Composite Structures 6 (1986) 295-318

The Damage Tolerance of Carbon Fiber Reinforced

Composites--A Workshop Summary

K. D. Challenger
Department of the Navy, Office of Naval Research, 223/231 Old Marylebone Road,
London NWl 5TH, UK

A BSTRA CT

The principal objective of this workshop was to identify the critical

problem areas associated with the damage tolerance of carbon-fiber
reinforced composite materials. The discussion was divided into six
areas: (1) damage tolerant materials; (2) testing methods; (3) structural
life prediction; (4) damage tolerant design concepts; (5) repair methods;
and (6) nondestructive testing. Approximately 1 h was devoted to the
discussion of each of these topics. Discussion was stimulated by having
one, two, or three introductory presentations by the discussion leaders
of each topic followed by open discussion. In this report, my
impressions of the discussion on each topic are combined with those of
the discussion leaders and presented in the following format: Summary
of Presentations; Summary of Discussion; and Critical Issues. It is
possible to obtain the final conclusions of the workshop by examining
the critical issues listed at the end of each section.
The discussion was lively, controversial, and open. The main
conclusions to be drawn from the workshop are: (1) damage from
impact is the worst type of damage for these materials---significant
reductions in the compressive strength will occur following impact; (2)
fatigue damage at the present time does not limit the use of these
materials but as new materials are developed fatigue failure may become
an issue; (3) very little is understood about the micromechanics of
damage, hence it is impossible to predict the effect of changing the
properties of the individual components (fibers, matrix, and fiber-
matrix interface) on the bulk material properties; (4) better analytical
models to describe the formation and growth of impact-damage are
295
Composite Structures 0263-8223186/$03.50 © Elsevier Applied Science Publishers Ltd,
England, 1986. Printed in Great Britain
296 K. D. Challenger

badly needed: and (5) rapid NDT methods to inspect large components
are required.

I INTRODUCTION

Composite materials are used in applications where they must be able to

absorb significant amounts of energy--such as for crash worthiness and
composite armor. In these applications, the requirements for the
materials are well defined, and the materials and components can be
designed accordingly. For structural applications, components are
designed to carry the service loads, and damage tolerance is more
difficult to define and allow for in the initial designs. This important fact
aside, one still needs to be able to assess the residual properties of a
damaged structure in order to make intelligent decisions regarding
whether to repair or replace the component.
After I had talked to many different people about this topic it became
clear to me that considerable confusion and a general lack of under-
standing exists in the field of damage assessment and damage
mechanisms in carbon-fiber reinforced plastic (CFRP) materials.
The following pages present the result of a workshop on the damage
tolerance of fiber reinforced composites. This workshop was organized
and sponsored by the London Branch Office of the US Office of Naval
Research. The workshop was held in Glasgow, Scotland, on 12
September 1985 as a follow-on to the Third International Conference on
Composite Structures held on the preceding days.* A list of participants
and their affiliations is given in the Appendix.

2 M A T E R I A L S ASPECTS TO D A M A G E T O L E R A N C E

2. I Summary of presentations

Professor D. Hull (University of Cambridge) began the workshop with

the message that we need to clearly define what is meant by damage
tolerance. One cannot state that one composite is more damage tolerant

*Available as Composite Structures 3, edited by I. H. Marshall,ElsevierApplied Science

Publishers, London, 1985.
 Damage tolerance of carbonfiber reinforced composites 297

than another because the resistance to damage depends on how the

loads get into the material to produce fracture. What we really mean by
damage tolerance is that we are able to predict the strength of the
materials---strength in the broadest sense; i.e., effects of environment,
temperature, stress state, damage state, etc. The definition of damage
can range from surface scratches to ballistic attack, depending on the
application. From the ensuing discussion it was agreed that for the
purpose of this workshop we would focus on the sources of damage in
large composite panels and the subsequent properties of the panel.
Two questions must be addressed by research: (1) how damage forms;
and (2) how damage grows under some external load. The nature of the
damage produced by a given load will depend on structural parameters
(lay-up sequence, shape, and size of the component) and on material
parameters (stiffness of the layers and the critical strain energy release
rates, G0, for inter- and intra-laminar fracture). The residual strength
will depend on various buckling instabilities and the growth of damage
which, in turn, depend on both structural and material parameters. The
energy absorbed in fracture will depend on the micromechanisms of
fracture, and this will depend on the properties of the matrix, fibers, and
fiber-matrix interface and on the local conditions of stress, which
determine which of a large number of different mechanisms can
operate. Predicting these local stresses, then, depends on a microscopic
understanding of how the loads get into a structure. In some instances it
may be desirable to minimize the initial damage and, in another
instance, to prevent subsequent propagation of damage. These are two
quite different circumstances in terms of a materials response to external
loads.
Composites can absorb vast amounts of energy provided one can
control the damage; the fiber orientation and the matrix and interface
properties must be optimized in order to achieve the minimum amount
of damage and thus absorb the maximum amount of energy.
Even though one would expect that the material properties such as
G,o, G,,c, and Gtuc (modes I, II, and III critical strain energy release
rates) are controlled by the properties of the fibers, matrix, and
fiber-matrix interface, it is not at all obvious how to change the values of
Go by varying the individual (fiber, matrix, and fiber-matrix interface)
properties. For example, what must be done to the matrix to increase
GH? This is still unknown. Much more research needs to be done in
developing micromechanical models of fracture in order to understand
298 K. D. Challenger

and predict the effect of changing the properties of the individual

components on the bulk properties of the composite.
Dr I. Wolock (Naval Research Laboratory [NRL], Washington, DC)
again expressed some concern over exactly what is meant bv damage
tolerance because failure--the avoidance of which is the basis of
developing damage-tolerant materials is a subjective term. Failure is
very much a function of the application. Damage of one kind may or
may not lead to failure. For example, a delamination may be vet,,'
deleterious to the compressive strength of a composite but have
essentially no effect on the tensile strength. Thus, a problem exists in
defining failure.
Energy absorption was proposed by Wolock as a convenient way to
measure damage tolerance~the ability to absorb energy without failing
(whatever is meant by failure in any specific instance). Energy absorp-
tion techniques for measuring damage tolerance are under development
at NRL.
R. Mousley (Royal Aircraft Establishment [RAE], Farnborough)

STRAIN AT
FAILURE IO 6
12000

N
,C/,6
8000 , uw
ill.
hi F-
0

z_.

4000 DUL.

Z!
u u
_ ~.,:

_o

• ~Z ~_
0,,. ,~.
J ZJ .J Z.J
,, ,~ ,, a~ .... • a

TENSION COMPRESSION

Fig, 1. Effects of various defects on the failure strain of C F R P .

 Damage tolerance of carbon fiber reinforced composites 299

indicated that for CFRP the most serious damage is that caused by
impact because this will normally create a region of delamination which
is surrounded by matrix and fiber cracks. Figure 1 summarizes results of
many tests from the RAE and clearly shows that the greatest loss in
strength (failure strain) of CFRP is due to impact damage. In fact, it is
the only type of defect that resulted in a residual strength less than the
design ultimate load (DUL). The form of the impact damage depends
on a variety of factors, including the material parameters, the impactor
characteristics (shape, velocity, mass, etc.), and the structural support

8
,'7'
o

,, 6

4 :'" ! :"-i
c

L L ~
20 t.O 60 8O

Damage width mm

Fig. 2. Strain to failureas a functionof the widthof impactdamagearea in large CFRP

panels subjectedto uniaxialcompressionloading.

of the laminate. Small but significant improvements for impact damage

tolerance of the current CFRP can be achieved by (1) the use of woven
- 4 5 ° plies in conjunction with unidirectional 0° plies; (2) the placement
of - 4 5 ° plies on the outside of the laminate to restrain the 0° plies, thus
providing artificial protection and reducing the ftexural strains in the 0°
plies; and (3) optimizing the fiber surface treatment to control the
energy of the fiber-matrix interface.
The development of higher strain-to-failure fibers and improvements
in matrix toughness have occurred, but without the understanding of
how these properties affect the bulk properties (as pointed out by Hull)
we cannot as yet optimize the combinations of fibers and matrices•
Figure 2 was presented and subsequently became the subject of
considerable discussion. The results are from uniaxial compression tests
300 K. D. Challenger

on current materials (i.e., no toughened epoxies) with lay-up construc-

tion that would be used for structural applications. Thus Gi~ for these
materials would tend to be in a narrow band. This curve suggests that, at
least in the short term, a set of semi-empirical assessment curves can be
developed to predict residual strength. But for the longer term.
modeling based on sound fundamental principals is required.

2.2 Summary of discussion

From the point of view of a structural designer, the solution to the

problem of damage tolerance is twofold: design better (more damage
tolerant) structures or use better materials. You can improve the
damage tolerance of a structure by either or both methods. However,
there is no single material that will exhibit the best damage tolerance for
all loading conditions. For example, tough matrices, such as the
thermoplastic, polyetheretherketone (PEEK), and toughened epoxies,
resist splitting within and delamination between plies. Thus, less damage
occurs in these materials as a result of impact and there is more
resistance to the growth of a delamination during subsequent loading in
compression. The stress concentration created by through-thickness
notches and broken fibers is reduced by the formation of a damage zone
in the vicinity of these defects when a tensile load is applied. This
damage zone consists of splitting within and delamination between the
plies. The tougher matrix materials resist this type of damage better than
the conventional lower toughness epoxies. Thus, the stress concen-
trations are not reduced for the tough matrix materials resulting in an
increase in the notch sensitivity for these tougher materials. The
decreased damage due to impact and the increased notch sensitivity tend
to cancel out and the residual tensile strengths are similar for P E E K
(APC-1), toughened epoxies, and standard epoxies. The newer PEEK
material, APC-2, tends to not only improve the post impact compres-
sion strength but also the post impact tensile strength. This is not clearly
understood yet, but seems to be related to the interracial bond
characteristics and the ability of PEEK to plastically deform.
The way to improve the damage tolerance of a material for a specific
application is first to predict the mode of failure. Once this is done, the
material should be optimized with respect to increasing the energy
absorbed by the predicted failure mode. The material properties which
are of significance for the specific application of interest need to be
 Damage toleranceof carbonfiber reinforced composites 301

identified and optimized. In this sense, it is not sufficient to evaluate a

material based only on the fracture energies G,,, G,~c, and Gm~ of
undirectional materials because the properties of interest (damage
formation and growth, and the residual strength) will involve other
modes of failure. The residual strength curve (Fig. 2) presented by
Mousley could not be correlated with any G~ value; however, many
different materials containing natural impact damage have been shown
to follow this behavior. Tougher resin matrix materials have been
developed, but the improvement in matrix toughness does not always
lead to a comparable improvement in toughness of the composite. The
lack of understanding of the relationship between the bulk properties of
the composite and the individual properties of the components (fibers,
matrix, and fiber-matrix interface) was identified as the largest problem
area in the development of improved materials. Once this under-
standing is achieved, then analytical models can be developed that will
allow one to optimize a composite for a given failure mode. But it will
still be necessary to first determine the controlling failure mode for the
specific application and then optimize the material.

2.3 Critical issues

1. Poor understanding of the micromechanisms of damage formation

and growth.
2. Lack of an analytical model to relate the properties of the fibers,
matrix, and fiber-matrix interface to the bulk properties of the com-
posite, and consequently, difficulty in defining a better material for any
specific application.

3 TESTING FOR DAMAGE TOLERANCE

Testing for damage tolerance is a major problem for composites. From

the previous section it is clear that given we know Gtc (even if it is a true
geometry-independent material parameter), we are still unsure of how
to improve it by modifying the properties of the components of the
composites and, further, we do not know what G~omeans or how to use
it for assessing the damage tolerance of a structure. The test results
compiled by the RAE (Mousley), Fig. 2, show that the post impact
uniaxial compression properties of large panels are insensitive to the
302_ K. D. Challenger

lay-up and matrix properties for current structural materials. Therefore,

we do not know what a high or low value of Gc really means in terms of
damage tolerance. Even before we begin to discuss testing methods, a
problem exists because we do not know how to interpret these results
with regard to damage tolerance. However, since this will not always be
the case (hopefully), test techniques to establish the material properties
G~¢, G~c, G,H~, and mixed mode loading must be developed which are
independent of specimen geometry. If the results cannot be scaled, up to
predict the properties of full-scale components, their value is severely
limited.

3.1 Summary of presentations

Ms G. Murri (USAUSCOM, National Aeronautics and Space Adminis-

tration [NASA]-Langley) pointed out that even low-energy impacts
such as tool drops or hail strikes can cause internal delaminations which
significantly reduce the compressive strength of laminated composites.
The degree of resistance to delamination growth can be assessed, at a
basic level, by simple edge delamination and double cantilever shear
tests which allow a comparison of available material systems in terms of
interlaminar toughness. More realistically, the problem can be
approached by a study of the mechanics of low-velocity impacts on
composite plates and their compression behavior after impact damage.
In the case of thin plates, the deformation patterns due to tow-velocity
impacts can be successfully simulated by static indentation loads. Test
results have demonstrated that the lowest bounds of residual compres-
sion strength are obtained after the impacting of compression-loaded
panels, and that low-mass/high-velocity impacts cause more extensive
damage than high-mass/tow-velocity impacts at the same energy levels.
Dr A. Kinloch (Imperial College, London) also identified potential
impact damage as a major obstacle to advanced applications of fiber-
reinforced composites. Currently used test methods aim at the deter-
mination of interlaminar toughness data, especially Go, which are
predicted on linear-elastic fracture mechanics principles. Such data can
be used for identification of the relationships between impact damage
and residual mechanical properties or for the evaluation of new resin
systems with improved damage tolerance. Some success has been
achieved in empirically correlating the interlaminar and bulk fracture
energies of toughened resins, and in also the interlaminar fracture
 Damage toleranceof carbonfiber reinforced composites 303
energy of impacted composites and the associated failure strains. Future
research efforts should address test methods which are independent of
specimen geometry and capable of supporting the investigation of
interlaminar crack stability and crack growth.
Dr Bergmann (German Aerospace Research Establishment
[DFVLR], Braunschweig, West Germany) emphasized the difficulty of
analytically predicting the growth of internal delaminations under
staticmand especially cyclicmloads. Experimentally, tolerance levels
can be identified below which internal delaminations do not grow and
can be safely ignored. In corresponding tests, difficulties are often
encountered in the form of unwanted edge delaminations of the test
coupons which disturb the aim of the investigations. These edge
delaminations can be avoided or minimized by surrounding the test
coupons with a pre-stressed woven glass-fiber hose. Test results show
that the suppression of edge delaminations tends to increase the static
strength by 10 to 15% and the fatigue life of unflawed test specimens by
an order of magnitude. It stands to reason that damage tolerance tests
are similarly affected and should be accordingly interpreted.

3.2 Summary of discussion

The measurement of the interlaminar toughness of fiber-reinforced

composites, mainly mode I values, is based on continuum fracture
mechanics. The issue of whether mode II and mode III values are as
relevant and should therefore also be measured could not be resolved.
Concern was raised that all measured properties reflect the cohesive
sffength of the resin as well as the adhesive quality of the interface, and
that the precepts of classical fracture mechanics may not be generally
applicable. Granting the applicability to purely interlaminar
phenomena, such as delaminations, the real problem will be the
establishment of mixed mode failure criteria in the form of interacting
fracture energies G,0, Guc, and Gm¢. The current testing of fracture
energy serves primarily as one of several screening techniques for the
fracture toughness of competing resin systems, although it is uncertain
whether G¢ values alone are sufficient. A correlation between the G¢ of
bulk matrix and the interlaminar toughness of the composite may also
support the optimization of the latter. By a similar line of reasoning,
measured values of G¢ may be needed for relating impact damage and
residual compressive strength.
304 K. D, Challenger

The discussion then turned to the question whether improved test

techniques might provide more meaningful information for the damage
tolerance of composites. It was agreed that fiber-reinforced resin
systems require a much more involved test philosophy than conven-
tional metals with respect to the constituent properties as well as to test
conditions. It follows that new test methods can hardly be developed
without a theoretical understanding of the interplay of the various
parameters. This, in turn, suggests closer communications between
analysts, designers, materials, and test engineers than is usually
practised in regard either to the characterization of material systems or
the improvement of design procedures.
An urgent need exists for the assessment of the significance of
delaminations caused by low-velocity impact as the governing damage
tolerance problem. For experimental investigations damage can be
introduced into the test specimens by impact or, artificially, by out-
gassing agents or by implantation of thin teflon disks. While the implant
allows direct control over delamination size and location and is well
suited for the support of analytical studies, it does not produce the
desired sharp crack front on the periphery of the disk. However. a
realistic crack front does exist once the delamination has commenced
growing, so this technique may be acceptable after all. To what extent
classical fracture mechanics concepts can be used for the prediction of
delamination growth is still an open issue. Pre-requisites seem to be the
development of geometry independent of standardized test methods for
GI~ Guc, and Gm~, and interracial failure criteria in the form of
interaction diagrams. Jones suggested that a unified test configuration
based on that proposed in AFWAL-TR-79-3095 be numerically studied
by a round robin testing program.
In summary, it is evident that the problem of damage tolerance testing
of fiber-reinforced composites has many facets and is far from being
understood at this time.

3.3 Critical issues

1. Lack of a standardized test method that produces fracture data

independent of specimen geometry.
2. Uncertainty about how the material parameters Gto, Gu,, and GIt~
affect damage tolerance.
 Damage tolerance of carbon fiber reinforced composites 305

4 STRUCTURAL LIFE PREDICTIONS

4.1 Summary of presentations

Dr G. Dorey (RAE, Farnborough) opened this session with the bold

statement that composite materials do not have a fatigue problem. As
an example, helicopter rotors have been designed and built using
composite materials where an infinite life is expected. However, Dorey
went on to say that composite materials can be degraded by cyclic
loading, and service fractures might result; therefore, design philoso-
phies should anticipate these problems.
With the fatigue behavior exhibited by metallic materials it is possible
to define the inspection periods such that any flaw (or crack) will be
found before it grows (by fatigue crack growth) to a known critical size.
Obviously this is a damage-tolerant design philosophy, and it requires
adequate inspection, a good description of the flaw growth, and
preferably a material with a slow crack growth rate and a large critical
flaw size. Composite materials are not damaged by fatigue in the same
way as metals. The degradation of composites under cyclic loading is
more by general growth of cracks through the matrix or at the
fiber-matrix interface, and is caused by secondary stresses in the system.
Such damage growth can be diffuse and widespread and is more
detrimental to compression and shear loading than to tensile loading.
This kind of behavior causes more scatter in fatigue lifetimes and a
greater sensitivity to stress. The time required to define failure proba-
bilities can become unacceptably long.
Impact damage tends to affect the static strength more than fatigue
life. Thus, the current design philosophy is to account for the effect of
impact damage on static strength and not worry about fatigue. In fact,
fatigue cycling of impact damaged specimens will often improve the
static tensile strength because the damage caused by the cyclic loading
tends to reduce stress concentration at the damage area caused by the
impact. The real concern is to predict the changes in static strength that
result from accidental damage; this causes more concern with inspection
than with a damage tolerant design. But a method based on inspection
will require frequent inspections, so rapid inspection methods are
necessary.
Dorey then pointed out that with the new tougher materials (such as
306 K. D. Challenger

PEEK) a change in failure mode may occur (they are more sensitive to
stress concentrations) and the S-N curves are steeper--perhaps indi-
cating a fatigue problem.
Dr I. H. Marshall (Paisley College of Technology) stated that present
research at Paisley into analytical methods of accounting for impact
damage has strongly underlined the extreme complexity of realistically
modeling this problem. Accordingly, although finite element methods
(FE) have considerable potential, it is considered that only a hybrid
technique which utilizes extensive experimental evidence along with FE
methods will ultimately solve the problem. A double-sided impact test
has been developed that, he claims, will provide a means to relate
externally applied impact energy to the ensuing damage in the com-
posite. Although this test is not representative of actual in-service
impact, it creates a symmetrical damage in the laminate which lends
itself more readily to analysis. Marshall believes that the relationship
between the external (macro) effects and the internal damage (micro)
can be developed for this more simple circumstance and should thus be
an important step towards gaining insight into the problem.

4.2 Summary of discussion

There was considerable debate about whether it is reasonable to

consider a damage-tolerant design concept for these materials. At
present there is really nothing to monitor, in the sense that one is
looking for a flaw of some size growing at a certain rate. Dorey pointed
out that many cyclic tests of impact damaged specimens have been
performed at RAE; sometimes the damage grows and sometimes it does
not, and they cannot always predict the behavior. A general agreement
was reached that with the existing fiber-critical CFRP materials, the
fatigue lifetime curve (S-N) is almost flat which, in e s s e n c e , is similar to
a static strength curve.
Concern over the definition of failure for a fatigue specimen exists. A
composite specimen may be extensively damaged and useless but still be
in one piece and thus, in some definitions, has not failed.
It was pointed out by several participants that the growth of impact
damage by cyclic loading in real structures has been shown to be very
much slower than for laboratory test specimens.
At this time, it seems safe to generalize that with the existing
carbon-fiber reinforced epoxies fatigue fracture is not an important
 Damage toleranceof carbonfiber reinforcedcomposites 307

issue. However, impact damage does grow (in some instances) when
subjected to cyclic loading, so more research needs to be performed in
order to understand the conditions which lead to and the mechanisms
which control this growth.

4.3 Critical issues

1. Need to predict static failure load in the presence of impact

damage.
2. Need to be able to predict the growth of impact damage during
cyclic loading.

5 DAMAGE TOLERANT DESIGN CONCEPTS

5.1 Summary of presentations

Dr E. Demuts (US Air Force Wright Aeronautical Laboratory

[AFWAL]) described the design methods used by industry for the US
Air Force to ensure that catastrophic failures due to unsuspected or
undetected damage do not occur. This is done by designing the structure
with an assumed initial damage; the size of this damage is set by the
maximum size of invisible damage that would be expected. If the ratio of
the maximum cyclic load to the static damage strength is below 0-6, the
fatigue life will exceed 106 cycles; hence, static strength governs design.
Large structures have been found to be more resistant to both the
initiation and growth of damage than small laboratory specimens. So
designs based on laboratory data are generally conservative. However,
much of the data generated in the laboratory cannot be used by a
designer because of the lack of standardization of testing methods.
Support fixtures, impactor size and shape, specimen size, and boundary
conditions all need to be standardized.
From the design viewpoint, many damage tolerant concepts are
available; all have been used with varying degrees of success: _+45° ,
10/80/10, stitching, fastener rows, buffer strips, softening strips, tear
straps, hybrid materials, fabric bonded reinforcement, 3-D reinforce-
ment, stitcbed-woven fabrics, adhesive interleaf.s, localized adhesive
strip interleaving, various stiffeners (I, blade, hat, isogrid), multiple
load paths, and hybrid structure boxes. However, at present a designer
308 K. D. Challenger

cannot reliably provide a design with a prescribed degree of damage

tolerance by currently available analytical means. Instead, experimental
data from typically expensive tests must be used. Current designs work
because of experience and ingenuity on the part of the designers, but we
need to know why certain concepts work. Hence, the significance of,
and the need for, a useful and affordable analytic capability.
Dr A. de Rouvray (Engineering Systems International, Paris)
discussed a finite element analysis model that has been developed to
predict the growth of damage in CFRP multilayered laminates when
subjected to tensile stresses. The model can analyze the matrix and
fibers separately, but can also simulate the interaction between them. A
value of Gtc determined from unidirectional tensile tests of uni-
directional fiber composites was used successfully to predict the onset
and growth of interlaminar cracking for the multilayered laminates. The
model calculates Gt along the crack front, comparing this to the critical
value, GLc, and assesses when the crack will propagate. The analytical
results have successfully predicted experimental results that have been
determined by Bergmann (DFVLR). de Rouvray also emphasized that
analytical models are desperately needed in order to reduce the need for
expensive structural tests. He also pointed out that the models should be
developed for specific problems, keeping them as simple as possible.

5.2 Summary of discussion

As most of the important issues had already been addressed, the

discussion was brief. However, Jones was emphatic that the situation is
not as bad as it may seem. He pointed out that. although there are
serious problems with G-based design concepts, a cooperative program
developed by the Commonwealth Advisory Aeronautical Research
Council has developed a useable S-based design. This is a strain energy
density criteria (see, for example. Ref. 1), and has been used to
successfully predict (within 10%) the residual compressive strength of
damaged CFRP laminates and to predict the shape of the curve shown in
Fig. 2. In fact, this generic shape (i.e.. residual compressive strength
approaching a limiting value that becomes independent of the size of the
damaged region) was predicted numerically before it was observed
experimentally.
Jones believes that it is possible to predict the onset of failure and the
direction of the growth of the damage, but that it is not currently
possible to predict the rate of growth of the damage.
 Damage toleranceof carbonfiber reinforced composites 309

5.3 Critical issues

1. Uniform test methods and standardized methods for damage

characterization are needed.
2. Analytical models based on the micromechanical mechanisms of
failure must be developed in order to assess the effect of damage in
composite structure, especially for the growth rate of the damage.

6 NONDESTRUCTIVE TESTING

6.1 Summary of presentation

W. N. Reynolds (NDT Center, Harwell Laboratory) described the

current state of development of each of the major nondestructive testing
(NDT) techniques applicable to this problem:
1. Radiography. The increasing availability of intense microfocus
sources in the appropriate energy range is bringing real-time projective
radiography, with the option of video recording, into wider use. In
particular, the possibility of high definition real-time analogue tomo-
graphy is now being investigated.
2. Ultrasonic C-scan. The practical difficulties of low scanning speed
and the need for water coupling can be overcome to some extent by the
use of special dry-coupled wheel transducers. Alternatively, eddy
current systems for the detection and assessment of damage can also
provide useful scans in carbon or graphite reinforced materials.
3. Thermography. Thermal pulses using flash tubes or other high-
power devices to produce a sudden change in surface temperature, when
used in conjunction with video-compatible infra-red imaging systems,
have been shown to provide good information about delaminations and
other features of barely visible impact damage. Images are similar in
many respects to C-scans but are obtained very rapidly and require no
physical contact. Many problems can be dealt with when access is
available from one side only. High quality imagers are fairly expensive
at present, but pyro-electric vidicon systems of much lower cost are still
being developed.
4. Mechanical response systems. These are systems in which higher
strains are required to assess directly the changes in mechanical consti-
tutive relationships due to damage. Additional infra-red techniques such
310 K. D. Challenger

as thermal stress analysis (the SPATE system) or vibrothermography

then become applicable.
5. Ultrasonic analysis. Computerized ultrasonic systems are now
available to measure the attenuation coefficient over a wide frequency
range (e.g., 2 to 20 M2Hz) in 1 to 2 min. This has made it possible to
measure moisture penetration and other types of damage in resins and
composites.
6. Stress wave emission. Current indications are that existing methods
of stress wave emission analysis are not yet adequate to isolate the
significant features from the highly complex data obtained from
d a m a g e d materials. However, useful future progress can be expected.

6.2 Summary of discussion

F r o m the general discussion it became clear that N D T is a significant but

not a primary concern for fabricators and users. However, a suitable
range of techniques with well defined performance, including operating
costs, will be needed for use under various conditions. The technique
which aroused most interest, as being the newest in the field, was the
pulse video thermography, and details of performance of the Barr and
Stroud IR18 imager were discussed. There were also other factual
questions about costs, emissivity and temperature measurements, and
image analysis. The other main points which arose were:
1. The primary purpose of N D T is to ensure that defects will not
become critical before the next inspection, i.e., that the vehicle will
survive the next mission. This is the essence of damage tolerant design,
but even with good N D T methods this question cannot be answered at
present.
2. Rapid inspection of a complete aircraft, or its critical areas, may be
essential. Tests can vary from hammer tapping to highly computerized
thermography.
3. It is assumed that defects 25 to 50 mm in diameter must be
detected, but there is no strong evidence for this figure.
4. Both thermoplastics and toughened thermosets are of interest, and
slightly different N D T approaches may be needed in each case.
5. Eddy current tests are sensitive to broken fibers, but it is not
known how small a number can be detected.
 Damage tolerance of carbonfiber reinforced composites 311

In conclusion, it was emphasized that performance data and costs for a

range of NDT approaches should be available so that the appropriate
choice of equipment could be made for specific circumstances.

6.3 Critical issue

Need to develop a rapid method for NDT of large structures.

7 REPAIR OF COMPOSITES

7.1 Summary of presentations

T. Hess (Naval Air Development Center [NADC], Warminster,

Pennsylvania) explained that graphite/epoxy composites have many
advantages for use as aircraft structural materials, including their high
specific strength and stiffness, resistance to damage by fatigue loading,
and immunity to corrosion. Extensive use of these composites should
reduce the high maintenance costs associated with repair of corrosion
damage normally encountered with conventional aluminum alloys,
particularly those exposed in a marine environment. Similarly, costs
associated with repair of damage due to fatigue should also be sub-
stantially reduced, since the composites do not in general suffer from the
cracking encountered with metallic structures, particularly cracking that
results from fretting around fastener holes or from corrosion pitting.
However, maintenance costs associated with repair of service contact
damage is expected to increase because graphite/epoxy is essentially an
unforgiving brittle material unable to yield plastically under overload.
Even quite modest impacts (by metallic standards) can lead to internal
damage in the form of delaminations which may result in a marked
strength reduction, particularly under compression loading. The impact
area may not be apparent from surface examination because of the
absence of permanent deformation.
The increased use of composite materials in Navy aircraft has focused
attention on issues associated with the maintenance and repair of
graphite/epoxy and other composite structures. These are both primary
and secondary structures involving several types of construction (Table
1), varied service conditions, and several types of manufacturing and
312 K. D. Challenger
TABLE 1
Typical Graphite/Epoxy Aircraft Structure; Ply Configuration Generally of the Quasi-
Isotropic _+45°/00/90° Varie W
(Note that ply thickness is usually about 0-13 mm)

Typical
Structures number of plies Applications

Honeycomb panels
Graphite/epoxy skins; 2 to 16 Control surfaces, fairings,
aluminum, fiber glass or access doors, flooring
nomex core
Sandwich panels
Graphite/epoxy skins; 2 to 6 As above
PVC foam core
Stiffened panels
Graphite/epoxy skins with 16 to 20 Fuselage shells
integral graphite/epoxy Tail skins, wing panels
stiffeners
Monolithic panels
Graphite/epoxy skins bolted 25 to 100 Main torque box, wing and tail
to aluminum alloy or
titanium substructure
Channels, beams
Graphite/epoxy 16 to 30 Spars (including sine wave
spars), ribs

in-service damage to be repaired (Tables 2a, 2b, and 3). Other, more
severe, handling and environmental damage will also occur; however,
this is also common to metallic structures, particularly those of honey-
comb construction.
R&D programs, under way for the past several years, have addressed
these issues in an attempt to develop repair methods and procedures,
and techniques for designing and analyzing bolted and bonded repairs.
The current US state-of-the-technology, as summarized in Table 4, is
such that some structures can be repaired and their strength restored
under specific circumstances which have been addressed in past R&D
programs. However, there are still many problems to solve and many
limitations to overcome. The increased sophistication of advanced
 Damage tolerance of carbon fiber reinforced composites 313

TABLE 2a
Typical Manufacturing Defects

Defect Typical causes

Voids Poor process control

Delarninations Inclusion of release film
Poor process control
Faulty hole formation procedures
Disbonds (in bonded joints) Poor fit of parts
Inclusion of release film
Poor process control
Surface damage Poor release procedure
Bad handling
Misdrilled holes Faulty jigging

TABLE 2b
Typical Service Mechanical Damage

Defect Typical causes

Cuts, scratches Mishandling

Abrasion Rain/grit erosion
Delaminations Impact damage
Disbonds Impact damage
Overload
Hole elongation Overload/bearing failure
Dents (with delaminations Impact damage
and crushed core) Walk in no-step regions
Runway stones
Edge damage Mishandling of doors and removable parts
Penetration Battle damage
Severe mishandling- e.g. fork lift
314 K. D. Challenger

TABLE 3
Typical Environmental Damage

Defect Typical causes

Surface oxidation Lightning strike

Overheat
Battle damage
Delamination Freeze/thaw stressing (due to moisture expansion)
Disbonds (in honeycomb panels) As for delaminations
Core corrosion Moisture penetration into honeycomb
Surface swelling Use of undesirable solvents,
e.g. paint stripper

TABLE 4
Current Repair Methods in US

Field repair Depot repair

Penetration damage to 4 inches Penetration damage to 8 inches wide and up to

Bonded/tapered external patches 0.5 inch thick
--skin thickness to 0.16 inch Scarf/step tapered flush repairs
Bolted external/internal metal Bolted/welded repairs to substructure
doublers
--skin thickness from 0-88 to
0.50 inch
Wet-layup fabric reinforcement
Resin injection
Oversized fasteners/sleeves

composite structures now being designed for new aircraft presents even
additional challenges and demands on composite repair technology.
D r R. Jones (Aeronautical Research Laboratory [ A R L ] , Melbourne,
Australia) presented the Australian view of repair technology, sum-
marized in Tables 5a and 5b.

7.2 Summary of discussion

Repairs described varied from the simple injection of resins into small
delaminations and disbonds, to a variety of patching procedures for
 Damagetoleranceof carbonfiber reinforcedcomposites 315

TABLE 5a
Non-Patch Repair Procedures for Minor Damage

Repairprocedure Applicable defect

Resin injection Connected voids

Potting or filling Minor depressions
Skin damage in honeycomb panels
Core replacement in honeycomb panels
Fastener hole elongation
Heat treatment Remove entrapped moisture in honeycomb panels
Dry out absorbed moisture
Surface coating Seal honeycomb panels
Restore surface protection

TABLE 5b
Patch Repairs for Major Damage: All are Capable of Restoring Ultimate
Strain Allowables to the Limits of Laminate Thickness Noted

Patchprocedure Application

Bonded external patch

Graphite/epoxy, Repairs to skins, particularly on honeycomb
(i) co-cured, (ii) pre-cured panels, up to 16 pries thick. Well suited for field
layers, (iii) pre-cured application
Titanium alloy foil
Bonded flush patch
Graphite/epoxy Repairs to skins 16 to 100 pries thick, holes up to
usually co-cured 100 ram. May be difficult to employ under field
Titanium alloy foil scarfed or conditions
stepped
Bolted external patch
Titanium alloy (usual) Repairs to monolithic skins 50 to 100 pries,
Aluminum alloy holes up to 100 mm
Suited for field applications

d a m a g e to areas up to 100 m m square. Emphasis was on repairs that can

be carried o u t u n d e r field or d e p o t conditions; however, s o m e of the
repairs described were m o r e appropriate to factory conditions. M a n y of
the repairs for composite-faced sandwich panels are similar to those for
316 K. D. Challenger

metallic honeycomb structures. This section highlights some of the

salient factors concerning damage, inspection, and repair criteria.
For field repairs to monolithic laminates, Dr Hess mentioned that the
current practice is to use a bolted external patch. There were several
comments as to the need to develop an alternative method and the need
to understand the mechanisms by which the repair restores the residual
strength. Dr Jones said that recent work performed in Australia has
shown that the increase in the residual strength is primarily due to the
reduction in the net section stress. Externally bonded scarf patches now
offer a viable alternative, even for monolithic structures. One major
disadvantage with mechanically fastened repairs is that they provide a
new path for moisture ingress. Indeed, tests are currently under way in
Australia to establish the feasibility of using a bonded external patch to
repair a 50 ply AS43501-6 laminate using the acrylic adhesive Flexon
241. Dr Kinlock mentioned that this adhesive has recently been used in
a similar fashion by British Airways.
In the case of impact damage to composites, it is first necessary to
assess a need to repair. It is fortunate that this assessment is simplified
by the fact that as the size of the damage increases the residual
compressive strength decreases to a constant value, Fig. 2.
Although, as indicated above, the majority of the discussion dealt
with the repair of graphite epoxy composites, Mr Mousley indicated that
repairs to P E E K may be much simpler in that any external patch could
be fironed' on to the structure. Dr Jones mentioned that composites
were also used to repair damaged metallic components and that work
has been done on this at the RAE (UK), NADC, AFML and ARL
(Australia). Indeed, in a series of tests recently undertaken at. ARL, this
technology, which had originally been developed for repairs to thin
sheet material, had been extended to enable repairs to surface cracks in
a 12.5 mm thick aluminum alloy.
In short, it was felt in this session that although current repair
techniques are adequate for the present, a greater understanding of the
structural aspects is required.

7.3 Critical issues

1. Need-to-repair criteria must be developed.

2. Reject-repair criteria should be established for the thermoplastics
and new tougher thermosets.
 Damage tolerance of carbonfiber reinforced composites 317

3. Analysis methods capable of assessing the behavior of repaired

regions need improvement.
4. Rapid low temperature cure materials to be used for repair are
necessary.
5. Future designs may involve higher design strain limits and different
materials. Methods to repair these structures must be established.

REFERENCE

Jones, R., Broughton, W., Mousley, R. F. and Potter, R. T., Compression

failures of damaged graphite epoxy laminates, Composite Structures, 3
(t985) 167-86.

APPENDIX: LIST OF PARTICIPANTS

Dr Kenneth D. Challenger, Dr E. Demuts,

Chairman, US Air Force Wright Aeronautical
Office of Naval Research, Laboratory/FIBA,
London, UK Wright-Patterson Air Force Base,
Ohio, USA
Dr I. H. Marshall, Vice Chairman,
Paisley College of Technology, Professor A. de Rouvray,
Paisley, UK Engineering Systems International
SA,
Dr Ing H. W. Bergmann, France
German Aerospace Research
Establishment, Dr G. Dorey,
West Germany Royal Aircraft Establishment,
Farnborough, UK
Dr W. S. Carswell, Mr T. Hess,
National Engineering Laboratory, Naval Air Development Center,
Glasgow, UK Warminster, Pennsylvania, USA
Dr G. Clark, Professor D. Hull,
Australian National Aerospace Department of Metallurgy and
Research Laboratory, Materials Science,
Melbourne, Australia University of Cambridge, UK
318 K. D. Challenger

Dr R. Jones, Dr F. Oertel,
Aeronautical Research US Army Research Office,
Laboratories, London, UK
Melbourne, Australia
Dr Y. Rajapakse,
Dr A. J. Kinloch, Office of Naval Research.
Imperial College of Science and Arlington, Virginia, USA
Technology,
London, UK Dr W. Reynolds,
AERE Harwell,
Mr R. F. Mousley, Oxfordshire, UK
Royal Aircraft Establishment,
Farnborough, UK Professor J. F. Williams,
University of Melbourne,
Ms G. B. Murri, Melbourne, Australia
USAUSCOM,
National Aeronautics and Space Dr I. Wolock,
Administration, Naval Research Laboratory
Langley Air Force Base, USA Washington, DC, USA