Composites Part B 109 (2017) 64e71

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Experimental and numerical investigation of the effects of stacking

sequence and stress ratio on fatigue damage of glass/epoxy
composites
Walid Roundi a, *, Abderrahim El Mahi b, Abdellah El Gharad a, Jean-Luc Rebie

re b
a
Department of Mechanical Engineering, Moroccan Laboratory of Innovation and Industrial Performance (LaMIPI), Higher School of Technical Education of
Rabat, (ENSET), Mohammed V University in Rabat, Morocco
b
Maine University, Acoustic Laboratory of Maine University (LAUM), CNRS UMR 6613, Avenue Olivier Messiaen, 72085 Le Mans Cedex 9, France

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

Article history: In the present paper fatigue behavior of glass/epoxy composite materials under different stress ratios and
Received 20 June 2016 for various plies orientation angles has been established. Experimental and numerical analysis (Finite
Received in revised form Element Method) were performed on various specimens subjected to cyclic tensile tests in order to
21 September 2016
outline the influences of stress ratios and the stacking sequence ([02/902]s; [902/02]s; [03/90]s; [903/0]s)
Accepted 16 October 2016
Available online 20 October 2016
on glass/epoxy fatigue properties. Static analysis is done experimentally to identify the stress-strain
diagrams, strain to failure, Young's modulus and tensile strength for each stacking sequence. Compari-
son between experimental and numerical prediction show good agreement. The Results of this inves-
Keywords:
Composite
tigation can lead us to choose the most optimal stacking sequence for giving boundary conditions to
Fatigue achieve the maximum fatigue life.
Stress ratio © 2016 Elsevier Ltd. All rights reserved.
Stacking sequence
Finite element method

1. Introduction interface properties). Many studies [1e3] shows that the fatigue
behavior of composite materials is mainly due to the stiffness
The increased demand for composites is mainly due to their changes during cycles, which can be divided into three stages: the
variant applications in several areas in which lightness, rigidity and first stage is characterized by a rapid decrease in stiffness and the
good resistance to damage are requested. Their use became more formation of ‘‘damage zones’’ with multiple microscopic crack.
and more important in a wide range of structural components After in the second stage the stiffness degradation became lower
which must withstand various types of constant and variable and substantially linear in function of the number of cycles, more
amplitude fatigue loads, such on the construction of several serious types of damage appears such emergence and growth of
frameworks in various areas (buildings, public works and delamination, this stage lasts about 90% of the total life. Finally the
transportations). third stage is characterized by a high level of stiffness degradation
The glass/epoxy material is one of the most used composite, associated with a rapid development of all types of fatigue damage,
especially in aerospace and automotive applications, due to their including fiber/matrix interface failure and fiber fracture leading to
high mechanical performances and low specific weight comparing an accelerated decline. Authors [4e9] have investigated the various
to other materials. In that fact, engineers and research scientists are parameters (ply-drop configuration, number of plies dropped at
required to reconsider fatigue loading as an important factor which one location, stress ratios, laminate thickness, fabric type, loading
may lead to the fracture of this composite. Therefore, mechanical condition, fiber content) influencing the fatigue properties of
properties of glass/epoxy composite were a subject of thorough and composite structures. Brillaud and El Mahi [10] show the significant
extensive studies. However, it's more complex to characterize their influence of stacking sequence on transverse ply cracking in com-
fatigue behavior comparing to metal materials due to their micro- posite laminates subjected to cyclic and repeated fatigue tests. Th.
structural composition (anisotropy, heterogeneity, fiber/matrix Johannesson [11] experimentally investigates the fatigue strain
limit of glass fiber/epoxy composite, in order to determine the in-
fluence of improved ductility of the matrix material on the strain
* Corresponding author. limits. F. Wu and Plumtree [12e14] have studied the damage

http://dx.doi.org/10.1016/j.compositesb.2016.10.044
1359-8368/© 2016 Elsevier Ltd. All rights reserved.
 W. Roundi et al. / Composites Part B 109 (2017) 64e71 65

Fig. 1. Schematic illustration of the specimen dimensions used for static and fatigue tests (in mm).

evolution in a [±45]2S CFRP laminate under block loading condi- description and characterization of effects of the applied stress
tions and damage development in carbon fiber epoxy laminates ratios on the glass/epoxy fatigue proprieties. In this way specimens
during fatigue tests. were submitted to cyclic loading under various stress ratios
In that reason the aim of this study is to highlight at first the (R ¼ 0.1; R ¼ 0.2; R ¼ 0.3; R ¼ 0.4; R ¼ 0.5). Afterward, a finite
influence of the stacking sequence on the fatigue behavior of a element analysis was carried out to predict the fatigue behavior of
glass/epoxy composite submitted to cyclic tensile tests. Specimens specimens under similar conditions used in the experimental part,
have been manufactured with different stacking sequences ([02/ with the same stacking sequences and applied stress ratio values.
902]s; [902/02]s; [03/90]s; [903/0]s), in order to observe the impact of Numerical obtained results will be compared and validate with
this geometric parameter in damage and crack development due to experimental data.
fatigue. The second part of the experimental study will focus on the

Fig. 2. Schematic illustration of the four studied stacking sequences (in mm).
66 W. Roundi et al. / Composites Part B 109 (2017) 64e71

Table 1 Three to five specimens were tested for each test in order to check
Mechanical properties of Glass fiber. the validity of the results. The tests were carried out with a tensile
Glass fiber properties machine equipped with a 100 kN load cell. The strains in the tensile
Areal Tensile Poisson's Strain to Tensile
direction were measured by means of an extensometer with a
density strength ratio failure modulus gauge length of 50 mm. The static tests were performed with a
displacement rate of 1 mm/min. The fatigue tests were performed
300 g/m2 2500 0.25 4.8% 74 GPa
MPa using sinusoidal waveform with load control at a frequency of 10 Hz
for all the tests. Specimens were tested under various maximum
stress levels (smax). The applied stress ratio RF (ratio between the
Table 2 minimum and maximum applied stress: RF ¼ smin/smax) was varied
Mechanical properties of Epoxy resin. from 0.1 to 0.5 (R1 ¼ 0.1; R2 ¼ 0.2; R3 ¼ 0.3; R4 ¼ 0.4; R5 ¼ 0.5). All
the fatigue tests were carried out until the failure of specimens. The
Epoxy resin properties
results allow us to plot the Wo €hler curves for the different com-
Young's Modulus Tensile strength Flexural strength
posites combinations and the fatigue life under the various stress
2900 MPa 74 MPa 310 MPa ratios. Afterward, special interest was focused on the analysis of all
the obtained results to deduce the influence of these parameters on
the fatigue behavior and damage development on the tested
2. Material and manufacturing specimens.

The material considered in this work is a cross-ply laminated

composites which are produced in the laboratory (LAUM) using a 4. Numerical analysis
vacuum bag moulding technique. Composite laminates were pro-
duced by a manual lay-up process from E-glass fibres of weight In the second part, a numerical study was performed with a
300 g m2 and epoxy resin of type SR1500/SD2505. They were finite element method in order to predict the evolution of the
elaborated with following proportion: equal part of fibres and mechanical properties and fatigue life of the glass/epoxy compos-
epoxy resin. Composite plates were cured at room temperature at a ites. Several geometrical models (3D models) for the four stacking
pressure of 30 kPa using a vacuum bagging technique. Before any sequences were created and subjected to similar boundary condi-
tests, the plates were left at room temperature for 2e3 weeks in tions used in the experimental analysis to ensure the validity of the
order to allow complete polymerization of the epoxy resin. The finite element analysis. Applying material (see Tables 1 and 2),
specimens (Fig. 1) were cut using a diamond saw from plates of three-dimensional element meshes and boundary conditions cor-
300  300 mm2. The dimensions of the specimens are: L ¼ 210 mm, responding to maximum loading were generated. Fatigue Analysis
w ¼ 20 mm and th ¼ 2.3 mm, where L, w and th are the length, has three main methods: strain life, stress life and fracture me-
width and thickness respectively. The test specimens are always chanics. In this article, the proposed fatigue study is the strain life
made with 8 layers of glass fibres for the four proposed stacking method: this approach is widely used, it's performed with the finite
sequences which are: [02/902]s, [902/02]s, [03/90]s and [90/03]s elements analysis using a technique known as Rainflow counting
(Fig. 2). Table 1 and Table 2 show the mechanical properties of the [15e17]. The literature offers a multitude of methods of counting
glass fibres/epoxy composite used during experimental and nu- cycles; the Rainflow method is the most used in the field of fatigue.
merical tests. This method defines a stress cycle as a closed hysteresis loop on the
stress-strain diagram i.e. converting the random loading to simple
cycles. The algorithm for this method was developed by Endo and
3. Experimental procedure Matsuishi [18] and allows the application of Miner's rule and strain-
life approach (Eq. (1)) to compute fatigue damage on each cycle (see
At room temperature, the samples of each stacking sequence Fig. 4). Note that the calculation is nonlinear and is solved by iter-
were subjected under uniaxial loading in static and fatigue ac- ative methods. Also note that 1 is the value used for the concen-
cording to the standard test method ASTM D3039/D3039 M (Fig. 3). tration factor, assuming that the mesh is refined enough to capture

Fig. 3. Experimental setup.

 W. Roundi et al. / Composites Part B 109 (2017) 64e71 67

any stress concentration effects [19]. Dowling, Amzallag and

Colombia [20e23] show that Rainflow counting is one of the best
fatigue methods that lead to more precise results. Since the study of
Dowling [20], Rainflow counting is considered the method leading
to best approximation of service life. This method can be found
using different algorithms in the ASTM E 1049-85 standard as well
as AFNOR A03-406.
The Strain Life Relation requires 6 parameters to define the
strain-life material properties: four strain-life parameter properties
and the two cyclic stress-strain parameters. The equation relating
total strain amplitude εa and life (N) is as follows:

s0 f
εa ¼ ð2NÞb þ ε0 f ð2NÞc (1)
E
Where:

 s0 f is the Fatigue Strength Coefficient;

 E is the Young's Modulus;
Fig. 4. Strain-Life approach [24].  b is the Fatigue Strength Exponent, slope of elastic strain line
(Basquin's exponent);
0
 εf is the Fatigue Ductility Coefficient;

Fig. 5. Stress e Strain curves for the different stacking sequences: a) [02/902]s, b) [902/902]s, c) [03/90]s and d) [903/0]s.
 Fig. 6. Static properties: a) Tensile Strength, b) Young's Modulus.

€ hler curves for different stacking sequences (R ¼ 0.1): a) [02/902]s, b) [902/02]s, c) [03/90]s and d) [903/0]s.
Fig. 7. Wo
 W. Roundi et al. / Composites Part B 109 (2017) 64e71 69

Table 3 results), which exceeds the results obtained for the three others
Experimental results of the mean values of fatigue life for the four stacking se- sequences (see Fig. 7). The Gap increase when specimens are sub-
quences (R ¼ 0.1).
mitted to lower cyclic loading. It's also observed that specimens
Applied Stress 130 MPa 175 MPa 260 MPa 305 MPa 350 MPa with the stacking sequences [903/0]s have the lowest fatigue
Sequences strength among the four composites (Table 3).
At applied stresses above the fatigue limit, fracture of specimens
[02/902]s 106 8.7 104 2 103 8 102 5 102
[902/02]s 106 6.2 104 1.5 103 5 102 3 102 was observed. The proposed results show that with decreasing of
[03/90]s 106 2.6 105 1.3 104 8 103 4.3 103 the applied stress, Wo€hler curves obtained from experimental and
[903/0]s 4 103 4 102 0 0 0 numerical fatigue tests for different stacking sequences converge
and the percentage gap will be less. Three fatigue life behaviors
namely low-cycle, high-cycle and infinite life can be distinguished.
 c is the Fatigue Ductility Exponent, slope of the static strain line All predictions of the applied stress versus cycles to failure
(Coffin eManson's exponent). curves are rearranged into the form of Basquin's formula, as
expressed in Eq. (2). Where sa is the applied stress, A and b are
respectively the constant and the curve exponent, N is the number
5. Results and discussions of cycles (Basquin's equations are shown on the curves).

5.1. Static tests

sa ¼ A$Nb 2
The proposed results in Fig. 5 depict the typical tensile stress
versus strain curves obtained for the glass/epoxy composites under Results of the influence of stress ratios on the fatigue behavior of
static loading. It is clearly shown that the stacking sequences glass/epoxy composite demonstrate that the fatigue strength in-
configuration has a significant influence on the static behavior of crease with R values for all fiber orientation angles. However, fa-
the tested composites. Different static properties are obtained for tigue life decreases significantly as the stress ratio was lowered.
each stacking sequence (strain to failure, tensile strength, and More lifespan was observed in specimens tested under a stress ratio
Young's modulus). It can be observed that the tested specimens do R ¼ 0.5, and the lower stiffness is perceived when R ¼ 0.1, which
not fail immediately under maximum loads and keep some residual indicate that specimens will be quickly broken. The predicted fa-
strength until final failure. The glass/epoxy composite [03/90]s tigue behavior under all loading conditions by the finite element
exhibit a higher tensile strength (610 MPa) and around 2.66% for analysis show good agreement with the experimental results,
the strain to failure. However, composite [903/0]s present the considering complex behavior and the bigger scatter of composites.
lowest static properties (tensile strength about 205 MPa) among The Fig. 8 show the result of life prediction of the composite [02/
the four stacking sequences (see Fig. 6). 902]s under smax ¼ 390 MPa, stress ratio R ¼ 0.1 and a frequency of
The results of monotonic tests clearly confirm that the most 10Hz. The lowest lifespan was observed in the elements with red
resistant laminate is the one with more plies (6 plies) in the loading color.
direction i.e. the [03/90]s; in contrast, the least resistant laminate is The fatigue life versus stress ratios level graph (Nr-R) is plotted
[903/0]s with only 2 plies in 0 . For the other two laminates, the according to experimental and finite element analysis (see Fig. 9) to
results of the applied stress are similar although the strain of the show the significant difference of fatigue behavior at various stress
collapse is different. ratios values. Stiffness of the specimens was observed to reduce
under higher fatigue cycles, and damages in the form of delami-
5.2. Fatigue tests nation and fiber rupture starts when specimens are near to the
rupture limit (see Fig. 10). The results of fatigue life obtained with
According to the experimental and numerical data it is notable the finite element analysis are slightly higher than values achieved
that specimens with the stacking sequences [03/90]s have a sig- from the experimental tests, this is due to the process used in the
nificant and more important lifespan. For instance, under an FEM which ignores effects of residual stresses on the specimen's
applied stress equal to 390 MPa, fatigue life goes between 650 and behavior. The gap between the two approaches is less important
940 cycles and around 13.103 cycles under 270 MPa (experimental with increasing of the applied stress ratios.

Fig. 8. Fatigue life under R ¼ 0.1 and smax ¼ 390 MPa for [902/02]s composite.
70 W. Roundi et al. / Composites Part B 109 (2017) 64e71

Fig. 9. Lifetime vs Apllied stress ratios for different stacking sequences: a) [02/902] s, b) [902/02]s, c) [03/90]s and d) [903/0]s.

6. Conclusion composites. For structures under higher applied stress, it's more
adequate to use the [03/90]s fiber orientations to achieve the
Fatigue tests under various stress ratio levels have been inves- maximum fatigue life. This study also highlights the strong influ-
tigated. Fatigue behavior of different stacking sequences has been ence of the stress ratio R on glass/epoxy lifetime, and put in light
compared. The Results show the significant effect of fiber orienta- the harmful impact of the stiffness degradation on fatigue strength.
tion angle on the static and fatigue performances of glass/epoxy Therefore, it's essential to choose the suitable glass/epoxy

Fig. 10. Damage evolution in the glass/epoxy specimens during fatigue tests.
 W. Roundi et al. / Composites Part B 109 (2017) 64e71 71

composite configuration for each application. sequence on transverse ply cracking in composite laminates. Compos Struct
1991;17(1):23e35.
Good agreement with experiment suggests that the glass/epoxy
[11] Johannesson Th. Fatigue strength of glass fiber/epoxy angle.ply laminates with
behavior is being correctly modeled by the numerical finite different matrix ductility. Mech Compos Mater 1993;28(5):439e44.
element analysis. [12] Wu F, Yao W. A fatigue damage model of composite materials. Int J Fatigue
2010;32(1):134e8.
[13] Plumtree A, Melo M, Dahl J. Damage evolution in a [±45]2S CFRP laminate
References under block loading conditions. Int J Fatigue 2010;32(1):139e45.
[14] Talreja R. Damage and fatigue in composites e a personal account. Compos Sci
[1] El Mahi A, Berthelot J-M, Brillaud J. Stiffness reduction and energy release rate Technol 2008;68(13):2585e91.
of cross-ply laminates during fatigue tests. Compos Struct 1995;30:123e30. [15] Downing SD, Socie DF. Simple rainflow counting algorithms. Int J Fatigue
[2] Kashtalyan M, Soutis C. Stiffness degradation in cross-ply laminates damaged 1982;4(1):31e40.
by transverse cracking and splitting. Compos Part A Appl Sci Manuf [16] Olagnon Gue de
 Z. Rainflow fatigue analysis for loads with multimodal power
2000;31(4):335e51. spectral densities. Mar Struct 2008;21(2e3):160e76.
[3] Zhang Y, Vassilopoulos AP, Keller T. Stiffness degradation and fatigue life [17] Marsh G, Wignall C,R, Thies P, Barltrop N, Incecik A, Venugopal V, et al. Review
prediction of adhesively-bonded joints for fiber-reinforced polymer compos- and application of Rainflow residue processing techniques for accurate fatigue
ites. Int J Fatigue 2008;30(10e11):1813e20. damage estimation. Int J Fatigue 2016;82(2):757e65.
[4] Khan R, Alderliesten R, Badshah S, Benedictus R. Effect of stress ratio or mean [18] Matsuishi M, Endo T. Fatigue of metals subjected to varying stress. Jpn Soc
stress on fatigue delamination growth in composites: critical review. Compos Mech Eng 1968:37e40.
Struct 2015;124:214e27. [19] Browell R, Hancq A. The ANSYS fatigue module: calculating and displaying
[5] Botting AD, Vizzini AJ, Lee SW. Effect of ply-drop configuration on delami- fatigue results. 2006.
nation strength of tapered composite structures. AIAA J 1996;34(8):1650e6. [20] Dowling NE. Fatigue failure predictions for complicated stressestrain his-
[6] Andersons J, Hojob M, Ochiaic S. Empirical model for stress ratio effect on tories. J Mater 1972;7(1):71e87.
fatigue delamination growth rate in composite laminates. Int J Fatigue [21] Amzallag C, Gerey JP, Robert JL, Bahuaud J. Standardization of the rainflow
2004;26(6):597e604. counting method for fatigue analysis. Int J Fatigue 1994;16(4):287e93.
[7] Mili F, Necib B. The effect of stacking sequence on the impact-induced damage [22] Colombia P, Dolin ~ skib K. Fatigue lifetime of welded joints under random
in cross ply E-glass/epoxy composites plates. Archive Appl Mech 2009;79(11): loading: rainflow cycle vs. cycle sequence method. Probabilistic Eng Mech
1019e31. 2001;16:61e71.
[8] Bedewi NE, Kung DN. Effect of fatigue loading on the modal properties of [23] Rychlik I, Gupta S. Rain-flow fatigue damage for transformed gaussian loads.
composite structures and its utilization for prediction of residual life. Compos Int J Fatigue 2007;29(3):406e20.
Struct 1997;37(3e4):357e71. [24] Martin JF. Cyclic Stress-Strain behavior and fatigue resistance of two struc-
[9] Rebie
re J-L, Gamby D. Analytical and numerical analysis of transverse cracking tural steels. University of illinois at urbana-champaign; 1973. Fracture control
in a cross-ply laminate. Influence of the constraining effect. Compos Struct program report no.9.
1992;20(2):91e101.
[10] Brillaud J. El Mahi a. Numerical simulation of the influence of stacking