Composites: Part A 34 (2003) 1125–1131

www.elsevier.com/locate/compositesa

A new mixed mode test for carbon/epoxy composite systems

Gregory D. Tracya, Paolo Ferabolib,*, Keith T. Kedwardb
a
Raytheon Infrared Operations, Goleta, CA 93107, USA
b
Department of Mechanical Engineering, University of California, Santa Barbara, CA 93106, USA
Accepted 26 April 2003

Abstract
The present research examines analytically and experimentally the mixed mode interlaminar fracture toughness of a resin film infused
(RFI) carbon fiber/epoxy laminate, namely a IM7-AS4/3501-6 hybrid composite system. The inability to develop representative interlaminar
failure in composites with current mixed mode test configurations motivated this particular investigation. The paper is part of a more
extensive research effort concerned with the effects of stitching upon the mixed mode fracture toughness of a RFI composite.
A new mixed mode test configuration is suggested, the Single Leg Four Point Bend (SLFPB), which provides a robust method with small
specimens and a simple apparatus. Closed form fracture mechanics-based strain energy release (SERR) calculations have been established
for this configuration. Finite element analysis was conducted to validate the closed form solution. Results show a very good agreement
between analytical solutions, numerical simulation and the newly designed SLFPB experimental test.
q 2003 Elsevier Ltd. All rights reserved.
Keywords: Mixed mode delamination; B. Fracture toughness

1. Introduction authors feel that delamination growth is the fundamental

issue in the evaluation of laminated composite systems
The primary concern with polymer composites is for durability and damage tolerance. In-service delamina-
delamination failure, which is debonding of adjacent tions often occur under complex load conditions and
plies, since the through-thickness properties of composite mixed mode failure phenomena are not yet well
laminates are typically matrix dominated and therefore, understood.
much weaker than the fiber dominated in-plane proper- The laminate of interest in this paper is a hybrid
ties. Inter-ply delaminations are usually initiated in one composite comprised of IM7 and AS4 fibers in a relatively
of three ways; (i) by process related defects, (ii) damage brittle 3501-6 epoxy matrix. The lay-up considered is
½^45=02 =90=02 = ^ 45ns with n ¼ 4; 6; which yields a
due to impact (handling or service related), (iii) direct or
[44/44/11] ply percentage in the 0/^ 45/908 orientation,
induced out-of-plane loads. Interlaminar shear or normal
respectively. The hybrid laminate calls for IM7 fibers in the
stresses are of particular importance in laminated
08 plies, and AS4 in the 45 and 908 plies.
composite structures, due to their highly anisotropic
While Mode I testing is usually performed with the aid of
nature. Interlaminar stresses originate because of a
the Double Cantilever Beam fixture (DCB), and Mode II
mismatch in the mechanical properties between individ-
testing with the End Notch Flexure setup, a review of
ual laminae within the laminate and develop at the free
existing mixed mode test methods was undertaken to
edge and at local discontinuities such as notches, ply-
identify the most suitable for this particular composite
drops, bonded and bolted joints, or when the laminate is
system [1]. Popular mixed mode tests are Crack Lap Shear
subject to hygro-thermal variations. These stresses need [2,3], Unsymmetric Double Cantilever Beam [2,4], Single
to be evaluated for structural applications and many Leg Bend (SLB) [7 – 9], and Four Point Bend (FPB)
[10 –13], but the most universally accepted is the Mixed
* Corresponding author. Tel.: þ 1-805-893-3381; fax: þ1-805-407-1123.
E-mail addresses: [email protected] (P. Feraboli);
Mode Bend [2,5,6].
[email protected] (G.D. Tracy); kedward@engineering. Each test has advantages and drawbacks. Table 1 lists the
ucsb.edu (K.T. Kedward). relative merits and disadvantages of each test while Fig. 1
1359-835X/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S1359-835X(03)00205-7
1126 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131

Table 1
Mixed mode test method comparison

Test method Advantages Disadvantages

Crack lap shear Simple fixture and coupon geometry; small Requires non-linear numerical analysis; due to large
crack opening displacement; constant mode ratio rotations at crack tip; different ply
lay-ups needed for different mode ratios
Unsymmetric DCB Simple coupon geometry; closed-form solution exists Requires complex fixture; requires bonded hinged tabs
Mixed mode bend Simple coupon geometry; variable mode mix ratio Requires complex fixture; requires bonded hinged
tabs; complex data reduction techniques
Single leg bend Simple fixture and coupon geometry; simple Mode ratio changes with crack length;
closed-form solution different coupon geometry needed for different
mode ratios
Four point bend Simple fixture and coupon geometry; simple Two cracks growing simultaneously at different
closed-form solution; constant mode ratio rates; different coupon geometry needed for
different mode ratios

shows a side-to-side comparison of specimen geometry and The analytical model will be first verified using a highly
loading configurations. Other less used test methods include meshed finite element model (FEM), then compared to
Edge Delamination Tension, Fixed Ratio Mixed Mode, experimentally obtained data.
Arcan and Variable Mixed Mode [2].
Due to disadvantages associated with the existing tests,
none of them was considered to be an optimal mixed mode 2. Analytical investigation
test solution for the evaluation of this material. As a result, a
new test fixture was developed to incorporate features from The fracture mechanics equations for GI and GII SERR
both the SLB and FPB tests. The new Single Leg Four Point expressions [14] for an isotropic material subjected to FPB
Bend (SLFPB) (Fig. 2) test eliminates both the double crack loading are
growth problem inherent to the FPB and the change in mode
3M 2 9M 2
ratio as a function of crack length typical of the SLB. It does GI ¼ ; GII ¼ ð1Þ
not require a complex test apparatus, nor does it need a non- Eh3 4Eh3
linear numerical analysis. The SLFPB test utilizes a small where h is the single leg thickness, M is the resulting
test coupon and a simple test fixture while providing a moment per unit width, and E is Young’s modulus. A
relatively simple method of obtaining controlled, mixed rescaling technique has been used to determine the SERR
mode crack propagation. for orthotropic laminates [14,15]. For the particular case
This paper will evaluate the SLFPB test as an alternative where the leg thickness is half the total thickness, h ¼ H=2;
to current mixed mode testing methods. A fracture the Mode I and Mode II SERR can be expressed as
mechanics-based analytic strain energy release rate
(SERR) solution that accounts for material anisotropy has GI ¼ b11 nl23=4 f2 ; GI ¼ b11 nl21=4 u2 ð2Þ
been previously established for the SLFPB configuration.
where f and u take the form
sffiffiffiffiffiffi sffiffiffiffiffiffiffi
3 9
f¼l M3=8
; u¼l M1=8
ð3Þ
nh3 4nh3

Fig. 2. Single Leg Four Point Bend coupon configuration.

Fig. 1. Common Mixed Mode test configurations. Fig. 3. ABAQUS Single Leg Four Point Bend finite element model.
 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131 1127

Table 2 different thicknesses were evaluated; 54-ply (H ¼ 8:64 mm

Material properties as measured and calculated (0.34 in.)) and 36-ply (H ¼ 5:59 mm (0.22 in.)). A total of
E1 ¼ 69:7 GPa E2 ¼ 36:5 GPa Eflex ¼ 66.2 GPa 28,231 nodes and 26,229 eight-node plane strain elements
(10.11 Msi) (5.3 Msi) (9. 60 Msi) were used. The FEM incorporates a focused mesh crack tip
G12 ¼ 19:8 GPa G13 ¼ 5:86 GPa G23 ¼ 4:482 GPa that allows the determination of the total SERR using the J-
(2.867 Msi), (0.85 Msi), (0.65 Msi) integral calculated over 10 contours around the crack tip.
n12 ¼ 0:388 n13 ¼ 0:3 n23 ¼ 0:3
SERR values were calculated for the SLFPB configuration
over a range of 0 – 11.12 kN (0 –2500 lb) applied load,
and the non-dimensional parameters n; r and l are measures which was the load range used in the experimental
of the material anisotropy investigation.
rffiffiffiffiffiffiffiffi The model is depicted in Fig. 3, where the mesh is shown
rþ1 2b þ b b
n¼ ; r ¼ p12ffiffiffiffiffiffiffiffi66 ; l ¼ 11 ð4Þ in the undeformed and deformed states (exaggerated for
2 2 b11 b22 b22
clarity). The geometry of the model is slightly different from
and can be evaluated for plane strain conditions using the test specimen, the difference being the overhang of
Hooke’s law material under one of the supporting rollers. The notch was
X si3 sj3 machined in the test coupon for sake of simplicity, avoiding
1ij ¼ bij sij ; where bij ¼ sij 2 ; the burden of machining the remaining dead material before
j¼1;2;6
s33
ð5Þ the crack tip. It can be seen that such an assumption has no
i; j ¼ 1; 2; 6 effect on the resulting moment at the crack tip. The
remaining dimensions used in the FEA are the same as the
The mode mix ratio can be determined using the following test coupon.
equation [14]: The flexural modulus of the laminate was previously
pffiffi determined through a series of experiments [16], while the
u 3 21=4
¼ l ð6Þ remaining material properties were calculated by means of
f 2
lamination theory (E1 ; E2 ; G12 ; G13 ; G23 ; n12 ; n13 ; n23 ; see
For the material in question, the mode mixity has been Table 2) and were then incorporated into the ABAQUS
calculated as 1.03, approximately 50% Mode I and 50% FEM input data.
Mode II. The analytic and FEM were evaluated in plane strain
An ABAQUS FEM of the SLFPB test configuration has conditions for two different specimen thicknesses: 36-ply
been generated to confirm the analytical solution. Two (H ¼ 5:59 mm (0.22 in.)) and 54-ply (H ¼ 8:64 mm

Fig. 4. Closed Form (CF) SERR vs. FEM SERR comparison.

1128 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131

(0.34 in.)). The FEM SERR values were compared to

analytic solutions and are graphically depicted in Fig. 4.
The comparison of the closed form solution and FEM
data yields that the analytic plane stress results are
uniformly 4% higher than FEM while the analytic plane
strain results are consistently 7% higher than FEM.
Given the consistency and the relative accuracy of the
Fig. 5. Picture of SLFPB test fixture. analytic results compared to the FEM, it can be concluded
that both the plane stress and plane strain analytic
solutions give a reasonably accurate description of the
mixed mode SERR for both the 54-ply and 36-ply
laminates studied.

3. Experimental investigation

Proof-of-concept testing of SLFPB mixed mode fracture

test has been completed. As mentioned previously, the
laminate tested has a ð^45=02 =90=02 = ^ 45Þns stacking
sequence, with n ¼ 4; 6 for a total thickness of 5.59 mm
(0.22 in.) and 8.64 mm (0.34 in.). The two thicknesses were
tested to determine if there was any dependence in SERR on
specimen thickness. The specimen geometry and loading
configuration are shown in Fig. 2.
The 54-ply coupons were cut from a 305 £ 305 mm2
Fig. 6. Diagram of SLFPB test fixture.
(12 £ 12 in.2) plate; while the 36-ply coupons came
from a separate 711 £ 711 mm2 (28 £ 28 in.2) panel.

Fig. 7. Applied load vs. crosshead displacement for 36-ply and 54-ply specimens.
 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131 1129

Fig. 8. 36-Ply strain energy release rate vs. crack length.

The delamination initiation region was obtained by The experimental test results for the 54-ply and 36-ply
embedding a strip of 25.4 mm (1 in.) wide and 5 mm specimens were consistent, and crack growth was con-
thick of Teflon at the mid-plane. trollable in both cases. Crack growth occurred in a ‘stick
A total of six specimens per thickness were cut to a slip’ manner, the crack advancing at finite increments,
nominal 25.4 mm (1 in.) width and 177.8 mm (7 in.) length which is the reason for the single, finite data points that
using a water-cooled circular saw with a diamond-coated characterize the SERR vs. crack length plot. The load –
tip. A 12.7 mm (0.5 in.) notch was machined into the displacement curves (Fig. 7) show that the maximum
surface to create the single leg. applied load for the 54-ply coupons averaged 2.67 kN
Due to the asymmetric configuration of the SLFPB (600 lb ft), while for the 36-ply coupons was consistently in
specimen, a test fixture was developed to ensure that equal the 1.34 kN (300 lb ft) range. Maximum load for the 54-ply
force was applied to the two contact points on the upper specimens was thus approximately double that of the 36-
surface. The upper loading truck is allowed to pivot about ply, but the crack length at maximum displacement was
the centerline, as shown in Figs. 5 and 6, and compensates significantly less. The reason for this is due to the greater
the loading asymmetry that appears as the crack advances, cross-sectional moment of inertia, hence increased bending
thus allowing to maintain a region of constant moment resistance of the thicker specimens.
between the inner rollers. The SERRs were calculated from the load data using the
In order to prevent the crack from propagating on the previously described analytic method. The SERR vs. crack
other side of the notch, a crack suppressing operation was length curves (Figs. 8 and 9) show that the SERR at which
performed by means of a simple clamp. steady state crack propagation occurred for the 54-ply
The tests were run on an Instron 1123 test machine material (, 1.7 kJ/m2 [10 in. lb ft/in.2]) was slightly higher
under displacement control at a crosshead feed rate of than for the 36-ply material (, 1.5 kJ/m2 [9 in lb ft/in.2]).
0.508 mm/min (0.02 in./min). Each coupon was Previous single mode fracture toughness testing of similar
painted white on the side and then scribed so that a (AS4/3501 and IM7/3501) material has produced Mode I
monoscope could be used to visually observe crack SERR values between 0.1 and 0.3 kJ/m2 and Mode II SERR
growth. values between 0.6 and 1.5 kJ/m2 [17,18].
1130 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131

Fig. 9. 54-Ply strain energy release rate vs. crack length.

4. Conclusions the primary structural material for the next generation of

commercial transport aircraft.
A new Mixed Mode Test has been developed, the
SLFPB, which has been shown to be a simple yet
effective method. Previously determined closed form
fracture mechanics-based SERR solutions for this con-
figuration have been verified through finite element References
modeling. A prototype SLFPB mixed mode test fixture
[1] O’Brien TK. Interlaminar fracture toughness: the long and winding
was manufactured for the testing of a hybrid IM7-
road to standardization. Composites, Part B (Engng) 1998;29B(1):
AS4/3501-6 composite system. Very good agreement 57– 62.
was obtained for all three approaches. It was furthermore [2] Reeder JR, Crews JH. Mixed mode bending method for delamination
shown that the SERR at which steady state crack testing. AIAA J 1990;28(7):1270– 6.
propagation occurred for the 54-ply specimens [3] Russell AJ, Street KN. Moisture and temperature effect on the mixed
mode fracture of unidirectional graphite/epoxy. ASTM STP 876;
(1.750 kJ/m2), was slightly higher than the 36-ply
1985. 349–70.
(1.576 kJ/m2), therefore suggesting a slight influence of [4] Sundararaman V, Davidson BD. Unsymmetric double cantilever
the specimen thickness on the SERR. beam test for interfacial fracture toughness determination. Int J Solids
Struct 1997;34(7):799– 817.
[5] Reeder JR, Crews JH. Redesign of mixed mode bending delamination
test to reduce nonlinear effects. J Compos Tech Res 1992;14(1):
Acknowledgements 12– 19.
[6] Kinloch AJ, Yayla P, Wang Y, Williams JG. Mixed mode
This research was funded in part by NASA through the delamination of fibre composite materials. Compos Sci Tech 1993;
47(3):225–37.
Boeing Company in support of the Advanced Composite [7] Yoon SH, Hong CS. Modified end notched flexure specimen for
Technology program. The ACT program evaluated the use mixed mode interlaminar fracture in laminated composites. Int J Fract
of advanced stitched/RFI composite laminates as 1990;43:R3–R9.
 G.D. Tracy et al. / Composites: Part A 34 (2003) 1125–1131 1131

[8] Davidson BD, Sundararaman V. A single leg bending test for [13] Feraboli P, Kedward KT. Four-point bend interlaminar shear testing
interfacial fracture toughness determination. Int J Fract 1996;78: of uni- and multi-directional carbon/epoxy composite systems.
193–210. Composites, Part A 2003; in press.
[9] Sundararaman V, Davidson BD. New test methods for determining [14] Suo Z, Bao G, Fan B, Wang TC. Orthotropy rescaling and
fracture toughness as a function of mode mix for bimateral interfaces. implications for fracture in composites. Int J Solids Struct 1991;
ASME 1995;EEP-11/MD-64:141–54. 28(2):235– 48.
[10] Charalambides PG, Lund J, Evans AG, McMeeking RM. A test [15] Bao G, Suo Z, Ho S, Fan B. The role of material orthotropy in
specimen for determining the fracture resistance of bimaterial fracture specimens for composites. Int J Solids Struct 1992;29(9):
interfaces. J Appl Mech 1989;56:77–82. 1105– 16.
[11] Charalambides PG, Cao H, Lund J, Evans AG. Development of a test [16] Tracy GD.PhD Dissertation. University of California, Santa Barbara;
method for measuring the mixed mode fracture resistance of 2001
bimaterial interfaces. Mech Mater 1990;8:269–83. [17] O’Brien TK. Composite interlaminar shear fracture toughness, GIIc:
[12] Charalambides PG. Steady state mechanics of delamination cracking shear measurement or sheer myth? ASTM STP 1330; 1998. p. 3– 18.
in laminated ceramic–matrix composites. J Am Ceram Soc 1991; [18] Daniel IM, Ishai O. Engineering mechanics of composite materials.
74(12):3066– 80. New York: Oxford University Press; 1993.