautics and Astronautics
VIII.
Summary
This work considered the effects of core density, core thickness, face sheet layup and indentor diameter on the
nature and extent of quasi-static indentation damage in aluminum core honeycomb sandwich laminates with eight
ply carbon fiber reinforced quasi-isotropic face sheets. It was observed that the core primarily controls the load
versus displacement response during QSI. The role of the face sheet appears to be primarily reflected in its
propensity for delamination, which has its strongest effect on the initial nonlinear region in the load-displacement
response. Following indentation, the permanent dent was found to always be larger than the contact area of the
indentor. The ratio of dent diameter to dent depth was found to depend on core density, where specimens with
denser cores show smaller dent diameters for a given depth. This is consistent with the previous observation. That is,
outside of the contact region, stiffer cores do a better job resisting crushing by the face sheet. Regardless of
specimen layup, delaminations were observed to occur essentially only at the 3rd, 5th, 6th and 7th interfaces.
Delaminations were oblong, with their major axis direction dictated by the angle of their back surface bounding ply.
However, deeper delaminations did not always span the dent, but rather could be split into two distinct areas on
either side of it. Stiffer cores, either in terms of a higher density or, for those cores considered, a smaller thickness,
meant more delamination would occur in the face sheets. Regardless of the core, larger delaminations were observed
in face sheets that contained only 90° angle changes between adjacent plies in comparison to those that contained
only 45° angle changes.
In order to assess the damage resistance of the various specimen types, the damage metrics of dent depth, dent
area and planar delaminated area were evaluated. If dent depth or dent diameter is chosen as the damage metric, then
more damage resistance is obtained with increasing core density, and a [±45/0/90]s face sheet was found to be best
choice of the four face sheets considered. Lower density cores resulted in a larger permanent dent for a given
indentation event, and a low density core with a [45/90/-45/0]s face sheet was found to be the least damage resistant
configuration of those evaluated. Conversely, if delamination area is the damage metric, then this low density core
and the [45/90/-45/0]s face sheet will provide the best damage resistance, i.e., the least amount of delamination for a
given indentation event. When considering compression-after-impact response, a deep dent can lead to a global
instability or kink band type of failure, whereas a large delamination can lead to failure controlled by delamination
buckling and fiber failures.7 This observation, along with the dramatically different results that were observed for
the damage modes among the different specimen types considered in this study, indicates the need for a closely
coupled approach to damage resistance and damage tolerance, where the effects of the damage induced in different
panel types by the same event can be predicted under the service loadings of interest. This knowledge may then be
used as guidance in the choice of the structural geometry.
Acknowledgements
This work was co-supported by the NASA Constellation University Institutes Project, Grant NCC3-989, Claudia
Meyer, Project Manager and by the Exploration Technology Development Program/Advanced Composites
Technologies Project, Mark Shuart, Project Manager.
References
1
Avery, W.B. and Grande, D.H., “Influence of Materials and Layup Parameters on Impact Damage Mechanisms,”
Proceedings of the 22nd International Society for the Advancement of Materials and Process Engineering Technical Conference,
Vol. 22, 1990, pp. 470-483.
2
Bull, P.H., “Damage Tolerance and Residual Strength of Composite Sandwich Structures,” Ph.D. Thesis, Department of
Aeronautical and Vehicle Engineering, Kungliga Tekniska Högskolan, Stockholm, Sweden, 2004.
3
Cantwell, W.J., Scudamore, R., Ratcliffe, J. and Davies, P., “Interfacial Fracture in Sandwich Laminates,” Composites
Science and Technology, Vol. 59, 1999, pp. 2079-2085.
4
Nicholas, T.N., Zukas, J.A. and Swift, H.F., “Impact Dynamics,” John Wiley & Sons, New York, 1982.
5
Xie, Z.H. and Vizzini, A.J., “A Feasible Methodology for Engineering Applications in Damage Tolerance of Composite
Sandwich Structures,” Journal of Composite Materials, Vol. 38, 2004, pp. 891–914.
6
Kang, K.W., Kim, J.K., Cheong, S.K. and Kim, H.S., “Residual Strength of Sandwich Structure Subjected to Low Velocity
Impact,” International Journal of Modern Physics B, Vol. 20, No. 25, 26, & 27, 2006, pp. 4384-4389.
7
Czabaj, M.W., Zehnder, A.T., Davidson, B.D., Singh, A.K. and Eisenberg, D.P., “Compression After Impact of Sandwich
Composite Structures: Experiments and Modeling,” Proceedings of the 51st AIAA/ASME/ASCE/AHS/ASC Structures, Structural
Dynamics, and Materials Conference, Orlando, FL, April 2010.
8
Williamson, J.E. and Lagace, P.A., “Response Mechanisms in the Impact of Graphite/Epoxy Honeycomb Sandwich Panels,”
Proceedings of the American Society for Composites Eighth Technical Conference on Composite Materials, Cleveland, OH,
1993, pp. 287-297.
16
American Institute of Aeronautics and Astronautics
9
Herup, E.J. and Palazotto, A.N., “Low-Velocity Impact Damage Initiation in Graphite/Epoxy/Nomex Honeycomb-Sandwich
Plates,” Composites Science and Technology, Vol. 57, 1997, pp. 1581-1598.
10
Ferri, R. and Sankar, B.V., “A Comparative Study on the Impact Resistance of Composite Laminates and Sandwich
Panels,” Thermoplastic Composite Materials, Vol. 10, 1997, pp. 304-315.
11
Oplinger, D.W. and Slepetz, J.M., “Impact Damage Tolerance of Graphite/Epoxy Panels,” Foreign Object Impact Damage
to Composites, ASTM STP 568, American Society for Testing and Materials, 1975, pp. 30-48.
12
ASTM Standard C364, “Standard Test Method for Edgewise Compressive Strength of Sandwich Constructions,” ASTM
International, West Conshohocken, Pennsylvania, 2007.
13
Davidson, B.D., Michaels, J.E., Sundararaman, V. and Michaels, T.E., “Ultrasonic Imaging of Impact Damaged Composite
Panels,” Acoustical Imaging, Volume 19, H. Ermert and H.-P. Harjes, Eds., Plenum Press, 1992, pp. 589-594.
14
Michaels, T.E., Krafchak, T.M. and Davidson, B.D., “Ultrasonic Inspection of Thin Walled Composite Tubes,” Review of
Progress in Quantitative Nondestructive Evaluation, Volume 12, D.O. Thompson and D.E. Chimenti, Eds., Plenum Press, 1992.
15
Kim, C.G. and Jun, E., “Impact Resistance of Composite Laminated Sandwich Plates,” Journal of Composite Materials,
Vol. 26, 1992, pp. 2247–2261.
17
American Institute of Aeronautics and Astronautics