Composite Structures 50 (2000) 17±27
www.elsevier.com/locate/compstruct
Experimental and numerical analysis of transverse stitched T-joints in
bending
P.B. Stickler a,b,*, M. Ramulu a, P.S. Johnson b
a
Department of Mechanical Engineering, University of Washington, Seattle, WA 98195-2600, USA
b
The Boeing Company, Seattle, WA 98124-2207, USA
Abstract
An experimental and numerical analysis has been undertaken to investigate the bending strength of transversely stitched T-joints
using a ®ber insertion process. Finite element analysis (FEA) was performed on the structural joint to predict failure loads and
experiments were conducted to determine the modes of failure and ultimate failure strength. Joint parameters including local web
thickness, ¯ange thickness, number of rows of stitching, and resin types were varied to determine the e€ect on joint performance.
Fractographic examination was performed to investigate the joint failure mechanisms. Signi®cant results of this experimental study
include: adding a local web pad-up to the T-joint bend specimen increases the bending strength of the joint; increasing the number of
rows of stitching increases bending strength; adding a local ¯ange pad-up to the T-joint bend specimen only marginally increases
joint failure strength; initial joint failure occurs by matrix cracking initiating at the resin rich corner ®llet and propagating inward
toward the transverse stitching; and ultimate joint failure occurs by ®ber breakage and/or ®ber pullout. Ó 2000 Published by
Elsevier Science Ltd.
Keywords: Composite T-joint; Transverse stitching; Fiber insertion; Bending strength; Resin transfer molding
1. Introduction
Composite materials, with their high speci®c strength
and sti€ness, are ®nding wide use in the aerospace industry. Applications of composite materials include
wing skins, spars, ribs, fuselage bulkheads, longerons,
and ¯oor beams. Current emphasis of composite design
is on increased performance with reduced material and
manufacturing cost. The ecient transfer of load
through composite assemblies requires the use of
bonded or fastened joints. One method of joining
composites is through the use of a structural T-joint.
Aerospace applications of T-joints include wing spar-toskin and sti€ener-to-skin interfaces. T-joints are also
used in fuselage bulkhead-to-skin and longeron-to-skin
interfaces. Current T-joint designs are typically fabricated by one of the two methods: either through the use
of opposing C-channels with a radius ®ller and cap plies
or through the use of three-dimensional woven ®ber
preforms and resin transfer molding (RTM) [1±3]. These
methods are structurally ecient but tend to have high
manufacturing and/or material costs. T-joints using
*
Corresponding author.
transverse stitching and two-dimensional ®ber preforms
have the potential of signi®cantly reducing the cost of
composite structure and improving the damage tolerance and ultimate failure strength.
Transverse stitching has been used as a method for
improving the damage tolerance and ultimate failure
strength of composite laminates and joints. Mignery
et al. [4] investigated the use of stitching to suppress
delamination in laminated composites. Their investigation showed that stitching e€ectively arrested delamination in composite laminates as the crack approached
the stitch line. Stitching was shown to have varying
e€ect on ultimate strength depending on the lay-up
orientation. Transverse stitching has also shown to improve the damage tolerance of single lap joints. Tong et
al. [5±8] analyzed and tested adhesively bonded composite lap joints with transverse stitching. They showed
that the ultimate tensile strength of stitched single lap
joints is 20% greater than unstitched specimens and
axial displacement of stitched specimens is 25% greater
than unstitched specimens. The observed failure modes
of the stitched specimens were ®ber breakage and ®ber
pullout. Drans®eld et al. [9,10] investigated the e€ect of
transverse stitching on delamination toughness of
polymer matrix composite materials. During this study,
0263-8223/00/$ - see front matter Ó 2000 Published by Elsevier Science Ltd.
PII: S 0 2 6 3 - 8 2 2 3 ( 0 0 ) 0 0 0 0 6 - 4
18
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
it was shown that the addition of through-the-thickness
reinforcement signi®cantly improved the mode I delamination toughness. The e€ect of stitching on both
modes I and II delamination toughness was also investigated [11±16]. These studies showed improvements in
both modes I and II delamination toughness through
the use of transverse stitching.
Stitching has also been shown to improve the damage
tolerance of composite T-joints. Tada and Ishikawa [17]
evaluated T-section sti€eners in compression. They
found that stitching resisted damage extension had the
ability to arrest cracks, and postponed ®nal fracture.
Young and Chuang [18] evaluated stitched and unstitched RTM composite T-joints in tension. These tests
were conducted on specimens with rows of Kevlar
stitches located equally spaced away from the center of
the rib-to-web interface. Results showed an improved
joint strength in tension with the addition of transverse
stitching. Rispler et al. [19] have investigated the failure
analysis of composite T-joints with and without inserts
(radius ®llers) in tension. A ®nite element analysis
(FEA) was performed and a truncated quadratic delamination failure criterion was used. Failure was shown
to initiate at and be dominated by the resin properties in
the radius ®ller ``resin rich area'' of the web-to-¯ange
interface. Theotokoglou and Moan [20±22] recently reported the strength of composite sandwich T-joints under pull-o€ load. Although signi®cant research has been
conducted on through-the-thickness reinforcement of
laminates and joints, research to date has not addressed
the ¯exural behavior of T-joints using transverse
stitching at the ¯ange to web interface.
The purpose of this paper is to report an experimental
and analytic characterization of T-joints in bending with
®ber insertion directly into the web to ¯ange interface.
This technique has the potential of creating a low cost
T-joint with improved damage tolerance and failure
strength. In order to evaluate the performance and
suitability of this approach, a study was undertaken to
investigate the bending ultimate failure strength and to
determine T-joint failure mechanisms. FEA and testing
were performed and fractographic examination was
used to determine failure modes.
2. Experiments and procedures
2.1. Material system
The T-joint specimens used for this evaluation were
fabricated using T-300-3k-4h satin weave two-dimensional preforms manufactured by Albany International
Techniweave. Amaco provided the T-300 ®bers. The
IM7 6k tow ®ber insertions used for stitching were
fabricated by Hexcel. Two resin systems were evaluated for these tests: Shell Epon 862/W and 3M PR520.
Table 1
T-joint constituent mechanical properties
Material
E (GPa)
Tensile strength
(MPa)
Elongation
(%)
IM7 6-k tow
Shell Epon
862/W resin
3M PR520 resin
275
2.72
5378
78.6
1.4
8.2
91
5.0
3.54
Table 1 shows a summary of the T-joint constituent
mechanical properties.
2.2. Specimen fabrication
T-joint specimens were fabricated using two-dimensional preforms consisting of multiple layers of woven
fabric. Prior to assembly, the fabric was treated with
tacki®er (a 2% blend of Shell 1001 and 828 resin). Then
fabric plies were cut for both the web and ¯ange elements. Next web plies were stacked using a quasi-isotropic lay-up [0/ + 45/ ) 45/90]2s and tacked by heating
and cooling (the ¯ange element was made the same way
as the preform web element). Then the web element was
placed into the molding tool and compacted to ®nal
dimensions as the tool was bolted together. Next the
¯ange element of the preform was placed in the tool and
compacted with metal inserts. The insert directly above
the joint was then removed while the inserts on the rest
of the ¯ange were maintained in their ®nal position. This
provided access to the joint so that the carbon tows
could be inserted through the ¯ange and into the web
element. Fiber insertion was then performed using IM7
6k tow. After the insertion process was completed, the
mold was sealed (bolted) and the RTM process initiated.
During RTM processing, the Epon 862/W resin was
degassed at 40°C (105°F) and pre-heated to 50°C
(122°F). The mold was then pre-heated to 100°C
(212°F). Next the resin was transferred at 138 kPa
(20 psi) while the mold was under vacuum. Then the
specimens were initially cured for 4 h at 120°C (248°F),
demolded and post-cured for 2.5 h at 175°C (347°F).
The specimens using PR520 resin had a similar cure
cycle. The cured specimen ®ber volume fraction was
found to be between 50% and 53%. Fig. 1 shows the Tjoint con®guration used in this investigation. The specimen geometry including web thickness, ¯ange thickness, and transverse-stitching location is shown
schematically in Fig. 1(a) and actual joint cross-section
in Fig. 1(b). The transverse stitches for all specimens
entered the web to a depth of 12.7 mm (0.5 in.).
2.3. Test matrix
An experimental test matrix was designed to determine the e€ect of varying key T-joint geometric
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
19
Web Thickness
Web
Local Flange
Pad-up
Local Web Pad-up
1.27 cm
Transverse Stitching
Flange
Flange Thickness
Fig. 1. Bend test specimen: (a) con®guration showing local web and ¯ange pad-up and (b) typical structural T-joint showing transverse stitching.
Table 2
T-joint bending test matrix
Specimen
con®guration
Con®guration description
Number of specimens
1A
2A
3A
4A
Baseline
Local web pad-up
Local ¯ange pad-up
PR520 matrix
4
4
4
4
1B
2B
3B
4B
Baseline
Local web pad-up
Local ¯ange pad-up
PR520 matrix
4
4
4
4
parameters on the load carrying capacity of the composite joint is shown in Table 2. The geometric parameters that were varied include local web thickness,
number of rows of stitching, ¯ange thickness, specimen
width, and resin type. Four specimens of each condition
were tested to failure and average values were recorded.
Specimens 1A and 1B were the baseline (no local padups) from which geometric parameters varied. Specimen
1A contained two rows of transverse stitching and
specimen 1B contained three rows of stitching. Specimen
2A and 2B included the use of a local web pad-up and
contained six rows of stitching each. Specimen 3A and
3B contained a ¯ange pad-up. Specimen 3A contained
two rows of stitching and specimen 3B contained three
rows of stitching. Specimen 4A and 4B used a PR520
resin system. Specimen 4A contained two rows of
stitching and specimens 4B contained three rows of
stitching. PR 520 resin was used in only specimens 4A
and 4B. The type ``A'' specimens had a thinner web than
the type ``B'' specimens.
Web thickness
(mm)
Flange thickness
(mm)
Specimen width
(mm)
4.993
9.652
4.629
4.578
3.810
3.810
6.350
3.810
50.756
50.692
50.838
50.635
7.709
12.681
7.652
7.620
3.810
3.810
6.350
3.810
50.749
50.698
50.806
50.762
designed and built. The upper ¯ange of each T-joint
specimen was bolted to a steel support ®xture at two
locations along the width as shown in Fig. 2. Prior to
attachment to the support ®xture, an aluminum backing
plate was bonded to each specimen using room temperature epoxy adhesive. During testing, load was applied normal to the web 25.4 mm (1.0 in.) from the
¯ange to web interface. All specimens were loaded at a
rate of 1.27 mm/min (0.05 in./min). Testing continued
through the initial damage to joint catastrophic failure.
2.4. Experimental procedure
The bend tests were conducted using a 445 kN (100
KIP) SATEC static load frame Model 100 UD with a
built-in cross-head displacement. The cross-head displacement uses an optical encoder to measure the displacement. A built-in load cell with strain gages was
used to measure the test load. A bending test ®xture was
Fig. 2. Bend test ®xture.
20
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
Fig. 4. Typical load±displacement curves for bending specimens 1A.
Table 4
Average ¯exural proportional limit and bending sti€ness for T-joint
specimens
Specimen ID
Proportional limit
(N)
Flexural sti€ness Eb ,
(GPa)
1A-1
2A-1
3A-1
1B-1
2B-1
3B-1
659
2083
1042
1227
3406
2196
15.5
5.9
15.9
7.2
4.4
7.2
the ®les to a spreadsheet program. Macro and microscopic examination of the fractured surfaces were conducted to delineate the failure modes and damage
progression using a Nikon DC digital camera with a
50´ Nikon macro lens.
2.5. Analysis procedure
Fig. 3. Finite element mesh for stitched joint bending specimen 1A.
Catastrophic failure was de®ned as a permanent drop in
the load carrying capability of the joint. During testing,
load and displacement were recorded digitally using
NuVision materials testing software for Windows 95/98.
Data post-processing was accomplished by importing
A linear FEA was performed using ANSYS [23]
software on the geometry of T-joint 1A and 1B. For this
analysis, a two-dimensional plane strain condition was
assumed. Plane strain was assumed because the specimen
width is signi®cantly greater than the ¯ange or web
thickness. Fig. 3 shows the ®nite element mesh for
specimens 1A. The resin rich ®llet region 2 was modeled
using 6-noded triangular elements. The quasi-isotropic
two-dimensional woven ®ber preform regions 3 and 4
were modeled using 8-noded quadrilateral elements. The
two ®ber insertions in region 1 were modeled using one-
Table 3
Material properties for FEA
Material
Region
Ex (GPa)
Ey (GPa)
Ez (GPa)
Gxy (GPa)
Gxz (GPa)
Gyz (GPa)
mxy
mxz
myz
IM7 6-k tow
862/W resin
T-300/862W
1
2
3 and 4
275
2.7
45.9
275
2.7
45.9
275
2.7
5.5
106
0.98
15.4
106
0.98
2.0
106
0.98
2.0
0.3
0.38
0.22
0.3
0.38
0.22
0.3
0.38
0.32
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
21
Fig. 5. Bend test results summary.
Fig. 6. FEA for stitched joint ®llet bending stress (MPa) with an applied load of 659 N for specimens 1A: (a) normal stress rx , (b) normal stress ry ,
(c) shear stress sxy , and (d) e€ective stress reff .
22
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
dimensional bar elements (Fig. 3(a) and (b)). A ®ner
mesh was used at the web-to-¯ange interface and in the
®llet areas (Fig. 3(c)) and a coarser mesh was used at the
distal ends of the web and ¯ange. The ®nite element
model (FEM) used in this analysis contains 3935 8-noded
quadrilateral elements, 54 6-node triangular elements,
and 116 bar elements. Table 3 shows the mechanical
properties used for the FEM for the various regions of
the mesh. Orthotropic material properties were used for
the composite web and ¯ange and isotropic material
properties were used for the resin rich ®llet and web-to¯ange interface. The tow insertions were modeled using
®ber properties. A nodal force of 659 N or the average
proportional limit for specimens 1A was applied normal
to the web 2.54 cm (1.0 in.) from the web-to-¯ange interface. Fixed boundary conditions (zero displacement)
were assumed at the ¯ange interface to represent the
bolted ¯ange-to-®xture attachment constraint.
A similar modeling approach was used for the FEA
of specimen 1B. In this case the FEM used 2188 8-noded
quadrilateral elements, 38 6-node triangular elements,
and 105 bar elements. The three ®ber insertions were
modeled using one-dimensional bar elements. A nodal
force of 1227 N or the average proportional limit for
specimens 1B was applied normal to the web 2.54 cm
(1.0 in.) from the web-to-¯ange interface.
3. Results and discussion
Fig. 4 shows the typical load±displacement curve for
the bend specimens. The initial load±displacement curve
is nearly linear until the onset of matrix cracking and
®ber insertion debonding. At the point of initial joint
failure, a leveling of the load±displacement curve is
observed. After the initial failure, the specimens continue to carry load and the behavior is nearly linear with
a reduced slope. Just prior to the ®nal failure, load±
displacement curves were associated with a slight
amount of non-linearity. This observation was believed
Fig. 7. FEA for stitched joint ®llet bending stress (MPa) with an applied load of 1227 N for specimens 1B: (a) normal stress rx , (b) normal stress ry ,
(c) shear stress sxy , and (d) e€ective stress reff .
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
to be due to geometric non-linearities and/or local
yielding of the matrix prior to cracking. Upon further
loading, failure progressed from initial matrix cracking
to ®nal failure as indicated by the ®nal drop in the load±
displacement curve. Similar and consistent observations
were made in all other joint types except specimens 4A
and 4B. It was discovered that manufacturing defects
were present in these specimens and further work with
PR520 is in progress and will be reported in the future.
The linear portion of the load±displacement curve for
each specimen was used to determine the ¯exural sti€ness for the T-joint specimens. The ¯exural sti€ness, Eb ,
of the specimens was approximated using equation
Eb ˆ
3 P
L
;
y
3I
where P is the applied load, y is the web displacement at
the applied load, L is the distance from the applied load
to the web-to-¯ange interface and I is the web moment
of inertia. Table 4 summarizes the calculated average
values of the proportional limit and ¯exural sti€ness for
the T-joint specimens. As expected, increasing the Tjoint web thickness increases the proportional limit and
reduces the ¯exural sti€ness. Also, increasing the ¯ange
thickness had the e€ect of increasing the proportional
limit and had little e€ect on the ¯exural sti€ness.
23
The ¯exural test results in terms of bending moment
per specimen width are presented and are shown in Fig. 5
for the six T-joint con®gurations. Both initial damage
moment and maximum moment are shown. These values
are used as the basis of design loads for calculating
maximum joint bending moments. As shown, the baseline specimens (1A and 1B) had an initial damage moment of 392 and 573 mm N/mm and a maximum
moment of 901 and 1616 mm N/mm, respectively. The
addition of the local web pad-up (2A and 2B) had the
e€ect of increasing the initial damage moment by 277%
and 424% and the maximum moment by 210% and
309%, respectively. Improvement in the bending moment
of specimens 2A and 2B over that of the baseline may be
attributed in part to the additional rows of stitching and
in part to the larger moment of inertia of the local web
cross-section. The addition of a local ¯ange pad-up (3A
and 3B) had the e€ect of increasing the initial damage
moment by 162% and 289% and the maximum moment
by 108% and 221%, respectively. In this case the results
are less dramatic. Improvements over the baseline may
be attributed to the thicker ¯ange providing additional
®ber matrix shear area at the ®ber insertions.
Detailed results of the FEA at the ¯ange to web interface for specimen con®guration 1A is shown in Fig. 6.
Local stresses are shown as contour lines with labels
indicating the magnitude of the stress in units of MPa.
Fig. 8. Displacement predictions by FEA and experimental results for specimens 1A and 1B.
Fig. 9. Failed specimen showing ®ber pullout and ®ber breakage.
24
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
For this analysis a nodal force of 659 N or the average
proportional limit for specimens 1A was applied. The
discontinuities in the contour lines may be attributed to
the di€erences in material properties in regions 1±4. The
local tensile stress in the web direction, rx is shown in
Fig. 6(a). The predicated stress in the matrix is shown to
just exceed the matrix allowable tensile strength of 78.6
MPa (11.4 ksi). The local tensile stress in the ¯ange direction, ry is shown in Fig. 6(b). The local stress just
exceeds the tensile strength of the cured matrix. The
local shear stress, sxy is shown in Fig. 6(c). This value is
below the failure strength of the matrix in shear. The
equivalent or e€ective stress, reff is shown in Fig. 6(d).
The e€ective stress also just exceeds the matrix allowable
strength at the onset of failure. As indicated by the
leveling-o€ of the load±displacement curve shown in
Fig. 4, the limiting stresses occur in the resin rich area in
the ®llet and under the web-to-¯ange interface indicating
matrix cracking and ®ber debonding.
FEA results at the ¯ange to web interface for specimen con®guration 1B are shown in Fig. 7. For specimens 1B a nodal force of 1227 N or the average
proportional limit was applied normal to the web
2.54 cm from the web-to-¯ange interface. The local
tensile stress in the web direction, rx is shown in
Fig. 7(a). Again, the predicated stress in the matrix is
shown to just exceed the matrix allowable tensile
strength. Similar trends, as indicated in Fig. 6 above, are
Fig. 10. Fractograph of failed specimens 25´.
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
shown for tensile stress in the ¯ange direction Fig. 7(b),
the shear stress Fig. 7(c) and the e€ective stress Fig. 7(d).
Predicted and average experimental displacement
values for T-joint con®guration 1A and 1B are shown in
Fig. 8. The FEM results agree well with the average
experimental values up to the point of matrix cracking
and ®ber matrix debonding for specimen 1A. Just prior
to the onset of matrix cracking, the experimentally determined displacement is slightly non-linear indicating a
local yielding or cracking of the matrix. Again, this may
25
also be observed in the load±displacement curve shown
in Fig. 4. As shown in Fig. 8 the predicted values for
specimen 1B are less consistent than that of 1A. The
experimental data indicates a more pronounced nonlinearity in the average load±displacement curve. This
may in part account for the di€erence in predicted and
experimental results for specimen 1B.
Fig. 9 shows a typical failed T-joint bend specimen
including a close-up view of the failure region. The rows
of transverse stitches are clearly visible. Fiber pullout
Fig. 11. Schematic of T-joint damage progression.
26
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
and ®ber breakage can be observed at the web-to-¯ange
interface. Fig. 10 shows photomicrographs for typical
bend specimens at a magni®cation of 25´. The two micrographs of specimen 1A show cracks at the ``resin
rich'' corner ®llet of the T-joint, a crack along the webto-¯ange interface, and a crack running parallel to the
®ber insertions. Similar cracking is shown for specimens
2A, 2B, 3A and 3B. Other observed mechanisms of
failure for the T-joint specimens in bending include ®ber
breakage, ®ber-matrix debonding, and ®ber bridging.
Identi®ed mechanisms of failure included crack initiation at the corner of the ¯ange to web interface and at
the core to ¯ange interface. All 32 specimens including
specimens 4A and 4B failed in a similar manner.
Based on the extensive fractographic analysis of the
failed specimens the damage progression in transversely
stitched T-joints subjected to bending loads is shown
schematically in Fig. 11. As shown, damage progresses
in a series of steps coinciding with increased load. As
sucient load is applied to the specimen, the local stress
in the matrix exceeds the matrix tensile strength and a
crack initiate at the ®llet or ¯ange-to-web interface (a).
Increased loading beyond this point then causes the
crack to propagate to the neighboring ®ber insertion (b).
The crack is initially arrested at the ®ber insertions until
the load is increased further where ®ber matrix-debonding and ®ber bridging occur due to high shear
stresses at the ®ber to matrix interface (c). Next the
crack arrests due to ®ber bridging at the ®ber insertion.
This has the a€ect of reducing the crack stress intensity.
This ®ber bridging gives the T-joint an increased damage tolerance or an ability to carry additional load after
®rst failure occurs. Increased loading then causes ®ber
elongation, followed by ®ber breakage and ®ber pullout
(d). Further loading causes additional matrix cracking
and propagation to the next ®ber insertion (e). Loading
beyond this point leads to catastrophic failure of the
T-joint.
4. Conclusions
An experimental and numerical investigation of
transverse stitched T-joints in bending was conducted.
Based on the results of this study, the following conclusions may be inferred from the T-joint analysis and
test:
· Adding a local web pad-up to the T-joint bend specimen has the greatest e€ect on increasing the failure
strength of the joint.
· Increasing the overall web thickness and number of
rows of stitching increases the failure strength of all
T-joint con®gurations.
· Adding a local ¯ange pad-up to the T-joint bend
specimen only marginally increases the joint failure
strength.
· Initial T-joint failure modes include matrix cracking
initiated at the resin rich corner ®llet and ®ber/matrix
debonding at the transverse stitching. Final T-joint
failure modes include ®ber breakage and ®ber pullout.
Acknowledgements
The authors would like to acknowledge The Boeing
Company for providing computing support and test
facilities and Keith Burgess and Albany International
Techniweave, for stitched T-joint development and
fabrication of test specimens.
References
[1] Wehrenberg RH. II. Composites get stronger 3-D with reinforcement. Mater Eng 1983;97:27±31.
[2] Soden JA, Weissenbach G, Hill BJ. The design and fabrication of
3D multi-layer woven T-section reinforcements. Compos Part A:
Appl Sci Manufact 1999;30:213±20.
[3] Baker A. Joining and repair of aircraft composite structures. In:
Mallick PK, editor. Composite materials handbook, New York:
Marcel Decker, 1997.
[4] Mignery LA, Tan TM, Sun CT. The use of stitching to suppress
delamination in laminated composites. Delamination and Dedonding ASTM STP 1985;876:371±85.
[5] Tong L, Jain LK. Analysis of adhesive bonded composite lap
joints with transverse stitching. Appl Compos Mater
1995;2(6):343±65.
[6] Tong L, Steven GP. A numerical study on stresses in RTM lap
joints reinforced with transverse stitching. Compos Struct
1996;36(1/2):91±100.
[7] Tong L, Jain LK, Leong KH, Kelly D, Hertzberg I. Failure of
transversely stitched RTM lap joints. Compos Sci Technol
1998;58(2):221±7.
[8] Tong L, Cleaves K, Kruckenberg T, Steven GP. E€ect of
transverse stitching on the fracture loads of RTM single lap
joints. Key Eng Mater 1998;137:195±202.
[9] Drans®eld KA, Jain LK, Mai YW. On the e€ects of stitching in
CFRPs ± I. Mode I delimitation toughness. Compos Sci Technol
1998;58:815±27.
[10] Jain LK, Drans®eld KA, Mai YW. On the e€ects of stitching in
CFRPs ± II. Mode I delimitation toughness. Compos Sci Technol
1998;58:829±37.
[11] Barrett DJ. The mechanics of Z-®ber reinforcement. Compos
Struct 1996;36:23±32.
[12] Jain LK, Mai YW. Determination of mode II delimitation
toughness of stitched laminated composites. Compos Sci Technol
1995;55:241±53.
[13] Jain LK, Mai YW. Analysis of stitched laminated ENF specimens
for interlaminar mode II fracture toughness. Int J Fract
1994;68:219±2443.
[14] Jain LK. On the e€ect of stitching on mode I delamination
toughness of laminated composites. Compos Sci Technol
1994;51:331±45.
[15] Shu D, Mai YW. E€ect of stitching on interlaminar delamination
extension in composite laminates. Compos Sci Technol
1993;49:165±71.
[16] Drans®eld KA, Baillie C, Mai YW. Improving the delimitation
resistance of CFRP by stitching ± a review. Compos Sci Technol
1994;50:305±17.
P.B. Stickler et al. / Composite Structures 50 (2000) 17±27
[17] Tada Y, Ishikawa T. Experimental evaluation of the e€ects of
stitching on CFRP laminate specimens with various shapes and
loadings. Key Eng Mater 1989;37:305±16.
[18] Young WB, Chuang MT. Fabrication of T-shaped composite
through resin transfer molding. J Compos Mater 1996;
29(16):2192±214.
[19] Rispler AR, Stevens GP, Tong L. Failure analysis of composite Tjoints including inserts. J Reinforced Plastics Compos 1997;
16(18):1642±58.
27
[20] Theotokoglou EE, Moan T. Experimental and numerical study of
composite T-joint. J Compos Mater 1996;30(2):190±209.
[21] Theotokoglou EE. Study of numerical fracture mechanics analysis
of composite T-joints. J Reinforced Plastics Compos 1999;
18(3):215±23.
[22] Theotokoglou EE. Strength of composite T-joints under pull-out
loads. J Reinforced Plastics Compos 1997;16(6):503±18.
[23] Users Manual, version 5.4. Canonsburge, PA: ANSYS, Inc.,
1998.