Accepted Manuscript

Buckling, postbuckling and progressive failure analyses of composite laminated plates

under compressive loading

Ömer Namdar, Haluk Darendeliler

PII: S1359-8368(17)30253-6
DOI: 10.1016/j.compositesb.2017.03.066
Reference: JCOMB 5000

To appear in: Composites Part B

Received Date: 20 January 2017

Revised Date: 25 March 2017
Accepted Date: 26 March 2017

Please cite this article as: Namdar E, Darendeliler H, Buckling, postbuckling and progressive failure
analyses of composite laminated plates under compressive loading, Composites Part B (2017), doi:
10.1016/j.compositesb.2017.03.066.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Buckling, Postbuckling and Progressive Failure Analyses of Composite Laminated Plates

under Compressive Loading

Ömer Namdara,*, Haluk Darendelilerb

a Structural Analysis Engineer, ELAN-AUSY GmbH, Channel 2, Harburger Schlossstrasse 24, 21079

PT
Hamburg, Germany, E-mail: [email protected]

b Department of Mechanical Engineering, Middle East Technical University, 06800, Ankara Turkey, E-

RI
mail: [email protected]

* Corresponding Author Tel: +49 (0) 40 28466660, E-mail address: [email protected]

SC
Abstract

U
In this study, buckling, post-buckling and progressive failure of composite laminated plates have
AN
been investigated numerically and experimentally. Buckling load, load-displacement relations for

post buckling and maximum out-of-plane displacements of the plates are determined. Furthermore,
M

the numerical results are compared with experimental findings for two different laminates made of

woven fabric and uni-directional tapes. The comparisons show that there is a good agreement
D

between numerical and experimental results obtained for buckling load and post-buckling behavior
TE

especially for the laminates with uni-directional tapes.

Keywords: Buckling, Post-buckling, Progressive Failure, Composite Structures

EP

1. Introduction
C

Investigation of the buckling and post-buckling behavior of the composite laminated plates is
AC

an important issue to observe their strength and stiffness characteristics since the structural

performance of a composite material depends on its composition, orientation, fiber shape, matrix

and fiber material properties and quality of bondings between fiber and matrix [1]. The critical

buckling loads of the composite structures have been investigated extremely in the literature and

there are many studies about post-buckling behavior of composite laminates [2-15]. The studied

structures were generally subjected to mono-axial compression although other types of loading

1
 ACCEPTED MANUSCRIPT

conditions, such as shear, were considered [19]. The shear contribution to the buckling response of

composite structures was also investigated [20]. Analytical and numerical methods have been

employed for predicting the buckling load and post-buckling behavior of the composite structures

[3,10,18,21,22] and the outcomes such as critical buckling loads, post-buckling behavior and failure

loads have been compared with the experimental results [6,7,9,12-18,22]. The effect of geometric

PT
imperfections on the buckling behavior has also been taken into consideration by carrying out

numerical analyses and experiments [3,11,13-15,23]. Additionally, the investigation of secondary

RI
instability and mode-jumping of composite structures have been performed in several studies

[3,4,17]. In some investigations, finite shell elements have been developed and used [6,12,15],

SC
whereas in most of the analyses commercial finite element programs have been employed by

implementing user-developed subroutines [2,5,9,13,14].

U
In literature, the post-buckling investigations have been performed mostly for the composite
AN
structures which have been manufactured from uni-directional and woven fabric pre-pregs. In

recent studies, the post-buckling behavior and progressive failure analyses of the variable-stiffness
M

composite structures which were produced by tow placement technology have also been studied

[9,13]. The progressive failure analyses have been performed using the continuum damage model
D

and propagation of damage has been correlated with experiments [5,8,9,24-28]. Moreover,
TE

analytical and closed-form solutions of functionally graded plates and beams have been studied for

various types of loading conditions [29-33].

EP

In this paper, buckling, post-buckling and failure characteristics of composite laminated plates

which made by uni-directional and woven fabric laminae have been investigated numerically and
C

experimentally. The composite specimens have been manufactured by using prepreg forms which
AC

consist of thermoset epoxy and carbon reinforcements and the specimens have been configured as

angle-ply symmetric and balanced laminates. Eigenvalue analysis, non-linear Riks and Newton

Raphson methods have been employed to analyze buckling, post-buckling behavior and failure [34].

The progressive failure calculations have been carried out by using the Hashin’s Failure Criterion

[35]. A test fixture was designed and manufactured to determine the large deformation of

composite laminated plates during post-buckling under compressive in-plane loading. The results

2
 ACCEPTED MANUSCRIPT

which had been obtained from numerical analyses and experiments were compared with each

other.

2. Materials

The laminated plates that used in this study had been manufactured by using two different

PT
prepreg laminae. These are AS4/8552 carbon fiber reinforced epoxy prepreg UD tape and

AS4/8552 carbon fiber reinforced epoxy prepreg 5HS Fabric. Both prepreg laminae consist of same

RI
fiber and matrix materials. AS4/8552 UD Tape is a uni-directional fiber reinforced composite which

all fibers are aligned in a single direction. AS4/8552 5HS is a fabric lamina, which the textile

SC
structure is formed by interlaced fibers that are 90° angle with each other [36]. The thickness

values and mechanical properties of each lamina are given in Table 1.

U
AN
3. Experimental Procedure

A number of experiments were performed to evaluate the results of the numerical analyses by
M

using four composite plates; UD-1 and UD-2 made from AS4/8552 UD Tape and FABRIC-1 and

FABRIC-2 made from AS4/8552 5HS Fabric. The stacking sequences and ply thickness values of the
D

manufactured laminated plates are given in Table 2. The stacking of the plies has been selected
TE

such that either of the plates are balanced or symmetric to prevent the secondary instabilities

under compressive load.

EP

The test fixture has been designed and manufactured to investigate the large displacement post-

buckling behavior of composite laminated plates under compressive in-plane loading (Figure 1).
C

The clamped boundary conditions were applied for the top and bottom edges of plates whereas the
AC

side edges of plates are simple supported. To provide the expected boundary condition in the tests,

30 mm length of the specimens from upper and lower edges were stuck in the metallic blocks of top

and bottom fixtures. Furthermore, knife edge supports were located 10 mm inward from the side

edges of the specimens. The supports were arranged such that the dimensions of the part of the

plate under the effect of buckling was 400 × 330 . This configuration was also applied

exactly into the FE model in order to simulate the realistic behavior of the specimen.

3
 ACCEPTED MANUSCRIPT

The composite laminates, which were tested in this study, had been cured at 175 ± 10

temperature and at 680 ± 50 pressure in an autoclave. Total duration of the autoclave

process was about 130-180 minutes. Furthermore, the vessel of autoclave had been purged of

oxygen using an inert gas to prevent thermal combustion or charring of the materials which were

cured [36]. After the curing process, specimens were cut with the help of a CNC milling machine.

PT
4. Numerical Method

RI
The buckling process of composite laminated plates has been analyzed by the commercial finite

element software Abaqus [34]. The critical buckling loads and mode shapes for the first buckling

SC
mode of laminates have been obtained by employing the Linear Eigenvalue Extraction method.

The first buckling mode shapes of the laminates with small amplitudes such as 0.1-5 % of the

U
panel thicknesses have been used as the initial geometric imperfections and implemented into FE
AN
model for the non-linear analyses. Riks and Newton-Raphson methods have been employed for the

non-linear analyses of composite laminated plates to investigate their buckling and post-buckling
M

performances. The progressive failure analyses have been carried out to examine the damages on

the laminates by using Newton-Raphson method since Riks method has long computation time and
D

convergence difficulties with progressive failure option.

TE

To prevent divergence of the computation, to mitigate instabilities and to eliminate rigid body

modes of the FE model, the adaptive automatic stabilization scheme has been employed for the
EP

buckling analyses which have been performed by using Newton-Raphson Method [34]. When the

FE solution goes unstable due to the buckling behavior, large displacements occur and
C

corresponding strain energy releases. Part of the strain energy is dissipated by the relevant scheme
AC

using the damping factor that is determined by means of the dissipated energy fraction. The

convergence is controlled by comparing the energy dissipated by viscous damping and the total

strain energy. The optimal dissipated energy fraction is determined through a trial and error

procedure by evaluating the FE results obtained from Riks method.

The analyses of progressive failure also create some convergence difficulties due to severe

softening behavior and stiffness degradation. Abaqus software provides a scheme called as viscous

regularization, which regulate the tangent stiffness matrix to be positive for small time increments

4
 ACCEPTED MANUSCRIPT

and stabilizes damage evolution problems by introducing viscosity coefficients into the material

model, to prevent divergence during the progressive failure analysis [34].

The built-in boundary condition was applied by restraining all degrees of freedom at the lower

edge of finite element model. At the top edge, only the vertical displacement was allowed and an

incremental vertical compressive load was applied. At both of side edges only the displacement in

PT
the out of plane direction is constrained and the rotation along the vertical edges was permitted to

simulate the knife edge supports.

RI
The composite laminates have been modeled by using quadrilateral shell element; S4R, reduced

integration. The S4R is a finite-membrane-strain element which is defined as general-purpose shell

SC
with 4 nodes and six degrees of freedom [34]. The 460 mm by 350 mm finite element model

consists of 6510 shell elements, and 6674 nodes. The elements are approximately 5 mm x 5 mm in

U
size however the regions of the plates between knife edges and metallic blocks are modeled with a
AN
finer mesh by using an element size of 3 mm x 5 mm.
M

5. Progressive Failure Analysis

The damage analysis used in this study has been based on material stiffness degradation of
D

model. The stress levels, which damage initiations were expected depending on the allowable
TE

strength values of materials in fiber, matrix and shear directions, have been determined by using

Hashin’s Failure Criterion [35] defined as follows:

EP

Tensile Fiber Mode, >0

+ ≥1
C

(1)

<0
AC

Compressive Fiber Mode,

≥1 (2)
!

Tensile Matrix Mode, >0

+ ≥1
"
(3)

Compressive Matrix Mode, <0

5
 ACCEPTED MANUSCRIPT

"!
# − 1& + + ≥1
"! 2 2
(4)

where is the normal stress in the fiber direction, is the normal stress in the transverse to the

fiber direction.

To determine the damage initiation, the stress level which indicates the initiation of material

stiffness degradation, the effective stress tensor '

( is calculated for all of the material points at each

PT
time step. The components of effective stress tensor are assumed to be stress values acting over the

area of a section that still remains undamaged. Hence the effective stress components )*+ are used

RI
in the Hashin’s criterion for *+ and computed from the following relation [34, 37]

SC
'
( = -' (5)

where '
( , ' and the damage operator M are given as

4
1
U 0 0 :
AN
) 351 − 6 7 9
3 1 9
( = . ) 0 , ' = .
' 0 , - = 3 0 0 9
51 − 6 7
̂ 3 9
(6)

3 1 9
0 0
M

2 51 − 6 78

The relationship between the strain tensor, ;, and effective stress tensor, '
( , is given as [34, 37]
D

; = <'
( = <-'
TE

(7)

where < is the compliance matrix. Then taking the inverse of the strain stress equation
EP

' = => ; (8)

where => is the constitutive tensor for the damaged laminae and given as:
C

51 − 6 7A 51 − 6 751 − 6 7B A 0
1
=> = @ 51 − 6 751 − 6 7B A 51 − 6 7A 0 D
AC

?
0 0 51 − 6 7C ?
(9)

In the above equation, ? = 1 − 51 − 6 751 − 6 7B B , A is the Young’s modulus in the fiber

direction, A is the Young’s modulus in the matrix direction, C is the shear modulus and B and

B are Poisson’s ratios. The variable of shear damage, 6 , is calculated using the damage variables

of fiber and matrix as follows [34].

6
 ACCEPTED MANUSCRIPT

6 = 1 − 51 − 6E 751 − 6F 751 − 6E 751 − 6F 7 (10)

The damage variables, 6 , 6 and 6 are determined at each integration point for all plies based

on effective stress levels and take different values for fiber and matrix for both tension and

compression [34,37]. The damage variables, 6E , 6F , 6E and 6F for a definite failure mode are

obtained by using following expression [26, 34].

GI̅ JG̅ − GK̅ L

6= , G̅ ≥ GK̅

PT
G̅JGI̅ − GK̅ L (11)

RI
where GK̅ and GI̅ are the equivalent displacement levels which correspond to initiation of damage

SC
and completely damaged material, respectively, for the related failure mode.

The slope of the equivalent stress, M, versus equivalent displacement, G̅, curve is positive for the

linear elastic material up to GK̅ , then a negative slope is achieved and the evaluation of the damage

U
AN
variables is started for the particular failure mode (Figure 2). Equivalent stress and equivalent

displacement values are calculated by using formulas given below [34].

Fiber tension 5 ) ≥ 07 6 = 6E , G̅ = G̅IE

M

G̅IE = NF O〈Q 〉 + Q (12)

D

〈 〉〈Q 〉 + Q
M IE =
G̅IE ⁄NF
(13)
TE

Fiber compression 5 ) < 07, 6 = 6F , G̅ = G̅IF

G̅IF = NF 〈−Q 〉
EP

(14)

〈− 〉〈−Q 〉
M IF =
G̅IF ⁄NF
C

(15)

Matrix tension 5 ) ≥ 07, 6 = 6E , G̅ = G̅ TE

AC

G̅ TE = NF O〈Q 〉 + Q (16)

〈 〉〈Q 〉 + Q
M TE =
G̅ TE ⁄NF
(17)

Matrix compression 5 ) < 07, 6 = 6F , G̅ = G̅ FE

7
 ACCEPTED MANUSCRIPT

G̅ TF = NF O〈−Q 〉 + Q (18)

〈− 〉〈−Q 〉 + Q
M TF =
G̅ TF ⁄NF
(19)

where NF represents the characteristic length of the element and is determined in accordance with

the element geometry and formulation. For S4R element, it is the square root of element area [34].

PT
The energy dissipated due to failure, C! , which corresponds to area of equivalent stress -

equivalent displacement curve given in Figure 2, has been specified for all failure modes to employ

RI
the damage evolution procedure. The value of C F affects the equivalent displacement, GI̅ .

SC
The energies dissipated for four failure modes of woven fabric and uni-directional tape

composite materials have been calculated by using FEA in this study. The 40 × 100 strips

U
with 0° and 90° material directions and one plied stacking sequence have been modeled to find the

equivalent stress versus equivalent displacement curves for fiber and matrix directions. Most of the
AN
degradation models are based on two approaches; instantaneous unloading and gradual unloading

[38]. In the analyses, the ply-discount theory has been applied to determine energies dissipated for
M

the relevant failure modes, which is a common instantaneous unloading methodology used for

degradation of material properties. Although in order to prevent convergence difficulties, the

D

values of C! have been determined by allowing a deformation which corresponds a small fraction
TE

(5 %) of the initial equivalent displacement (Table 3), the stiffness degradation led to convergence

difficulties during the progressive failure. Therefore, viscous regularization scheme has been
EP

employed to regulate tangent stiffness matrix during the progressive failure analyses [34]. Figure 3

represents the energy dissipation of the matrix compression failure for UD Tape which has been
C

obtained from numerical analysis with viscous regularization scheme. After the damage initiation
AC

was achieved in matrix due to the compression, the material stiffness started to degrade with

respect to specified failure energy for matrix compression (Table 3) during damage evolution

procedure. Therefore, the viscous regularization scheme started to regulate the tangent stiffness of

the material after the sharp stiffness decrease as seen in matrix compression of UD laminate

(Figure 3).

8
 ACCEPTED MANUSCRIPT

6. Results and Discussion

6.1 Comparison of FE results and experiments

Figure 4 compares the experimental results with the finite element solutions obtained from two

unidirectional laminates UD-1 and UD-2 for maximum out-of-plane displacement, respectively.

PT
Finite element analyses have been carried out by using 5 % imperfection. The first buckling mode

shape of the laminates has one longitudinal half-wave and gives the maximum out-of-plane

RI
deflection at midpoint of the plates. The critical buckling load that has been obtained numerically

agrees well with the experimental results for UD-1 whereas buckling of UD-2 was observed earlier

SC
in experiments than determined by the finite element solution (Table 4, Figure 4). Post-buckling

behavior of UD-1 that has been obtained in the experiment is similar with the finite element results

U
up to 45 U load level. However, after the load level of 45 U, the out-of-plane deflections obtained
AN
by the experiments have been found greater than the finite element result. For UD-2 the out of-

plane displacement deviates with a constant amount compared to UD-1. The UD-1 specimen failed
M

at the same load level that was observed in the experiments while the failure load of UD-2 is lower

than the experimental value (Table 4).

D

The first ply failure was observed in the outer layer of UD laminate due to compression as the
TE

compressive stress in the matrix exceeded the allowable limit (Table 1). Figure 5 shows the damage

variable field of matrix compression, ωW , for the mentioned ply. The same result was observed in
EP

the experiments; the damage initiation was first occurred at the upper end of the knife edges and

then propagated towards to center of specimens (Figure 6).

C

The experimental and numerical load vs. maximum out-of-plane curves obtained for two fabric
AC

specimens, FABRIC-1 and FABRIC-2 are given in Figure 7, respectively. For the first buckling mode

shape of FABRIC laminates (one longitudinal half-wave), the maximum out-of-plane deflection is

observed at center of plates in experiments. The critical buckling load, which is obtained from

linear eigen value analysis, agreed well with the experimental results for FABRIC-1 and FABRIC-2

specimens (Table 5, Figure 7). Maximum out-of-plane deflections of FABRIC laminates which have

been obtained from experiments are similar with the finite element results up to a load level of

50 U (Figure 7). Then, the out-of-plane deflections determined by the finite element analyses are

9
 ACCEPTED MANUSCRIPT

found smaller than the experimental results. On the other hand, the failure loads obtained from the

finite element analyses are found less than the experimental results for both FABRIC-1 and FABRIC-

2 (Table 5).

6.2 Numerical Studies

PT
A number of numerical analyses have been realized to evaluate the effect of ply angle

orientations on the composite laminates. Results are presented to investigate buckling loads, post-

RI
buckling behavior and failure characteristics of different laminates. The finite element models of UD

and FABRIC laminates were modified for various ply angles and stacking sequences. Same

SC
boundary conditions, loading type and element types were used in all of the numerical analyses.

The amplitude of initial geometric imperfections was assumed 1% of the panel thicknesses for

these numerical examples.

U
AN
6.2.1 Variation of critical buckling load with Ply Angle
M

The variation of critical buckling load with ply angle have been determined for the UD and

FABRIC laminates having the same stacking sequence, thickness and boundary conditions that were
D

used in the experiments. As shown in Figure 8, ply angle variation does not affect significantly the
TE

critical buckling load of FABRIC laminates due to quasi-isotropic material properties of AS4/8552

5HS Fabric prepreg. However, the critical buckling loads of UD laminates decrease with increasing
EP

ply angle since the stiffness in the fiber direction of AS4/8552 UD tape is greater than transverse

stiffness.
C

6.2.2 Effects of Ply Angle Variation on Buckling for UD Laminates

AC

Effects of ply angle on buckling and failure loads are presented in Tables 6-8 considering

AS4/8552 Carbon Fiber Reinforced UD laminate with 0.184 mm ply thickness for the stacking

sequences [θ/-θ]S, [θ/-θ]2S and [θ/-θ]S4, respectively. Table 6 shows that [0/-0]S UD laminate has

highest buckling load and the buckling load decreases as ply angle increases. The highest failure

load is obtained for [15/-15]S UD laminate, whereas the failure loads of the laminates with the

stacking sequences [0/-0]S, [30/-30]S and [45/-45]S are close to each other. The failure load

10
 ACCEPTED MANUSCRIPT

decreases significantly for ply angles greater than 60 . The decrease of the buckling load with

increasing ply angle is observed also for composite laminates with stacking sequence of [θ/-θ]2S

and [θ/-θ]4S and almost a similar tendency is observed for the variation of failure load as shown in

Tables 7 and 8. The relations between the load and out-of-plane displacement for different ply

angles are presented in Figure 9 for AS4/8552 Carbon Fiber Reinforced UD laminate with the

PT
stacking sequence [θ/-θ]4S. The first buckling mode shapes of laminates with stacking sequences

[60/-60]S, [75/-75]S, [90/-90]S, [60/-60]2S, [75/-75]2S, [90/-90]2S, [60/-60]4S, [75/-75]4S and [90/-

RI
90]4S have two longitudinal half waves while the other laminates have one. The differences

between mode shapes which have one longitudinal half wave and two longitudinal half waves are

SC
shown in Figure 10. On the other hand, for the [45/-45]S and [75/-75]2S laminates that were

observed to buckle in one half wave mode-jump to two half-waves occurred at the loads of 2.9 kN

U
and 11 kN, respectively. First buckling mode shapes of all 16 plied [θ/-θ]S4 UD laminates have one
AN
longitudinal half wave pattern and mode jump is not observed in any stacking sequence.

Tables 9-11 show the effects of ply angle on buckling and failure loads using AS4/8552 Carbon
M

Fiber Reinforced 5HS Fabric laminate with 0.28 mm ply thickness for the stacking sequences [θ/-

θ]S, [θ /- θ / θ /- θ / θ]S and [θ/-θ]4S, respectively. It is seen from Tables 9-11 that, the buckling and
D

failure loads are maximum for 0 and 90 ply angles, and decreases gradually and reaches the

minimum for 45 ply angle. The maximum differences are about 10 % and 30 % for the bucking
TE

and failure loads, respectively. Figure 11 shows the load versus out-of-plane displacement for
EP

different ply angles for [45/-45]4S stacking sequence. Mode-jump occurred for [0/-0]S and [0/-

0/0/-0/0]S laminates from one half wave to two half-waves at the load levels of 4.7 kN and 65 kN,
C

respectively. No mode jump is observed for 16 plied [θ/-θ]S4 FABRIC laminates in any stacking
AC

sequence.

7. Conclusions

This study presents the numerical analyses of buckling, post-buckling and failure of composite

laminated plates built with different carbon-epoxy laminates and stacking sequences which are

validated by experimental results. The buckling behavior of plates that has been determined

experimentally are found quite similar with the FE results that were obtained by using 5 % initial

11
 ACCEPTED MANUSCRIPT

geometric imperfection amplitude. The following conclusions have been acquired by the current

study.

1. The buckling loads observed by the experiments are in good agreement with the numerically

determined values for UD-1, FABRIC-1 and FABRIC-2 specimens. However, UD-2 specimen

predicted a lower value than the experimental one.

PT
2. Until about a load level 50 kN, the post-buckling behavior of the test specimens are quite similar

with the analyses except the UD-2 specimen. For the same load level, out of plane displacements

RI
are found approximately 1 mm greater for the UD-2 specimen compared to UD-1 specimen.

After 50 kN, the results deviate with small amounts from the experimental predictions.

SC
3. Though the FABRIC specimens have failed at lower load levels than the values determined by FE

analyses, failure prediction of the experiments and FE analyses are found close to each other for

both UD and FABRIC laminates.

U
AN
4. In the experiments, after the damage initiation, a sudden decrease was observed in the stiffness

of the specimens.
M

5. Analyses and experiments showed that the studied plates can withstand the loads which are

about four times higher than their critical buckling loads during the post-buckling.
D

6. Numerical studies showed that the critical buckling loads of UD laminates decreases
TE

considerably with increasing ply angle. However, ply angle variation does not affect the critical

buckling load values of FABRIC laminates significantly.

EP

7. Highest post buckling stiffness values for UD laminates are observed at [15/-15] stacking

sequence for [θ/-θ]S, [θ/-θ]2S and [θ/-θ]S4 laminates. Then stiffness decreases with increasing

ply angle giving the minimum at 90 . For FABRIC laminates, it is observed that the post buckling
C

stiffness decreases with increasing ply angle giving the minimum at 45 .

AC

8. Mode-jumps are observed for [45/-45]S UD, [75/-75]2S UD laminates and [0/-0]S FABRIC, [0/-

0/0/-0/0]S FABRIC laminates during the deformation.

9. Mechanical stiffness of UD and FABRIC laminates are found to increase at higher orders than the

increase of laminate thickness.

This study shows that the stacking sequence and the thickness of the laminates have significant

influences on buckling, post buckling and failure performances of the composite laminated

12
 ACCEPTED MANUSCRIPT

structures. Although the laminated structures withstand higher loads during post buckling, the

sudden stiffness loses due to material failure and secondary instabilities must be taken into

consideration.

References

PT
[1] M. C. Y. Niu, Composite Airframe Structures: Practical Design Information and Data, Conmilit

Press, 1992.

RI
[2] U. Farooq and P. Myler, "Finite element simulation of buckling-induced failure of carbon

fibre-reinforced laminated composite panels embedded with damage zones," Acta

SC
Astronautica, vol. 115, p. 314–329, 2015.

[3] R. Vescovini and C. Bisagni, "Two-step procedure for fast post-buckling analysis of composite

U
stiffened panels," Computers and Structures, vol. 128, p. 38-47, 2013.
AN
[4] H. Hofmeyer and J. Courage, "Analytical and finite element modelling of long plate mode

jumping behaviour," Thin-Walled Structures, vol. 73, p. 101–111, 2013.

M

[5] S. Rivallant, C. Bouvet and N. Hongkarnjanakul, "Failure analysis of CFRP laminates subjected

to compression after impact: FE simulation using discrete interface elements," Composites:

D

Part A, vol. 55, p. 83–93, 2013.

TE

[6] D. Kumar and S. Singh, "Stability and failure of composite laminates with various shaped

cutouts under combined in-plane loads," Composites: Part B, vol. 43, p. 142–149, 2012.
EP

[7] L. Boni, D. Fanteria and A. Lanciotti, "Post-buckling behaviour of flat stiffened composite

panels: Experiments vs. analysis," Composite Structures, vol. 94, p. 3421–3433, 2012.
C

[8] D. Xie and S. B. Biggers Jr., "Postbuckling Analysis with Progressive Damage Modeling in
AC

Tailored Laminated Plates and Shells with A Cutout," Composite Structures, vol. 59, p. 199-

216, 2003.

[9] C. S. Lopes, P. P. Camanho, Z. Gürdal and B. F. Tatting, "Progressive Failure Analysis of Tow-

placed, Variable-stiffness Composite Panels," International Journal of Solids and Structures,

vol. 44, p. 8493-8516, 2007.

[10] A. K. Soh, L. C. Bian and J. Chakrabarty, "Elastic/plastic Buckling of A Composite Flat Plate

Subjected to Uniform Edge Compression," Thin-Walled Structures, vol. 38, p. 247-265, 2000.

13
 ACCEPTED MANUSCRIPT

[11] J. Reinoso, A. Blázquez, F. París, J. Cañas and J. Meléndez, "Postbuckling behaviour of a

pressurized stiffened composite panel – Part I: Experimental study," Composite Structures,

vol. 94, p. 1533–1543, 2012.

[12] E. Gal, R. Levy, H. Abramovich and P. Pavsner, "Buckling Analysis of Composite Panels,"

Composite Structures, vol. 73, p. 179-185, 2006.

PT
[13] S. C. White, P. M. Weaver and K. C. Wu, "Post-buckling analyses of variable-stiffness

composite cylinders in axial compression," Composite Structures, vol. 123, p. 190–203, 2015.

RI
[14] C. Bisagni, "Numerical Analysis and Experimental Correlation of Composite Shell Buckling

and Post-buckling," Composites, vol. 31, p. 655-667, 2000.

SC
[15] P. Kere and M. Lyly, "On Post-buckling Analysis and Experimental Correlation of Cylindrical

Composite Shells with Reissner–Mindlin–Von Karman Type Facet Model," Computers and

Structures, vol. 86, p. 1006-1013, 2008.

U
AN
[16] B. G. Falzon and M. Cerini, "An Automated Hybrid Procedure for Capturing Mode-jumping in

Postbuckling Composite Stiffened Structures," Composite Structures, vol. 73, p. 186-195,

M

2006.

[17] B. G. Falzon, K. A. Stevens and G. O. Davies, "Postbuckling Behaviour of A Blade-stiffened

D

Composite Panel Loaded in Uniaxial Compression," Composites, vol. 31, p. 459-468, 2000.
TE

[18] P. Pevzner, H. Abramovich and T. Weller, "Calculation of the Collapse Load of An Axially

Compressed Laminated Composite Stringer-stiffened Curved Panel–An Engineering

EP

Approach," Composite Structures, vol. 83, p. 341-353, 2008.

[19] A. d. P. G. Villani, M. V. Donadon, M. A. Arbelo, P. Rizzi, C. V. Montestruque, F. Bussamra and M.

C

R. Rodrigues, "The postbuckling behaviour of adhesively bonded stiffened panels subjected

AC

to in-plane shear loading," Aerospace Science and Technology, vol. 46, p. 30–41, 2015.

[20] G. Mancusi, L. Feo (2013) “Non-linear Pre-buckling Behavior of Shear Deformable Thin-

walled Composite Beams with Open Cross-section” Composites Part B: Engineering (ISSN:

1359-8368), vol. 47, p. 379-390, 2013.

[21] C. Kassapoglou, "Composite Plates with Two Concentric Layups under Compression,"

Composites, vol. 39, p. 104-112, 2008.

14
 ACCEPTED MANUSCRIPT

[22] Q. J. Yang, B. Hayman and H. Osnes, "Simplified buckling and ultimate strength analysis of

composite plates in compression," Composites: Part B, vol. 54, p. 343–352, 2013.

[23] M. W. Hilburger and J. H. Starnes Jr., "Effects of Imperfections of the Buckling Response of

Composite Shells," Thin-Walled Structures, vol. 42, p. 2004, 369-397.

[24] P. P. Camanho, P. Maimi and C. G. Davila, "Prediction of Size Effects in Notched Laminates

PT
using Continuum Damage Mechanics," Composites Science and Technology, vol. 67, p. 2715-

2727, 2007.

RI
[25] S. Mukhopadhyay, M. I. Jones and S. R. Hallett, "Compressive failure of laminates containing

an embedded wrinkle; experimental and numerical study," Composites: Part A, vol. 73, p.

SC
132–142, 2015.

[26] P. Camanho and C. G. Davila, "Mixed-Mode Decohesion Finite Elements for the Simulation of

U
Delamination in Composite Materials," NASA/TM-2002–211737, p. 1–37, 2002.
AN
[27] P. Maimi, P. Camanho, J. Mayugo and C. Davila, "A continuum damage model for composite

laminates: Part I – Constitutive model," Mechanics of Materials, vol. 39, p. 897–908, 2007.
M

[28] P. Maimi, P. Camanho, J. Mayugo and C. Davila, "A continuum damage model for composite

laminates: Part II – Computational implementation and validation", Mechanics of Materials,

D

vol. 39, p. 909–919, 2007.

TE

[29] R. Barretta, R. Luciano, "Exact solutions of isotropic viscoelastic functionally graded

Kirchhoff plates", Composite Structures, vol. 118, p. 448-454, 2014.

EP

[30] A. Apuzzo, R. Barretta, R. Luciano, "Some analytical solutions of functionally graded Kirchhoff

plates", Composites: Part B, vol. 68, p. 266-269, 2015.

C

[31] R. Barretta, R. Luciano, "Analogies between Kirchhoff plates and functionally graded Saint-
AC

Venant beams under torsion", Continuum Mechanics and Thermodynamics, vol. 27, p. 499-

505, 2015.

[32] R. Barretta, L. Feo, R. Luciano, F. Marotti de Sciarra, R. Penna, "Functionally graded

Timoshenko nanobeams: A novel nonlocal gradient formulation", Composites Part B, vol.

100, p. 208-219, 2016.

[33] R. Barretta, L. Feo, R. Luciano, "Some closed-form solutions of functionally graded beams

undergoing non-uniform torsion", Composite Structures, vol. 123, p. 132-136, 2015.

15
 ACCEPTED MANUSCRIPT

[34] Abaqus Analysis User's Manual Version 6.10, SIMULIA, 2010.

[35] Z. Hashin, "Failure Criteria for Unidirectional Composites," Journal of Applied Mechanics, vol.

47, p. 329-334, 1980.

[36] "Composite Materials," in MIL-HDBK-17, US Department of Defense, 1990.

[37] A. Matzenmiller, J. Lubliner and R. L. Taylor, "A Constituve Model for Anisotropic Damage in

PT
Fiber-composites," Mechanics of Materials, vol. 20, p. 125-152, 1995.

[38] D. W. Sleight, "Progressive Failure Analysis Methodology for Laminated Composite

RI
Structures," NASA/TP-1999-209107, p. 1-67, 1999.

U SC
AN
M
D
TE
C EP
AC

16
 ACCEPTED MANUSCRIPT

Table 1
Properties of each lamina.
AS4/8552 UD Tape Elastic AS4/8552 5HS Fabric Elastic
Properties Properties
Ply Thickness (mm) 0.184 0.28
E11 (GPa) 130.0 61.0
E22 (GPa) 8.5 61.0
G12 (GPa) 4.2 4.2
ν12 0.35 0.05

PT
XT (MPa) 1530.0 647.0
Xc (MPa) 770.0 657.0
YT (MPa) 64.0 647.0
YC (MPa) 175.0 657.0

RI
S (MPa) 95.0 109.0

Table 2

SC
Lay-up and ply thicknesses of specimens
Specimen Stacking Sequence Thickness(mm)
UD-1, UD-2 [45/-45/0/45/90/90/-45/0]S 2.94
FABRIC-1, FABRIC-2 [45/0/45/0/45]S 2.80

U
Table 3
Energies Dissipated due to Failure (N/mm)
AN
Gcft Gcfc Gcmt Gcmc
AS4/8552 UD Tape 49.5 12.5 1.3 9.5
AS4/8552 5HS Fabric 19.7 19.9 19.7 19.9
M

Table 4
Buckling and Failure Loads of UD Laminates
Laminate Buckling Load (kN) Failure Load (kN)
D

UD-1(FE) 22.2 82.9

UD-1 (Experiment) 22.2 84.4
TE

UD-2 (Experiment) 16.2 78.0

Table 5
Buckling and Failure Loads of Fabric Laminates
Laminate Buckling Load (kN) Failure Load (kN)
EP

FABRIC-1 (FE) 16.3 91.8

FABRIC-1 (Experiment) 16.3 77.8
FABRIC-2 (Experiment) 16.3 71.0
C

Table 6
Buckling and Failure Loads of [θ/-θ]S UD Laminates for different ply angles
AC

Laminate Buckling Load (kN) Failure Load (kN)

[0/-0]S 0.41 5.3
[15/-15]S 0.36 6.2
[30/-30]S 0.26 5.6
[45/-45]S 0.22 5.0
[60/-60]S 0.21 4.3
[75/-75]S 0.14 3.5
[90/-90]S 0.12 2.9

21
 ACCEPTED MANUSCRIPT

Table 7
Buckling and Failure Loads of [θ/-θ]2S UD Laminates for different ply angles
Laminate Buckling Load (kN) Failure Load (kN)
[0/-0]2S 3.3 18.7
[15/-15] 2S 3.1 26.0
[30/-30] 2S 2.4 23.2
[45/-45] 2S 2.2 18.5
[60/-60] 2S 2.1 14.6
[75/-75] 2S 1.3 12.4

PT
[90/-90] 2S 0.9 9.6

Table 8
Buckling and Failure Loads [θ/-θ]4S UD Laminates and Specimens for different ply angles

RI
Laminate Buckling Load (kN) Failure Load (kN)
[0/-0]4S 26.0 73.0
[15/-15] 4S 24.7 78.9
[30/-30] 4S 19.7 83.0

SC
[45/-45] 4S 18.0 68.0
[60/-60] 4S 17.1 54.0
[75/-75] 4S 10.4 43.0
[90/-90] 4S 7.4 35.6

Table 9
U
AN
Buckling and Failure Loads (kN) of [θ/-θ]S Fabric Laminates for different ply angles
Laminate Buckling Load (kN) Failure Load (kN)
[0/-0]S 0.95 18.6
[15/-15] S 1.00 18.5
M

[30/-30] S 1.02 16.4

[45/-45] S 0.92 14.3
[60/-60] S 1.02 16.4
[75/-75] S 1.00 18.5
D

[90/-90] S 0.95 18.6

TE

Table 10
Buckling and Failure Loads (kN) of [θ/-θ/θ/-θ/θ]S Fabric Laminates for different ply angles
Laminate Buckling Load (kN) Failure Load (kN)
[0/-0/0/-0/0]S 14.9 101.6
EP

[15/-15/15/-15/15]S 15.6 100.8

[30/-30/30/-30/30]S 15.9 87.1
[45/-45/45/-45/45]S 14.2 75.9
[60/-60/60/-60/60]S 15.9 87.1
C

[75/-75/75/-75/75]S 15.6 99.3

[90/-90/90/-90/90]S 14.9 99.5
AC

Table11
Buckling and Failure Loads (kN) of [θ/-θ]4S Fabric Laminates for different ply angles
Laminate Buckling Load (kN) Failure Load (kN)
[0/-0]4S 60.7 245.1
[15/-15]4S 63.6 238.2
[30/-30]4S 64.8 205.7
[45/-45]4S 57.6 182.6
[60/-60]4S 64.8 205.7
[75/-75]4S 63.6 238.2
[90/-90]4S 60.7 245.1

22
 ACCEPTED MANUSCRIPT

PT
RI
U SC
Figure 1
AN
Equivalent Stress (MPa)

σ0
M
D
TE

0 δ0 δf
Equivalent Displacement (mm)
EP

Figure 2

200
C Equivalent Stress (MPa)

160
AC

Matrix Compression
120

80

Gcmc
40

0
0.0 0.1 0.2 0.3 0.4
Equivalent Displacement (mm)

Figure 3

17
 ACCEPTED MANUSCRIPT

100
FEM (%5 imperfection)
experiment UD-1
80 experiment UD-2

Load (kN)
60

40

PT
20

RI
0
0 5 10 15 20
Out-of-Plane Displacement (mm)

SC
Figure 4

U
AN
M
D
TE

(a) (b) (c)

Figure 5
EP
C
AC

Figure 6

18
 ACCEPTED MANUSCRIPT

100
FEM (%5 imperfection)
experiment FABRIC-1
80 experiment FABRIC-2

Load (kN) 60

PT
40

20

RI
0
0 5 10 15 20 25

SC
Out-of-Plane Displacement (mm)

Figure 7

U
30
AN
25 16 ply UD
10 ply Fabric

20
Load (kN)

15

10
D

5
TE

0
0 15 30 45 60 75 90
Ply Angle
EP

Figure 8

100
Teta=0
C

Teta_15
80 Teta=30
Teta=45
AC

Teta=60
Load (kN)

60 Teta=75
Teta=90

40

20

0
0 5 10 15 20
Out-of-Plane Displacement (mm)

Figure 9

19
 ACCEPTED MANUSCRIPT

PT
(a) (b)

RI
Figure 10

SC
250

200

U
Load (kN)

150
AN
100
Teta=0 & 90
Teta=15 & 75
M

50 Teta=30 & 60
Teta=45

0
D

0 5 10 15 20 25
Out-of-Plane Displacement (mm)
TE

Figure 11

Figure Legends
EP

Fig. 1 Experimental set-up used for UD and FABRIC test specimens

Fig. 2 Eq. Stress versus Eq. Displacement
Fig. 3 Energy Dissipation of Matrix Compression for AS4/8552 UD Tape with Viscous
C

Regularization Scheme
Fig. 4 Load vs. maximum out-of-plane displacement curves of UD specimens
AC

Fig. 5 Matrix Compression Damage Variable ( ) plot of UD Laminate. (a) First ply failure at
maximum load level, P=82.9 kN. (b) Damage evolution of UD laminate of matrix compression at 13
mm out-of-plane displacement and (c) Damage evolution of UD laminate of matrix compression at
14 mm out-of-plane displacement
Fig. 6 Failure in UD-1 Specimen
Fig. 7 Load vs. maximum out-of-plane displacement curves for Fabric Specimens
Fig. 8 Variation of buckling load for [θ/-θ]4S UD laminates and [θ/-θ/θ/-θ/θ]S fabric laminates
Fig. 9 Load versus out-of-plane displacements of [θ/-θ]4S UD laminates
Fig. 10 Buckling mode shapes of UD Laminates (out-of-plane displacement) for (a) [45/-45]4S (b)
[60/-60]4S stacking sequences
Fig. 11 Load versus out-of-plane displacements of [θ/-θ]4S FABRIC Laminates

20