Composites Science and Technology 80 (2013) 93–100

Contents lists available at SciVerse ScienceDirect

Composites Science and Technology

journal homepage: www.elsevier.com/locate/compscitech

Interlaminar shear fatigue behavior of glass/epoxy and carbon/epoxy

composites
Andrew Makeev
Department of Mechanical and Aerospace Engineering, University of Texas at Arlington, Arlington, TX, USA

a r t i c l e i n f o a b s t r a c t

Article history: This work addresses a strong need in accurate fatigue properties of glass- and carbon-fiber reinforced
Received 29 December 2012 polymer–matrix composites. In particular, interlaminar shear fatigue properties of glass/epoxy and car-
Received in revised form 18 February 2013 bon/epoxy composite tape material systems used in aircraft fatigue-critical applications are needed. The
Accepted 13 March 2013
interlaminar shear fatigue material properties, essential for the development of the analysis methods able
Available online 21 March 2013
to capture fatigue delamination failure onset in composite structures, are presented in this work. S–N
curves are generated based on custom short-beam shear (SBS) fatigue tests. The custom SBS test config-
Keywords:
urations ensure a consistent interlaminar shear failure mode. This work extends the recently developed
A. Polymer–matrix composites
B. Fatigue
methodology, published in Composites Science and Technology to characterize nonlinear shear proper-
B. Delamination ties of composite materials using digital image correlation (DIC) and finite element analysis, to fatigue
D. Life prediction loading. Test data sets used to develop the fatigue properties, include approximately 20 glass/epoxy
and 30 carbon/epoxy SBS coupons. All tests were run in load control at 0.1 load ratio. To better under-
stand material behavior under cyclic loading, surface shear strain was monitored using the DIC technique.
Accurate shear stress approximation resulted in similar S–N curve shapes for the glass-fiber and the car-
bon-fiber composites.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction is required in the public domain. As composite materials could ex-

The ability to predict failure of composite structures under cyc-
lic loads is one of the major challenges in the aerospace engineer-
ing community for both fixed-wing and rotary-wing aircraft
applications. In particular, helicopter rotor systems are subject to
extreme fatigue environments. Fatigue capability is a limiting fac-
tor for high-cycle components found in the helicopter dynamic
systems [1].
Glass-fiber and carbon-fiber reinforced epoxy-matrix tape com-
posites are frequently used in the design of helicopter rotor sys-
tems. Typical examples of the fatigue-critical primary composite
structures include main rotor blade spars and yokes in commercial
and military helicopters [2]. Primary failure mechanism for such
structures is delamination. A major barrier to accurate fatigue life
prediction for composites has been the lack of material properties
that could be used as a basis for the development of the failure cri-
teria [3–5]. As fatigue tests are expensive and time-consuming,
companies which own such data often keep them proprietary,
and there is a significant lack of quality data on this subject avail-
able in the open literature. To accelerate the development of rigor-
ous fatigue life prediction methods for composite structures,
accurate characterization of the material fatigue failure behavior
E-mail address: makeev@uta.edu
hibit complex failure phenomena including multiple failure modes
and their interactions, appropriate consistent failure modes in the
fatigue test configurations are critical for measurement of specific
basic material properties. And test results must reflect true mate-
rial characteristics and not just coupon behavior.
Following numerous requests from the engineering community,
to expand the methodology for characterization of nonlinear shear
properties for composite materials using digital image correlation
and finite element analysis, recently published in Composites Sci-
ence and Technology [6], this work develops interlaminar shear fa-
tigue properties representative of glass/epoxy and carbon/epoxy
tape material systems. In particular, Cytec E-Glass/5216-Epoxy
and Hexcel IM7-Carbon/8552-Epoxy prepreg tape composites
[7,8] used in rotorcraft applications, are studied. Interlaminar shear
S–N curves [1] for such materials are generated based on custom
short-beam shear (SBS) fatigue tests. The custom test configura-
tions [6,9,10] ensure consistent interlaminar shear failure mode
in the unidirectional SBS coupons. To better understand material
behavior under cyclic loading, surface shear strain was monitored
using the Digital Image Correlation (DIC) technique. Periodic
images taken throughout the fatigue history also ensured accurate
number of cycles to failure in the SBS tests. The failure was defined
as the onset of visually detectable delamination.
Test data presented in this work are essential for the develop-
ment of the analysis methods able to capture fatigue delamination

0266-3538/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.compscitech.2013.03.013
94 A. Makeev / Composites Science and Technology 80 (2013) 93–100
failure phenomena in composite aircraft structures. In particular,
References [3,5] presented a comprehensive structural analysis
methodology to capture matrix-dominated failure modes and their
interaction in composites, and predict initiation and progression of
structural damage under cyclic loads, without a priori assumptions
of the initial damage or the damage path. The methodology is
based on 3D solid finite element models (FEMs) that simulate the
initiation and progression of structural damage to detectable size.
Stress-based and fracture-based criteria are used to predict initia-
tion of ply-cracks and delaminations as well as their progression
[3,5]. The failure criteria relied upon matrix-dominated shear and
tensile S–N curves to predict the onset of the fatigue damage
growth [3,5] in composite structures.
This work expands results of Ref. [6] which developed a SBS
based method for assessment of interlaminar shear material prop-
erties for composites under quasi-static loads, to fatigue loading.
The Digital Image Correlation (DIC) technique for full-field mea-
surement of surface deformation was successfully coupled with
three-dimensional finite element stress analysis to develop the
nonlinear interlaminar shear stress–strain curve for a 350°F cured
IM7-Carbon/8552-Epoxy tape composite.
DIC was also utilized in Refs. [9,10] to generate the nonlinear
interlaminar shear stress–strain relations based on SBS tests of
250°F cured glass/epoxy tape composites. Full-field deformation
measurements overcame the conventional strain gage limitations
in the SBS tests and proved the validity of simple stress models
to characterize the shear stress–strain curves. While a finite ele-
ment-based shear stress model was used to generate accurate non-
linear portion of the shear stress/strain curve for the carbon/epoxy
composite up to the material failure [6], a simple closed-form
stress approximation was acceptable for the glass/epoxy material
systems [9,10].
It is worth noting that before Ref. [9], use of the SBS test method
for the development of design allowables for structural design cri-
teria for composites was discouraged in the literature due to com-
plexity of stress distribution not being reflected in the simple beam
equation for peak shear stress. According to Ref. [11], instrumenta-
tion of SBS test coupons was not practical; therefore shear modulus
and stress–strain data could not be obtained using conventional
strain gages. Indeed, a conventional strain gage suitable for a SBS
test must be very narrow to minimize the effect of the strong gra-
dient of the strain in the coupon thickness direction. Also, the
strain gage must be perfectly placed along the neutral axis corre-
sponding to the maximum absolute value of the shear strain
throughout the coupon thickness. However, the location of the
neutral axis might be unknown prior to the test. One reason for a
shift of the neutral axis could be the difference in the axial tensile
and compressive moduli, which is the case for unidirectional car-
bon/epoxy tape systems [10]. Another deficiency of the strain
gages is their inability to capture the effect of the out-of-plane dis-
placement on the in-plane strain tensor components during finite
deformation. Three-dimensional deformation measurement would
be required to account for such effect.
SBS specimens are prismatic coupons with uniform rectangular
cross-sections and subject to three-point bending. The American
Society for Testing and Materials (ASTM) standard [11] guidelines
for geometry of polymer–matrix composite SBS specimens include
the width W twice the thickness T, and the support length L 4 times
the thickness. Consistent with DIC surface strain assessment, the
surface strain must represent the strain distribution throughout
the specimen width away from support locations. Based on
three-dimensional finite element analysis, the width of the SBS
coupons has been reduced from the ASTM recommended 200% to
a 100% of the specimen thickness (square cross-section) for more
uniform strain distribution through the specimen width [6,10]. It
is worth noting that the author initially used W = 0.8T [9] but found
that W  T was also acceptable. The square cross-section is conve-
nient as the SBS test coupons could be machined in the 0° and 90°
directions from a single unidirectional panel and loaded in the 1–2
(in-ply), 1–3 (interlaminar) and 2–3 principal material planes to
characterize 3D constitutive relations for composite materials
[10]. The fiber direction is denoted as 1 (0°); the in-ply transverse
direction as 2 (90°); and the laminate thickness direction as 3
(interlaminar material direction.)
Guidance to assessment of fatigue properties for composite
materials, available in public domain, is limited [1,12]. As listed
in Ref. [1]: fatigue testing is performed by cyclic loading of a test
specimen below static failure load to determine time or number
of cycles to failure. The loading is often characterized by the ratio
of the minimum to maximum load, for example, R = 0.1. A series
of tests are usually conducted at a loading frequency chosen to
be low enough to avoid heating of the specimen. This heating
can lead to thermally-induced failure. Load is cycled between se-
lected values until failure, and maximum load or some other indi-
cation of load intensity is plotted against the log of the number of
cycles to failure. Multiple tests are performed and the plot of the
results of these tests is referred to as an S–N curve.
It is also worth noting that Ref. [13] evaluated the ASTM stan-
dard [11] SBS test configuration to generate interlaminar shear
S–N curves for carbon/epoxy. Consistent with Ref. [10], Ref. [13]
also reported matrix cracks in unidirectional carbon/epoxy tape
coupons under the ASTM standard 0.25-in diameter loading nose
[11], causing the subsequent delamination shift from shear to
mixed-mode failure closer to the loading nose. Therefore, the
author believes the ASTM standard SBS test setup is not adequate
for the development of appropriate interlaminar shear fatigue
properties for carbon-fiber material systems. In this work, the
loading nose diameter in the unidirectional IM7-Carbon/8552-
Epoxy tape SBS tests was increased compared to the ASTM stan-
dard [11] to eliminate compressive damage under the loading nose
before shear delamination failure. Fig. 1 shows an example of the
modified ASTM standard SBS test fixture with a 2-in diameter load-
ing nose. However, no increase of the ASTM standard loading nose
diameter was required for more compliant unidirectional E-Glass/
5216-Epoxy SBS coupons to exhibit shear delamination.
2. Stress approximation
In the development of techniques to measure accurate stress–
strain curves in composite materials with strong anisotropy and
complexity of the stress–strain constitutive relations, the ability
to eliminate the need in ad hoc assumptions regarding formulation
of the nonlinear constitutive models is desired. If stress calculation
is independent of the deformation measurement, such ad hoc
assumptions are not needed.
The stress state in a unidirectional SBS specimen is complex.
Stress concentrations exist at the loading nose and support
Fig. 1. A custom SBS test with 2-inch diameter loading nose.
 A. Makeev / Composites Science and Technology 80 (2013) 93–100 95

locations [6,10]. However, the unidirectional SBS coupons do exhi-
bit regions of pure shear strain and shear stress states along the
coupon neutral axis, all the way until the shear failure. The follow-
ing simple closed-form approximation of the maximum shear
stress in the critical sections located mid-way between the loading
nose and the lower supports has been derived [10]
3P
r13 ¼ ; A ¼ TW
4A
where P is applied force. Such approximation is derived using the
DIC based observations of linear axial strain distribution through
the specimen thickness (far away from the loading nose and sup-
port locations) in the SBS tests; linear relationship between axial
normal stress and strain; and equilibrium assumption [10]. The
observations of the linear axial strain distribution through the SBS
specimen thickness; or the linear axial stress–strain behavior; does
not impose any restrictions on the nonlinearity of the shear stress–
strain response in the unidirectional material systems.
Can Eq. (1) be used in the SBS coupons, together with DIC-based
shear strain assessment, to generate accurate shear stress–strain
curves? Finite element analysis-based studies [6,9,10,15–17] found
that nonlinear shear stress–strain response reduces the maximum
shear stress compared to the simple beam theory approximation
(1). Three-dimensional solid finite element models (FEMS) showed
that such reduction was small for glass/epoxy tape composites
[9,10]. Therefore, the simple stress approximation (1) was accept-
able for the glass/epoxy material systems [9,10]. However, the
shear stress reduction was significant for a unidirectional carbon/
epoxy tape composite close to material failure [6]. An iterative
FEM-based stress approximation procedure developed in Ref. [6]
resulted in the accurate nonlinear portion of the shear stress/strain
curve for the carbon/epoxy composite material system. The itera-
tive procedure started with the DIC measurement-based shear
strain and the closed-form stress approximation (1), without ad
hoc assumptions about the form of the nonlinear stress–strain con-
stitutive model.
Fig. 2 compares the closed-form approximation (1) and the FEM
results for the maximum shear stresses in the unidirectional glass/
epoxy [10] and carbon/epoxy [6] SBS coupons, midway between
the loading nose and lower supports. All stress results are in close
agreement at small loads corresponding to the linear elastic behav-
ior. Such agreement is not intuitive as constitutive properties of
anisotropic materials contribute to elasticity equations even for
stress-bound problems [18]. The ratio of Young’s modulus in the fi-
ber direction and the transverse material directions is approxi-
mately four for glass/epoxy compared to approximately 20 for
the carbon/epoxy materials [6,9,10]. However, despite such a large
difference in the material anisotropy, both glass/epoxy and carbon/
epoxy SBS coupons exhibit similar shear stress behavior.
The difference between the closed-form approximation (1) and
ð1Þ the FEM results becomes more pronounced close to material fail-
ure as nonlinearity in the shear stress–strain response becomes
stronger [6]. The relative difference between finite element-based
stress and Eq. (1) remains within 5% till failure of glass/epoxy
SBS coupons, but increases up to 15% for carbon/epoxy. Again, both
glass/epoxy and carbon/epoxy SBS coupons exhibit similar shear
stress behavior. Fig. 2 shows that larger errors in the closed-form
approximation (1) of the maximum shear stress in the carbon/
epoxy SBS coupons compared to glass/epoxy coupons came from
the higher shear strength.
A bilinear model
(
3 P 3 P
;
4 A 4 A
6 6:8 ksi
r13 ¼ ð2Þ
0:8ð34 P
A
 6:8Þ þ 6:8; 34 AP > 6:8 ksi
represents a good engineering approximation which keeps the error
in the maximum interlaminar shear stress below 5% in both glass/
epoxy and carbon/epoxy SBS coupons. Eq. (2) will be used to in-
crease the accuracy of the shear stress approximation in the car-
bon/epoxy SBS coupons.
3. Interlaminar shear fatigue properties of E-Glass/5216-Epoxy
tape
This Section presents the interlaminar shear (1–3 principal
material plane) S–N curve for the E-Glass/5216-Epoxy tape system.
To generate the S–N curve, unidirectional SBS coupons were placed
in a uniaxial servohydraulic load frame with 5.5 kip (25 kN) load
cell capacity, and subject to cyclic loads, at 0.1 load ratio and
5 Hz frequency, to the delamination failure. The coupons are 1.5-
inch (38 mm) long, 0.23-inch (5.8 mm) thick, and 0.18-inch
(4.6 mm) wide. Support length is 1 inch (25.4 mm). The ASTM
Standard [11] loading nose diameter is 0.25 inches and the lower
support diameter is 0.125 inches. Approximately 20 SBS coupons
were tested. The tests were conducted at a room temperature of

100%
20000
90%
0.75P/A
y = 1.15x -0.3493
FEM Carbon/Epoxy [6] 80%
Peak Stress / Static Strength

R2 = 0.9059
FEM Glass/Epoxy [10]
15000 70%
BiLinear Approximation
Shear Stress (psi)

60%

50%
10000
40%

30%
5000
Material Strength 20%

10%

0 0%
0 5000 10000 15000 20000 1 2 3 4 5 6 7
0.75P/A (psi) Log (Cycles)

Fig. 2. Comparison of FEM results and closed-form stress approximations for Fig. 3. Interlaminar shear (1–3 principal material plane) S–N curve for E-Glass/
carbon/epoxy [6] and glass/epoxy [10] SBS coupons. 5216-Epoxy tape system.
96 A. Makeev / Composites Science and Technology 80 (2013) 93–100
P
Shear
Delamination
P/2 P/2
Fig. 4. E-Glass/5216-Epoxy SBS test specimen failure at 2.315 million cycles.
72 °F. An infrared thermometer was used to monitor coupon tem-
perature. No heating of the coupons was observed.
Fig. 3 shows the interlaminar shear S–N curve. To generate the
S–N curve, the ratio of the peak shear stress and the interlaminar
shear strength was plotted against the log of the number of cycles
to failure. The failure was defined as the onset of a visually-detect-
able delamination. Similar to Ref. [9], the interlaminar shear
strength was determined based on five static SBS tests. The average
interlaminar shear strength value was 10 ksi (69.1 MPa.) Fig. 3 lists
the power law coefficients determined based on a least-squares
approximation of the fatigue test data. And the R2 value greater
than 0.9 indicates an acceptable correlation of the S–N curve and
the test data.
Fig. 4 shows a typical failure in one of the coupons at 2.3 million
cycles. The loading conditions are also shown. The failure mode is a
shear delamination starting between the loading nose (upper
y,3
ε11
γ 13 2 mm
ε 13 =
2
ε 33
Fig. 5. Lagrange strain tensor components for a fatigue SBS test at peak cyclic stress equal to 60% of static strength and 2 million cycles.
support) and a lower support and quickly propagating to the edge.
All coupons exhibited similar behavior.
In this work, surface shear strain was monitored in a limited
number of SBS coupons subject to fatigue loading. A continuous
analog output from the load frame controller was used in the DIC
image acquisition software to capture stereo images at specified
loads during fatigue tests. Although the images could be captured
at 1 Hz or lower frequency, the time-delay for the individual
images was of the order of a millisecond. Therefore, the deforma-
tion corresponding to the specified loads could be captured in
the fatigue tests.
The DIC full-field surface strain assessment is based on the anal-
ysis of stereo images of the specimen surface with a random tex-
ture [14]. Fig. 4 shows an example of the random texture created
using black and white spray paints. A simple single-action air
brush has been successfully used to spray the black paint with a
good control of size of the black speckles by adjusting the nozzle
opening and air pressure.
DIC software VIC-3D [14] has been used in this work for assess-
ment of Lagrange strain tensor components on specimen surface.
The VIC-3D software determines three-dimensional positions be-
fore and after deformation by tracking the gray value pattern in
small subsets throughout the acquired stereo image sequence
[14]. Fig. 5 shows results of strain analysis which uses a subset size
of 27  27 pixels, corresponding to approximately 0.36 mm2 for
these particular tests. Data was obtained on six-pixel centers,
resulting in approximately 11,000 data points per load case. The
x,1
2 mm
 A. Makeev / Composites Science and Technology 80 (2013) 93–100 97

0.045
Peak Stress 6 ksi (60% Static Strength)

0.04
γ 13
ε13 =
0.035 2

0.03
2,314,600 Cycles
Peak Strain

0.025
γ 13
ε13 =
2
0.02

0.015 2,314,200 Cycles

γ 13
0.01 ε13 =
2
0.005
2,313,400 Cycles
0
0 500,000 1,000,000 1,500,000 2,000,000 2,500,000
Cycles

Fig. 6. Interlaminar shear engineering strain history for a fatigue SBS test at peak cyclic stress equal to 60% of the interlaminar shear strength.

0.03 100%
y = 1.284x -0.3414
Peak Stress 6.5 ksi (65% Static Strength) R2 = 0.9062
90%
0.025 Peak Stress 6.4 ksi (64% Static Strength)
80%
Peak Stress / Static Strength

0.02
Peak Strain

70%
0.015 60%
Batch 1 (T 0.15 in; Loading Nose D 0.5 in) Runout

0.01 50% Batch 2 (T 0.25 in; Loading Nose D 2 in)

40% Batch 3 (T 0.26 in; Loading Nose D 4 in)

0.005
0
0 100,000 200,000 300,000
Cycles
Fig. 7. Typical interlaminar shear engineering strain histories for fatigue SBS tests
at peak cyclic stresses equal to 64% and 65% of the interlaminar shear strength.
displacements measured in this manner were then numerically
differentiated using the strain computation algorithm [14] built
in the VIC-3D software to compute the Lagrange surface strain ten-
sor components. The Lagrange shear strain contour plot in Fig. 5
shows 2-mm long gage sections located along the coupon neutral
axis. Average shear strain in these sections was monitored during
fatigue testing.
Fig. 5 illustrates typical DIC analysis results for the three surface
strain components in a fatigue test of a E-Glass/5216-Epoxy unidi-
rectional SBS coupon, loaded in the 1–3 principal material plane, at
the peak cyclic stress equal to 60% of static strength and 2 million
cycles. The shear delamination failure that is shown in Fig. 4 oc-
curred in that test at 2.315 million cycles. The axial, transverse nor-
mal, and the interlaminar shear Lagrange strain tensor components
plotted in Fig. 5 are denoted as e11, e33, and e13 ¼ c213 (a tensor shear
strain component equals half of the engineering shear strain com-
ponent), respectively. The deformation measurements show close
to linear axial strain distribution through the SBS specimen thick-
ness midway between the loading nose and lower support loca-
tions, which supports the simple closed-form approximations for
the maximum shear stresses listed in the previous Section.
Figs. 6 and 7 show the typical interlaminar shear engineering
strain (c13) histories at the delamination onset location, mid-way
30%
400,000 500,000 20%
10%
0%
1 2 3 4 5 6 7
Log (Cycles)
Fig. 8. Interlaminar shear (1–3 principal material plane) S–N curve for IM7-Carbon/
8552-Epoxy tape system.
between the loading nose and a lower support, at the specified
peak interlaminar shear stresses, as functions of cycles during fati-
gue tests. The last point in the fatigue history corresponds to the
onset of a visually detectable delamination. A 5–10% increase in
the peak shear strain can be observed during the first 90% of the
fatigue life, and a fast and strong increase of the peak shear strain
occurs shortly before the delamination onset. A likely reason for
this response is the increase in the non-visible damage accumula-
tion in the resin prior to detectable cracking. It is worth noting that
the S–N curve shown in Fig. 3 has been generated in terms of stres-
ses due to the creeping peak strain values during the constant load
amplitude fatigue tests.
4. Interlaminar shear fatigue properties of IM7-Carbon/8552-
Epoxy tape
To generate the interlaminar shear S-N curve for IM7-Carbon/
8552-Epoxy tape in the 1–3 principal material plane, constant load
amplitude unidirectional SBS fatigue tests were conducted at 0.1
98 A. Makeev / Composites Science and Technology 80 (2013) 93–100
y,3
ε11 = − ky − b
γ 13
ε13 =
2 2 mm
ε 33
Fig. 9. Lagrange strain tensor components for a Batch 2 fatigue IM7/8552 SBS test at peak cyclic stress equal to 68% of static strength and 2 million cycles.
load ratio and 10 Hz frequency. Three batches of unidirectional SBS
coupons were tested. The dimensions of the coupons in the first set
(Batch 1) were 0.15 inch (3.8 mm) thickness, 0.12 inch (3.0 mm)
width, and 0.76 inch (19.3 mm) support length; and the corre-
sponding loading nose diameter was 0.5 inches (12.7 mm.) The
dimensions of the second set of SBS coupons (Batch 2) were
0.25 inch (6.4 mm) thickness and width, and 1.2 inch (30.5 mm)
support length; and the loading nose diameter was 2 inches
(50.8 mm.) The dimensions of the third set of SBS coupons (Batch
3) were 0.26 inch (6.6 mm) thickness, 0.25 inch (6.4 mm) width,
and 1.2 inch (30.5 mm) support length; and the loading nose diam-
eter was 4 inches (102 mm.) The lower support diameter was
0.125 inches (3.2 mm) for Batches 1 and 2, and 0.25 inches
(6.4 mm) for Batch 3 tests. It is worth noting that all coupons in
the same batch were machined from a single unidirectional panel
manufactured per specifications listed in Ref. [8], and the panels
were manufactured at least six months apart from each other.
Similar to the glass/epoxy SBS coupons, the carbon/epoxy cou-
pons were tested in the uniaxial servohydraulic load frame with
5.5 kip (25 kN) load cell capacity. The tests were conducted at
the room temperature of 72 °F; and the infrared thermometer
was used to monitor coupon temperature. No heating of coupons
was observed during fatigue tests.
Fig. 8 shows the interlaminar shear S–N curve. To generate the
S–N curve, the ratio of the peak shear stress and the interlaminar
x,1
2 mm
shear strength was plotted against the log of the number of cycles
to failure. The failure was defined as the onset of a visually-detect-
able delamination. The interlaminar shear strength value for each
batch of the SBS coupons was determined based on five static tests.
The average interlaminar shear strength value was 14 ksi
(96.3 MPa) for Batch 1; 14.3 ksi (98.9 MPa) for Batch 2; and
14.1 ksi (97.5 MPa) for Batch 3 of the SBS coupons. Fig. 8 lists the
power law coefficients determined based on a least-squares
approximation of the fatigue test data. And the R2 value greater
than 0.9 indicates an acceptable correlation of the SN curve and
the fatigue test data. The 10 million cycle runouts, shown in
Fig. 8, are not used in the SN curve approximation. The data gen-
erated from the different batches of the IM7/8552 SBS coupons of
different sizes follow the same SN curve.
The interlaminar shear SN curves developed for the glass/
epoxy and the carbon/epoxy composites exhibit similar shapes. In-
deed, the exponent in the power law SN curve approximation for
the IM7-Carbon/8552-Epoxy is only 2% lower compared to the
exponent for the E-Glass/5216-Epoxy system. The exponent for
the carbon/epoxy would be 12% higher compared to the glass/
epoxy if less accurate Eq. (1) were used for stress calculation in
the carbon/epoxy material system.
The similarity in the fatigue curve shapes might be useful to re-
duce the number of tests required to determine fatigue properties
of different composite material systems. It is worth noting that the
 A. Makeev / Composites Science and Technology 80 (2013) 93–100
0.03
0.025
0.02
Peak Strain
0.015
0.01
0.005
0
0 2,000,000 4,000,000 6,000,000 8,000,000
Cycles
Fig. 10. Interlaminar shear strain history for two fatigue IM7/8552 SBS tests at peak
cyclic stresses equal to 68% and 70% of the static interlaminar shear strength.
0.025
0.02
Peak Strain
0.015
0.01
0.005
0 properties essential for understanding fatigue delamination onset
0 20,000 40,000 60,000 80,000
Cycles
Fig. 11. Interlaminar shear strain history for a fatigue IM7/8552 SBS test at peak
cyclic stress equal to 73% of the static interlaminar shear strength.
static interlaminar shear strength of IM7-Carbon/8552-Epoxy is
more than 40% higher compared to the E-Glass/5216-Epoxy mate-
rial system.
Fig. 9 illustrates typical DIC analysis results for the three surface
strain components in a fatigue IM7/8552 SBS test at peak cyclic
interlaminar shear stress 68% of static strength and 2 million cy-
cles, about 6 million cycles before the onset of delamination.
Similar to the glass/epoxy SBS coupons, the deformation measure-
ments for the carbon/epoxy coupons confirm linear axial strain dis-
tribution through the SBS specimen thickness midway between the
loading nose and lower support locations. The Lagrange shear
strain contour plot in Fig. 9 shows 2-mm long gage sections located
along the coupon neutral axis. Similar to the glass/epoxy SBS cou-
pons, average shear strain in these sections was monitored during
fatigue testing.
Figs. 10 and 11 show the typical interlaminar shear engineering
strain (c13) histories at the delamination onset location, mid-way
between the loading nose and a lower support, at the specified
peak interlaminar shear stresses, as functions of cycles during fati-
gue tests. The last point in the fatigue history corresponds to the
onset of delamination failure. The short-cycle SBS test (72,000 cy-
cles to failure) had a slight (less than 5%) increase in the peak shear
strain during the first 90% of the fatigue life, and a fast and strong
increase of the peak strain close to failure. However, longer-cycle
SBS tests (Fig. 10) had more significant (up to 20%) increase in
the interlaminar shear strain during the first few hundred thou-
sand cycles of the fatigue loading. Although the peak strain in-
crease under the constant-stress-amplitude fatigue tests does
99
12000
Fatigue Initial
Fatigue 10000 Cycles
Fatigue 20000 Cycles
10000 Fatigue 30000 Cycles
Fatigue 40000 Cycles
Shear Stress (psi)
Fatigue 60000 Cycles
8000 Fatigue 80000 Cycles
Fatigue 110000 Cycles
Fatigue 140000 Cycles
6000 Fatigue 170000 Cycles
Fatigue 220000 Cycles
4000
2000
0
0 0.005 0.01 0.015 0.02
Shear Strain
Fig. 12. Shear stress–strain response at early stage of fatigue loading.
appear as creep-like response, the physical mechanism of such
behavior is yet to be investigated and will be the subject of a future
work. Fig. 12 shows that the increase in the peak strain at the early
stage of the fatigue history is not due to the shear modulus change
with cycles, but mostly due to the kinematic shift of the fatigue
stress–strain behavior. It took about 500 cycles (50 s) to complete
each 50-point stress–strain loop in Fig. 12 as stereo images for DIC-
based strain analysis were taken at one-second intervals (see
Fig. 11).
5. Conclusions
This work provided the engineering community with material
in glass/epoxy and carbon/epoxy composite systems. Interlaminar
shear SN curves were developed for E-Glass/5216-Epoxy and
IM7-Carbon/8552-Epoxy tape composites popular in aerospace
applications.
The custom SBS fatigue test design resulted in the consistent
interlaminar shear failure mode for the glass/epoxy and carbon/
epoxy material systems. Among the key elements of the fatigue
data generated in this work are a large number of test coupons
and low scatter. Large sample sizes are extremely useful for subse-
quent statistical simulations. Low scatter was achieved based on
the custom SBS test design to enable a consistent interlaminar
shear failure mode.
Despite significant difference in the static interlaminar shear
strength, the glass/epoxy and carbon/epoxy materials studied in
this work exhibit similar SN curve shapes. The similarity of the fa-
tigue curve shapes was established based on accurate stress calcu-
lation in the carbon/epoxy SBS coupons. A reduced stress model
was derived from three-dimensional FEM-based iterative stress
calculation. If proven general, the similarity of the fatigue curve
shapes can potentially reduce the number of tests required for
characterization of fatigue properties of composite material
systems.
The Custom SBS test design not only resulted in the similar
interlaminar shear failure mode but also showed that coupons of
various sizes followed the same S–N curve. Such observation is
essential to support a statement that S–N curves generated in this
work reflect true material fatigue properties and not coupon-spe-
cific behavior.
The ability to measure deformation in the SBS coupons was use-
ful to support the validity of shear stress approximations. Also, sig-
nificant creep in the interlaminar shear strain was observed in the
carbon/epoxy composite material at an early stage of long-cycle
constant stress amplitude fatigue fests. No change in the shear
modulus was observed. Constant modulus during most of the
100 A. Makeev / Composites Science and Technology 80 (2013) 93–100
fatigue life simplifies formulation of the models able to capture fa-
tigue delamination onset in composite structures.
Acknowledgements
This work is sponsored by the US Office of Naval Research
(ONR) and the US Army Vertical Lift Research Center of Excellence
(VLRCOE.) In particular, the development of test data and analysis
for the E-Glass/5216-Epoxy composite material was sponsored by
the VLRCOE; and the development of test data and analysis for
the IM7-Carbon/8552-Epoxy composite material was sponsored
by the ONR. Such support is gratefully acknowledged. The author
also thanks Mr. Christopher Cline and Mr. Russ Fay at Boeing for
manufacturing the glass/epoxy test specimens; and Mr. Edward
Lee at Bell Helicopter Textron for manufacturing the carbon/epoxy
SBS test specimens for this work. And special thanks to Dr. Yihong
He at the University of Texas Arlington for his assistance with finite
element models, and to Mr. Brian Shonkwiler at the University of
Texas Arlington, for his assistance with running the tests.
References
[1] US Department of Defense. Military handbook – MIL-HDBK-17-1F: Composite
materials handbook, Polymer Matrix composites guidelines for
characterization of structural materials; 2002: vol. 1.
[2] Dobyns A, Rousseau CQ, Minguet P. Helicopter applications and design. Del
Compos Des Encyclopedia 2000;5:223–42.
[3] Nikishkov Y, Makeev A, Cline C, Beasley J, Fay R. Finite element-based damage
tolerance methods for aircraft composites. In: Proceedings of the American
Helicopter Society 67th Forum, (Updated Cheeseman Award Paper of the 36th
European Rotorcraft Forum, Paris, France, 2010); 2011.
[4] Makeev A, Nikishkov Y. In: Komorowski J, editor. Fatigue Life Assessment for
Composite Structure, ICAF 2011 Structural Integrity: Influence of Efficiency
and Green Imperatives, Springer; 2011.
[5] Nikishkov Y, Makeev A, Seon G. Progressive fatigue damage simulation method
for composites. Int J Fatigue 2013;48:266–79.
[6] He Y, Makeev A, Shonkwiler B. Characterization of nonlinear shear properties
for composite materials using digital image correlation and finite element
analysis. Compos Sci Technol 2012;73(2012):64–71.
[7] Cytec engineered materials. CYCOM 5216 Modified Epoxy Resin, Technical
Data Sheet, <http://www.cemselectorguide.com/pdf/CYCOM_5216_040412.
pdf>; 2012.
[8] Hexcel. HexPly 8552 Epoxy Matrix, Product Data, available online at <http://
www.hexcel.com/Resources/DataSheets/Prepreg-Data-Sheets/8552_us.pdf>;
2012.
[9] Makeev A, Ignatius C, He Y, Shonkwiler B. A test method for assessment of
shear properties for thick composites. J Compos Mater 2009;43(25):3091–105.
[10] Makeev A, He Y, Carpentier P, Shonkwiler B. A method for measurement of
multiple constitutive properties for composite materials. Composites: Part A
2012;43(12):2199–210.
[11] American society for testing and materials. Standard test method for short-
beam strength of polymer matrix composite materials and their laminates,
ASTM Standard D 2344/D 2344M, ASTM International; 2006.
[12] Brack RA, Chris RM, Sims DF. Material and process development of a fiberglass/
carbon-reinforced toughened epoxy for tiltrotor components. In: Proceedings
of the 39th SAMPE conference; 1994: vol. 39. p. 236–250.
[13] May M, Hallett SR. An assessment of through-thickness shear tests for
initiation of fatigue failure. Composites: Part A 2010;41:1570–8.
[14] Sutton MA, Orteu J-J, Schreier HW. Image correlation for shape, motion and
deformation measurements. Springer; 2009.
[15] Cui WC, Wisnom MR. Contact finite element analysis of three- and four-point
short-beam bending of unidirectional composites. Compos Sci Technol
1992;45(4):323–34.
[16] Xie M, Adams DF. Contact finite element modeling of the short beam test for
composite materials. Comput Struct 1994;57(2):183–91.
[17] Xie M, Adams DF. Study of three- and four-point shear testing of unidirectional
composite materials. Compos 1995;26(9):653–9.
[18] Lekhnitski SG. Theory of elasticity of an anisotropic body. Moscow: Mir
Publishers; 1981.