Observation of Damage
Growth
in
Compressively Loaded Laminates
The phenomenological aspects of composite-panel-failure
under compressive in-plane loading and low-velocity
transverse impact are studied experimentally via real-time
recording of the failure-propagation event
by
H.
Chai,
W.G.
Knauss
and
C.D.
Babcock
ABSTRACT-An experimental program to determine tie
pi-enomenological aspects of composite-panel failure under
simultaneous compressive n-plane loading and low-velocity
transverse impact
[C-75
m/s
(0-250
ftls)] is described. High-
speed photograohy coupled with the shadow-moire techniq~e
is usec to record the phenomenon of failure propagatior. The
informat on gained from these records, supplemented
by
plate sectioning ard observatior for interior damage, has
provided informaticn regarding the failure-propagation
mechanism.
The results show that
the failure process can be divided
roughly into two phases.
In
the first phase the plare is in-
pacted, and tie resulting resoonse causes interla,ninar
separation.
in
the second phase the local damage spreads to
the unaamaged portion of tie plate through a combination of
laminae
b~ckling and further delanination.
List
of
Symbols
n
=
fringe number
V
=
impact velocity
w
=
out-of-plane displacement
w/n
=
fringe constant
At
=
time interval between frames
eo
=
compression strain ('load')
Introduction
Graphite/epoxy laminates are a special class of com-
posites which enjoy a definite strength-to-weight advantage
over many standard engineering materials used in aero-
space applications. This advantage is somewhat lessened,
however, by their sensitivity to operational hazards such
as low-velocity impact by foreign objects. Impact may
cause delamination damage since the strength of the plate
in its thickness direction is quite low. Under compressive
loadings this local delamination may grow and lead to a
global failure of the structure at a load well below the
design level. Structures may be subjected to both
high-
compression load and impact damage, possibly simul-
taneously. This situation is exemplified, for instance, by
H.
Chai, formerly Graduare Student and siibsequently Research Fellow at
California Instilute of Technology, is Visiting Scientist,
AFWAL/IMLBM,
Wrighr-Patterson
4FB,
Dayron,
OH
45433.
W.G. Knauss (SESA Member)
and
C.D.
Babcock (SESA Member) are Professors of Aeronautics,
Culifornra Institute of Technology, Pasadena,
CA
91 125.
Original manuscript submilted: July 13, 1982. Final version received:
A@
25, 1983.
a stone, kicked up on a runway, impacting
a
lower-wing
skin during landing or take-off. The wing motion generated
by the landing induces compression load cycled in the
skin so that the local damage induced
by impact can
propagate to the undamaged portions of the skin.
The effect of impact damage on the strength degrada-
tion in compressively loaded laminates can be charac-
terized
by
a threshold for catastrophic failure such as that
shown by curve
A
or
B
in Fig.
1.
This curve is generated
by impacting panels preloaded in compression and assessing
the post-impact results. Panels that fail completely upon
impact generate data for strains ('loads') and impact
velocities above the threshold curve whereas those
which did not fail generate data points below this curve.
Also shown in Fig.
1
is a graph of the damage area
Control O
Panels
Ref.
2
9
1
Fa~lure Threshold
Curve
A,
ref
Curve
B,
set
g,
€0
Impact
Velocity, V (mkec)
Fig. 1-Impact conditions for T300/5208 laminates.
Impactor is 12.7-mrn aluminum sphere. Triangular symbol
represents panel (set No.
2)
which failed due to global
plate response
Reprinted from EXPERIMENTAL MECHANICS,
Vol.
23,
No.
3,329-337,
September
1983
caused upon impact vs. impact velocity. This curve was
generated by impacting unloaded panels and determining
the extent of damage using the C-scan technique. Figure
1
shows that for small impact velocities [i.e.,
V
<
40
m/s
(130 ft/s)] there is no damage and no apparent strength
reduction in the laminate. For larger impact velocities, the
damage area increases while the plate strength decreases.
At impact velocity on the order of 100 m/s (330 ft/s) the
plate strength approaches a minimum which is only about
30 percent of the undamaged plate strength. It should be
noted that variations in test parameters such as plate
thickness,
layup and projectile mass can lead to variations
in the nominal value of projectile speed at which strength
degradation begins to take place or strength degradation
approaches a maximum.'
Several investigations on this topic have been per-
formed also by other
researcher^.'-^
In all these works
only postmortem examination of the impacted plates was
considered. While the effect of impact damage was fairly
well established in Refs. 1-5, little information was
revealed concerning the pocess of damage spreading. The
large amount of energy released during catastrophic
failure usually leads to a considerable disintegration of the
test specimen making it difficult if not impossible to
deduce the failure process from such examinations. The
mechanism of delamination growth in unloaded glass/
epoxy laminates subjected to impact was studied in Ref.
6
using high-speed photography. Unlike in Ref.
6
where the
growth of the delamination is attributed to the stress
waves generated by impact, here we are primarily con-
cerned with the spread of delamination under the action
of the compressive load.
In the present work, which is condensed from parts of
Refs. 8 and
9,
we determine the damage-growth mechanism
via real-time recording of the impact event. The material,
a T300/5208 graphite/epoxy laminate, is typical of the
configuration proposed for future heavily loaded primary
structures and has stiffness properties similar to those of
wing skins in existing transport aircrafts. The model
projectile is a 12.7-mm (0.5-in.) diam aluminum sphere
propelled normal to the plane of the plate at a velocity
in the range of 0-75 m/s (0-250
ft/s), its mass and speeds
simulating momenta typical of low-velocity impact hazards
that can occur in service. High-speed photography is used
to record the propagating failure from several positions.
In some tests back surface (unimpacted surface) and front
surface, or back surface and edge surface of the plate are
taken simultaneously using two high-speed cameras. The
records of the back and front surface are made in con-
junction with the
shadow-moirC technique to obtain a
full-field, out-of-plane deformation history of the im-
pacted plates.
Test Apparatus
A photograph of the test setup is shown in Fig. 2 and
includes the panel support fixture, the loading machine,
the impact device (air-gun), the light sources and two
high-speed cameras. A review of the main components
follows.
Fig.
2-Impact and recording test setup
Panel Fixture
Fig. 3-Test-panel support fixture
/
Flexed
Panel
Most& Grid/
1
/
Grid Shadow
.
.
Sect
ion
A
-
A
Fig. 4-shadow-moirdoptical setup
Mercury
lam^
4bEE
Spherical Condensing Coll
i
mating
Mirror Lens Lens
Mirror
(b)
Pulsed Mercury Lamp
Two fixtures are used to support two sets of graphite/
epoxy panels having different sizes, i.e., 20.3
x
10.2
x
0.60 cm (8
x
4
x
0.24 in.) in set No. 1, and 25.4
x
15.2
x
0.60 cm (10
x
6
x
0.24 in.) in set No. 2. The
'small' fixture is shown in Fig.
3.
The upper and lower
edges of the panel which transmit the load are bonded, to
aluminum plates in grooves filled with a bonding agent
(Devcon A). The depth of the grooves in the 'small' and
'large' panel fixtures are 1 cm (0.4 in.) and 2.5 cm
(1.0 in.), respectively. Both sides of the specimens are
held by narrow strips. These strips are adjustable along
the length of the specimen to ensure a close, but non-
binding, support.
Loading Machine
The load is applied to the specimen by a hydraulic
testing machine capable of exerting loads up to 150 metric
tons. A spherical base support transmits the load to the
specimen to assure uniformity in the load distribution.
The latter, measured by as many as eight strain gages, is
within +3 percent of the mean.
Impact Device
The projectile is accelerated by a simple air-gun con-
sisting of a compressed-air reservoir, solenoid activated
valve and a 38 cm (15 in.) long barrel. The projectile
velocity is controlled by adjusting the pressure in the
reservoir and is measured by twice interrupting a light
beam monitored by a photodiode. The time interval
between beam interruptions is measured by a counter.
With this device, the scatter in the impact velocity is less
than
+
1 percent.
Shadow Moire
Moir6 is an optical phenomenon based on the inter-
action of light formed with two gratings; one grating is
stationary and undeformed (the master grating) while the
other deforms with the specimen. As the specimen grating
Ground
0.2
in.
Flash -+-/f,=4,5in.~f2
=
1
l
in.---=-
Tube
Time
__I_)
(0)
Xenon Flash Tube
Fig. 5-Optical
arrangement
(upper parts) and
light-intensity
history (lower
parts) of two
light sources
deforms, a fringe pattern related to the surface deforma-
tion of the specimen is observed. In the shadow moirC, a
grid placed close to the specimen surface serves as the
master grating while its shadow on the matte-reflective
surface of the specimen, formed by an incident collimated
light beam, serves as the specimen grating. The normal
displacement
w
is given by (for details, see for example
Ref.
7)
=
-
P
=fringeconstant (I),
n
tan
0,
+tan
0,
where
n
is the fringe number,
p
is the grating pitch and
0,
and
0,
are the angles formed between the normal to the
plate and the directions of the incident and viewing light,
as shown in Fig. 4.
A
Ronchi-ruled grid with 4 P/mm (100 !/in.) is used as
the master grating. The grid is placed close to the speci-
men surface [2-4 mm (0.08-0.16 in.)] for sufficient fringe
contrast. Because the grating is destroyed in each test,
replicas of the grating are formed by contact printing the
master onto high-contrast, high-resolution Kodak
Ortho
Plate PFO. The matte reflective surface for the shadow
moire is achieved by painting the plate surface with Kry-
lon 'silver' spray paint.
Light
Source
A
Xenon flash tube, powered by a bank of capacitors
which is capable of discharging 4000 joules is used in
recording the impact event. The optical setup shown in
Fig.
5(a) is used in order to provide a collimated and
uniform light beam at the specimen surface. The flash
history, controlled by adjusting the discharging rate of the
capacitors, is selected to produce a pulse as shown in the
record in Fig.
5(a). The flash tube is triggered by the
projectile using a simple wire-breaking circuit.
A
200-w
Mercury bulb is used in conjunction with the Xenon flash
tube on occasions where multiple but simultaneous re-
cordings are made. The bulb is pulsed to
800
w during the
Fig. 6(a) and
(b)-
Impact damage
in
a
6-mm thick graphite1
epoxy laminate:
(a) C-scan of an
unloaded plate
impacted at 76 mls,
(b)
through-the-
thickness section
along
90
deg of the
panel in (a)
i
impact to increase its light intensity, according to the
record shown in Fig. 5(b).
High-speed Camera
One, and on occasions two, 16-mm Hycam (Red Lake
Laboratory, Model K2054BE) high-speed framing
cameras were used to record the impact event. In one
camera a quarter-frame assembly has been installed to
TABLE
1
-TEST CONDITIONS'
--
Set No. l(20.3~ 10.2x0.6crn) Set No. 2(25.4x 15.2x0.6crn)
Panel
V
(mls)
€0
Panel
V
(rnls)
co
'Except where otherwise mentioned, back-surface moire' is employed,
and the panels failed completely upon impact.
'Back surface and edge of panel are recorded simulfaneously.
3~ack and front surfaces of panel are recorded simultaneously.
4Panel did not fail
upon
impact.
increase its recording rate up to 40,000 frameds. The
exposure time, controlled by an interchangeable shutter,
is selected in the range of 2.5-5 ps. In cases where two
cameras are used simultaneously, a common timing light
pulse displayed on both films synchronizes the two
records (Kodak pan film, 2484 high-speed film, rated
at 800 A.S.A.).
Test Procedure and Test Data
The panels tested in this investigation were furnished
through NASA-Langley.* The specimens were fabricated
from commercially available tapes made from unidirec-
tional Thornel-300 graphite fibers pre-impregnated with
Narmco-5208 epoxy resin (28-percent resin content). The
tapes were laid up to form a 48-ply
(+45/ -45/0/0/ +45/
-45/0/0/ +45/ -45/0/90),, laminate.
**
The panels, supported by the fixtures, are impacted at
the center while under compressive load applied along the
0-deg ply direction. The impact event is photographed.
Table
1
.summarizes the test conditions and the photo-
graphic procedure. The test results are displayed in Fig. 1
along with a rough estimate of a curve indicating the
failure threshold of panels belonging to set No. 2 (curve
B). The lower segment of this curve is drawn relatively
accurately. The high-speed photographs of the two 'large'
panels closest to this segment
(i.e., panel No. 3 and
No. 6 in Table 1) exhibit a phase of initial delay in the
damage growth. This is in contrast to the rest of the
panels in that segment of the curve which show damage
growth without such a delay, the growth rate which
increases with increasing strain. Therefore, the test
conditions for the two panels under discussion represent
nearly threshold data. The rest of the curve could not be
drawn with such an accuracy. Note first that establishing
*Courtesy of J.H. Starnes, Jr
**The nominal ply properties are: longiludinal modulus
=
131.0 GN/m2
(19.0
x
106 PSI), transverse modulus
=
13.0 GN/mZ (1.89
x
lo6 PSI),
shear modulus
=
6.4 GN/m2 (0.93
x
lo6 PSI), major Poisson's ratio
=
0.38. The calculated composite stiffnesses can be found in Refs. 8 and 9.
a failure threshold curve was not our prime motivation in
this work. Secondly, we recognize that the limited supply
of test material is inadequate to define such a curve well.
The available data, however, are sufficient to indicate the
existence of limit behavior in the threshold curve at both
relatively low and relatively high projectile speeds; at
these velocity ranges the failure strain seems to be little
affected by changes in the impact speed. It is also clear
that in between these two extremes a transition region
exists in the threshold curve where the failure strain
degrades rapidly with increasing impact velocity. Also
shown in Fig.
1
is a failure threshold curve obtained in
Ref.
2
(curve A) for the same material as in this work but
for a slightly different plate size [25
x
13
x
0.68 cm
(10
x
5
x
0.27 in.)]. The differences observed between
the data of Ref. 2 (curve A) and our data (curve B and
the data for the panels of set No. 1) are probably due to
lot variations in material properties.
Interior-damage Characterization
Some panels that did not fail catastrophically upon
impact were examined ultrasonically and/or sectioned
through the plate thickness at different locations and
examined microscopically for interior damage. Figure 6(a)
shows an ultrasound C-scan* record of a plate-impacted
while under zero load. From this record, the damage area
appears roughly circular in shape with approximately
3 cm (1.2 in.) in diameter. A section of this panel along
the 90-deg ply direction and through the impact point is
shown in Fig.
6(b). The damage in Fig. 6(b) is charac-
terized by delamination and intraply cracking extending
across the full thickness of the plate. This damage is
extended laterally more toward the unimpacted panel
face than toward the impacted face.
A section through the impact point and along the 0-deg
ply direction of a plate which was loaded during impact
*The aulhors are grateful to J.H.
Starnes, Jr. of NASA-Langley for
providing the C-scans.
Fig. 6(c) and
(d)-
Two sections
along 0 deg,
12.7-mm apart, of
a panel loaded to
,,
E,,
=
0.0035 and
impacted at 65 mls
