Journal of Sandwich Structures and Materials-2002-Moody-71-82

http://jsm.sagepub.
com/
and Materials
Journal of Sandwich Structures
http://jsm.sagepub.com/content/4/1/71
The online version of this article can be found at:
DOI: 10.1177/1099636202004001037
2002 4: 71 Journal of Sandwich Structures and Materials
R. Clifton Moody, James S. Harris and Anthony J. Vizzini
Composite Sandwich Panels
Scaling and Curvature Effects on the Damage Tolerance of Impacted
Published by:
http://www.sagepublications.com
at:
can be found Journal of Sandwich Structures and Materials Additional services and information for
http://jsm.sagepub.com/cgi/alerts Email Alerts:
http://jsm.sagepub.com/subscriptions Subscriptions:
http://www.sagepub.com/journalsReprints.nav Reprints:
http://www.sagepub.com/journalsPermissions.nav Permissions:
http://jsm.sagepub.com/content/4/1/71.refs.html Citations:
What is This?
- Jan 1, 2002 Version of Record >>

at Dokuz Eylul Universitesi on April 16, 2014 jsm.sagepub.com Downloaded from at Dokuz Eylul Universitesi on April 16, 2014 jsm.sagepub.com Downloaded from
Scaling and Curvature Effects on
the Damage Tolerance of Impacted
Composite Sandwich Panels
R. CLIFTON MOODY, JAMES S. HARRIS AND ANTHONY J. VIZZINI*
Composites Research Laboratory, Department of
Aerospace Engineering, University of Maryland, College Park,
Maryland 20742, USA
ABSTRACT: An experimental program was conducted to investigate the effects of
specimen width and specimen curvature on the residual compression strength of
composite sandwich panels after low-energy impact. Flat and curved (1.1 m radius)
specimens of three widths (83, 152, and 305 mm) were manufactured from thin-gage
(four plies) graphite/epoxy facesheets and 25.4-mm-thick Nomex core. A drop
mechanism was used to create barely visible impact damage on one facesheet of the
panels (convex face of the curved panels) maintaining the energy level as constant in
all cases. All specimens were then loaded axially in compression to failure. Above
a certain panel width, the panel width and curvature had no effect on the ultimate
strength of the panel. Although width and curvature increased the size of the damage
caused by an identical energy impact, the indentation growth was delayed with
respect to farfield strain (stress). Because indentation growth is chiefly a local
phenomenon, the delay was probably due to the variations in the damage state. The
baseline specimen (flat, 83 mm in width) serves as an adequate coupon-level test even
though it is conservative when compared to wider plates.
KEY WORDS: damage tolerance, sandwich structures, barely visible impact
damage, CAI, curvature, scaling effects.
BACKGROUND
S
ANDWICH STRUCTURES WITH thin composite facesheets are used
extensively in aerospace structures because of their high bending effi-
ciency (stiffness/weight). A disadvantage of these types of structures is their
susceptibility to low-velocity-impact damage. Low-energy-impact damage
Journal of SANDWICH STRUCTURES AND MATERIALS, Vol. 4 January 2002 71
1099-6362/02/01 007112 $10.00/0 DOI: 10.1106/109963602021037
2002 Sage Publications
*Author to whom correspondence should be addressed. E-mail: vizzini@eng.umd.edu
at Dokuz Eylul Universitesi on April 16, 2014 jsm.sagepub.com Downloaded from
can cause a significant reduction in the ultimate compression strength of a
composite sandwich panel and yet not be easily detected using visual
inspection methods. Primary aircraft structures are routinely inspected using
visual methods. Barely visible impact damage (BVID) is the damage level
corresponding to the threshold of detectability by a visual inspection.
Studies have been conducted to model the impact event in terms of the
resulting damage state and to predict the residual strength of the damaged
panels. Kassapoglou conducted several studies related to the damage
tolerance of composite laminates and thin-gage sandwich panels by
modeling the damage as a delamination and a debond [14]. Minguet
presented a different approach to modeling the residual strength of damaged
sandwich structures in which the damage is modeled as a facesheet
indentation with core damage [5].
In the literature, damage resistance refers to the ability of the structure to
remain undamaged after an impact event. Damage tolerance refers to the
ability of the structure to withstand load after being damaged. Cairns
conducted an extensive analysis of the impact response of composite
laminates [6]. Damage is modeled as a reduced stiffness elliptical inclusion.
Lie developed analytical models to describe the impact event and
compression failure of thin-gage composite sandwich panels [7]. As an
extension to the work by Cairns, the damaged panels are modeled as an
intact panel with an elliptical hole representing the damage. Lagace et al.
investigated the damage resistance and damage tolerance of some specific
layups and materials [8]. In their study, panels with low-energy-impact
damage showed no reduction in tensile strength, but had a significant loss of
compressive strength.
A typical approach in design has been to characterize the response of
sandwich structures with a given coupon-level specimen. The data derived
from such tests are then applied to more general structures that are
substantially larger and are no longer flat. Hopefully this process results in
a conservative design assuming that any phenomena that arise due to
geometry effects occur at higher load levels. However, such conservatism
prevents the full use of the composite materials and thus results in a heavier
and costlier structure. The objective of this study is to determine the
applicability of flat coupon-level specimens on the design of more general
sandwich panels.
EXPERIMENTAL APPROACH
The overall approach in this study was to determine the baseline response
of flat 83-mm-wide specimens. Wider specimens and specimens with
72 R. CLIFTON MOODY ET AL.
curvature were impacted and loaded to failure under uniaxial compression
to explore for additional phenomena and to determine the associated
effects of width and curvature. Thus, the energy level required to cause
BVID in the baseline 83-mm-wide flat test specimens was determined
initially by using 20 test panels. The impact energy level was varied for each
panel until the BVID level was determined by visual inspection. Once
chosen, this energy level was kept constant throughout the test program
regardless of the level of damage associated with it.
Flat and curved panels were made in the same fashion. Both were cured as
larger panels with the curved panels being fabricated on an aluminum cure
plate with a 1.1 m radius of curvature. The inside radius was placed against
the aluminum cure plate.
The sandwich panels were produced using a 25.4-mm-thick, 48-kg/m
3
Nomex honeycomb core with 3.2 mm hexagonal cells. The facesheets were
manufactured from preimpregnated unidirectional graphite/epoxy AS4/
3501-6 with a [90/0]
S
layup with respect to the loading axis. The outer-most
layer was chosen to be transverse to the applied loading to prevent the
occurrence of ply buckling. FM300 film adhesive was used to bond the
facesheets to the core.
The facesheets were cured and bonded to the core concurrently. The
manufacturers recommendations were followed for the cure in the
autoclave with one exception; a reduced autoclave pressure of 276 KPa
was used during the cure so as not to crush the honeycomb core. Aluminum
bars were used around the free edges of the panel to prevent core damage
at the edges of the panel during the cure due to vacuum bag pressure.
The nominal size of the cured panels was 508 mm-wide by 356 mm-long.
The cured panels were then cut to the appropriate specimen size using
standard diamond saw tooling on a milling machine with water irrigation.
In this test program, panels were cut to widths of 83, 152, and 305 mm as
given in Table 1. All test specimens were 356 mm-long.
To validate the manufacturing process used in this program, some of the
panels were non-destructively examined using a pulse-echo ultrasonic scan.
Any voids or other flaws in the facesheets could be detected. This initial
Table 1. Testing matrix.
Specimen Width
83mm
Undamaged BVID Level Impacted 152mm 305mm
Flat 4 20 16 7 7
Curved 7 7
Scaling and Curvature Effects 73
screening of specimens indicated uniform adhesion of the facesheets to
the core.
IMPACT DAMAGE
The impact damage was imposed using a standard drop tower with a 25.4-
mm-diameter stainless steel hemispherical tup. The mass of the impacting
apparatus was 1.31 kg. The specimens were clamped across the width (top
and bottom) with a span of 229 mm and impacted at the center of the
specimen.
The BVID was characterized as the energy level required to cause the
indentation left by the impact and the associated local damage to be barely
visible from a distance of 1.2 m
4
. In this study the required energy level was
determined by manufacturing 20 83-mm-wide impact test specimens. These
panels were impacted at increasing energy levels by increasing the drop
height until the BVID was reached. The energy level required to cause BVID
in the 83-mm-wide specimens is 5.88 J (drop height of 0.46 m) based on the
method recommended by Kassapoglou. On these specimens, the BVID is
typically a very small indentation (<0.5 mm in depth) with no visible
cracking of the matrix. This energy level was used for all specimen widths
both flat and curved. In all of the drop tests, the weight was caught after it
rebounded from the specimen.
To further investigate the damaged region, several of the impacted flat
specimens were scanned ultrasonically to determine the amount of internal
damage within the sandwich panel. The results from individual scans show
that the overall size of the elliptical damage area increases (Figure 1). The
damage areas were approximated from the scans and are plotted in Figure 2
versus the panel width. Clearly the damage area is repeatable and increases
with specimen width primarily due to greater transverse stiffness of the
panels with increasing width. Because of their increased complexity, no
scans were performed on the curved panels.

25.4 mm
83-mm specimen 152-mm specimen 305-mm specimen
25.4 mm 25.4 mm
Figure 1. Ultrasonic scan of BVID area for 83-, 152-, and 305-mm-wide test specimens.
INSTRUMENTATION
Strain gages were placed on the test panels to capture the response to
loading as shown in Figure 3 for the 305-mm-wide specimens. The gages
were placed near the damage and in line with the expected indentation
propagation. Gages were also placed on the back facesheet to determine the
amount of load transferred to the back facesheet during loading. In
Figure 2. Increasing damage area with specimen width.
Figure 3. Strain gage locations for 305-mm-wide specimens (dimensions in mm).
particular, the strain gage in location #2 was used to indicate growth of the
indentation by measuring a local softening. In the 152-mm-wide specimens,
no gage was placed in the #5 location. In the 83-mm-wide specimens, no
gages were placed in the #4 and #5 locations.
TEST FIXTURES
An appropriate gripping system for the test specimens that could be used
for flat or curved panels up to 305 mm in width was developed. Because of
the relatively low out-of-plane strength of the Nomex honeycomb core,
wedge grips cannot be used without crushing the core. In addition, the
available width of the testing machine grip and the curvature of the test
specimens necessitated an alternative method. This very limitation is one of
the main reasons why many designers rely on coupon level data.
Many alternative methods to grip specimens involve the potting of the ends
of the test specimens in the following manner. Typically, the core is hollowed
out in the grip area, and a resin material is poured into the core. Alternatively,
the entire end can be placed into resin and then the ends are machined to be
parallel and compatible with the testing machine. These methods are costly
and subsequently result in awkward handling of failed specimens. In this
program, the specimens were potted in a low melt (70
C) Bismuth metal alloy.

The melting temperature of the alloy was sufficiently low so as not to cause
damage to the composite facesheets or Nomex core. This potting system had
several advantages. The test specimens could be removed intact to be
realigned prior to testing or to be inspected after testing; no machining was
required between tests; and the potting material was reusable.
The fixed end condition was accomplished by using aligned aluminum
end molds to contain the ends of the test panels and the potting material.
These end molds were attached to the testing machine to apply the
compression loading. The test panel was centered in the fixture with tabs
and secured in place. To prevent voids due to rapid cooling of the potting
alloy upon contact with the molds, the end molds were heated to the melting
temperature of the alloy. The alloy was heated to 82
C and poured around

the test panel. The mold was allowed to cool before the process was repeated
for the other end. The plates in Figure 4, in addition to aligning the top and
bottom fixtures, prevent the test specimen from being loaded by the end
fixtures prior to and after testing. Figure 5 indicates the loading direction
with respect to the curvature of the curved specimens.
Once the ends of the specimen were potted in the aluminum end molds,
the end molds were fastened to the end plates and round steel bars. These
round bars were then gripped in the testing machine so load passes from the
testing machine grips to the round bars to the end plates. The alignment
plates were removed before the compression test.
After testing, the broken panels were removed from the fixture by heating
the end molds and draining out the potting material.
Figure 4. Gripping system (flat panel).
Figure 5. Axial loading direction.
COMPRESSION TESTING
The test specimens were loaded to failure using a MTS uniaxial testing
machine. The test was stroke controlled with a ramp rate of 0.0254 mm/s.
Load, stroke, and strain gage data were collected at 0.2 s intervals.
Under compression, there appeared to be a point at which the indentation
caused by the impact began to grow perpendicular to the loading direction.
This observable indentation slowly propagated to approximately 25.4 mm
from the centerline of the panel. At this point, it appeared as if the
indentation propagation stopped.
The final failure of the panel occurred when the front facesheet
experienced a localized buckling (crippling) possibly due to a stress
concentration at the edge of the damage indentation. The front facesheet
crippled suddenly across the width of the specimen. The weakening of the
front facesheet caused the load to be transferred to the back facesheet where
the high eccentric load caused a global type buckling of the back facesheet.
Typically, the back facesheet then debonded from the core.
RESULTS
The maximum load divided by the cross-sectional area of the facesheets
determines the ultimate stress for the panel. The results are then plotted in
Figure 6 where the flat specimens are indicated with squares and the curved
specimens with circles. Lines are drawn between the average values to show
trends in the data. Note that the data are offset from the actual widths (flat
panels to the left and curved panels to the right of the actual width) to
increase the readability of the data. There is no significant difference in the
Figure 6. Ultimate strength as a function of specimen width for flat and curved panels.
ultimate strength of the 152- and 305-mm-wide flat and curved specimens. If
the one low strength 152-mm-wide curved specimen is excluded the averages
would nearly be the same. The 83-mm-wide specimens fail at a lower stress
than the wider specimens. This is probably due to the inability of the
indentation to arrest itself in the 83-mm-wide specimens as is the case in the
152- and 305-mm-wide specimens.
The strain gage data were then examined to look for evidence of
indentation propagation. The instantaneous load is again divided by the
cross-sectional area of the facesheets to determine the panel stress and
plotted against the strain-gage readings as in Figure 7. Under compressive
load, the indentation resulting from the impact propagates perpendicular to
the loading direction. When the indentation begins to spread towards
location #2, it affects the local strain. As the #2 gage becomes involved with
the deflection of the indentation, the local strain increases faster than the
far-field strain. This strain increase results in an apparent softening in the
local modulus of elasticity. In contrast, in Figure 7, the strain at location #3
(38.1 mm from the centerline) does not show a significant change in the
apparent modulus of elasticity until the end of the test. This is consistent
with the observation that the indentation propagates and then arrests itself
prior to ultimate failure in the 152- and 305-mm-wide specimens.
To determine a point at which the behavior in the stressstrain curve
changes, the algorithm LIN6 is used [9]. LIN6 determines contiguous
regions in a data set that have the highest linear correlation coefficients. For
example, in Figure 8, LIN6 indicates a 28% change in the tangent apparent
Figure 7. Stressstrain curves indicating growth point at location #2.
modulus of elasticity from just before to immediately after the knee point.
The first such knee point is defined as the growth point and is dependent on
the size and location of the strain gage.
The growth points are shown on Figure 9 for flat and curved specimens of
different widths. Again the squares and the circles, representing the flat and
curved specimens respectively, are offset to the left and right of the actual
specimen width to increase readability. Also, the lines that connect the
average values indicate the trends in the data. Growth occurs at a greater
Figure 8. Location of knee point in a stressstrain curve.
Figure 9. Growth stress as a function of specimen width for flat and curved panels.
stress as the thickness increases or when curvature is introduced. This is
further emphasized in Figure 10 where the stressstrain curves at location #2
from four different panels are plotted side by side. On each curve the initial
knee point is indicated.
CONCLUSIONS
Flat and curved composite sandwich panels of different widths were
impacted with an identical level of energy and loaded to failure in uniaxial
compression. As the transverse stiffness of the panels increased (increasing
width or finite curvature), the damage resulting from the impact increased.
Under compressive load, the damage tended to grow perpendicular to the
applied load by increasing the depth and area of the indentation.
Subsequently, ultimate failure of the panel occurred. Based on the
experimental observations, the following conclusions can be made:
1. Growth of the indentation occurs at a greater far-field strain (greater
stress) with increasing width or finite curvature. Because such parameters
as width and curvature are global to the local phenomena, the delay in
the growth point is most likely dependent on the damage state which
itself is dependent on the global geometry.
2. When the indentation grows, it reaches a point at which it grows no
further provided that the specimen is sufficiently wide. At this point
failure occurs as a catastrophic failure initiated at the indentation.
Because of the lack of dependence on width and curvature (provided that
Figure 10. Strain gage data from flat 305-mm-wide test specimen.
the specimen is sufficiently wide to allow for arrest to take place), the
damage must grow to an effectively similar condition independent of the
initial damage state.
3. The use of the 83-mm-wide specimen provides a conservative estimate of
the ultimate-load-carrying ability of realistic panels that are larger and
contain curvature. For the specific facesheet studied, the damage will
grow to between 50 and 75 mm in width prior to arrest. The 83-mm-wide
specimen is too narrow for the indentation growth to arrest itself.
ACKNOWLEDGMENTS
The authors would like to thank the following organizations and
individuals for their support of this work: Boeing-Philadelphia, Dr. Pierre
Minguet; the National Rotorcraft Technology Center, Dr. Yung Yu; and
the Federal Aviation Administration, Dr. Peter Shyprykevich.
REFERENCES
1. Kassapoglou, C. (1988). Buckling post-buckling and failure of elliptical delaminations in
laminates under compression. Composite Structures, 9: 139155.
2. Kassapoglou, C. (1988). A correlation parameter for predicting the compressive strength of
composite sandwich panels after low speed impact. Proceedings of the AIAA/ASME/ASCE/
AHS/ASC 29th Structures, Structural Dynamics and Materials Conference. Williamsburg,
Virginia. pp. 642649.
3. Kassapoglou, C., Jonas, P.J. and Abbott, R. (1988). Compressive strength of composite
sandwich panels after impact damage: an experimental and analytical study. Journal of
Composites Technology and Research, 10: 6573.
4. Kassapoglou, C. (October, 1996). Compression strength of composite sandwich panels after
barely visible impact damage. Journal of Composites Technology and Research, 18(4):
274284.
5. Minguet, Pierre J. (1991). A model for predicting the behavior of impact-damaged minimum
gage sandwich panels under compression. Proceedings of the AIAA/ASME/ASCE/AHS/
ASC 32nd Structures, Structural Dynamics and Materials Conference. Baltimore, Maryland.
pp. 11121122.
6. Cairns, Douglas S. (1987). Impact and post impact response of graphite/epoxy and kevlar/
epoxy structures. Technology Laboratory for Advanced Composites, Report 87-15,
Massachusetts Institute of Technology.
7. Lie, Simon C. (1989). Damage resistance and damage tolerance of thin composite facesheet
honeycomb panels. Technology laboratory for advanced composites, Report 89-3,
Massachusetts Institute of Technology.
8. Lagace, P., Tsang, W. and Williamson, J. (1990). Damage resistance and damage tolerance
of AMOCO ERLX-1983 3 K-70 PW sandwich panels. Technology Laboratory for Advanced
Composites, Report 90-1, Massachusetts Institute of Technology.
9. Vizzini, A. J. (March, 1992). An efficient algorithm to characterize stress-strain data using
piece-wise linear curves. Journal of Testing and Evaluation, 20(2): 126131.

Journal of Sandwich Structures and Materials-2002-Moody-71-82

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Journal of Sandwich Structures and Materials-2002-Moody-71-82

Uploaded by

Copyright:

Available Formats

http://jsm.sagepub.

http://jsm.sagepub.com/cgi/alerts Email Alerts:

- Jan 1, 2002 Version of Record >>

C) Bismuth metal alloy.

C and poured around

You might also like