International Journal of Mechanical Sciences 241 (2023) 107966

Contents lists available at ScienceDirect

International Journal of Mechanical Sciences

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

Investigation on the compressive mechanical properties of ultra-thick

CFRP laminates
Yu Gao a, b, 1, Jian Wang a, b, 1, Xiaowen Song a, b, Huiming Ding a, b, *, Han Wang a, b, c, YunBo Bi a, b,
Yinglin Ke a, b
a
State Key Laboratory of Fluid Power and Mechatronic System, School of Mechanical Engineering, Zhejiang University, Hangzhou 310027, China
b
Key Laboratory of Advanced Manufacturing Technology of Zhejiang Province, School of Mechanical Engineering, Zhejiang University, Hangzhou 310027, China
c
Hangzhou AME Aerospace Manufacturing Equipment, Co, Ltd., Hangzhou 311200, China

A R T I C L E I N F O A B S T R A C T

Keywords: The damage mechanism of ultra-thick laminates widely used in aerospace and marine industries was obviously
Ultra-thick laminate different from that of thin laminates. However, compression testing and simulation methods were generally
Carbon fibre limited to relatively thin laminates. To explore the compressive failure behavior of ultra-thick laminates, Firstly,
Damage mechanics
a layered pre-compaction scheme was proposed to ensure the curing quality of ultra-thick laminates. A heavy-
Finite element analysis (FEA)
duty compression test equipment (accuracy reached 0.001 mm) was then developed, and a novel compression
Fractography
test method for ultra-thick composite laminates was proposed. The results showed that delamination played a
key role in the failure behavior of ultra-thick laminates. The low interlaminar normal strength of ultra-thick
laminates (55% of the 2 mm thin laminate) led to premature delamination, and wrinkle defects seriously
reduced the compressive strength (11%). The compressive behavior for the matrix-dominant failure specimen
predicted by the Puck criterion was the smallest (0.07%), while Tsai-Wu 3D predicted the smallest error (5.1%)
for the delamination-dominant specimen. Finally, these findings are helpful for the design of ultra-thick com­
posite structures.

1. Introduction assessment of ultra-thick composite structures.

The compressive failure mechanism of thick laminates differs
In recent years, compared with traditional metal alloys, the appli­ depending on the type of layups. The research on compressive failure
cation of carbon fiber composites in marine structures and underwater modes of unidirectional thick CFRP composite laminates mainly focuses
vehicles has grown rapidly due to their higher specific stiffness, specific on kink band fracture [28–31], which is caused by local buckling of the
strength, and superior corrosion resistance. For deep-sea submersibles, fibers. The resin first undergoes shear yielding because of the initial fiber
the buckling stress of thin shell structures in deep sea environment is misalignment. Subsequently, the restraint of the fiber bending defor­
much lower than the static strength of the materials. In order to with­ mation due to the resin is weakened, the fibers undergo micro-buckling,
stand huge hydrostatic pressure, thick composite structures with high which eventually leads to the fracture of the laminate. Unlike thin
load-bearing become a cost-effective alternative [1–9]. However, as the laminates, the thickness of thick laminates is much greater than the kink
thickness increases, the reduced fiber volume fraction and fiber wavi­ band width. When the kink band grows, it is more likely to grow in the
ness have negative effect on the compressive properties of the com­ vertical direction with a constant angle away from its assumed straight
posites. Test results of thick composite cylinders under hydrostatic line due to weaker interfacial properties. Although the compressive
pressure showed that failure stress of thick composites is far below the fracture initiation and evolution of thick laminates are difficult to esti­
compressive strength of thin laminates [10–27]. Therefore, a better mate, the initial damage of tends to occur at the free edge of the spec­
understanding of the compressive mechanical behavior of ultra-thick imen, and delamination as well as fiber micro-buckling are the dominant
composites is of particular importance for the design and safety failures mechanism [32]. In contrast, "brooming" fractures are often

* Corresponding author at: State Key Laboratory of Fluid Power and Mechatronic System, School of Mechanical Engineering, Zhejiang University, Hangzhou
310027, China.
E-mail address: pangding@zju.edu.cn (H. Ding).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.ijmecsci.2022.107966
Received 4 October 2022; Received in revised form 17 November 2022; Accepted 20 November 2022
Available online 22 November 2022
0020-7403/© 2022 Elsevier Ltd. All rights reserved.
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

observed in thick multi-directional CFRP laminates [33–39]. Matsuda Table 1

et al. summarized the fracture process of quasi-isotropic carbon fiber Mechanical properties of GW800G/GS-13.
laminates through experiments and simulations. They found that the Elastic properties
initial failure occurred at the free edge of the innermost 0◦ ply due to the
E1[GPa] E2[GPa] G12[GPa] G23[GPa] ν12 ν23
concentration of the out-of-plane shear and compression stresses [34]. 147 9.32 4.53 3.9 Gpa 0.3 0.3
The initial failure behavior of the 12.35 mm thick laminates was Strength properties
observed by Goto et al. in two cases starting from the innermost 0◦ ply XT[MPa] YT[MPa] XC[MPa] YC[MPa] S12[MPa] Failure strain [%]
and the fixed section of the specimen [35]. Tsampas et al. comparatively 2820 60.2 1538 190 125Mpa 0.99
pc⊥‖ pt⊥‖ pt,c v⊥‖f mσf
analyzed the uniaxial compression mechanical behavior of thick
θ ⊥⊥
53◦ 0.25 0.30 0.22 0.2 1.1
quasi-isotropic and [0/90] cross-ply laminates. Delamination and
in-plane shear fracture were the main failure mechanisms. Compared
with the cross-ply laminates, the quasi-isotropic laminate showed
stronger response and higher failure load, but it was also more prone to Table 2
delamination and post-failure damage [36]. CV. Opelt et al. proposed a Test matrix.
new classification for the compressive mode [39], based on the phe­ Specimen code Layup Laying process
nomena observed during the test and the fracture study of the failed 30-A {[45/02/-45/90]11+[45/0/-45/902]6}s AFP laying
specimens (by scanning electron microscopy), the failure modes were 30-C {[45/902/-45/0]11+[45/90/-45/02]6}s AFP laying
divided into four main types: shear failure, interlaminar failure, inter­ 30-C-W {[45/902/-45/0]11+[45/90/-45/02]6}s AFP laying
(Fiber wrinkle)
facial failure and kink band failure. However, the failure mechanism of
ultra-thick laminate has not been uniformly defined and classified.
In fact, the above-mentioned compression failure modes of thick sample design and the 800t high-precision compression test equipment
laminates are all found under specific test methods. At present, the was introduced. Finally, the effectiveness of the test method was verified
widely used composite compression test standards are aimed at thin by experiments.
laminates [40–44]. In contrast, the test methods for ultra-thick lami­
nates have not been widely developed and applied. Different from thin
laminates, the compressive failure load of thick laminate is larger. 2.1. Specimen preparation
Buckling and end fracture of the specimen often occur, resulting in
premature failure of the specimen [40–42]. Therefore, prior to testing, All the specimen materials were GW800G/GS-13 provided by China
the design of the specimen and test fixture should ensure a uniform Weihai Guangwei Composite Materials Co., Ltd., with a single layer
one-dimensional compressive stress field, and careful consideration thickness of 0.18 mm. The properties of the materials were all measured
should be given to the thickness and bonding characteristics of the by the Testing and Evaluation Center for High Performance Carbon Fiber
end-tabs. In addition, the compression test is sensitive to many experi­ of Beihang University. The parameters are shown in Table 1. According
mental parameters, such as specimen geometry, loading method, stress to the bearing characteristics of the underwater vehicle that mainly
concentration, buckling stability, and specimen end crushing. In bears bidirectional loads in the deep sea, {[45/02/-45/90]11+ [45/0/-
response to these problems, Hsiao et al. [41], Componeschi [44], Oga­ 45/902]6} S axial layup and {[45/902/-45/0]11+ [45/90/-45/02]6} S
sawara and Ishikawa [45] and Lee and Soutis [46] developed circumferential layup were designed. The test matrix was shown in
compression test methods suitable for thick composites. However, the Table 2, in which the letter "A" and "C" represented axial and circum­
developed ICSTM test fixture or NAL-II test fixture need to accurately ferential specimens, respectively, and "W" represented specimens con­
control the tightening force of the fixture bolts to ensure that the taining manufacturing defects.
compressive stress is transmitted to the middle area of the sample As shown in Fig. 1(a), all the samples were laid by automatic laying
through a part of the shear stress, which is prone to artificial assembly equipment. In order to ensure the size and shape of the sample and
errors. On the other hand, the uniaxial compression of the ultra-thick eliminate pores, a layered pre-compaction method was used during the
laminate is a structural test rather than material test. The manufacturing process (see Fig. 2(b)). Before formal curing, every 34
length-to-thickness ratio of the samples in these test methods is too layers were laid as a prefabricated unit, and the prefabricated unit was
small, and the measured results are more of the ultimate strength of the placed in an autoclave at 80◦ C, 0.6MPa for 30min. For thicker laminates,
material, while the size effect of the actual structure needs to be the curing process must stay in an autoclave for a long time (80◦ C,
considered during the structure test. Thus, there is a need to develop a 0.6MPa for at least 1h) to distribute the heat evenly throughout the
reliable compression test method for ultra-thick laminates. panel and reduce the possibility of exothermic reactions because ther­
The purpose of the present work is to study the uniaxial compressive mal energy could cause uncontrollable temperature rise inside the
mechanical behavior and failure mechanism of CFRP laminates with a laminate (see Fig. 2(c)). After curing, the Olympus ultrasonic C-scan
thickness of 31.79 mm. Firstly, an 800t high-precision compression test equipment was used to detect the defects of the specimens, as was shown
equipment for composite materials was built, and a new uniaxial in Fig. 2(d), the interlayer compaction was very good without obvious
compression test methods for ultra-thick CFRP laminates was proposed. pore or bubble defects.
Two kinds of specimens with different layups were then prepared by the
AFP technology. The effects of layups and manufacturing defects on the 2.2. New uniaxial compression test method for ultra-thick laminate
compressive properties of ultra-thick laminates were analyzed, and the
evolution mechanism was concluded by observing the fracture surface of 2.2.1. Specimens design
the laminate. Finally, the mechanical behavior and typical failure po­ The size design of the specimen was one of the key factors to ensure
sition of the ultra-thick laminate were predicted using several classical the rationality and scientificity of the test. Firstly, in order to ensure that
failure criteria. the load of the testing machine could be effectively transmitted to the
test area, both ends of the specimen needed to be protected, and suffi­
2. Experimental procedures cient restraint should be provided to avoid end crush. On the other hand,
the gauge length of the specimen L was the most critical parameter, and
In this section, the preparation method and design method of com­ the gauge length L must ensure that the specimen undergoes obvious
posite ultra-thick plate samples were firstly presented, and then the compressive failure before buckling [40].
uniaxial compression test method of ultra-thick laminate including Firstly, the critical buckling gauge length Lbuckle needed to be

2
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 1. Manufacturing process of ultra-thick specimens: (a) Process of automated fiber placement (AFP); (b) Process pre-compaction; (c) Process of curing the
specimens; (d) Nondestructive testing including A-scan, B-scan and C-scan.

Fig. 2. (a) Dimensions of the compression specimens; (b) Micrographs of layups at the outer free edges of 30-A and 30-C specimens; (c) 800t high-precision
compression test equipment; (d) Composition of uniaxial Loading unit.

determined. According to the Euler column buckling principle, the shear

0.67kc π 2 EI
deflection caused by transverse shear force of anisotropic material was σ cr ≤ (1)
considered while the cross-section of the specimen was assumed to be AL
rectangular, the ultimate compressive strength σcr must satisfy [47]: where E was the longitudinal Young’s modulus, I was the second-order

3
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 3. Comparison of Longitudinal strains (Y) and transverse strains (X) of 30-C and 30-A specimens along the surface to the interior of A and B specimens.

bending moment area, A was the cross-sectional area of the specimen, the thickness of the specimen exceeded 30 mm, and the damage load
and kc was a constant reflecting the end support condition. When the reached a hundred tons, the existing test platform could not meet the test
compression specimen had a rectangular cross-section, I = bt3 /12 (b and requirements. Therefore, an 800t high-precision compression test
t were the width and thickness of the specimen, respectively), then the equipment was developed, as shown in Fig. 2(c). Different from the
critical length Lbuckle of the specimen could be expressed as: traditional hydraulic control equipment, servo motor was adopted to
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅ drive lead screw movement, and the large load was transmitted to the
0.67kc E end of the specimen through the thrust platform. The control accuracy
Lbuckle ≤ πt (2)
12σcr reached 0.001 mm. Further, the horizontal level was adjusted through
the Z-direction support platform, as shown in Fig. 2(d). Three-
For muti-directional laminate, kc was assumed to be 5/6, and the
directional strain gauges were placed on the surface and thickness of
values of E and σcr were shown in Table 1. On the other hand, when the
the specimen to monitor the failure process, as shown in Fig. 2(e).
gauge length was too short, the insufficient load transmission distance of
the testing machine could easily lead to interlaminar failure rather than
3. Results and discussion
compression failure. Thus, the minimum gauge length Lmin could be
expressed as
In this section, the uniaxial compression mechanical behavior of the
( )
t σ cr ultra-thick laminate was first discussed, and then the influence of the
Lmin ≤ ln t0 (3)
t0 τc manufacturing process and the layups on the failure mechanism of the
ultra-thick laminates was discussed. The dominant compressive failure
where τcwas the in-plane shear strength. t/t0 was the stress concentra­ behavior was summarized through the electron microscope analysis of
tion factor, which was taken as 1.5 here. To satisfy Eqs. (2) and (3) at the the fracture surface. Finally, the compressive mechanical behavior of the
same time, the gauge length L must be within the following interval, and ultra-thick laminate was predicted by FEM method.
L was selected 180 mm here.
Lmin ≤ L ≤ Lbuckle (4) 3.1. Validation of test method

2.2.2. Test fixture and mechanical tests In order to verify the validity of the compression test method, a very
In this study, both ends of the ultra-thick specimens were restrained important criterion was whether the specimen was bent or Euler buck­
by metal frames. The resin and quartz sand mixture were filled between ling. As shown in Fig. 3, the longitudinal strains (Y) and transverse
the specimen to increase the contact area of the end and prevent the end strains (X) from the surface to the interior of samples 30-C and 30-A
from being damaged prematurely. The specific dimensions of the spec­ were extracted by FEM method. The longitudinal strains from the sur­
imen were shown in Fig. 2(a). It could also be seen from Fig. 2(b) that the face to the center of the specimens showed good consistency, while the
ultra-thick laminate had better interlayer compaction after curing. Since transverse strains (X) of the two specimens changed at the thickness of
22 mm, which was mainly due to the selection of two different ply

4
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

of this delamination led to the premature loss of the load-bearing ca­

pacity. Goto et al. [35] used the NAL-II test fixture to perform uniaxial
compression tests on a 12 mm thick T800 quasi-isotropic laminate, and
the compressive strength and failure strain were 605 MPa and 1.3%,
respectively. Although the compressive testing method was different, it
could be inferred that the compressive strength decreased gradually
with increasing thickness.

3.3. Detailed description of failure mechanisms

3.3.1. Effect of manufacturing defects

Fiber wrinkle was a common geometric defect in AFP process, which
was characterized by fiber deviating from the fiber direction designed in
path planning [52,53]. Fig. 6 compared the cross-sections of the two
specimens with and without fiber wrinkle. During the AFP process, the
placement force, temperature, and speed sometimes did not match the
temperature-viscosity profile of the prepreg well, which led to gaps and
overlaps in the tows, resulting in extra-length out-of-plane fiber wrinkle,
Fig. 4. Representative stress-strain curves of strain gauges SG-1, SG-2 and SG-3 see Fig. 6(b). Although fiber wrinkle was also a common defect during
for 30-C, and the red and grey lines represent 30-A and 30-C specimen, the curing process of thick laminates [53], the out-of-plane fiber bending
respectively. due to insufficient curing pressure was avoided because the laminates in
this study were subjected to a hot pre-compression process before formal
curing. The fiber wrinkle was usually characterized by the deflection
angle of the aligned fibers in the reference plane, and for any given
n-layer fiber wrinkle angle θ could be expressed as [54]:
ntply 180
θ = (5)
Lg π

where tplywas the thickness of the single-layer prepreg, Lgwas the lateral
distance between the two symmetrical layers, Lgwas measured to be
5.14 mm here, thus, θ = 4.17◦ .
The stress-displacement curves of the two specimens with and
without wrinkle defects were compared in Fig. 7. The displacement
shown in the figure was the relative displacement of the two ends in the
loading direction, and the point of occurrence for critical damage was
marked on the loading curve. Both specimens exhibited linear me­
chanical behavior in the initial loading stage. As the loading continued,
Fig. 5. Comparison of modulus, compressive strength and failure strain of the the specimen with wrinkle defects first suffered shear cohesive failure
30-C and 30-A specimens. between wavy layers. Before reaching the peak load point, delamination
between wavy layers began to flex and expand toward the free edge,
sequences in this paper. Each ply sequence contains two different ply leading to fiber micro-buckling failure of the wavy layers [55,56]. This
blocks, and the strain changes at the ply changes due to the Poisson’s micro-buckling failure then expanded from the inside to the outside at
ratio effect [37]. In fact, for the structural designer, the strain test results higher compressive stress, eventually leading to a significant drop in
of the two specimens along the longitudinal strains (Y direction) are load. Therefore, for the specimens with wrinkle defects, the failure was
more concerned. Therefore, three strain gauges (SG-1, SG-2 and SG-3) mainly dominated by the micro-buckling of the fibers. Compared with
were also arranged on the surface and thickness of the specimen to the full brittle fracture of the wrinkled specimens, for the 30-C specimen,
monitor whether the specimen was bent. As was shown in Fig. 4, the an off-axis split was first observed in the 45◦ layer near the outermost
monitoring data of the three strain gages were in good agreement for fixed end, and with a slight load drop. Subsequently, microbuckling of
both the 30-C and 30-W specimens, indicating that the specimen did not the 0◦ ply occurred in the intermediate region of the specimen, which
bend. induced matrix shear failure of the 90◦ layer. Finally, the crack propa­
gated from the inside to the outside leading to complete fracture of the
specimen, which was also found in the Ref. [34]. The entire failure
3.2. Mechanical response process of 30-C specimen was shown in Supplementary Material 1. For
the 30-C specimen, although the crack propagation was caused by the
Fig. 5 compared the average values of compressive strength, internal delamination, the failure was mainly dominated by the internal
modulus and failure strain for various compression specimens. During matrix shear failure. On the other hand, due to the premature fiber
the AFP process of ultra-thick laminates, there were laying defects such micro-buckling, the failure stress and strain of the 30-C-W specimen
as tow gaps and overlaps, resulting in the fiber wrinkle after curing were far away lower than 30-C specimen.
[48–54]. (For a discussion of the specific reasons, see chapter 3.2.1.) The
wrinkle seriously reduced the compressive strength by about 11%, while 3.3.2. Effect of layups
this defect had little effect on the modulus. Compared to the 30-C Compared with the 30-C specimen, the initial damage of the 30-A
specimen, the compressive strength decreased by about 4% despite the specimen also occurred on the outside, but it was delamination be­
higher number of 0◦ plies for the 30-A specimen. This phenomenon tween the 45◦ and 0◦ layers on the upper and lower surfaces. This
occurred mainly due to the premature delamination of the 30-A spec­ resulted in two slight drops in the compressive load, respectively. Then,
imen during the compression process, and the gradual inward expansion under the combination of high bending stress and compressive stress

5
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 6. Morphology of the ultra-thick specimen without (a) and with (b) fiber wrinkle in the thickness section.

Fig. 7. Stress-displacement curves of 30-C and 30-C-W specimens, and the typical failure behavior of the two specimens was shown at the point where the
load changes.

generated in the outermost layer, delamination occurred in multiple 3.3.3. Characteristics of fracture surface
places along the thickness direction, and the laminate was divided into To further investigate the formation mechanism of initial damage
multiple sub-layers. Eventually, the crack propagated to the middle re­ and damage evolution, the fracture surface of the sample was analyzed
gion, resulting in matrix shear failure in the 90◦ layer and fiber breakage by S-3700N scanning electron microscope, as shown in Fig. 9(a). "①"
in the 0◦ layer, and the structure lost its bearing capacity (see Fig. 8). The represented the location of initial damage, and "②" was the location of
entire failure process of 30-A specimen was shown in Supplementary crack propagation. For the wrinkled specimen 30-C-W, obvious delam­
Material 2. Therefore, the failure mechanism of the 30-A specimen was ination was first observed inside, and the dominant failure causing the
mainly driven by delamination. load drop was microscopic kink-band of the 0◦ layer. The formation
mechanism of this microscopic kink-band was explained in detail in

6
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 8. Stress-displacement curves of 30-C and 30-A specimens, and the typical failure behavior of the two specimens was shown at the point where the load changes.

Fig. 9. (a) Comparison of SEM images of fracture surface at typical locations for the three kinds of specimens, where "①" represented the location of initial damage,
and "②" was the location of crack propagation; (b) formation mechanism of typical failure modes.

7
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 10. FE model of thick laminate, showing the cross section of the half of the specimen along the thickness.

Fig. 9(b), the wavy layer underwent obvious bending deformation due to 3D criteria (Tsai Wu 3D and Hashin 3D) was lower than the corre­
the action of shear stress, and the high bending stress caused by this sponding 2D criterion, mainly because the 3D criterion considered the
deformation led to fiber fracture. stress state in 3 directions, which led to relatively conservative predicted
Compared with the thin laminate, the out-of-plane shear stress of the results. On the other hand, based on the Mohr-Coulomb model, Puck and
ultra-thick laminate along the thickness direction could not be ignored, Schuermann distinguished different fracture modes according to the
and a high stress concentration was generated at the free edge during the stress combinations appearing on the detected fracture surface, which
loading process [35]. Although the initial damage of 30-C and 30-A further predicted the matrix failure between fibers. According to the
specimens occurred at the outermost side, the subsequent crack propa­ previous summary, the failure of the 30-C specimen was mainly domi­
gation directions were different. For the 30-C specimen, serious matrix nated by the matrix fracture, so the strength value predicted by the Puck
damage occurred in the inner 90◦ layer, and then the cracks gradually criterion was the closest, with an error of only 0.07%. In contrast, the
spread from inside to outside. In contrast, the cracks in specimen 30-A strength predicted by the LaRC05 was lower, mainly because LaRC05
gradually spread from the outside to the inside after delamination took into account the local fiber kinking and splitting caused by the
occurred on the outside (see Fig. 9 (b)). matrix failure on the basis of the matrix failure predicted by Puck,
making the prediction result more conservative. For the prediction of
3.4. Prediction of compressive behavior failure strain, except for Hashin3D and LaRC05 criteria, the rest of the
prediction errors were not more than 6%.
For ultra-thickness composite components, there is currently no The eight failure criteria only consider in-plane damage and did not
applicable failure model that can accurately predict their mechanical involve delamination failure criteria. For the 30-A specimen, the failure
properties. Therefore, in order to provide a reference for the design of was mainly dominated by delamination, so it can be seen from Table 3
ultra-thick composite structures, several classical failure criteria were that the prediction and test error of compressive strength was large. The
used here to predict the in-plane performance of the laminates: compressive strength predicted by Tsai-Wu 2D/3D was the closest, with
maximum stress/strain [55,56], Tsai-Wu2D/3D [57], Tsai-Hill [58], errors of 10.1% and 5.1%, respectively. In addition, for the 30-A spec­
Hashin2D/ 3D [59–63,27], Puck [64], and LaRC05 [65], while the imen, the delamination from the outside to the inside was mainly caused
stress-based Ye secondary stress separation criterion [66] was used to by the normal stress in the 3 directions, which were considered in the
analyze the interlaminar damage. These failure criteria were imple­ failure judgment of the 3D criterion compared with the 2D criterion
mented in ABAQUS by UVARM subroutine because UVARM subroutine (S13, S23, S33) on the structure, so the prediction results of the 3D
was more suitable for structural engineering analysis or strength pre­ criterion were closer. For the prediction of failure strain, except that the
diction in the early stage of structural design, and had a wide range of error of Puck criterion reached 14%, the error between the other pre­
application [55]. The GETVRM of UVARM subroutine was used to dicted results and test was not more than 10%.
obtain the stress and strain information at the material integration point In general, each failure criterion focused on the evaluation of
in the current analysis step, and then the state variable UVARMn (n = 1, different failure modes, so the predicted results were different. For the
2,3…) of different failure criteria was defined with this information, 30-C specimen, the dominant failure mode was matrix fracture, the
where n represented the number of failure modes in the failure criteria. strength and failure strain predicted by the Puck criterion were therefore
Once the state variable UVARMn was equal to 1, the structure was the closest. The mechanical behavior predicted by the Tsai-Wu 3D and
considered to be damaged. As the layup was symmetric, only half of the Hashin 3D criteria considering the stress state in 3 directions was closer
sample model was established along the Z direction (see Fig. 10). The to that of the 30-A specimen with delamination as the dominant failure
laminate used 8-noded reduced integration solid elements (C3D8R), and mode.
the whole model contained 135,120 elements. One end was fixed, the
other end adopts displacement loading. 3.4.2. Evaluation of failure modes
The traditional Tsai-Wu and Tsai-Hill criteria combined the different
3.4.1. Evaluation of mechanical response fundamental failure mechanisms through only one equation that in­
Firstly, the stress-strain data at four points on the surface of the cludes all stress components. Different from them, the failures of lami­
sample were compared with the FEM results. As shown in Fig. 11, the nates were classified as fiber-dominated and matrix-dominated modes
simulation results were in good agreement with the test results in the by Hashin, Puck, and LaRC05 criteria. It could be seen from Fig. 12 that
linear segment. The specific values were compared with the test in three failure criteria accurately predicted the initial damage of fiber
Table 3. The moduli predicted by the simulation of the two specimens fracture in the outermost 45◦ layer of the 30-C specimen, and the fiber
had a small error compared with the test, and the error did not exceed splitting failure of the 45◦ layer was analyzed by the LaRC05 criterion.
4%. For the 30-C specimen, the strength predicted by the two selected On the other hand, the simulation accurately captured the matrix

8
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 11. Comparison of the test results and FEM results of the Representative stress-strain curves of four strain gauges (SG-4, SG-5, SG-6 and SG-7) at the outer
surface for 30-C and 30-A specimens.

damage of the inner 90◦ layer and the subsequent 0◦ -layer fiber predict the interlaminar failure. However, during the design of large-
breakage, which was consistent with the test. Similarly, for 30-A spec­ thickness composite structures, the number of cohesive elements inser­
imen, the outside matrix damage and the internal fiber breakage that ted layer by layer led to a large number of mesh elements in the thick­
ultimately resulted in the loss of bearing capacity were all observed by ness direction, which seriously reduced the computational efficiency.
three failure criteria (see Fig. 13). Therefore, based on the Ye secondary stress separation criterion, addi­
In addition to predicting the damage behavior of the ultra-thick tional failure criteria were added to the UVARM subroutine to initially
laminate, the interlaminar failure was also a key factor affecting the judge the delamination damage position. As shown in Fig. 14(a), the
mechanical properties. Insertion of cohesive elements was often used to initial delamination damage positions of 30-C and 30-A specimens were

9
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Table 3 inside and outside the laminate respectively, which was mainly because
Comparison of compressive strength, modulus and failure strain predicted by high normal stress S33 generated in the inner and outer 0◦ layers during
different failure criteria with experimental results. deformation (see Fig. 14(b)). Through the out-of-plane tensile test of the
Specimen Failure Compressive strength Failure Modulus ultra-thick laminate (see Appendix A), it was found that the interlaminar
Criteria (MPa) strain (GPa) normal strength of the ultra-thick plate was only 55% of that of the thin
(%) laminate. The main reason for this phenomenon was that the thickness
30-C Test 525.5 1.30 65.9 of the laminate was too large, and the internal heat dissipation effect was
Max stress/ 519.4(1.2%) 1.25 63.9(3.0%) poor during the curing process, resulting in the low interlaminar
strain (4.00%)
strength [67]. Further, as shown in Fig. 14(c), high stress concentration
Tsai-Wu 2D 515.5(1.9%) 1.24
(4.67%) generated in the middle layer and outer side of the inner surface of 30-C
Tsai-Wu 3D 509.7(3.0%) 1.23 and 30-A specimen, respectively. This also verified the "inside-to-out­
(5.38%) side" direction of crack propagation for the 30-C specimen and "out­
Tsai-Hill 512.3(2.5%) 1.23 side-to-inside" for the 30-A specimen mentioned earlier.
(5.38%)
Hashin 2D 512.3(2.5%) 1.24(4.6%)
Hashin 3D 440.7(16.3%) 1.18(9.2%) 4. Conclusions
Puck 529.3(0.07%) 1.30(0%)
LaRC05 490.2(6.7%) 1.2(7.7%)
The uniaxial compression mechanical behavior and failure mecha­
30-A Test 504.5 1.00 75.4
Max stress/ 603.5(19.6%) 1.10 72.7(3.6%)
nism of 31.79 mm ultra-thick composite laminates were studied by
strain (10.00%) means of experiments and numerical simulations. The effects of layups
Tsai Wu 2D 555.7(10.1%) 0.98 and manufacturing defects on the compressive properties of ultra-thick
(2.00%) laminate were compared and analyzed, and the failure evolution
Tsai Wu 3D 530.4(5.1%) 0.96
mechanism was summarized by observing the fracture surface of ultra-
(4.00%)
Tsai Hill 591.9(17.3%) 1.08 thick specimens.
(9.00%) A heavy-duty compression test equipment was developed, and the
Hashin 2D 602.5(19.4%) 1.08 control accuracy reached 0.001 mm. Further, a uniaxial compression
(8.00%)
test method for ultra-thick composite laminates was also proposed.
Hashin 3D 582.3(15.4%) 1.03
(3.00%)
The fiber wrinkle defect seriously reduced the uniaxial compressive
Puck 621.3(23.2%) 1.14 strength by about 11%, but had little effect on the modulus. Delami­
(14.00%) nation played a key role in the compressive failure behavior of ultra-
LaRC05 598.5(18.6%) 1.09 thick laminate. Due to the premature delamination, the compression
(9.00%)
strength of axial specimens with larger modulus was lower than that of
Note: The value in () represents the deviations between FEM and Test. circumferential specimens with lower modulus, which decreased by
about 4%. The failure of the axial specimens expanded from outside to
inside, which was dominated by the lateral lamination, while the crack
propagation of the circumferential specimens was from inside to outside
due to the internal matrix fracture.

Fig. 12. Comparison of typical failure behavior locations for the 30-C specimen predicted by different failure criteria (Hashin, LaRC05 and Puck) and test results.

10
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

Fig. 13. Comparison of typical failure behavior locations for the 30-A specimen predicted by different failure criteria (Hashin, LaRC05 and Puck) and test results.

Fig. 14. Prediction of delamination location showing (a) distribution of normal stress S33 (b) and contours of maximum principal stress on the inner surface of 30-C
and 30-A configuration.

The error between the simulation and experimental results of The test method and results of ultra-thick laminate proposed in this
compression modulus was less than 4%. For the circumferential speci­ study can provide guidance for the testing of deep-sea composite ultra-
mens with matrix-dominant failure, the strength predicted by the Puck thick structures, and the results predicted by various failure criteria can
failure criterion was the closest, with an error of only 0.07%. For axial provide a theoretical basis for the design. Further research can be carried
specimens where delamination dominated failure, the Tsai-Wu 3D cri­ out on how to suppress delamination and improve the interlaminar
terion prediction the closest, with an error of 5.1%. On the other hand, normal strength of ultra-thick laminates.
the large out-of-plane normal stress S33 inside the ultra-thick laminate
was the main cause of delamination, and the interlaminar normal Supplementary material
strength of the ultra-thick laminate was only 55% of that of the 2 mm
thin laminate. Supplementary Material 1. Video of failure process of 30-C specimen

11
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

propagation, white paint was applied to the surface of the specimen [69,
25]. Finally, the assembled specimen was connected to an Instron5985
universal testing machine with a loading speed of 0.1 mm/min (see
Fig. A1). It was observed that the smooth horizontal delamination
appeared in the middle plane of the specimen, and there was no phe­
nomenon of migration (see Fig. A1(b)). The failure mode of debonding
and fiber fracture coupling appeared on the cross section (see Fig. A1
(c))). On the other hand, it could be seen from Table A1 that the
out-of-plane tensile strength of the ultra-thick laminate decreased by
about 44% compared with that of the thin laminate.

References

[1] Rajapakse YDS, Hui D. Marine composites and sandwich structures. Compos Part B
Eng 2008;39:1–4. https://doi.org/10.1016/j.compositesb.2007.02.003.
Fig. A1. (a) Set-up for through-thickness tensile test showing the crack path (b)
[2] Sutherland LS, Guedes Soares C. Impact behaviour of typical marine composite
and the fracture surface of the ultra-thick laminate (c). laminates. Compos Part B Eng 2005;37:89–100. https://doi.org/10.1016/j.
compositesb.2005.09.001.
[3] Gao Y, Song X, Ding H, Wang H, Wang H, Ke Y. Influence of fiber cutting at the
Table A1 composite grid intersection on the compressive performance of laminate. Compos
Struct 2021;267. https://doi.org/10.1016/j.compstruct.2021.113859.
Comparison of test results between thin laminate and ultra-thick laminate.
[4] Kaybal HB, Ulus H, Eskizeybek V, Avcı A. An experimental study on low velocity
Out-of-plane strength (MPa) impact performance of bolted composite joints-part 2: influence of long-term
seawater aging. Compos Struct 2021;272:113571. https://doi.org/10.1016/j.
Thin laminate 53.3 compstruct.2021.113571.
Ultra-thick laminate 29.8 [5] Li Z, Guedes Soares C, Pan G. Buckling prediction for composite laminated
cylindrical shells in underwater environment. Ocean Eng 2022;258:111244.
https://doi.org/10.1016/j.oceaneng.2022.111244.
Supplementary Material 2. Video of failure process of 30-A specimen [6] Wang J, Zhang Y, Cheng L, et al. Effect of riveting parameters on the forming
quality of riveted lap joints with reduced countersunk head half-crown rivet. Int J
Adv Manuf Technol 2022;121:8083–98. https://doi.org/10.1007/s00170-022-
CRediT authorship contribution statement 09905-7.
[7] Kolat K, Neşer G, Özes Ç. The effect of sea water exposure on the interfacial
fracture of some sandwich systems in marine use. Compos Struct 2007;78:11–7.
Yu Gao: Methodology, Validation, Formal analysis, Data curation, https://doi.org/10.1016/j.compstruct.2006.01.009.
Writing – original draft. Jian Wang: Writing – review & editing. [8] Zhang K, Li H, Cheng H, Luo B, Liu P. Combined effects of seawater ageing and
Xiaowen Song: Conceptualization, Writing – review & editing, Visual­ fatigue loading on the bearing performance and failure mechanism of CFRP/CFRP
single-lap bolted joints. Compos Struct 2020;234:111677. https://doi.org/
ization. Huiming Ding: Resources, Writing – review & editing. Han 10.1016/j.compstruct.2019.111677.
Wang: Formal analysis, Validation, Project administration. YunBo Bi: . [9] Adams RD, Singh MM. The effect of immersion in sea water on the dynamic
Yinglin Ke: Project administration, Resources. properties of fibre-reinforced flexibilised epoxy composites. Compos Struct 1995;
31:119–27. https://doi.org/10.1016/0263-8223(95)00007-0.
[10] Jia L, Yu L, Zhang K, Li M, Jia Y, Blackman BRK, et al. Combined modelling and
Declaration of Competing Interest experimental studies of failure in thick laminates under out-of-plane shear. Compos
Part B Eng 2016;105:8–22. https://doi.org/10.1016/j.compositesb.2016.08.017.
[11] Sachse R, Pickett AK, Middendorf P. Simulation of impact and residual strength of
The authors declare that they have no known competing financial thick laminate composites. Compos Part B Eng 2020;195:108070. https://doi.org/
interests or personal relationships that could have appeared to influence 10.1016/j.compositesb.2020.108070.
[12] Hasan Z, Rader J, Olson A, Turpin D, St Onge R, Amback J. Design, analysis and
the work reported in this paper. fabrication of thick co-cured wing structures. Compos Part B Eng 2019;177:
107335. https://doi.org/10.1016/j.compositesb.2019.107335.
Acknowledgments [13] Fathallah E, Qi H, Tong L, Helal M. Design optimization of lay-up and composite
material system to achieve minimum buoyancy factor for composite elliptical
submersible pressure hull. Compos Struct 2015;121:16–26. https://doi.org/
The author(s) disclosed receipt of the following financial support for 10.1016/j.compstruct.2014.11.002.
the research, authorship, and/or publication of this article: This work [14] Rango RF, Bellomo FJ, Nallim LG. A variational Ritz formulation for vibration
analysis of thick quadrilateral laminated plates. Int J Mech Sci 2015;104:60–74.
was supported by the Science Fund for Creative Research Groups of https://doi.org/10.1016/j.ijmecsci.2015.09.018.
National Natural Science Foundation of China (No. 51821093) and the [15] Jin G, Ye T, Ma X, Chen Y, Su Z, Xie X. A unified approach for the vibration analysis
Key Research and Development Plan of Zhejiang (No. 2020C01039). of moderately thick composite laminated cylindrical shells with arbitrary boundary
conditions. Int J Mech Sci 2013;75:357–76. https://doi.org/10.1016/j.
ijmecsci.2013.08.003.
Supplementary materials [16] Zhou G, Sun Q, Li D, Meng Z, Peng Y, Chen Z, et al. Meso-scale modeling and
damage analysis of carbon/epoxy woven fabric composite under in-plane tension
and compression loadings. Int J Mech Sci 2021;190:105980. https://doi.org/
Supplementary material associated with this article can be found, in
10.1016/j.ijmecsci.2020.105980.
the online version, at doi:10.1016/j.ijmecsci.2022.107966. [17] Bateman MG, Miller OH, Palmer TJ, Breen CEP, Kingston EJ, Smith DJ, et al.
Measurement of residual stress in thick section composite laminates using the
deep-hole method. Int J Mech Sci 2005;47:1718–39. https://doi.org/10.1016/j.
Appendix A. Through-thickness tensile test of ultra-thick ijmecsci.2005.06.011.
laminate [18] Sundaresan P, Singh G, Rao V. Buckling of moderately thick rectangular composite
plates subjected to partial edge compression. Int J Mech Sci 1998;40:1105–17.
https://doi.org/10.1016/S0020-7403(98)00009-5.
The out-of-plane tensile test of the ultra-thick laminate was designed
[19] Ye T, Jin G, Su Z, Chen Y. A modified Fourier solution for vibration analysis of
according to the ASTM D7291 standard [68]. In order to ensure that the moderately thick laminated plates with general boundary restraints and internal
fracture of the specimen occurred in the middle region, the "Spool"-­ line supports. Int J Mech Sci 2014;80:29–46. https://doi.org/10.1016/j.
ijmecsci.2014.01.001.
shaped specimen in the standard was used. The specimens were bonded
[20] Manalo A, Mutsuyoshi H, Matsui T. Testing and characterization of thick hybrid
to the two end tabs using an adhesive film (J272B, supplied by Q-SHY fibre composites laminates. Int J Mech Sci 2012;63:99–109. https://doi.org/
Company, China) with a thickness of 0.1 mm. Binding fixtures and 10.1016/j.ijmecsci.2012.07.002.
positioning fixtures were used to ensure the coaxiality of the tab block [21] Wang X, Huang XH, Yang J, Hou X, Zhu Y, Xie D. Experimental and analytical
study on the pre-crack impact response of thick multi-layered laminated glass
and the specimen after bonding. To better capture the crack

12
Y. Gao et al. International Journal of Mechanical Sciences 241 (2023) 107966

under hard body impact. Int J Mech Sci 2021;206:106613. https://doi.org/ [45] Ogasawara T, Ishikawa T. A simple test method, NAL-II, to evaluate the strength
10.1016/j.ijmecsci.2021.106613. and elastic modulus of polymer matrix composite laminates. Adv Compos Mater
[22] Liao B, Zhou J, Ai S, Lin Y, Xi L, Cao Y, et al. Comparison of laminate thickness on 2016;25:131–45. https://doi.org/10.1080/09243046.2016.1226689.
the low velocity impact behaviors for Z-pinned composite laminates. Int J Mech Sci [46] Lee J, Soutis C. Thickness effect on the compressive strength of T800/924C carbon
2021;204. https://doi.org/10.1016/j.ijmecsci.2021.106567. fibre-epoxy laminates. Compos Part A Appl Sci Manuf 2005;36:213–27. https://
[23] Liao B, Zhou J, Li Y, Wang P, Xi L, Gao R, et al. Damage accumulation mechanism doi.org/10.1016/j.compositesa.2004.06.010.
of composite laminates subjected to repeated low velocity impacts. Int J Mech Sci [47] Fenwick RC. Shear deformation in seismic frame structures. Journal of Structural
2020;182:105783. https://doi.org/10.1016/j.ijmecsci.2020.105783. Engineering 1983;109:965–76. https://doi.org/10.1061/(ASCE)0733-9445(1983)
[24] Chen X, Njomo-Wandji W, Miao XY. A robust and automated method for geometric 109:4(965).
modelling of thick laminates with multiple and asymmetric ply wrinkles. Compos [48] Boisse P, Colmars J, Hamila N, Naouar N, Steer Q. Bending and wrinkling of
Struct 2022;287:115319. https://doi.org/10.1016/j.compstruct.2022.115319. composite fiber preforms and prepregs. A review and new developments in the
[25] Ge X, Zhang P, Zhao F, Liu M, Liu J, Cheng Y. Experimental and numerical draping simulations. Compos Part B Eng 2018;141:234–49. https://doi.org/
investigations on the dynamic response of woven carbon fiber reinforced thick 10.1016/j.compositesb.2017.12.061.
composite laminates under low-velocity impact. Compos Struct 2022;279:114792. [49] Mukhopadhyay S, Jones MI, Hallett SR. Compressive failure of laminates
https://doi.org/10.1016/j.compstruct.2021.114792. containing an embedded wrinkle; experimental and numerical study. Compos Part
[26] Zhang Z, Hou S, Mao Y, He L, Han X. Rate-related study on the ply orientation of A Appl Sci Manuf 2015;73:132–42. https://doi.org/10.1016/j.
carbon fiber reinforced epoxy composite laminates. Int J Mech Sci 2020;188: compositesa.2015.03.012.
105968. https://doi.org/10.1016/j.ijmecsci.2020.105968. [50] Thompson AJ, McFarlane JR, Belnoue JPH, Hallett SR. Numerical modelling of
[27] Lv J, Xiao Y, Zhou Y, Xie Y. Characterization and modeling of the creep behavior of compaction induced defects in thick 2D textile composites. Mater Des 2020;196:
fiber composites with tension and compression asymmetry. Int J Mech Sci 2020; 109088. https://doi.org/10.1016/j.matdes.2020.109088.
170:105340. https://doi.org/10.1016/j.ijmecsci.2019.105340. [51] Xie N, Smith RA, Mukhopadhyay S, Hallett SR. A numerical study on the influence
[28] Bergan AC, Herráez M, González C, Lopes CS. A constitutive model for fiber of composite wrinkle defect geometry on compressive strength. Mater Des 2018;
kinking: Formulation, finite element implementation, and verification. Compos 140:7–20. https://doi.org/10.1016/j.matdes.2017.11.034.
Part A Appl Sci Manuf 2020;129. https://doi.org/10.1016/j. [52] Rajan S, Sutton MA, Wehbe R, Tatting B, Gürdal Z, Kidane A, et al. Experimental
compositesa.2019.105682. investigation of prepreg slit tape wrinkling during automated fiber placement
[29] Matsuo T, Kageyama K. Compressive failure mechanism and strength of process using StereoDIC. Compos Part B Eng 2019;160:546–57. https://doi.org/
unidirectional thermoplastic composites based on modified kink band model. 10.1016/j.compositesb.2018.12.017.
Compos Part A Appl Sci Manuf 2017;93:117–25. https://doi.org/10.1016/j. [53] Rajan S, Sutton MA, Sockalingam S, McMakin W, Gurdal Z, Kidane A. Simulations
compositesa.2016.11.018. and experiments for automated fiber placement of prepreg slit tape: wrinkle
[30] Kabiri Ataabadi A, Hosseini-Toudeshky H, Ziaei Rad S. Experimental and analytical formation and fundamental observations. Compos Part B Eng 2020;201:108287.
study on fiber-kinking failure mode of laminated composites. Compos Part B Eng https://doi.org/10.1016/j.compositesb.2020.108287.
2014;61:84–93. https://doi.org/10.1016/j.compositesb.2014.01.039. [54] Nartey M, Zhang T, Gong B, Wang J, Peng S, Wang H, et al. Understanding the
[31] Liu XH, Moran PM, Shih CF. The mechanics of compressive kinking in impact of fibre wrinkle architectures on composite laminates through tailored gaps
unidirectional fiber reinforced ductile matrix composites. Compos Part B Eng 1996; and overlaps. Compos Part B Eng 2020;196. https://doi.org/10.1016/j.
27:553–60. https://doi.org/10.1016/S1359-8368(96)00008-X. compositesb.2020.108097.
[32] Sun Q, Zhou G, Guo H, Meng Z, Chen Z, Liu H, et al. Failure mechanisms of cross- [55] Ahmadi Jebeli M, Heidari-Rarani M. Development of Abaqus WCM plugin for
ply carbon fiber reinforced polymer laminates under longitudinal compression progressive failure analysis of type IV composite pressure vessels based on Puck
with experimental and computational analyses. Compos Part B Eng 2019;167: failure criterion. Eng Fail Anal 2022;131:105851. https://doi.org/10.1016/j.
147–60. https://doi.org/10.1016/j.compositesb.2018.12.041. engfailanal.2021.105851.
[33] Daniel IM, Hsiao HM, Wooh SC. Failure mechanisms in thick composites under [56] Lemanski SL, Wang J, Sutcliffe MPF, Potter KD, Wisnom MR. Modelling failure of
compressive loading. Compos Part B Eng 1996;27:543–52. https://doi.org/ composite specimens with defects under compression loading. Compos Part A Appl
10.1016/1359-8368(95)00010-0. Sci Manuf 2013;48:26–36. https://doi.org/10.1016/j.compositesa.2012.12.007.
[34] Matsuda N, Wakayama K, Nishikawa M, Hojo M, Arai M, Goto K, et al. [57] Greenhalgh ES. 3 - Fibre-dominated failures of polymer composites. Failure
Experimental observation and FEM analysis of initial fractures in quasi-isotropic analysis and fractography of polymer composites. Woodhead Publishing; 2009.
CFRP laminates subjected to compressive loading. Adv Compos Mater 2021;30: p. 107–63. https://doi.org/10.1533/9781845696818.107.
50–70. https://doi.org/10.1080/09243046.2020.1802543. [58] Fuchs C, Bhattacharyya D, Fakirov S. Microfibril reinforced polymer–polymer
[35] Goto K, Arai M, Kano Y, Hara E, Ishikawa T. Compressive fracture aspect of thick composites: application of Tsai-Hill equation to PP/PET composites. Compos Sci
quasi-isotropic carbon fiber reinforced plastic laminates. Compos Sci Technol Technol 2006;66:3161–71. https://doi.org/10.1016/j.compscitech.2005.02.023.
2019;181:107706. https://doi.org/10.1016/j.compscitech.2019.107706. [59] Tsai SW, Wu EM. A general theory of strength for anisotropic materials. J Compos
[36] Tsampas SA, Greenhalgh ES, Ankersen J, Curtis PT. On compressive failure of Mater 1971;5:58–80. https://doi.org/10.1177/002199837100500106.
multidirectional fibre-reinforced composites: a fractographic study. Compos Part A [60] TSAI WS. Strength theories of filamentary structure. Fundam Asp Fiber Reinf Plast
Appl Sci Manuf 2012;43:454–68. https://doi.org/10.1016/j. Compos 1968.
compositesa.2011.11.013. [61] Hashin Z, Rotem A. A fatigue failure criterion for fiber reinforced materials.
[37] Hara E, Katoh H, Iwahori Y, Yokozeki T. Influence of specimen geometry on J Compos Mater 1973;7:448–64. https://doi.org/10.1177/002199837300700404.
apparent Poisson’s ratio evaluated under the compressive loading of quasi- [62] Hashin Z. Failure criteria for unidirectional fiber composites. J Appl Mech 1980;47:
isotropic CFRP laminate. Adv Compos Mater 2021;30:96–110. https://doi.org/ 329–34. https://doi.org/10.1115/1.3153664.
10.1080/09243046.2020.1753637. [63] Xu Y, Gao Y, Wu C, Fang J, Sun G, Steven GP, et al. On design of carbon fiber
[38] Standard test method for compressive properties of polymer matrix composite reinforced plastic (CFRP) laminated structure with different failure criteria. Int J
materials using a combined loading compression (CLC). Test Fixt 2014:1–13. Mech Sci 2021;196:106251. https://doi.org/10.1016/j.ijmecsci.2020.106251.
https://doi.org/10.1520/D6641. i. [64] Puck A, Schürmann H. Failure analysis of FRP laminates by means of physically
[39] Opelt CV, Cândido GM, Rezende MC. Compressive failure of fiber reinforced based phenomenological models1. Compos Sci Technol 1998;58:1045–67. https://
polymer composites – a fractographic study of the compression failure modes. doi.org/10.1016/S0266-3538(96)00140-6.
Mater Today Commun 2018;15:218–27. https://doi.org/10.1016/j. [65] Pinho ST, Darvizeh R, Robinson P, Schuecker C, Camanho PP. Material and
mtcomm.2018.03.012. structural response of polymer-matrix fibre-reinforced composites. J Compos Mater
[40] Lee J, Soutis C. A study on the compressive strength of thick carbon fibre-epoxy 2012;46:2313–41. https://doi.org/10.1177/0021998312454478.
laminates. Compos Sci Technol 2007;67:2015–26. https://doi.org/10.1016/j. [66] Ye L. Role of matrix resin in delamination onset and growth in composite
compscitech.2006.12.001. laminates. Compos Sci Technol 1988;33:257–77. https://doi.org/10.1016/0266-
[41] Hsiao HM, Daniel IM, Wooh SC. A new compression test method for thick 3538(88)90043-7.
composites. J Compos Mater 1995;29:1789–806. https://doi.org/10.1177/ [67] Esposito L, Sorrentino L, Penta F, Bellini C. Effect of curing overheating on
002199839502901307. interlaminar shear strength and its modelling in thick FRP laminates. Int J Adv
[42] Precision WS. Standard test method for compressive properties of polymer matrix Manuf Technol 2016;87:2213–20. https://doi.org/10.1007/s00170-016-8613-5.
composite materials using a combined loading compression (CLC). Annu B ASTM [68] ASTM D. 7291. Stand test method through-thickness “flatwise” tensile strength
Stand 2011;97:1–11. https://doi.org/10.1520/D5467. elastic modul a fiber-reinforced. Polym Matrix Compos Mater Am Soc Test Mater
[43] Standard test method for compressive properties of rigid plastics. ASTM Int 2010: 2007.
1–8. https://doi.org/10.1520/D0695-15.2. i. [69] Kalantari M, Dong C, Davies IJ. Effect of matrix voids, fibre misalignment and
[44] Camponeschi ET. Compression testing of thick-section composite materials DTRC- thickness variation on multi-objective robust optimization of carbon/glass fibre-
SME-8. David Taylor Research Center; 1989. DTRC-SME-8. reinforced hybrid composites under flexural loading. Compos Part B Eng 2017;123:
136–47. https://doi.org/10.1016/j.compositesb.2017.05.022.

13