You are on page 1of 11

International Journal of Mechanical Sciences 161–162 (2019) 105012

Contents lists available at ScienceDirect

International Journal of Mechanical Sciences


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

Influence of in-plane tensile preloads on impact responses of composite


laminated plates
Cunxian Wang a, Tao Suo a,b,c,∗, Chao Hang d, Yulong Li a,b,c, Pu Xue a,b, Qiong Deng a,b
a
School of Aeronautics, Northwestern Polytechnical University, 127 West Youyi Road, Beilin District, Xi’an Shaanxi, 710072, China
b
Fundamental Science on Aircraft Structural Mechanics and Strength Laboratory, Northwestern Polytechnical University, Xi’an 710072, China
c
Shaanxi Key Laboratory of Impact Dynamics and Engineering Application(IDEA), Northwestern Polytechnical University, Xi’an 710072, China
d
Commercial Aircraft Corporation of China Ltd, Shanghai 200126, China

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

Keywords: Ballistic impact tests on composite laminated plates were carried out with a preloading device to investigate the
Laminated plates impact responses at nominal velocities of 50 m/s and 70 m/s. Not only the center locations but also the near-edge
In-plane preload regions of the laminated plates were impacted. Furthermore, finite element models were developed using three-
Impact response
dimensional solid elements and cohesive elements, and a user subroutine VUMAT of Ladevèze failure criterion
Delamination
was used in support of simulations in ABAQUS Explicit, providing detailed insights into the characteristics of
Interlaminar stress
delamination and target response. Results showed that there was a significant promotion of delamination resis-
tance in the center location when applying appropriate preloads, but some weakening parts were also generated
over large areas at the same time. A competition mechanism between the increased bending stiffness induced
enhancement effect and the interlaminar stresses induced weakening effect was proposed to explain the influence
of the biaxial in-plane tensile preloads on the delamination resistance of laminated plate. The work in this study
may provide guidance for the impact resistance design of composite aircraft wing skin.

1. Introduction the anisotropy of composites, and results clearly showed typical fail-
ure modes of CFRP when subjected to impact loading. Studies also con-
Composites are widely used in aeronautics and astronautics due to cerned the anisotropic damage especially for the shape and distribu-
their advantages such as excellent mechanical properties, high specific tion of delamination [11, 12]. Yew and Kendrick [13] developed the
strength, good stiffness and light-weight [1]. Since aeronautic structures cross section staining method and ultrasonic C-scan method to detect
tend to suffer impact from falling tools, gravel, birds, hail and weapon the damage of composites after impact, and put forward that the im-
debris during the flight [2], the ballistic tests on composites become pact damage of composites was anisotropic. With the development of
more extensive in preliminary research. Currently the carbon fiber rein- computer technology, finite element analysis has been widely used in
forced composites are sensitive to impact, and the impact damages such most investigations on impact response of composites. Riccio et al. [14,
as delamination, cracking in the matrix and fracture of brittle fibers can 15] conducted a numerical study on the impact behavior of a stiffened
significantly degrade the mechanical property, even result in a reduction composites panels with omega shaped stringers, and the experimental
of the structural integrity [3, 4]. non-destructive evaluation techniques of impact damage was also inves-
Over years, the efforts to explore the impact responses of compos- tigated. Zhang et al. [16, 17] developed a series of numerical methods
ites almost never stopped. It has attracted more than forty years of to study the impact failure behavior of composite structures, with de-
widespread concern on the damage detection and analysis. By evalu- tailed descriptions and discussions on matrix cracking, free-edge effect
ating the bearing capacity and determining the failure characteristics and interface delamination of braided composites. Zhang and coworkers
of composites, the designing of structures manufactured by compos- [18, 19] developed a multi-scale framework to model the impact failure
ites can be guided eventually [5–7]. Generally, for carbon fiber rein- behavior of composites subjected to high-speed impact, and predicted
forced composites under impact loading, the main failure mechanisms the failure pattern and impact threshold well.
were known as local deformation, delamination, fiber breakage and Since 1990s, the researches around the effects of preloads on com-
debonding [8]. Takashima et al. [9] and Johnson and Sims [10] car- posites were widely concerned since the aeronautic structures exten-
ried out some tests, in which the layer angles were changed to increase sively suffer from high-velocity impact under a certain preloaded state.


Corresponding author at: School of Aeronautics, Northwestern Polytechnical University, 127 West Youyi Road, Beilin District, Xi’an Shaanxi, 710072, China.
E-mail address: suotao@nwpu.edu.cn (T. Suo).

https://doi.org/10.1016/j.ijmecsci.2019.105012
Received 1 April 2019; Received in revised form 27 June 2019; Accepted 9 July 2019
Available online 11 July 2019
0020-7403/© 2019 Elsevier Ltd. All rights reserved.
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Advisory Circular 20–107B pointed out that there was a need to con- cepts and mechanisms of the preloads induced damage. A special test
sider the actual loads (tension, compression, or torsion) on the analyses rig was designed to apply biaxial in-plane tensile preloads on [0/90]8
of unexpected impact damage when designing the structures. With the and [0/90/+45/−45]2S laminated plates. Afterwards, a series of ballis-
guidance, lots of attentions were devoted to the influences of preloads on tic impact tests were performed launching steel sphere against both the
impact response of composites. Research by Chiu et al. [20] conducted preloaded and unloaded laminated plates. In the experiments, both the
some ballistic impact tests and found that the impact resistance of com- center location and the near-edge region of the laminated plates were
posites can be reduced by large tensile preload. Afterwards, Kelkar et al. impacted by using a gas gun, and the ultrasonic C-scan was applied to de-
[21] and Nettles [22] applied several kinds of tensile preloads on com- tect the delamination. Moreover, numerical simulations and analytical
posites to guarantee that no preloads induced damage was generated be- modeling were conducted to gain additional insight into the character-
fore impact, and results showed that the impact resistance of composites istics of damage generated from the impact. A user subroutine VUMAT
was improved by appropriate tensile preloads. Furthermore, researches with revised Ladevèze failure criterion was developed and implemented
on impact responses of composites with compressive preloads were con- in ABAQUS Explicit, and the cohesive elements were employed to an-
ducted to gain additional insight into the effect of preloads. Zhang et al. alyze delamination. By comparing the experimental and numerical re-
[23] presented that the compressive preload tended to cause the buck- sults, the influences of preloads on laminated plates were investigated
ling for relatively thin composite plates, which may bring the complex- and discussed. Both the in-plane preloads induced interlaminar stresses
ity to this research. Subsequent efforts by Heimbs et al. [24–28] reduced and stress stiffening were used to explain the special phenomenon. The
the complexity and successfully kept the compressive preload at a stable work can help to obtain a better understanding of the impact responses
level, and then put forward that the stiffness and impact damage resis- of preloaded laminated plates, also provides guidance for the designing
tance of composites were reduced by the compressive preloads. Previous of composites aircraft wing skin.
studies are reasonable to describe the effects of preloads on composites
by analyzing the impact responses of center location, but the near-edge 2. Material preparation and experimental procedure
impact performances are rarely investigated. Actually, not only the cen-
ter location but also the near-edge region of composite structures may 2.1. Material preparation
suffer impact loads at times, like gravel impact or debris impact on the
aircraft wing skin. The impact responses of preloaded composites, espe- In the ballistic impact tests, the material used was T700/epoxy resin
cially for the damages of preloaded carbon fiber reinforced laminated M10R laminated plates with two types of stacking sequences which were
plates, are still far from full understanding. depicted as [0/90]8 and [0/90/+45/−45]2S . All specimens were cut to
In the present work, in support of designing the composite wing plates with size of 300 × 300 mm2 and attached to two 30 mm width
skin, the ballistic impact responses of preloaded carbon fiber reinforced support frames on all parallel sides, thus leaving a 240 × 240 mm2 free
laminated plates were investigated by using both experimental and nu- target area, as shown in Fig. 1(a). The water jet cutting was employed
merical method. The purpose was to identify and analyze both con- to avoid the initial damage as possible. Fig. 1(b) showed the preloading

Fig. 1. Strain measurement scheme, clamping system and test devices; (a) strain measurement scheme; (b) preloading device; (c) schematic diagram of testing
device–pneumatic gun system.
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Table 1
Experimental conditions.

NO. Layups Impact point Off-central distance [x, y] (mm) Actual Velocity(m/s) Pre-strain

1–2# [0/90/+45/−45]2S Point 1 [−5, 0] (1#), [13, −5] (2#) 49.33(1#), 68.72(2#) N/A
3–4# [0/90/+45/−45]2S Point 1 [8, −2] (3#), [7, −2] (4#) 52.45(3#), 68.95(4#) 1000𝜇𝜀
5–6# [0/90]8 Point 1 [9, 0] (5#), [14, 0] (6#) 49.42(5#), 68.18(6#) N/A
7–8# [0/90]8 Point 1 [28, −9] (7#), [12, −7] (8#) 49.33(7#), 69.51(8#) 1000𝜇𝜀
9–10# [0/90/+45/−45]2S Point 2 [5, 67] (9#), [2, 64] (10#) 51.34(9#), 50.78(10#) 1000𝜇𝜀
11–12# [0/90]8 Point 2 [2, 69] (11#), [4, 66] (12#) 47.51(11#), 50.42(12#) 1000𝜇𝜀

device which consisted the outer frame, loading blocks and bolts. The small tensile signal. After the small in-plane tensile wave, a compres-
specific in-plane tensile preloads can be applied by adjusting the bolts sive component of an oncoming transverse flexural wave arrives at the
connecting the outer frame with the loading blocks. A total of six strain strain gauges, followed by a large tensile component of the transverse
gauges were stacked on the back side of the laminated plate not only flexural wave [29]. Actually, if the projectile exactly impacts the point
to measure the in-plane strain histories during impact but also to give 1 where the strain gauges 1 and 2 are mounted, the first in-plane ten-
feedback for preloading control before impact. The C-scan tests were sile wave and the following compressive component of the transverse
conducted on the target plates after applying different biaxial in-plane flexural wave will not be collected by the strain gauges 1 and 2. In the
preloads, and results showed that there was no initial damage on the ballistic impact tests, the impact locations are more or less deviated from
laminated plates when no more than 2000𝜇𝜀 biaxial pre-strain was ap- the point 1 (see Fig. 2(e) and Table 1), thus causing the in-plane ten-
plied. Therefore, considering the stress concentration of opening holes sile and compressive components of the transverse flexural waves can
on the plates and the cyclic-loading results in the material tests which be detected during the impact. Theoretically, the largest flexural wave
were used to support the parameters determination of Ladevèze model, component is supposed to be the most important factor which results in
about 1000𝜇𝜀 biaxial pre-strain was applied on the laminated plate in delamination.
the ballistic tests. Fig. 3(a) shows the delamination and peak deflection under differ-
ent impact conditions. As shown, for the laminated plates with same
2.2. Instruments and measurements layups and preloads, higher impact energy tends to result in larger de-
lamination area and larger deflection, and the [0/90/+45/−45]2S plates
Ballistic impact tests were performed with a one stage gas gun as are found to exhibit more delamination than [0/90]8 plates when sub-
shown in Fig. 1(c). The gas gun consisted of a pressure vessel with a jected to almost exactly the same impact load. Moreover, results also
volume of 0.35 m3 , a gun barrel with a length of 5.2 m and an inner di- show that the biaxial in-plane tensile preloads (1000𝜇𝜀 biaxial pre-
ameter of 25 mm. Steel spheres with a diameter of 10 mm corresponding strain) can enhance the delamination resistance of the center location
mass of 4.1 g were selected as projectiles to conduct the experiments. for both [0/90]8 plates and [0/90/+45/−45]2S plates. It can be sup-
Each projectile was supported by a wooden shell and accelerated by posed that the limited in-plane tensile preloads can improve the bending
compressive air in the gun barrel. During the impact, the wooden shell stiffness of plates to some extent, thus causing the decreased peak val-
was removed by the shell removal device, which can help to avoid some ues of in-plane strains and normal deflections during the impact. Such
influences on the measurement accuracy of grating laser velocimeter. phenomenon which is called stress stiffening can also be observed in
Meanwhile, two V711 high-speed cameras were installed in the experi- tightropes and taut drums.
ments where the frontal one was used to capture the impact location and In Fig. 3(b), the delamination and peak deflections of the near-edge
the back one was applied for capturing the normal deflection, respec- region (point 2) are illustrated to make contrast to the results from cen-
tively. In addition, both the center location (point 1) and the near-edge ter location. Results show that both the [0/90/+45/−45]2S plates and
region (point 2) of the laminated plates were impacted at the velocities the [0/90]8 plates appear more delamination when changing the im-
of 50 m/s and 70 m/s. After the ballistic impact tests, the delamination pact location from point 1 to point 2 even though the peak deflections
area of each laminated plate was measured by using a PAC ultrasonic decrease at the same time. In addition, the strain variation signals from
C-scan. Table 1 listed all the experimental conditions for ballistic impact strain gauge 3 and 4 on 3# and 9# plates are illustrated in Fig. 4(a)
tests. and (b). Results show that the peak values of in-plane strain on 9# plate
are higher than those of 3# plate, causing by the fact that the impact
location of 9# plate is near to the strain gauge 3 and 4. However, the
3. Experimental results peak in-plane strains of strain gauge 3 and 4 on 9# plate are still smaller
than those of strain gauge 1 and 2 on 3# plate, which is corresponding
The in-plane strain histories of point 1 and 2 on the target plates with the changing trends of peak deflections between 3# plate and 9#
2#, 4#, 6# and 8# are illustrated in Fig. 2. As shown, the peak in- plate, as shown in Fig. 3(b). Moreover, the C-scan results indicate the
plane strains of preloaded target plates are relatively smaller than those shapes of delamination are elongated along the direction which perpen-
of unloaded target plates at velocities of 50 m/s and 70 m/s. In addi- dicular to the nearest edge, as shown in Fig. 4(c). Previous works and
tion, no matter which impact velocity is applied, the in-plane strains the results from the plates 1#−8# indicate that the stiffening effect in-
of [0/90]8 plates are still smaller than those of [0/90/+45/−45]2S duced by the biaxial in-plane tensile preloads can reduce the degree of
plates. Taking the target plates 6# and 8# as the examples, the peak bending deformation and then typically prevent delamination propaga-
strain of unloaded plate 6# impacted at velocity of 70 m/s is close tion [30]. Nevertheless, the results from plates 9#−12# show different
to 4500𝜇𝜀; however, the peak strain of preloaded plate 8# is only tendency by only changing the impact location. Consequently, the ef-
3000𝜇𝜀. fect of biaxial in-plane preloads on delamination resistance cannot be
It is deserved to be noticed that the strain histories from strain gauge simply judged by considering the results from the ballistic impact tests
2 show a tendency that a very small in-plane tensile wave arrives at in which only the center locations of the target plates are impacted. In
the strain gauges right after the contact of steel projectile with the tar- the following section, the influences of biaxial in-plane preloads on de-
get plates. However, since the peak value of the tensile wave is much lamination at different impact locations will be discussed by combining
weaker than that of following stress wave, it is difficult to observe this with simulation results.
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 2. Strain variation signals and typical high-speed photographs; (a–b) strain signals of 2# and 4# plates; (c-d) strain signals of 6# and 8# plates; (e) high-speed
photographs for 8# plate impacted by steel sphere.
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 3. Delamination and peak deflection of [0/90/+45/−45]2S plates and [0/90]8 laminated plates; (a) laminated plates impacted at the center loacation; (b)
laminated plates with different impact loacation.

Fig. 4. Strain variation signals of 3# and 9# plates and C-scan results; (a) strain gauge 3; (b) strain gauge 4; (c) C-scan results of 3#, 9#, 7#, 12# plate.

4. Numerical modeling and validation models in the simulation, which is caused by the fact that element ero-
sion would occur unsymmetrically once impact failure initiates because
4.1. Finite-Element models and material models of the presence of off-axial lamina and the variation of impact locations.
All the laminated plates were modeled with three-dimensional solid el-
Since the excellent nonlinear analysis ability of commercial software ements with reduced integration (C3D8R) and 8-node hexahedral cohe-
ABAQUS, different finite element (FE) models were developed to simu- sive elements (COH3D8) while the projectile was set as a rigid body.
late the detailed impact responses of these target plates. The mesh pat- The in-plane mesh pattern was kept to be uniform as well as each ply
tern and size dependency were examined at first for purpose of finding was modeled with one element in the thickness direction. In detail, each
an optimized mesh for stability, accuracy, and efficiency of the impact model consisted sixteen solid plies (0.1 mm) and fifteen cohesive plies
analysis. Noted that the full FE models were used instead of one-quarter (0.001 mm), and the mesh density was gradually coarsening from the
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 5. Strain distributions of preloaded laminated plates.

inner region (the potential impact region) to the outer. Mesh transitions where T is the thickness of single laminate layer, and 𝛼 denotes an ex-
between regions were good enough to prevent stress wave reflections perience parameter, 1 represents the fiber direction; 2 represents the
from the boundaries of regions. Simply refining the mesh did not neces- transverse direction; 3 represents the through-thickness direction. The
sarily improve the accuracy since the material model parameters were effective stiffness for the composite along the thickness direction can be
calibrated for a specific mesh size and the mode of failure changed dur- measured by
ing the simulation. Furthermore, there was no clear theoretical guide- 𝐸33
line on the required mesh density for the range of impacts covered in 𝐸𝑒𝑓 𝑓 = (2)
1
1+ 𝛼
the simulation. Considering the softening effects and mesh dependent
failure algorithms, the only reasonable methodology to find the most Considering the accuracy and time cost of simulation, 𝛼 is taken as
appropriate mesh size was trial and error while comparing the results 50, thus causing the loss of stiffness due to the existence of the interface
against a controlled test data. Subsequently, it was possible to arrive at is less than 2% [32]. The strength of layers can also be calculated by in-
an optimum after experimenting with different meshing and draw some plane parameters based on the method from De Moura [33]. In addition,
guidelines for that particular case. Thus, each target thickness was mod- the mixed mode of secondary stress criterion was selected to estimate
eled with one mesh densities in this study while many more were ex- the initiation of delamination as well as the coupling secondary critical
perimented on during the calibration runs. A viscous based stabilization energy release rate criterion was employed to simulate the evolution
method was used during the simulations to prevent hourglass modes of of delamination. Table A.2 lists the constitutive parameters of cohesive
the reduced integration elements. Contact behavior between the projec- element which are calculated by using the anisotropic parameters.
tile and the target was attained by using a penalty based single surface
type contact algorithm that uses a nodal constraint formulation with an 4.2. Method of applying preloads
element erosion scheme.
A more sophisticated material model was necessary for the targets The impact analysis of preloaded laminated plate was performed by
to simulate ballistic impact responses, where the global Ladevèze con- establishing a preloading step followed by an impacting step. However,
stitutive model [31] was employed to describe the impact response of the explicit solver faced the influence of the inertia force when apply-
laminated plates. Since the global Ladevèze constitutive model can de- ing the in-plane preloads, thus requested a smooth loading method to
scribe the fiber-matrix coupling effect, fiber brittle rupture, matrix micro conduct the quasi-static preloading step [34]. Consequently, the dis-
cracking, debonding of fiber/matrix and effect of strain rate, a user sub- placement loading mode combined with smooth step amplitude curve
routine VUMAT with revised Ladevèze failure criterion was employed were employed to avoid the fluctuations caused by discontinuous load-
in ABAQUS to predict the damage evolution of laminated plates. Ta- ing rate. Fig. 5 illustrates the in-plane strain distributions along hori-
ble A.1 lists the Ladevèze constitutive parameters of the unidirectional zontal and vertical direction of the preloaded laminated plates. Results
carbon-epoxy material. show that, even the strain distributions near the edge show some ups
Theoretically, there are some difficulties to determine interlaminar and downs, but those of most regions especially for the center location
stiffness due to the relatively thin thickness of cohesive layer. Thus, the (point 1) and the near-edge region (point 2) were practically uniform
method from Turon et al. [32] was employed to calculate the interlam- and acceptable. After checking the strain distributions and the value
inar normal stiffness (Knn ) and two interlaminar shear stiffness (Kss and of in-plane preloads, the restart analysis was conducted to support the
Ktt ): impacting step.

4.3. Numerical results and model validation


𝛼𝐸33
𝐾𝑛𝑛 = The simulation results (delamination and peak deflection) of the
𝑇
laminated plates with different in-plane preloads (−1000𝜇𝜀, 0, 1000𝜇𝜀
2𝛼𝐺13
𝐾𝑠𝑠 = and 2000𝜇𝜀) are shown in Fig. 6(a) and (b) in which the negative pre-
𝑇
strain represents the in-plane compressive preloads while the positive
2𝛼𝐺23 pre-strain means in-plane tensile preloads. Moreover, the strain histo-
𝐾𝑡𝑡 = (1)
𝑇 ries of the corresponding points from the simulations are illustrated to
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 6. Delamination areas, peak deflections and strain histories of the laminated plates; (a) delamination areas under various preload; (b) peak deflections under
various preload; (c-d) comparison of strain histories between experimental and numerical results at corresponding points of [0/90]8 laminated plate 8#.

compare with the experimental results, which is supported to the further in-plane tensile preloads can help to reduce the bending degree of the
validation of the numerical model, as shown in Fig. 6(c) and (d). Results laminated plates under normal impact load to some extent, thus plays
show that not only the level but also the tendency of the strain histories a positive role in the delamination resistance of the laminated plates.
from simulations coincides well with the experimental results. In addi- However, the effect of interlaminar stresses between adjacent layers in-
tion, both the delamination areas and peak deflections decrease with duced by biaxial in-plane tensile preloads cannot be ignored. Since the
the increase of in-plane tensile preloads. It can be concluded that the difference of ply orientation between each two adjacent layers, there
in-plane tensile preloads tend to improve the delamination resistance of always exist tensile-shear coupling or the mismatch of Poisson’s ratio
the laminated plates at the center location while the in-plane compres- in the laminated plate. Consequently, as shown in Fig. 8(a) and (b), the
sive preloads seem to promote the generation of delamination. More- biaxial in-plane tensile preloads tend to induce different degrees of inter-
over, the comparisons of delamination shapes are illustrated in Fig. 7, laminar stresses between adjacent layers in the laminated plate before
in which SDEG means the scalar stiffness degradation in ABAQUS. Re- impact, which also plays an important role in delamination resistance
sults also indicates that the results simulated by FE model show good of preloaded laminated plates when subjected to impact loading.
agreement with those obtained by the C-scan measurements, proving By using the layerwise laminate theory (LWT) method from Robbins
that the simulation method in this study is appropriate and feasible. and Reddy [35] and the first-order shear deformation theory, the dis-
placements of each random point in laminated plates, i.e. u, v and w,
can be expressed as
5. Effect of in-plane tensile preload on delamination
𝑛+1

Not only the experimental results but also the numerical results in- 𝑢(𝑥, 𝑦, 𝑧) = 𝑢0 (𝑥, 𝑦) + 𝑧𝜑𝑥 (𝑥, 𝑦) + 𝑈𝑘 (𝑥, 𝑦)𝜓𝑘 (𝑧)
dicate the delamination resistance of the center location on laminated 𝑘=1
plates tends to be improved by applying the biaxial in-plane tensile 𝑛+1

preloads, which was also presented by Nettles [22] and Zhang et al. 𝑣(𝑥, 𝑦, 𝑧) = 𝑣0 (𝑥, 𝑦) + 𝑧𝜑𝑦 (𝑥, 𝑦) + 𝑉𝑘 (𝑥, 𝑦)𝜓𝑘 (𝑧)
𝑘=1
[23]. However, results from the experiments also show that the delami-
𝑛+1

nation areas are increased when the impact location moves to the near-
𝑤(𝑥, 𝑦, 𝑧) = 𝑤0 (𝑥, 𝑦) + 𝑊𝑘 (𝑥, 𝑦)𝜓𝑘 (𝑧) (3)
edge region in spite of the same in-plane tensile preloads and impact
𝑘=1
loads are applied.
Theoretically, it can be considered that the bending stiffness of lam- where u0 , v0 and w0 represent the initial displacements of each
inated plates is enhanced to a certain extent by applying the in-plane point, U, V and W denote the displacements of the correspond-
tensile preloads, which is supported by the reduced peak in-plane strains ing points on the specified layer, n is the total number of lay-
and decreased peak deflections during the impact. Therefore, the biaxial ers, 𝜓 k (z) is the first order Lagrange interpolation functions which
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 7. Comparisons of delamination shapes between experimental and numerical results.


C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 8. Schematic diagram of interlaminar stresses; (a) applying biaxial in-plane tensile preloads; (b) preloads induced interlaminar stresses; (c–d) typical interlaminar
stresses of composite laminated plates with stacking sequences of [0/90]8 and [0/90/+45/−45]2S .

can be defined as 𝜓 k (z) = (z − zk )/(zk + 1 − zk ),zk − 1 ≤ z ≤ zk and tion, and then gradually converts from compressive stress into tensile
𝜓 k (z) = (zk + 1 − z)/(zk + 1 − zk ),zk ≤ z ≤ zk + 1 , for z=else, 𝜓 k (z) = 0. stress when changing from 0.75b to the edge. Besides, the interlami-
Besides, the displacements of the corresponding point for isoparamet- nar shear stresses produced by in-plane tensile preloads in the preload-
𝑚
∑ 𝑚
∑ 𝑚
∑ ing process can be considered as two important factors affecting the
ric element can be depicted as 𝑈 = 𝑈 𝑖 𝜃𝑖 , 𝑉 = 𝑈 𝑖 𝜃𝑖 , 𝑊 = 𝑈 𝑖 𝜃𝑖 ,
𝑖=1 𝑖=1 𝑖=1 delamination resistance of laminated plates when subjected to impact
where m is the number of nodes for one unit, Ui , Vi , Wi are the loading. Theoretically, the interfacial strength of the laminated plates
displacements of the corresponding node. By using Lagrange isopara- are weakened by the interlaminar shear stresses. Thus, the in-plane ten-
metric element with 3 nodes, 𝜃 i can be calculated as 𝜃1 = 12 𝜇(𝜇 − 1), sile preloads induced interlaminar shear stresses can be assumed to pro-
𝜃 2 = (1 + 𝜇)(1 − 𝜇), and 𝜃3 = 12 𝜇(𝜇 + 1), where − 1 ≤ 𝜇 ≤ 1. The values mote the delamination of laminated plates when subjected to impact
of 𝜀x and 𝜀y are equal to 1000𝜇𝜀 which refer to the in-plane preloads loading.
applied in the experiments. The other strains of the corresponding point For further discussion, Fig. 9(a) illustrates the stiffness degradations
on the specified layer can be calculated by ignoring w0 (x,y) when tak- of all the cohesive elements on the [0/90/+45/−45]2S preloaded lam-
ing a derivative to z, as shown in Eq. (4). At last, 𝜎 z , 𝜏 yz , 𝜏 zx can be inated plate 3# before impact by using the toggle global translucency
obtained by employing the classical laminated plate theory [36]. Based function in ABAQUS. Results show that most of cohesive elements on
on MATLAB and the parameters of anisotropic model from Table A.2, the preloaded laminated plates appear different degrees of stiffness
the interlaminar stresses were calculated by setting the length and width degradation before impact, which can help to explain the different ten-
of the laminated plate as 2b. dencies of delamination resistance between center location impact and
near-edge region impact by combining the effect of increased bending
𝑛+1
∑ 𝑚
𝜕𝜓 ∑ stiffness.
𝜀𝑧 = ( 𝑘 )( 𝑊𝑘𝑖 𝜃𝑖 )
𝑘=1
𝜕𝑧 𝑖=1 For [0/90/+45/−45]2S laminated plate 3#, the impact location is
𝑚 𝑛+1 𝑚
near to the center location where the initial stiffness degradation is rel-
𝜕 ∑ 𝑖 ∑ 𝜕 ∑ 𝑖 atively small (almost zero). Consequently, in the center location, the
𝛾𝑥𝑦 = 𝑧 ⋅ ( 𝜑 𝑥 𝜃𝑖 ) + 𝜓𝑘 ⋅ ( 𝑈 𝑘 𝜃𝑖 )
𝜕𝑦 𝑖=1 𝑘=1
𝜕𝑦 𝑖=1 weakening effect of interlaminar stresses on delamination resistance
𝑚 𝑛+1
[ 𝑚 𝑚
] is negligible, thus making the increased bending stiffness becomes the
∑ ∑ 𝜕 𝜓𝑘 ∑ 𝑖 𝜕 ∑
𝛾𝑦𝑧 = 𝜑𝑖𝑦 𝜃𝑖 + ( 𝑉𝑘 𝜃𝑖 ) + 𝜓𝑘 ( 𝑊𝑘𝑖 𝜃𝑖 ) dominant factor which can influence the delamination resistance of the
𝑖=1 𝑘=1
𝜕𝑧 𝑖=1 𝜕𝑦 𝑖=1
preloaded laminated plates during impact. It should be noted that the
𝑚 𝑛+1
[ 𝑚
]
∑ ∑ 𝜕 𝜓𝑘 ∑ 𝑖 delamination area is relatively small and locates at the center location
𝛾𝑧𝑥 = 𝜑𝑥 + 𝜑𝑖𝑥 𝜃𝑖 + ( 𝑈 𝑘 𝜃𝑖 ) (4) since the relatively small impact energy of projectile. Therefore, the de-
𝑖=1 𝑘=1
𝜕𝑧 𝑖=1
lamination area of the preloaded laminated plates 3# is reduced by ap-
The typical distributions of interlaminar stresses for both [0/90]8 plying the biaxial in-plane tensile preloads (1000𝜇𝜀) in both the exper-
and [0/90/+45/−45]2S laminated plates after applying the in-plane ten- iment and the simulation.
sile preloads are shown in Fig. 8(c) and (d). Results indicate that all Fig. 9(b) and (d) also show the stiffness degradations of all the co-
the interlaminar stresses are relatively small (close to zero) at the cen- hesive elements on the preloaded [0/90/+45/−45]2S laminated plates
ter location of preloaded laminated plates. Furthermore, 𝜎 z decreases on which different locations are impacted. Different with the case
to the minimum at the location where is 0.75b from the center loca- of [0/90/+45/−45]2S laminated plate 3#, the impact location of
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Fig. 9. Stiffness degradation of the laminated plates; (a) [0/90/+45/−45]2S laminated plate 3# before impact; (b) [0/90/+45/−45]2S laminated plate 3# after
impact; (c) [0/90]8 laminated plate 7# after impact; (d) [0/90/+45/−45]2S laminated plate 9# after near-edge impact.

[0/90/+45/−45]2S laminated plate 9# locates at the weakening area 6. Summary and conclusions
(near-edge region) where the initial stiffness degradations are relatively
higher than those of stiffening area (center location), thus causing the The influence of biaxial in-plane preloads on high-velocity impact
interlaminar stresses play an important role in delamination. On the responses of laminate composite plates was investigated in this work.
basis of stiffness degradation around the impact location, the tolerance Ballistic impact tests were conducted launching steel sphere against
of stiffness degradation in the process of impact for laminated plate 9# both the preloaded and unloaded laminated plates by using a gas gun.
is smaller than that of laminated plate 3#, which results in the reduction Specially, not only the center location but also the near-edge region
of delamination resistance of laminated plate 9#. Moreover, the weak- of preloaded plates was impacted. Meanwhile, numerical simulations
ening effect of interlaminar stresses on the delamination resistance is were performed to gain additional insight into the impact responses of
much stronger than the strengthening effect of increased bending stiff- preloaded laminated plates by employing the revised Ladevèze failure
ness. Therefore, the delamination area significantly increases when im- criterion in ABAQUS. The simulation results showed good agreement
pacting the near-edge region of preloaded laminated plate. with experiments. By using the LWT method combined with the simula-
In general, when impacting the center location on the target plate, tion results, the influence of biaxial in-plane preloads on the delamina-
the weakening effects of interlaminar stresses on delamination resis- tion resistance of laminated plate was discussed. The main conclusions
tance of the preloaded laminated plates are almost negligible, thus caus- were drawn:
ing the increased bending stiffness becomes the dominant factor which The delamination resistance of laminated plate can be influenced by
can influence the delamination resistance of the preloaded laminated the biaxial in-plane tensile preloads under high-velocity impact loading.
plates. However, when impacting the near-edge region on the target On one hand, the applied biaxial in-plane tensile preloads can enhance
plate, the weakening effects of interlaminar stresses on delamination the out of plane bending stiffness of the laminated plate, thus provide
resistance are gradually enhanced, which results in a competition be- a positive effect on the delamination resistance of laminated plate. On
tween the bending stiffening effect and weakening effects of interlami- the other hand, the interlaminar stresses generating in the laminated
nar stresses. Such competing mechanism is considered as the reason of plate when applying the biaxial in-plane tensile preloads tend to result
different delamination area when impacting distinct locations of the tar- in the weakening effect on the delamination resistance of laminated
get plates. It is worthy to mention that the special delamination tendency plate when subjected to high-velocity impact loading. Therefore, the
of laminated plate 7# since its relatively large off-central distance. On influence of the biaxial in-plane tensile preloads on the delamination
plate 7#, the stiffening effect and the weakening effect on delamination resistance of laminated plate is considered as the competition between
resistance nearly cancel each other out, which results in almost the same the increased bending stiffness induced enhancement effect and the in-
delamination area of preloaded plate and unloaded plate. terlaminar stresses induced weakening effect.
C. Wang, T. Suo and C. Hang et al. International Journal of Mechanical Sciences 161–162 (2019) 105012

Acknowledgments [16] Zhang C, Binienda WK, Morscher GN, Martin RE, Kohlmand LW. Experi-
mental and FEM study of thermal cycling induced microcracking in car-
bon/epoxy triaxial braided composites. Compos Part A 2013;46:34–44.
This work was performed under the financial support from the doi:10.1016/j.compositesa.2012.10.006.
National Natural Science Foundation of China (grant Nos.11772268, [17] Littell JD, Binienda WK, William AA, Gary DR, Goldberg RK. Effect of microscopic
11522220, 11272267 and 11527803) and 111 Project (grant No. damage events on static and ballistic impact strength of triaxial braid composites.
Compos Part A 2009;40(12):1846–62. doi:10.1016/j.compositesa.2009.08.001.
B07050) of Northwestern Polytechnical University. [18] Zhang C, Li N, Wang WZ, Binienda WK, Fang HB. Progressive damage simulation
of triaxially braided composite using a 3D meso-scale finite element model. Compos
Supplementary materials Struct 2015;125:104–16. doi:10.1016/j.compstruct.2015.01.034.
[19] Zhao ZQ, Dang HY, Zhang C, Gun JY, Li YL. A multi-scale modeling framework
for impact damage simulation of triaxially braided composites. Compos Part A
Supplementary material associated with this article can be found, in 2018;110:113–25. doi:10.1016/j.compositesa.2018.04.020.
the online version, at doi:10.1016/j.ijmecsci.2019.105012. [20] Chiu ST, Liou YY, Yuan CC, Ong CL. Low velocity impact behavior of
prestressed composite laminates. Mater Chem Phys 1997;47(2–3):268–72.
doi:10.1016/S0254-0584(97)80063-6.
References
[21] Kelkar A, Sankar J, Rajeev K. Analysis of tensile preloaded composites subjected
to low-velocity impact loads. In: Proceedings of the AIAA/ASME/ASCE/AHS/ASC
[1] Chaudhry MS, Czekanski A, Zhu ZH. Characterization of carbon nanotube enhanced structural dynamic, and mater conference and exhibit; 1998.
interlaminar fracture toughness of woven carbon fiber reinforced polymer compos- [22] Nettles A.T. Effects of tensile preloads on the impact response of carbon/epoxy lam-
ites. Int J Mech Sci 2017;131-132:480–9. doi:10.1016/j.ijmecsci.2017.06.016. inates, https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19960012504.pdf;
[2] Heimbs S, Heller S, Middendorf P, Hahnel F, Weisse J. Low velocity impact on CFRP 1995.
plates with compressive preload: test and modelling. Int J Impact Eng 2009;36(10– [23] Zhang X, Davies GAO, Hitchings D. Impact damage with compressive preload
11):1182–93. doi:10.1016/j.compstruct.2009.11.001. and post-impact compression of carbon composite plates. Int J Impact Eng
[3] Johnson AF, Holzapfel M. Influence of delamination on impact dam- 1999;22(5):485–509. doi:10.1016/S0734-743X(99)00003-2.
age in composite structures. Compos Sci Technol 2006;66(6):807–15. [24] Heimbs S, Heller S, Middendorf P, Hahnel F, Weisse J. Low velocity impact on CFRP
doi:10.1016/j.compscitech.2004.12.032. plates with compressive preload: test and modelling. Int J Impact Eng 2009;36(10–
[4] Pagano NJ, Pipes RB. Some observations on the interlaminar strength of composite 11):1182–93. doi:10.1016/j.ijimpeng.2009.04.006.
laminates. Int J Mech Sci 1994;15(8):679–88. doi:10.1016/0020-7403(73)90099-4. [25] Sun CT, Chen JK. On the impact of initially stressed composite laminates. J Compos
[5] Shyr TW, Pan YH. Impact resistance and damage characteristics of composite lami- Mater 1985;19(6):490–504. doi:10.1177/002199838501900601.
nates. Compos Struct 2003;62(2):193–203. doi:10.1016/S0263-8223(03)00114-4. [26] Choi IH. Low-velocity impact analysis of composite laminates under initial in-plane
[6] Chen CY, Zhang C, Zhao ZQ, Wang YP, Shing-Chung Wong, Li YL. Effect of fiber load. Compos Struct 2008;86(1–3):251–7. doi:10.1016/j.compstruct.2008.03.042.
reinforcement and fabrication process on the dynamic. Int J Mech Sci 2019;155:170– [27] Khalili SMR, Mittal RK, Panah NM. Analysis of fiber reinforced composite plates
7. doi:10.1016/j.ijmecsci.2019.02.034. subjected to transverse impact in the presence of initial stresses. Compos Struct
[7] Nezhad HY, Egan B, Merwick F, McCarthy CT. Bearing damage characteristics of 2007;77(2):263–8. doi:10.1016/j.compstruct.2005.08.027.
fibre-reinforced countersunk composite bolted joints subjected to quasi-static shear [28] Pickett AK, Fouinneteau MRC, Middendorf P. Test and modelling of im-
loading. Compos Struct 2017;166:184–92. doi:10.1016/j.compstruct.2017.01.029. pact on pre-loaded composite panels. Appl Compos Mater 2009;16(4):225–44.
[8] Morye SS, Hine PJ, Duckett RA, Carr DJ, Ward IM. Modelling of the energy doi:10.1007/s10443-009-9089-3.
absorption by polymer composites upon ballistic impact. Compos Sci Technol [29] Takeda N, Sierakowski RL, Malvern LE. Wave propagation experiments on bal-
2000;60(14):2631–42. doi:10.1016/S0266-3538(00)00139-1. listically impacted composite laminates. J Compos Mater 1981;15(Mar):157–74.
[9] Takashima H, Miyagai K, Hashida T, Li VC. A design approach for the me- doi:10.1177/002199838101500206.
chanical properties of polypropylene discontinuous fiber reinforced cementi- [30] Heimbs S, Bergmann T, Schueler D, Toso-Pentecote N. High velocity im-
tious composites by extrusion molding. Eng Fract Mech 2003;70(7–8):853–70. pact on preloaded composite plates. Compos Struct 2014;111:158–68.
doi:10.1016/S0013-7944(02)00154-6. doi:10.1016/j.compstruct.2013.12.031.
[10] Johnson AF, Sims GD. Mechanical properties and design of sandwich materials. Com- [31] Ladevèze P, Ledantec E. Damage modeling of the elementary ply
posites 1986;17(4):321–8. doi:10.1016/0010-4361(86)90749-4. for laminated composites. Compos Sci Technol 1992;43(3):257–67.
[11] Hufenbach W, Bohm R, Kroll L, Langkamp A. Theoretical and experimental investi- doi:10.1016/0266-3538(92)90097-M.
gation of anisotropic damage in textile-reinforced composite structures. Mech Com- [32] Turon A, Davila CG, Camanho PP, Costa J. An engineering solution for mesh size
pos Mater 2004;40(6):519–32. doi:10.1007/s11029-005-0022-z. effects in the simulation of delamination using cohesive zone models. Eng Fract Mech
[12] Chen JK, Medina DF. The effects of projectile shape on laminated 2007;74(10):1665–82. doi:10.1016/j.engfracmech.2006.08.025.
composite perforation. Compos Sci Technol 1998;58(10):1629–39. [33] de Moura MFSF, Goncalves JPM. Modelling the interaction between matrix cracking
doi:10.1016/S0266-3538(97)00230-3. and delamination in carbon-epoxy laminates under low velocity impact. Compos Sci
[13] Yew CH, Kendrick RB. A study of damage in composite panels pro- Technol 2004;64(7–8):1021–7. doi:10.1016/j.compscitech.2003.08.008.
duced by hypervelocity impact. Int J Impact Eng 1987;5(1):729–38. [34] Zhuang Z, Zhang F, Cen S. ABAQUS nonlinear finite element analysis and instance.
doi:10.1016/0734-743X(87)90087-X. Beijing: Sci Press Co; 2005.
[14] Riccio A, Ricchiuto R, Saputo S, Raimondo A, Caputo F, Antonucci V, et al. Im- [35] Robbins DH, Reddy JN. Modeling of thick composites using a layerwise laminate
pact behaviour of omega stiffened composite panels. Prog Aerosp Sci 2016;81:41 -8. theory. Int J Numer Meth Eng 1993;36(4):655–77. doi:10.1002/nme.1620360407.
doi:10.1016/j.paerosci.2015.11.004. [36] Pipes RB, Pagano NJ. Interlaminar stresses in composite laminates
[15] Riccio A, Saputo S, Sellitto A, Lopresto V. Characterisation of the impact induced under uniform axial extension. J Compos Mater 1970;34(4):234–45.
damage in composites by cross-comparison among experimental non-destructive doi:10.1016/j.compscitech.2010.05.026.
evaluation techniques and numerical simulations. Proc Inst Mech Eng Part C J Mech
Eng Sci 2017;231(16):3077–90. doi:10.1177/0954406216681595.

You might also like