Sei sulla pagina 1di 25

International Journal of Fatigue 70 (2015) 297–321

Contents lists available at ScienceDirect

International Journal of Fatigue


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

Fatigue behavior and modeling of short fiber reinforced polymer


composites: A literature review
Seyyedvahid Mortazavian, Ali Fatemi ⇑
Mechanical, Industrial and Manufacturing Engineering Department, The University of Toledo, 2801 West Bancroft Street, Toledo, OH 43606, USA

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

Article history: Applications of short fiber reinforced polymer composites (SFRPCs) have been rapidly increasing and
Received 7 July 2014 most of the components made of these materials are subjected to cyclic loading. Therefore, their fatigue
Received in revised form 25 September behavior and modeling have been of much interest in recent years. This literature review presents a broad
2014
review of the many factors influencing cyclic deformation, fatigue behavior, and damage development in
Accepted 9 October 2014
Available online 19 October 2014
SFRPCs. These include microstructural related effects as well as effects related to loading condition and
their service environment. Microstructural related effects include those related to fiber length, content
and orientation, surface treatment, and failure mechanisms. Cyclic deformation and softening, viscous
Keywords:
Short fiber polymer composite
characteristics, and dissipative response used to characterize and model their fatigue damage behavior
Thermoplastic and accumulation are discussed. The effects of stress concentrations and their gradient on fatigue behav-
Fatigue ior are also discussed, due to their significant influence. The effects related to the loading condition
Cyclic loading include mean stress effects which may be accompanied by cyclic creep, variable amplitude loading,
Deformation and multiaxial stress effects. Since fatigue behavior is substantially influenced by the testing frequency
with self-heating as the primary consequence of increased frequency, this effect is also investigated. Envi-
ronmental effects considered include the effects of moisture content and temperature, as well as thermo-
mechanical fatigue behavior. The effect of welded joints in manufactured components made of SFRPCs
and fatigue analysis and life estimation techniques used for such components are also included.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction discontinuous fiber composite depends on fiber length and orien-


tation distributions. To maximize mechanical properties, optimum
Industrial application of short fiber reinforced polymer compos- nominal fiber length to diameter ratio should be selected.
ites (SFRPCs) is rapidly growing due to their remarkable properties. Injection molding is a cost effective and high volume processing
The capability to be molded in complex geometries, low manufac- technique for SFRPCs. A fiber weight fraction of 20–40% and vol-
turing cost, high production rate, and significantly low weight to ume fraction of 10–25% is typically utilized, depending on the
strength ratio are among the interesting features of these compos- length of the fibers and requirements. Glass, carbon, and boron
ites. Fiber reinforced polymer composites are fabricated with con- are some typical fiber reinforcements. Among these, glass is the
tinuous or discontinuous fiber reinforcements. Continuous fiber most common fiber and is being employed in almost 95% of fiber
composites have superior strength and stiffness over discontinu- reinforced plastics [1] due to its low cost, high strength, high
ous fiber composites. However, discontinuous fiber composites chemical resistance, and insulating properties. However, glass
are much less expensive to manufacture, are suitable for rapid fibers have low modulus, high density and hardness, and low fati-
and high volume manufacturing, and are more easily reprocessed. gue strength. S-glass and E-glass are the two commonly used glass
Discontinuous fibers with length of about 2 cm are referred to fiber reinforcements. S-glass has a higher content of silica and
as long fiber reinforcements and fibers with approximate length higher tensile strength, modulus, and fatigue performance, as com-
on the order of 1 mm are categorized as short fiber reinforcements pared to E-glass [1]. The advantages of carbon fibers over glass
[1]. Short fiber composites are more suitable for reprocessing com- fibers are their higher strength and modulus, more dimensional
pared with long fiber composites. The mechanical performance of a stability due to their low coefficient of thermal expansion, higher
thermal conductivity and fatigue strength. Low impact strength
⇑ Corresponding author. Tel./fax: +419 530 8213.
is a disadvantage of carbon fibers. Boron fibers have good resis-
E-mail addresses: seyyedvahid.mortazavian@utoledo.edu (S. Mortazavian),
tance to buckling due to their large diameter. The modulus of
afatemi@eng.utoledo.edu (A. Fatemi).

http://dx.doi.org/10.1016/j.ijfatigue.2014.10.005
0142-1123/Ó 2014 Elsevier Ltd. All rights reserved.
298 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Nomenclature

a maximal relative stiffness loss Tg glass transition temperature


b characteristic energy density DT temperature change
c specific heat capacity Win inelastic strain energy density
d fiber diameter W_d dissipated energy density per unit time
D fatigue damage DW strain energy density range
E apparent stiffness Y thermodynamic force
E0 initial stiffness a material damage constant
E0 storage modulus b material damage constant
E00 loss modulus e strain
E⁄ stiffness in the 10th cycle ed strain amplitude
f cyclic frequency em0 mean strain rate
Hs,av average hysteretic energy loss for structural change h fiber orientation angle or stabilized temperature rise
K heat transfer coefficient heq equivalent temperature
Kt stress concentration factor q mass density
L specimen length r stress
lc critical fiber length r1 maximum absolute principal stress
n, N number of applied cycles r1,a normal stress in fiber direction
Q specific heat loss r1,fat longitudinal fatigue strength
R stress ratio r2,fat transverse fatigue strength
Sa, ra stress amplitude r3 minimum absolute principal stress
Sc creep rupture strength ra,n nominal stress amplitude
Sm, rm mean stress rf tensile fatigue limit
Smax, rmax maximum stress rfu fiber tensile strength
SNf fully-reversed fatigue strength rfat fatigue strength
Su, St, rUTS tensile strength rnom nominal stress
DS stress range s12,fat shear fatigue strength
DSED strain energy density range sa,n nominal shear stress amplitude
t time seq characteristic time
T temperature sf shear fatigue limit
tf fatigue lifetime sy shear stress at fiber–matrix interface

boron fibers is above glass and below carbon fibers, although they cessability. PET has high strength and modulus but low ductility.
are more costly than carbon fibers. It is widely used for packaging of materials and it is characterized
Thermoplastics and thermosets are the two main categories of as a low gas and solvent permeable polymer [3,4].
polymeric materials. Cross-linking in between individual mole- In fiber composites, the boundary between fiber and matrix is
cules of thermosets significantly changes their properties, as com- called the interface where a gradient of properties exists. Specific
pared with thermoplastics. Thermosets cannot be melted by heat. mechanical properties can be obtained in the interface by treating
Thermal stability, chemical and creep resistance are some advanta- a subsurface with a silane, also called a coupling agent. Covalent
ges of thermosets and long fabrication time and low ductility are bonds can be formed by reaction of silane, which has an inorganic
their drawbacks [2]. In thermoplastics the polymer chains are not composition, with either the finished polymer or in the copolymer-
chemically attached and they are held by secondary bonds or inter- ization process [5].
molecular forces. Therefore, with heating or cooling, weak second- In automotive applications, under the hood is the harshest envi-
ary bonds can temporary break or restore in a new configuration, ronmental condition for plastic-based materials. During the nor-
resulting in reprocessability of thermoplastics. Higher impact mal working of the engine, the temperature reaches up to 130 °C,
strength and ductility and shorter fabrication time are advantages which might be dropped down to 40 °C in winter. Plastic materi-
of thermoplastics over thermosets [2]. als undergo significant inelastic deformations at high temperatures
Polyamide (PA), polypropylene (PP), and thermoplastic polyes- and, conversely, they behave in a brittle manner with considerable
ters such as polybutylene terephthalate (PBT) and polyethylene increase in strength and stiffness at temperatures below the glass
terephthalate (PET) are the most common thermoplastics, and transition temperature. Addition of fillers such as talc and rubber
epoxy and thermoset polyesters are the most frequently used ther- impact modifiers significantly affects mechanical properties of
mosets. PAs are the most common thermoplastics processed with unfilled (or neat) plastic materials at all temperatures.
short fiber reinforcement. High tensile strength, impact strength Cyclic loading occurs in most engineering applications and the
and modulus, as well as improved bondability and ease of process- great majority of mechanical failures are attributed to this type
ing are of characteristics of PAs. Dimensional instability and reduc- of loading. The cyclic behavior of SFRPCs has been discussed in sev-
tion of properties due to moisture absorption are the main eral studies. Due to the complexity and a high number of parame-
disadvantages of these thermoplastics. PP is a low weight polymer ters which affect the microstructure of these materials, cyclic
with low degree of water absorption. It has good chemical resis- behavior has traditionally been evaluated through experiments,
tance and good insulator properties. One of the main difficulties although empirical relations have also been developed in a few
with PP is its low temperature brittleness where at 0 °C it becomes works.
brittle. PBT has high strength and stiffness, good dimensional sta- Self-heating occurs under cyclic loading and makes character-
bility, low water absorption, chemical resistivity, and good pro- ization of SFRPCs under such loading challenging. Self-heating pri-
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 299

marily depends on the viscoelastic nature of the matrix and is also


related to the frictional heating. Surface contact under crack clo-
sure condition in fatigue crack growth process and friction
between the fiber and the matrix are the main sources of frictional
heating in SFRPCs.
Geometric discontinuities such as holes, sharp corners, and
changes in the cross section area increase the local intensity of
the stress field. The stress field can be either uniaxial or multiaxial
with constant or variable loading amplitude. The aforementioned
conditions are common under service loads and their effects need
to be evaluated to avoid or minimize probability of fatigue failure.
This review aims to compile the information from the literature
on fatigue behavior of SFRPCs. This includes microstructural related
factors, damage development and modeling, loading conditions,
environmental effects, and fatigue of components. Cyclic deforma-
tion behavior of these materials is first discussed. The anisotropy
developed in the material during the fabrication process as well as
fiber length, surface treatment and variation and their effects on
fatigue performance are then considered. Fatigue damage mecha-
nisms and evolution from both macro-mechanical and micro-
mechanical points of view are discussed next. The effect of stress
ratio and mean stress models, as well as cycling frequency effect
on fatigue behavior of SFRPCs are then discussed. Other effects dis-
cussed in this review include those of stress concentrations, accu-
mulation of fatigue damage under variable amplitude loading, and
multiaxial states of stress. This is followed by a discussion of envi-
ronmental effects including humidity or moisture and temperature,
as well as thermo-mechanical fatigue. Fatigue behavior of some
components made of SFRPCs is also briefly reviewed. The emphasis
of this review is on fatigue crack initiation behavior of synthetic
short fiber reinforced polymer composites, mainly processed with
injection molding and mostly used in automotive applications due
to their cost effective and high volume production features.

2. Cyclic deformation behavior Fig. 1. Stress-controlled cyclic deformation response (a) and variation of cyclic
stress–strain curve with cycling (b) for a randomly oriented short glass fiber
In studying the cyclic behavior of materials, it is common to polyester composite [7].
develop a cyclic stress–strain curve. A set of stabilized hysteresis
loops at different strain amplitudes is typically used to establish
Cyclic deformation behavior was also evaluated in [8] through
this curve [6]. Wang and Chim [7] investigated cyclic deformation
conducting cyclic tension–tension stress- and strain-controlled
behavior of short glass fiber polyester composite with random fiber
tests. Due to stress relaxation mechanism, a compressive stress
orientation under tension–tension loading. Due to initiation and
developed in the strain-controlled test, although it was under the
growth of microcracks during cycling, a stable hysteresis loop
tension–tension straining condition. With decreasing the stress
was not reached and continuous cyclic softening was observed,
rate in stress-controlled tests, larger hysteresis loops were observed
as shown in Fig. 1(a). Cyclic stress–strain curves were obtained
due to short term viscosity. Cyclic softening occurred in both stress-
at different cycles, as shown in Fig. 1(b). In matrix-dominant areas
and strain-controlled tests. This phenomenon was attributed to
with less dispersed fibers, cracks were formed normal to the load-
physical sources such as transformation of semi-crystalline matrix,
ing direction. In fiber-dominant areas, cracks were detected in
formation of voids at the fiber ends or matrix, and fiber–matrix deb-
matrix and along the interfacial bonding for aligned and off-axis
onding. Apparent stiffness evolution as a measure of cyclic soften-
fibers, respectively.
ing was correlated with inelastic energy density as:
Launay et al. [8] studied cyclic behavior of a short glass fiber
reinforced polyamide-6.6. Sinusoidal displacement amplitude in a
tensile mode resulting in strain less than 0.1% was applied to the E ¼ E0 ½1  að1  eW in =b Þ ð1Þ
specimens machined in the mold flow direction. Viscoelastic
parameters including the storage modulus and the phase-lag were where E0 is the initial stiffness, E is the apparent stiffness, a is the
recorded for a wide range of frequencies (0–500 Hz) and tempera- maximal relative stiffness loss (%), b is a characteristic energy den-
ture (50 to 250 °C). Viscous characteristics increased as the fre- sity, and Win is the inelastic strain energy density.
quency increased and no dissipative peak was observed. Launay et al. [9] developed a nonlinear constitutive model for
Temperature strongly influenced the viscoelastic behavior with cyclic behavior of short glass fiber reinforced polyamide-6.6 based
two dissipative peaks. The first peak was at the glass transition on the theory of standard generalized materials. This theory is
temperature of the material and the second peak was due to fusion described by the elastic energy density and a dissipation potential
of the crystalline phase at about 250 °C. Cyclic creep and tension with the application in finite strain elastoplasticity [10,11]. Evolu-
relaxation recovery tests were also performed to evaluate long tion of the mechanical response during the fatigue cycles showed
term viscosity effects. The Armstrong–Frederick kinematic harden- that 20 cycles is sufficient for stabilization of some macroscopic
ing law was fitted to the data to predict the residual plastic strains. quantities such as the viscoplastic strain increment and the
300 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

dissipated energy density. The model was validated for different were observed across the thickness of the material, consisting of
loading histories and fiber orientations. surface, shell, and core layers. In the surface layers fiber orientation
Janzen and Ehrenstein [12] characterized cyclic stress and strain was slightly random. Shell layers, as the dominant layers, showed
sensitivity of short glass fiber reinforced polybutylene terephthal- highly oriented fibers in the mold flow direction. A thin core layer
ate specimens with axis in the mold flow direction with fiber con- was also formed which most of its fibers were perpendicular to the
tents ranging between 15 and 44 wt%. Stepwise load-controlled mold flow direction. In axial load-controlled fatigue test of speci-
and strain-controlled tests were conducted under R = 0.01 condi- mens machined perpendicular to the mold flow direction and
tion. Damping was defined as the ratio of the lost or dissipated tested under the R = 0.1 condition, damage mainly developed in
energy (i.e. the area of hysteresis loop) to the strain energy. The the interfacial bonding of the fiber and the matrix, although matrix
dissipated energy mechanisms include viscoelastic deformation, microcracks were dominant at the core layer.
crack formation and grow, and friction and wear. Damping was Zago and Springer [20] investigated axial fatigue behavior of a
plotted against the maximum applied stress and maximum applied short glass fiber co-polyamide with samples machined in the mold
strain in load-controlled and strain-controlled tests, respectively. flow direction. Fatigue strengths normalized by the tensile
At lower stress/strain levels, damping increased linearly with strength were similar for both 2 mm and 4 mm thickness samples,
stress/strain. Increasing the maximum applied stress/strain, damp- as shown in Fig. 3. They collapsed fatigue strengths normalized by
ing increased abruptly at a specific stress/strain level. This level the tensile strength for this material and other SFRPCs from the lit-
was characterized as cyclic stress/strain sensitivity. The stress sen- erature on a master equation, given by:
sitivity increased with fiber weight fraction, but the strain sensitiv-
Sa
ity showed a decrease with increasing the fiber weight fraction. ¼ A  B log Nf ð2Þ
Su
3. Anisotropy effects due to fiber orientation where A and B are constants which depend on R ratio with A = 1.078
and B = 0.133 for the R = 0 loading condition and A = 0.986 and
The injection molding process is widely used to produce B = 0.136 for the R = 1 loading condition.
SFRPCs. During the injecting process, the complex flow field in Yau and Chou [21] evaluated bending fatigue behavior of sev-
the cavity controls the orientation of fibers and introduces anisot- eral randomly oriented short glass fiber reinforced thermoplastics.
ropy in SFRPCs [13]. Factors influencing the orientation distribu- Considered thermoplastics included polyetherimide (PEI), poly-
tion of fibers consist of the geometry of fibers (i.e. fiber aspect ether sulphone (PES), and PEEK and were reinforced by 30 wt%
ratio and cross section geometry), the viscoelastic behavior of short glass fibers. Specimens were tested under the fully-reversed
matrix which depends on glass transition, melt and mold temper- load-controlled condition. Tensile strength of PEI was higher than
atures, and shape or geometry of the component [14]. Orientation PEEK and PES. However, fatigue strength of PEEK was considerably
distribution of fibers is one of the main parameters influencing the higher than PEI and PES composites and no correlation was found
mechanical properties of SFRPCs. between fatigue strength and tensile strength of these composites.
A common technique in evaluating the effect of anisotropy is to Zhou and Mallick [22] studied the effect of specimen orienta-
use test specimens in different locations and at different orienta- tion on axial fatigue of short glass fiber reinforced polyamide-6.6.
tions with respect to the injection molding flow direction. Anisot- Stress-controlled tests were conducted under R = 0.1 condition on
ropy can also be evaluated by characterizing the orientation both longitudinal and transverse specimens. Fatigue strength in
distribution of fibers throughout the thickness. Vélez-García [15] the transverse direction was about 50% lower than that in the lon-
evaluated the orientation distribution of fibers in injection molded gitudinal direction, as shown in Fig. 4.
short glass fiber polymers using surface imaging techniques. A Cosmi and Bernasconi [23] conducted an experimental study on
detailed sample preparation procedure was developed to avoid dis- axial fatigue behavior of edge-notched specimens made of short
turbing the fibers and to obtain high quality images from polished glass fiber polyamide-6. Load-controlled fatigue tests were carried
surfaces. out under the R = 0.1 condition with mold direction longitudinal
Horst and Spoormaker [16] evaluated axial load-controlled fati- and transverse to the specimen axis. The longitudinal specimens
gue behavior of a short glass fiber polyamide-6. Tension–tension showed higher fatigue strength than the transverse specimens, as
tests were performed under the stress ratio of R = 0.1 with speci- shown in Fig. 5. The cell method [24] was applied over a 5 mm por-
mens in the mold flow direction. They found the fatigue strength tion of specimen gage section in the vicinity of the notch to
to depend on the location of the specimen in the injection molded develop a relationship between the fiber orientation and the elastic
plaques. Fatigue strength increased about 50% from the middle to material properties. Material properties in each cell of the mesh
the edge of the rectangular plaques. This effect was attributed to were scaled based on properties of fiber and matrix and elastic
different orientation distributions of fibers in the edge and middle material properties were computed by simulation.
of the injection mold plaques. A shell and core morphology exists Guster et al. [25] studied axial fatigue behavior of a composite
in injection molded SFRPCs where the mid-thickness (core layer) made of short glass fiber reinforced polyamide-6T/6I using stress-
is between two near the wall layers (shell layers). In the core layer controlled tests under the R = 0.1 condition on longitudinal and
the majority of fibers are oriented perpendicular to the mold flow transverse specimens at several temperatures. Fatigue strength in
direction, while in the shell layers fibers are mainly oriented in line the transverse specimen was significantly lower than the fatigue
with the mold flow direction, as seen in Fig. 2 from [17]. It was also strength of the longitudinal specimens at each temperature. The
observed that the core layer is thinner in specimens taken from the slopes of the S–N curves were also steeper in the transverse sam-
edge of a plaque compared to specimens taken from the middle. ples, as compared to the longitudinal samples at each temperature,
The formation of core layer is due to uniform profile of velocity as can be seen from Fig. 6.
in this layer and no shear orientation which results in displace- Brunbauer et al. [26] studied effects of fiber orientation on ten-
ment of polymer melt perpendicular to the mold flow direction sile and bending fatigue behaviors of short glass fiber reinforced
[18]. It was found that, regardless of the specimen location, the polyamide-6.6 with specimen axis in 0° and 90° directions with
fatigue strength is directly proportional to the tensile strength. respect to the mold flow direction. As specimen angle increased
Arif et al. [19] employed the X-ray micro-computed tomogra- from 0° to 90° and temperature increased from 23 °C to 150 °C, fati-
phy technique to evaluate the fiber orientation distribution of gue strength decreased. Specimen with a gage length of 80 mm
short glass fiber reinforced polyamide-6.6. Three sets of layers showed higher fatigue strength as compared with specimen with
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 301

Fig. 2. SEM micrograph of fracture surface (a) with higher magnification of shell layer (b) and core layer (c) for a short carbon fiber composite [17].

Fig. 3. Stress amplitude normalized by tensile strength versus cycles to failure for a
short glass fiber co-polyamide using 2 mm and 4 mm thickness specimens
machined in the mold flow direction and tested under R = 0 condition [20].

20 mm gage length, due to more fiber alignment in load direction.


Slope of S–N curves was found to be independent of temperature,
Fig. 4. Effect of the specimen orientation and weld line on fatigue behavior of short
specimen angle and geometry. Due to the stress gradient effect, glass fiber polyamide-6.6 tested at room temperature under R = 0.1 loading
flexural fatigue strength was 40% higher than tensile fatigue condition [22].
strength.
Wang et al. [27] studied the effect of specimen size on axial
R = 0.1 load-controlled fatigue behavior of random short glass fiber 90° directions. Material anisotropy was not evident for the 3 mm
epoxy composite. Specimens of nine different sizes with test sec- thickness samples, whereas by increasing the specimen off-axis
tions ranging from 7.5 to 22.5 cm in length and 1.25 to 5 cm in angle, a significant fatigue strength reduction was observed in
width were tested. Shorter and wider specimens with length to the 1 mm thickness samples. This effect was observed for both
specimen width ratio of equal or less than three provided higher fully-reversed and R = 0 conditions, as shown in Fig. 7 for the
fatigue strength. R = 0 condition. The anisotropic material response was found to
De Monte et al. [28] investigated load-controlled axial fatigue be less pronounced under cyclic loading rather than quasi-static
behavior of a short glass fiber reinforced polyamide-6.6. Tension– loading. This effect was even more significant with increasing the
tension loads were applied under R = 1 and R = 0 conditions to specimen thickness. The authors also explain the possible cause
1 mm and 3 mm thickness specimens machined in 0°, 30°, and of this phenomenon to be that in the high cycle regime, damage
302 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Fig. 5. Effects of the injection molding direction on fatigue behavior of notched


short glass fiber polyamide-6 specimens under R = 0.1 loading condition [23].

Fig. 6. Effects of testing temperature and specimen orientation on fatigue behavior Fig. 7. Effect of the specimen orientation on fatigue behavior of 1 mm thickness (a)
of a short glass fiber polyamide-6T/6I obtained from tests under R = 0.1 loading and 3 mm thickness (b) specimens of short glass fiber polyamide-6.6 tested at room
condition [25]. k1 values on the graph are reciprocals of the slope of S–N lines. temperature under R = 0 loading condition [28].

process is controlled by the weakest region across the thickness determined by the S–N curves for specimens taken in 0° and 90°,
(i.e. the core layer). As a result, the entire cross section does not respectively, and s12,fat(N) was obtained by minimizing the devia-
uniformly influence the damage process under cyclic loading. tion of calculations for 30° or 60° fatigue strengths utilizing Eq.
However, Bernasconi et al. [29] found a significant anisotropy (3), as seen in Fig. 8(c). De Monte et al. [28] also fitted the Tsai–Hill
effect in 3.2 mm thickness specimens made of short glass fiber criterion to their fatigue data of short glass fiber reinforced polyam-
reinforced polyamide-6 (see Fig. 8(a)). ide-6.6 tested at three different fiber orientations with respect to
Bernasconi et al. [29] performed an experimental study on axial the mold flow direction.
fatigue behavior of short glass fiber reinforced polyamide-6 using
load-controlled fatigue tests under R = 0.1 condition. Specimens
were machined in different directions relative to the mold flow 4. Fiber length, surface treatment and variation effects
direction (0°, 30°, 60°, and 90°). The fatigue strength decreased with
increasing the specimen angle and the S–N curves were nearly par- In short fiber composites, the applied load transfers from the
allel for different orientation angles, as seen from Fig. 8(a). Fatigue matrix to the fibers through the fiber ends and the fiber–matrix
strengths were proportional to the corresponding tensile strength interface. The transferability of the load from the matrix to the
values. A good correlation of data at different mold flow directions fiber strongly depends on the fiber length. The critical fiber length,
was obtained by normalizing with the tensile strength, as can be lc, at and beyond which the maximum fiber stress is achievable is
seen from Fig. 8(b). They suggested the use of Tsai–Hill criterion defined as [30]:
[30] to relate fatigue strength to the specimen orientation: rfu d
lc ¼ ð4Þ
"
2 4 2
#12 2sy
cos2 ðhÞðcos2 ðhÞ  sin ðhÞÞ sin ðhÞ cos2 ðhÞ sin ðhÞ
rfat ðhÞ ¼ þ 2 þ ð3Þ
r21;fat ðNÞ r2;fat ðNÞ s212;fat ðNÞ where rfu is the tensile strength of the fiber, d is the diameter of the
fiber, and sy is the shear stress at the cylindrical fiber–matrix inter-
where r1,fat(N), r2,fat(N), and s12,fat(N) are the experimental fatigue face. For composites with fibers shorter than the critical length, the
strengths for a specimen life of N cycles. r1,fat(N) and r2,fat(N) were fiber may not be stressed but concentrate the local stress in the
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 303

age fiber lengths in the ranges of 0.16–0.28 mm and 1.18–1.37 mm.


For these composites, the critical fiber length is in the range of 0.6–
1.1 mm. For each of the considered materials, in LCF regime, the
longer fiber composite displayed superior axial fatigue strength
than the shorter fiber composite, since the stress level was high
enough to induce fiber breakage. However, in HCF regime, axial fati-
gue strengths of short and long fiber composites were the same,
because under low stresses the failure is mainly characterized by
matrix or fiber–matrix debonding. Influence of the fiber length on
flexural fatigue behavior was observed in both low and high cycle
regions, with superior fatigue strength for the longer fiber compos-
ites. Under the flexural fatigue condition, various levels of fiber
breakage and fiber debonding were observed due to tensile and
compression stresses at the outer regions of the specimens.
Lavengood and Gulbransen [32] evaluated the effect of fiber
aspect ratio on three point flexural fatigue behaviors of a short
boron fiber epoxy composite stressed in the mold flow direction.
Fatigue life was shown to be strongly dependent on the fiber aspect
ratio with fatigue life rapidly increasing with increasing the aspect
ratio up to of 200, and then reaching a plateau, as shown in Fig. 9.
Matrix crazing was found to originate at the fiber ends and grow
along the fiber interface. In addition to crazing, fiber brittle failure
in 45° direction with respect to fiber length was also observed.
Meneghetti et al. [33] studied the effect of fiber length on axial
fatigue behavior of short glass fiber reinforced polypropylene (PP).
Fully-reversed load-controlled fatigue tests were conducted with
two different nominal fiber lengths of 1 mm and 10 mm. The mea-
surement of fiber length was performed after molding process by
using a metrological computed tomography system and average
fiber lengths of 0.72 and 3.43 mm were reported. Degradation of
fiber length after processing was attributed to interaction between
fibers and matrix, screw, barrel, nozzle, runners and gates. Fatigue
strength of the short fiber PP was higher than the long fiber PP due
to more scattered fiber orientation for longer fiber specimens.
Scanning electron microscopy indicated fiber–matrix deboning as
the predominant fatigue damage mechanism for both the short
and the long fiber composites.
Subramanian and Senthilvelan [34] investigated the effect of
fiber length on hysteresis heating of a short glass fiber polypropyl-
ene composite. Axial displacement-controlled tests were carried
out under R = 0 condition in the mold flow direction. Two compos-
ites with average fiber lengths of 0.28 and 1.38 mm were considered.

Fig. 8. Effect of specimen orientation on fatigue behavior (a), fatigue data


normalized by tensile strength (b), and Tsai–Hill criterion and fatigue data as a
function of fiber orientation angle (c) for 3.2 mm thickness samples of short glass
fiber reinforced polyamide-6 under R = 0.1 loading condition [29].

matrix. The initial length of fibers can be reduced during the mold-
ing process depending on the chosen fiber type and length, the fiber
volume fraction, the geometry of mold, and the processing condi-
tion. Grove and Kim [31] conducted an experimental study on axial
and flexural fatigue behaviors of a short glass fiber polyamide-6 and
Fig. 9. Fatigue life versus fiber aspect ratio for applied stresses of 60%, 70%, 80% and
a short glass fiber polypropylene with fibers in the mold flow direc- 90% of the tensile strength for a short boron fiber epoxy composite with specimens
tion. Each of these materials was fabricated with two different aver- machined in the mold flow direction [32].
304 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

For neat thermoplastics, the temperature rise is only due to the hys- and glass and carbon fiber PA6.6 composites collapsed on a single
teresis heating in the matrix. For reinforced thermoplastics, the curve (see Fig. 10) when normalized with tensile strength. Fatigue
presence of fibers and fiber–matrix interfaces increases the gener- fracture surface of glass fiber PEEK composite indicated a poorly
ated heat in composites due to stress concentration effect at the fiber bonded system. S–N data of this composite converged to the fati-
ends and friction between fiber and matrix. In a fixed volume of gue strength of neat PEEK in HCF regime, but significant difference
material, a composite with lower average fiber length has larger in LCF regime was observed. Converge of fatigue strength in HCF
number of fiber ends as compared to a composite with higher aver- was attributed to loss of glass fiber reinforcements due to the
age fiber length. As a result, higher temperature rise due to the fric- breaking of interfacial bonding. Polysulfone matrix and its glass
tional heating was observed in the composite with shorter fiber and carbon reinforced composites showed a transition in S–N
length, compared to the composite with longer fiber length. curves due to different failure mechanisms at high and low stress
Takahara et al. [35] studied fully-reversed axial strain-con- levels. In HCF and LCF regimes, failure mechanism was character-
trolled fatigue behavior of a neat and 30 wt% randomly oriented ized by crazing and matrix yielding, respectively.
short glass fiber reinforced polybutylene terephthalate. The effect In the study of Janzen and Ehrenstein [12] on R = 0.01 fatigue of
of fiber surface modification was evaluated utilizing a silane cou- short glass fiber reinforced polybutylene terephthalate specimens
pling agent. Surface modified glass fiber composite displayed with axis in the mold flow direction mentioned earlier, effects of
higher fatigue strength than the unmodified glass fiber composite fiber content was studied. LCF strength increased with increasing
by 15%. This effect was attributed to the strengthening of weak the fiber weight fraction from 15 to 44 wt%. However, in HCF
interfacial bonding in the surface unmodified glass fiber composite. regime, fatigue strengths were nearly the same for different fiber
The generated heat under fatigue loading was always higher in the contents. The ratio of shell to core layer thicknesses were 7.3,
composite with surface unmodified glass fibers compared to the 5.6, and 3 for 15, 27, and 44 wt% fiber, respectively. As the fiber
composite with surface modified glass fibers. This was attributed content increased the fiber alignment in shell layers increased.
to more fiber–matrix debonding resulting from the concentration Fotouh and Wolodko [38] studied axial load-controlled fatigue
of the stress near surface unmodified fibers. behavior of short hemp fiber with high density polyethylene com-
Yamashita et al. [36] studied effect of fiber surface treatment posite. Specimens axis was in the injection molding direction and
consuming an amino silane coupling agent on fully-reversed axial tests were performed under R = 0.1 condition. Increasing the fiber
strain-controlled fatigue strength of short glass fiber reinforced content from 0 to 40 wt%, the S–N line slope was nearly constant
polyamide-6. It was concluded that the short glass fibers with and fatigue strength improved slightly. Hence, fatigue behavior
unmodified surface act as stress concentrators and reduce the fati- was matrix dominant and S–N curves were normalized by the ten-
gue life of the composite. sile strength. An empirical fatigue life estimation model was devel-
Mandell et al. [37] investigated effects of matrix and interface oped considering the effects of frequency, stress ratio, fiber volume
properties on axial fatigue behavior of short glass fiber and short fraction, and moisture.
carbon fiber thermoplastic composites. Performances of a brittle
matrix system, polyphenylene sulfide (PPS), and two ductile matrix
systems, polyamide-6.6 (PA 6.6) and polyether ether ketone 5. Micro-damage mechanisms
(PEEK), were studied. Specimen axis was in the mold flow direction
and they were tested under R = 0.1 load-controlled condition. In Under axial or bending fatigue loading, similar fatigue damage
glass fiber PPS composite with good interfacial bonding, behavior mechanisms have been reported in different studies. Matrix rup-
was fiber dominated, while in carbon fiber PPS behavior of com- ture, fiber debonding, and crazing are the most common fatigue
posite was dominated by matrix. They also concluded that in a failure mechanisms in SFRPCs. These mechanisms cause initiation
ductile matrix, yielding or necking of the matrix is the major failure and growth of micro- and macro-cracks prior to fracture.
mechanism and, thereby, ductile matrix is dominant in fatigue Zhou et al. [39] conducted an experimental study on four-point
behavior of composites with perfect interfacial bonding, such as bending fatigue behavior of short glass fibers with a blend of poly-
PA6.6, with either glass or carbon fibers. S–N fatigue data of neat phenylene ether ketone and polyphenylene sulfide. Specimens
with axis along the fiber direction were tested under R = 0.1, 0.3,
and 0.5 load-controlled conditions. In S–N curves, a shift in the
slope between low and high cycle regimes was observed. This shift
was attributed to a transition in the fatigue failure mechanism. At
high stress levels, failure was mainly due to matrix rupture or fiber
debonding, whereas at low stress levels crack propagation caused
the failure. No influence of frequency was observed in the fre-
quency range of 0.89–7 Hz.
In the study of Horst and Spoormaker [16] on axial load-con-
trolled fatigue testing of a short glass fiber reinforced polyamide-
6 under the R = 0.1 condition on specimens in the mold flow direc-
tion mentioned previously, material behaved in a ductile manner
due to 1.5% absorbed water after about one year. Under cyclic load-
ing, fiber ends debonding occurred due to the presence of stress
concentrations and shear stresses at the fiber ends. As a result of
unloading of fiber ends, load increases on the matrix. Because the
lateral contraction of matrix is restricted due to the bonding with
surrounded fibers, the interface undergoes tensile stresses. There-
fore, they concluded the tensile strength of the interface is of
importance, in addition to the shear strength of the interface.
Fig. 10. Normalized S–N data for an unreinforced polyamide-6.6, a short glass fiber
Casado et al. [40] assessed fatigue damage mechanism of a short
polyamide-6.6, and a short carbon fiber polyamide-6.6 obtained from tests under glass fiber reinforced polyamide-6.6 through surface roughness
R = 0.1 loading condition at frequencies of 1–2 Hz [37]. evolution. Specimens were machined in the mold flow direction
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 305

and tested under axial load-controlled tension–tension condition. common changes that occur in composite materials. The Wang
Evolutions of both the deformation rate and surface roughness and Chim study mentioned earlier [7], investigated cyclic degrada-
were recorded throughout the test. Three different stages were tion in random short glass fiber polyester composite. Continuing
identified. Stage I was characterized by a decrease of the strain rate the cyclic loading caused crack initiation and growth and conse-
with no development of crazing. A constant strain rate along with quently increase in the size of hysteresis loops (see Fig. 1(a)). Based
uniform development of crazes and an increase of strain rate with on the experimental results, a damage parameter was introduced
formation of the main crack were considered as stages II and III, to quantify the fatigue damage as:
respectively. Fatigue damage evolution was quantified by measur-
ing the degree of specimen surface roughness. As the stress ampli- E
D¼1 ð5Þ
tude decreased, the level of roughness increased due to increased E
craze formations at lower stress levels. A linear proportional rela-
where E⁄ is the stiffness of the 10th cycle and E is the stiffness in
tionship between the temperature rise and the surface roughness
cycle N. A linear relationship between the rate of damage and fati-
was also observed.
gue life was obtained in log–log scale, expressed as:
Hitchen and Ogin [41] investigated fatigue damage mechanism
of a mold flow oriented short glass fiber reinforced polyamide-6 in dD
axial load-controlled tests under the R = 0.1 condition. The core– ¼ AðDÞNB ð6Þ
dN
shell morphology was clearly visible throughout the 1 mm speci-
men thickness where the shell comprised 80% of the thickness of where B is the slope of the curve which was found to have a poly-
the specimen. Samples containing both shell and core layers nomial relationship with maximum stress, and A is related to the
showed a modulus reduction throughout the test. A lower modulus damage rate in a reference state during cycling. Fatigue life is then
reduction was revealed for samples machined from the shell layer. obtained by integrating Eq. (6).
It was, therefore, concluded that the overall modulus reduction is a In the study of De Monte et al. [28] mentioned earlier, stiffness
consequence of damage in both layers. Formation of cracks was degradation in a short glass fiber reinforced polyamide-6.6 was
evident in the core layer, but no such cracks were observed in also used for fatigue damage evolution. Under the fully-reversed
the shell layers. These cracks were perpendicular to the loading loading condition, variations of tensile and compressive moduli
direction and propagated to the core–shell boundary, where they with applied cycles are different. This is due to more damage
stopped. The stop of crack growth in the boundary was related to occurring in tensile mode than in compression. They also mention
the oriented fibers in shell layers restricting the growth of cracks. that, in some cases, the hysteresis loops cannot fully characterize
Microscopy of the shell layers showed a limited number of frac- the fatigue damage progression because fatigue is a local phenom-
tured fibers. Since unreinforced polyamide-6 did not show any enon and the hysteresis loops are measured in the middle part of
stiffness reduction under cyclic loading, it was concluded that fiber the specimen.
fracture or matrix failure could not be the cause of damage evolu- Wang et al. [43] developed a constitutive equation for effective
tion in shell layers. The accumulation of damage in this layer was elastic properties of the damaged material based on self-consistent
attributed to fiber–matrix debonding. mechanics theory, which is an approximate method for treating an
Kabir et al. [42] proposed that short fiber composite fatigue fail- inhomogeneous system. In determining the effective elastic prop-
ure occurs as a result of the accumulation of statistically distrib- erties of the damaged material, probability density functions for
uted flaws in both fiber and interfacial bonding. They suggested a crack length and orientation are used. The effective elastic proper-
fatigue damage model based on the statistical microscopic Weibull ties are then substituted in Eq. (5) to obtain the fatigue damage
damage law. The evolution of damage was quantified and com- parameter. A good agreement between the theory and experiments
pared with the experimental data of a randomly oriented short was obtained.
glass fiber reinforced polypropylene. Even though matrix cracking Wang et al. [44] evaluated shear fatigue degradation in random
was not considered as a failure mechanism, good agreement was short glass fiber reinforced polyester. Load-controlled tests were
found between the model and the experimental results. performed on a double V-shaped notched specimen under pure
In the study of Brunbauer et al. [26] on effects of fiber orienta- shear fatigue loading. As the number of cycles increased, the slope
tion on R = 0.1 axial and bending fatigue behavior of short glass of hysteresis loops decreased and their size enlarged. A shear fati-
fiber polyamide-6.6 mentioned earlier, matrix brittle failure and gue damage parameter was then developed, similar to Eq. (5), by
fiber pull-out were observed in mold flow oriented samples, substituting the shear modulus for the Young’s modulus.
whereas in samples perpendicular to the mold flow direction brit- Klimkeit et al. [45] evaluated the fatigue damage propagation in
tle fracture surface was observed. Also, as temperature increased, a a short glass fiber reinforced with a blend of polybutylene tere-
more crazed and fibril structure was observed. phthalate and polyethylene terephthalate. Specimens were
machined in the longitudinal and transverse directions. Axial
load-controlled fatigue tests were conducted under R = 0.1 and
6. Fatigue damage development and modeling
R = 1 conditions. In the longitudinal direction, the evolution of
hysteresis loops at R = 1 was nearly unaffected by the fatigue
Various mechanical parameters based on dissipative response of
loading, although a small drop of stiffness was observed in the
SFRPCs have been used to characterize and model damage accumu-
transverse direction. In the case of R = 0.1 loading, the hysteresis
lation and fatigue life estimation under cyclic loading. Progressions
loops translated along the strain axis, reflecting considerable cyclic
of the hysteresis loop, strain energy, nonlinear viscoelasticity, and
creep, which was more pronounced in the transverse direction.
hysteretic energy dissipation are the typical mechanical parameters
Cyclic stiffness decreased, while the mean strain and damping
used.
increased with continued cycling. They observed that the fatigue
life was not governed by the propagation of a single macroscopic
6.1. Evolution of hysteresis loop crack, as a single crack appeared in the last 5% of the total life.
Micro-cracking at fiber ends and around the fibers were predomi-
Changes in the shape of the stress–strain hysteresis loop indi- nant failure mechanisms observed in the material. These mecha-
cate a thermodynamically irreversible damage in the material. nisms result from the stress concentrations at the fiber ends and
Evolutions of either the size or the slope of hysteresis loop are the spatial distribution of fibers in the matrix.
306 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Hour and Sehitoglu [46] studied the fatigue damage develop-


ment of a short glass fiber composite with a mix of polyester, cal-
cium carbonate, and some other additives. Axial strain-controlled
tests were conducted under R = 0 condition and axial load-con-
trolled loading conditions were used under R = 0 and R = 0.46
conditions. Significant scatter was observed in strain-controlled
tests, compared to the load-controlled tests. This was attributed
to stress relaxation in the first few cycles of strain-controlled tests,
the level of which varied in different tests under the same loading
conditions. The evolution of hysteresis loops for axial stress versus
axial, lateral, and thickness strains indicated anisotropic character-
istics of the damage accumulation in such materials. In stress-con-
trolled tests, the thickness strain was initially compressive and
gradually increased in the tensile direction. Therefore, the thick-
ness strain was found to have direct relationship with the damage
state in the material.
Hoppel [47] evaluated fatigue modulus reduction of a short
glass fiber reinforced thermoplastic under axial load-controlled Fig. 11. Axial fatigue data of short glass fiber reinforced polyamide-6.6 at room
R = 0.1 condition. ‘‘Fatigue modulus’’ was defined as the slope of temperature [51] rearranged in terms of averaged strain energy density over a
a line from the maximum stress–strain point to the origin. Fatigue control volume [50].
damage (moduli reductions) was accumulated at three stages char-
acterized by a rapid damage growth, a steady state damage growth,
and a rapid damage growth prior to failure. Ladeveze and LeDantec [52] developed a phenomenological
Esmaeillou et al. [48] studied fatigue damage development of a fatigue damage model for laminate composites in the framework
short glass fiber reinforced polyamide-6.6 under axial stress-con- of continuum damage mechanics. The parameter introduced ear-
trolled R = 0.1 cyclic loading conditions. Evolution of Young’s mod- lier in Eq. (5) in terms of stiffness reduction to characterize damage
ulus along with the specimen surface temperature showed three accumulation of material was used in terms of elastic strain
stages during the fatigue life. These three stages consisted of a energy. This model was extended to SFRPCs by Nouri et al. [53].
decrease in the modulus with an increase in the temperature, pla- In short glass fiber reinforced thermoplastics the fatigue damage
teaus for both the modulus and the temperature, and a drop in the evolution occurs under three stages. In the first stage, a high stiff-
modulus with elevating the temperature. It was concluded that ness reduction occurs during the initial cycles due to the formation
fatigue damage development was possible only after stabilization of matrix micro-voids. In the second stage, the coalescence of
of the temperature and modulus. The last two stages disappeared microcracks takes place which gives a gradual stiffness reduction
with increasing the level of applied stress and thermal failure of the material. In the last stage, fiber fracture and macroscopic
was observed. Micro-ductile areas were observed around the fibers crack propagation occurs, resulting in a rapid stiffness reduction
from scanning electron microscopy of the fractured surfaces. The and leading to final fracture. Therefore, damage rate in SFRPCs
size of these micro-ductile areas was related to the maximum was assumed to be the sum of two components, one power law
applied stress and these areas disappeared at higher stresses. and one exponential law, to capture all the three stages. The devel-
In the study of Meneghetti et al. [33] mentioned earlier, under oped model was able to describe all three stages for tensile and
R = 1 loading condition the stiffness reduction occurred mainly shear damages. Longitudinal, transverse, and shear damage param-
in tension and stiffness was nearly constant in compression. The eters of this model were identified from experimental data col-
area inside the midlife hysteresis loop, as a measure of total lected from axial strain-controlled behavior of short glass fiber
mechanical energy loss in the material, was utilized to correlate reinforced polyamide-6 under the R = 0.3 condition [54,55].
the stress-life data. This parameter synthesized data in a single Avanzini et al. [56] applied the Nouri et al. damage model for a
scatter band. The hysteresis energy loss was found to be almost polyether ether ketone reinforced by carbon micro-fibers by add-
the same for frequencies of 3 and 22 Hz. As a result, the dissipated ing graphite and PTFE as fillers. Axial load-controlled fatigue tests
energy due to viscoelasticity was negligible relative to the total were performed under R = 0 condition. The orthotropic model pro-
hysteresis energy. posed by Nouri et al. was simplified to a one damage state variable
(Y) and one damage, D, evolution:

D ¼ a lnðNÞY bð1YÞ ð7Þ


6.2. Strain energy
where coefficients a and b are the material damage constants and Y
Strain energy is a common parameter for qualifying damage of is the thermodynamic force. A good correlation of the model and
homogeneous isotropic materials such as metallic alloys and elas- the experimental data was shown.
tomers. Lazzarin and Zambardi [49] suggested a strain energy den-
sity (SED) based method for estimating fatigue life of a linear 6.3. Nonlinear viscoelasticity
elastic isotropic material by averaging the SED over a control vol-
ume. This method was extended to SFRPCs [50]. It was assumed Polymeric materials exhibit a time dependent or viscoelastic
that high cycle fatigue failure occurs if the average SED over a con- response to mechanical loading. At lower stress levels viscoelastic
trol volume around the notch tip under a linear elastic field behavior reveals linear relationship between stress and strain,
exceeds a critical value. Due to the presence of notch in specimens whereas at higher stress levels this behavior can be nonlinear. Fati-
the state of stress is multiaxial and the distribution of SED was gue behavior has been evaluated by means of nonlinear viscoelas-
computed by FEA. Fatigue data from load-controlled notched tests tic change with cycling in some studies.
of short glass fiber polyamide-6.6 with five different notch radii Noda et al. [57] conducted an experimental study on fatigue
and two different load ratios [51] collapsed on a narrow scatter behavior of short glass fiber reinforced polyamide-6.6. Axial
band by using this parameter, as shown in Fig. 11. tension–tension tests with loading along the fibers direction were
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 307

performed in a range of temperature below and above the glass ether ketone and polyphenylene sulfide. Specimens were
transition temperature. The degree of nonlinear viscoelasticity machined with axis in the mold flow direction and tested under
(NVP) was determined as follows: R = 0.1, 0.3, and 0.5 conditions. The stress version of Smith–Wat-
e2 þ e3 þ    þ e10 son–Topper equation was applied to correlate the fatigue data at
NVP ¼ ð8Þ different stress ratios. This equation is expressed as:
e1
where e1 is the fundamental strain amplitude and e2, . . . , e10 repre- pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
SNf ¼ ðDS=2ÞðSmax Þ ð10Þ
sent higher order harmonic strain amplitudes in Fourier expanded
series of the strain response during the load-controlled test. Optical
where DS and Smax are the range and the maximum of the applied
microscope observations during the cycling indicate the increase in
stress, respectively. This parameter successfully correlated the
NVP was related to the propagation of microcracks. Thus, it was
experimental data.
concluded that the evolution of NVP can possibly indicate the
Sonsino and Moosbrugger [59] evaluated the effect of the stress
mechanisms of failure and could be used to estimate the fatigue life.
ratio on axial load-controlled fatigue strength of short glass fiber
Fatigue failure mechanisms were classified depending on whether
reinforced polyamide-6.6. Unnotched and notched specimens were
the test was performed above or below the glass transition temper-
machined in the mold flow direction and tested with R ratios of 1
ature. At temperatures below the glass transition temperature, fail-
and 0. Fatigue strength reduced with increasing the tensile mean
ure mechanisms and stages consisted of formation of voids and
stress. This decrease was attributed to both the mean stress sensi-
initiation of microcracks at the fiber ends, propagation of micro-
tivity and cyclic creep. The reduction of fatigue strength was less
cracks around the fiber ends, growth of cracks between the fiber
for the notched specimens compared to the smooth specimens,
ends, followed by fast crack growth in a brittle manner. At temper-
as can be observed in Fig. 12. This effect was attributed to the pres-
atures above the glass transition temperature, failure mechanisms
ence of the stress gradient in the vicinity of the notch.
and stages consisted of formation of local defects at the fiber ends,
Avanzini et al. [60] conducted an experimental study on axial
fiber debonding and deformation of the matrix between the fibers,
load-controlled fatigue behavior of a carbon micro-fiber polyether
growth of cracks between the fibers, and fast crack growth in a duc-
ether ketone composite under R = 0 condition. Specimen axis was
tile manner.
in the mold flow direction. The mean strain values progressively
Komatsu et al. [58] investigated strain-controlled fatigue
shifted along the strain axis with different rates, depending on
behavior of a short glass fiber reinforced polyamide-6 in tension–
the stress level. The rate of mean strain, defined as the cyclic creep
tension (T–T), tension–compression (T–C), and compression-com-
rate, indicated a linear relationship with the number of cycles to
pression (C–C) testing with loading along the fiber direction. NVP
failure in log–log scale, as seen in Fig. 13.
was calculated from the coefficients of Fourier series of the stress
Mallick and Zhou [61] studied effect of R ratio or mean stress on
response signal, similar to Eq. (8). For all loading conditions, NVP
axial stress-controlled fatigue behavior of short glass fiber rein-
increased with applied cycles. Void formation, fiber fracture, and
forced polyamide-6.6. Tests were carried out under several stress
crack growth contributed to the increase of NVP in T–T condition.
ratios, ranging between 0.1 and 0.8 using specimens machined in
In C–C condition, interfacial debonding was the major cause of the
the mold flow direction. The presence of mean stress resulted in
increase in NVP. Under the fully-reversed condition both crack
a significant effect on fatigue life, as can be seen from Fig. 14(a).
growth and interfacial debonding resulted in a rise in NVP.
A substantial increase of the mean strain throughout the test was
attributed to cyclic creep. They suggested the use of Gerber mean
6.4. Hysteretic energy dissipation
stress equation normalized with the creep rupture strength given
by:
The energy loss in viscoelastic materials such as SFRPCs is
divided into energy dissipated as heat and energy used for struc-  2
tural changes. The amount of energy consumed for structural Sa Sm
þ ¼1 ð11Þ
changes in the material can be related to fatigue life of SFRPCs. SNf Sc
Takahara et al. [35] established a fatigue fracture criterion based
on the dissipated energy consumed in structural changes. This cri- where Sa is the stress amplitude, Sm is the mean stress, SNf is the
terion provides a relationship between the dissipated energy for fully-reversed fatigue strength, and Sc is the creep rupture strength.
structural changes and the fatigue lifetime as follows: This parameter predicted the effect of mean stress well, as shown in
ðHs;av  H0 Þt f ¼ C ð9Þ Fig. 14(b). Sauer et al. [62] suggested the use of creep rupture
strength, Sc, instead of the tensile strength in modified Goodman
where Hs,av is the average hysteretic energy loss for structural equation, to represent the effect of mean stress in fatigue design
change during cyclic straining and tf is the fatigue life time. H0 of polymeric components. The resulting equation is similar to Eq.
and C were obtained from the curve fitting of the experimental data. (11), except without the exponent in the second term. They did
This criterion was applied to randomly oriented short glass fiber not verify this model due to the lack of experimental data.
reinforced polybutylene terephthalate axial strain-controlled fati- Oka et al. [63] investigated effect of mean stress on fatigue
gue tests performed under R = 1 condition. Yamashita et al. [36] strength and cyclic ratcheting of short glass fiber reinforced poly-
applied the fatigue fracture criterion established by Takahara butylene terephthalate with specimens machined in the mold flow
et al. [35] to short glass fiber reinforced polyamide-6 axial strain- direction. Experiments were conducted under several stress ratios
controlled tests conducted under R = 1 condition. ranging between 1 and 0.9. Fatigue life reduced as stress ratio
increased. At stress ratios as high as 0.7, maximum stress versus
7. Mean stress effects time to fracture curve approached creep curves. As stress ampli-
tude increased, at the same mean stress, more cyclic ratcheting
Most service load histories consist of cycles with mean stress. was observed. Modified Goodman non-conservatively estimated
The effect of mean stress on fatigue behavior of SFRPCs has been the fatigue life at stress ratios above 0.7, due to significant creep
evaluated by several researchers. Zhou et al. [39] conducted an deformation developed at high mean stresses. Similar to the
experimental study on load-controlled flexural fatigue behavior Mallick and Zhou [61] study, Equation (11) was employed to cor-
of a short glass fiber composite with a blend of polyphenylene rect the effect of mean stress at all stress ratios.
308 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Fig. 12. Woehler-lines of smooth and notched short glass fiber reinforced polyamide-6.6 for R = 1 and R = 0 loading conditions [59].

forced polyamide-6.6 mentioned previously, using smooth and


notched specimens prepared in the mold flow direction and tested
under R = 0 and R = 1 conditions. Under the fully-reversed condi-
tion, increasing frequency from 4 to 5 Hz increased smooth speci-
men temperature from 30 to 72 °C and reduced the fatigue life by a
factor of about 10. Under R = 0 condition, increasing frequency
from 5 to 8 Hz, resulted in an increase of smooth specimen temper-
ature from 30 to 90 °C and reduction of the fatigue life by a factor
of 20. For notched specimens higher frequencies used did not have
a detrimental effect on fatigue life since the maximum stressed
volume of the material in the notch tips is smaller and heat dissi-
pates faster. In smooth (unnotched) specimens frequency effects
can be pronounced due to the high maximum stressed material
volume. Therefore, for practical applications results obtained with
Fig. 13. Mean strain rate versus cycles to failure for short carbon fiber reinforced unnotched specimens should be relativized with respect to the
PEEK composite for R = 0 loading condition [60].
highest stressed volume at the stress concentrations in the
component.
8. Cyclic frequency and self-heating effects Esmaeillou et al. [65] also studied the effect of frequency on
fatigue behavior of short glass fiber reinforced polyamide-6.6.
Cyclic frequency can have a much more significant effect on Load-controlled axial tests were conducted under R ratios of 0.1
fatigue behavior of SFRPCs than on metallic alloys. Various studies and 0.3 and strain-controlled flexural tests were performed under
have evaluated the effect of testing frequency on fatigue behavior R = 1 condition. Frequencies of 2, 10, and 20 Hz were selected for
of SFRPCs. Depending on the mode of loading, the stress level, and testing. For the same maximum applied stress in either of axial or
the material characteristics, frequency affects the cyclic behavior of bending test, in LCF regime the same fatigue lives were achieved at
materials to various degrees. Self-heating is the major effect of different frequencies, whereas in HCF regime fatigue life decreased
cycling frequency resulting from viscoelastic nature of the matrix with increasing frequency. For the same number of cycles to fail-
and frictional heating in the composite. ure, temperature rise due to self-heating was higher in axial fatigue
Bellenger et al. [64] studied the effect of frequency on bending tests than in flexural fatigue tests. Under both axial and flexural
fatigue behavior of a short glass fiber reinforced polyamide-6.6. fatigue tests with temperature elevation of 70–90 °C, fracture sur-
Specimens were machined in the mold flow direction and tested faces were observed by SEM. In axial fatigue tests matrix plastic
under fully-reversed strain-controlled condition. Frequencies of deformation and broken fibers were distinguished, while in flex-
2 Hz and 10 Hz were utilized to characterize thermal and mechan- ural fatigue experiments broken fibers and cavities around fibers
ical fatigue of the material. At 2 Hz, the specimen temperature were observed.
raised 10 °C whereas at 10 Hz the temperature elevated about In the study of Zhou and Mallick [22] on axial fatigue behavior
100 °C. The high temperature rise at 10 Hz was due to damping of short glass fiber reinforced polyamide-6.6, the effect of fre-
induced by crack initiation and change of the matrix state from quency was evaluated. Stress-controlled tests were performed
glassy to rubbery. Variation of the maximum stress showed differ- under the R = 0.1 condition on mold flow oriented specimens. A
ent trends in these experiments. At 10 Hz, a plateau of maximum beneficial effect of frequency was observed at frequencies lower
stress followed by a steep drop. This effect is believed to be due than 2 Hz, so that the fatigue life increased proportionally with fre-
to the transition of matrix from glassy state to rubbery state. There- quency, as seen in Fig. 15. In this range of frequency, the mean
fore, both mechanical and thermal fatigue existed at the frequency strain values increased with applying cycles, but the strain ampli-
of 10 Hz. tude did not change appreciably. At frequencies higher than 2 Hz,
Sonsino and Moosbrugger [59] evaluated the effect of testing with increasing frequency, fatigue life decreased due to the influ-
frequency on axial fatigue behavior of the short glass fiber rein- ence of hysteresis heating, mentioned earlier, as can be seen in
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 309

Fig. 15. Effect of cyclic frequency on fatigue life of short glass fiber reinforced
polyamide-6.6 under R = 0.1 loading condition [22].

lic straining was divided into heat generation energy and energy to
cause structural changes. By balancing the generated heat and the
transferred heat to the surrounding environment, the equivalent
temperature was expressed as:

heq ¼ vpf e2d E00 =K ð12Þ

where v is the ratio of generated heat to the total energy loss, f is


cycling frequency, ed is strain amplitude, E00 is nonlinear dynamic
loss modulus, and K is the heat transfer coefficient. Axial strain-con-
trolled fully-reversed fatigue data of randomly oriented short glass
fiber reinforced and unreinforced polybutylene terephthalate (PBT)
show a linear relationship between the equilibrium temperature
and the square of the strain amplitude in Fig. 16.
Handa et al. [67] evaluated axial stress-controlled fatigue
behavior of short glass fiber reinforced polyamide-6.6 under R = 0
condition. Specimens with axis in the mold flow direction were
tested with frequencies of 5, 20, and 50 Hz. All the S–N curves were
approximated by straight lines, and S–N line slopes and intercepts
Fig. 14. Effect of mean stress on axial fatigue behavior (a) and relationship between were found to decrease with increasing the frequency, as shown in
stress amplitude and mean stress (b) for short glass fiber polyamide-6.6 with
specimens machined in the mold flow direction [61].

Fig. 15. In this case both mean strain and strain amplitude
increased with applying cycles.
Lang and Manson [66] investigated crack tip heating sources for
short glass fiber reinforced polyamide-6.6 and short glass fiber
reinforced polystyrene in axial load-controlled fatigue tests under
R = 0.1 condition. Specimens with axis parallel and perpendicular
to the mold flow direction were used. Hysteretic and frictional
heating were considered as the major crack tip heating sources.
Hysteretic heating is mainly due to plastic or viscoelastic deforma-
tions in the matrix. Frictional heating is related to the surface con-
tact under crack closure or friction between the fibers and the
matrix. Significantly higher crack tip heating in polystyrene com-
posite was observed, as compared to the polyamide-6.6 composite.
Frictional heating was thought to have a superior effect on crack
Fig. 16. The relationships between equilibrium temperature and the square of the
tip heating, as compared to the hysteretic heating.
strain amplitude for unreinforced and reinforced short glass fiber polybutylene
Takahara et al. [35] developed a temperature rise model based terephthalates from fatigue tests performed under strain control and R = 1
on the hysteretic energy loss. The hysteretic energy loss under cyc- condition [35].
310 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Fig. 18. Effect of the cycling frequency on fatigue behavior (a) and specimen surface
temperature versus maximum applied stress (b) for short glass fiber polyamide-6
specimens immersed in water from tests performed at room temperature under
R = 0.1 loading condition [68].

Fig. 18(a). Surface temperature rise was found to be dependent


Fig. 17. Effect of cycling frequency on fatigue behavior (a) and temperature rise on stress and frequency, as shown in Fig. 18(b). Dissipated energy
versus a fatigue parameter (b) of short glass fiber polyamide-6.6 from tests density per unit time was computed as the area inside the hyster-
performed in stress control under R = 0 loading condition [67]. esis loop multiplied by frequency:
I
_ d¼f
W rde ð14Þ
Fig. 17(a). A model was derived to predict the temperature rise
from fatigue data and viscoelastic properties of the material,
At different frequencies, a linear relationship between average tem-
expressed as:
perature rise and dissipated energy density per unit time was
pLf E00 r2 found. The area inside the hysteresis loop increased during the test
DT ¼ ð13Þ showing some extent of cyclic softening. In addition, the hysteresis
4KE02
loop translated along the strain axis, indicating significant amount
where L is specimen length, K is heat transfer coefficient, and f, E00 , of cyclic creep. In log–log scale, the cyclic creep mean strain rate
E0 , and r are frequency, loss modulus, storage modulus, and maxi- was linearly related to the number of cycles to failure. All fatigue
mum stress, respectively. A good correlation between the prediction data at frequencies of 1, 2, and 4 Hz overlapped on the same curve.
model and the experimental data was found, as seen in Fig. 17(b). A frequency superposition method based on the Larson–Miller
Bernasconi and Kulin [68] studied the effect of frequency on parameter related the cyclic creep mean strain rate, the surface
axial load-controlled fatigue behavior of conditioned short glass temperature rise, and the mean stress.
fiber reinforced polyamide-6. Specimens in the mold flow direction Jegou et al. [69] applied a method based on dissipated energy to
were immersed in room temperature water for 7 days and then estimate the axial fatigue life of a short glass fiber reinforced poly-
allowed to reach equilibrium with ambient moisture at 23 °C and amide-6.6 under R = 0 condition. A thermo-mechanical problem
relative humidity of 50% for several weeks. Tests were conducted defined by both thermal and mechanical equations was solved to
under R = 0.1 condition with frequencies between 1 and 4 Hz. S– estimate dissipated energy from thermal measurements [70]. The
N curves were nearly parallel at different frequencies and fatigue dissipated energy was then correlated to the number of cycles to
strength decreased with increasing frequency, as shown in failure, expressed as:
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 311

qch b gue tests were performed with R ratios of 1 and 0. A decrease in


N ¼C ð15Þ
f seq nominal fatigue strength with increasing the severity of the notch
was observed, as can be seen in Fig. 20. Specimens with the central
where q is mass density, c is specific heat capacity, h is stabilized sharp notch were tested at room temperature and at 130 °C. The
temperature rise, and f is frequency. seq is characteristic time which effect of notch nearly vanished at 130 °C, and the distance between
depends on thermal boundary conditions and sample geometry and S–N lines of smooth and notched specimens substantially
obtained utilizing cooling curves, and b and C are the model param- decreased, as compared to at room temperature. This results from
eters. Temperature rise tests at a frequency of 1 Hz were performed a significant reduction of notch sensitivity due to increased mate-
at multiple stress levels to obtain stabilized temperature rise, and rial ductility. A transition from nominal stress system to local
temperature rise was correlated with number of cycles to failure. stress system was explained and it was shown that due to stress
The same values of b and C were applied to two different smooth gradient effect and reduced highly stressed material volume in
specimen geometries and correlations were found between the notches, much higher stresses are supported compared with
model and experimental data. unnotched state.
Haldar and Senthilvelan [71] studied the effect of notch size on
9. Notch or stress concentration effects axial load-controlled fatigue behavior of neat and short glass fiber
reinforced polypropylene (PP). Reinforced PP was fabricated with
The stress concentration effect is a key factor in studying the average fiber lengths of 0.44 and 1.25 mm. A rectangular specimen
fatigue behavior and application to design. This is due to the fact geometry with axis in the mold flow direction with central circular
that most components manufactured from SFRPCs contain geomet- hole diameters of 2 or 4 mm were made for testing under R = 0.1
ric contours or features resulting in stress concentrations. Most condition. Nominal fatigue strength of neat PP was nearly the same
fatigue behavior investigations are carried out with unnotched with both stress concentrations. In LCF both short and long fiber
specimens, while fatigue test results with notched specimens are composites indicated higher fatigue performance with decreasing
more relevant to the component level performance. the notch size. In HCF, the short fiber composite indicated no effect
Effects of notch or stress concentration on fatigue behavior of of notch size, but the long fiber composite showed significant effect
SFRPCs have been studied by several investigators. Zhou and Mal- of notch size. Fiber pullout was the major failure mode in the short
lick [22] studied effect of notch size on R = 0.1 axial fatigue behav- fiber composite due to low capacity of load bearing of fibers,
ior of short glass fiber reinforced polyamide-6.6 using mold flow whereas in the long fiber composite most of the fibers bent near
oriented specimens with a width of 12.7 mm with a central circu- the notch.
lar hole with diameters of 1.58, 3.17, and 6.35 mm. Stress concen- Yamamoto and Hyakutake [72] investigated effect of fiber
tration factor (Kt) for the specimens used were 2.64, 2.43, and 2.18, length on notched fatigue behavior of short glass fiber polypropyl-
respectively. Fatigue strength of notched specimens based on nom- ene. Double-edge U-shaped notched specimens with notch root
inal gross stress was lower than that of the smooth specimen. radii of 0.5, 1 and 2 mm and with specimen axis in mold flow direc-
However, the nominal net area-based S–N curves of different hole tion were tested under R = 0.1 condition. Samples with average
diameter specimens collapsed on the same S–N curve, as shown in fiber length of 3.5 mm indicated a factor of about 5 longer fatigue
Fig. 19. fracture life at a given nominal stress, as compared with samples
Sonsino and Moosbrugger [59] evaluated the influence of with average fiber length of 0.4 mm. Initiation and growth of dam-
notches on axial fatigue behavior of short glass fiber reinforced age near the notch root were similar in both long and short fiber
polyamide-6.6. Rectangular bars with axis in the mold flow direc- composite such that microcracks initiated at fiber ends and con-
tion with width of 4 cm and either a central circular or a slit notch nected to form the main crack. However, fatigue damage was ini-
with Kt in the range of 2.5–9.8 were utilized. Load-controlled fati- tiated in a larger area near the tip of notch in long fiber
composite, compared to in short fiber composite. Notch root radii
did not influence cycles to fatigue fracture. However, cycles to
damage initiation defined as fatigue damaged area of 0.1 mm2
increased a factor of about 100 as notch root radius decreased from
2 mm to 0.5 mm. Slopes of stress versus cycles to damage initiation
did not exhibit notch root radius sensitivity.
Nisitani et al. [73] compared the effect of stress concentration
on rotating bending fatigue behavior of neat and short carbon fiber
polyetheretherketone (PEEK). Specimens with a central circular
hole or circumferential V-shaped notch with Kt values ranging
between 1.04 and 4.48 were considered. Nominal fatigue strength
of short fiber PEEK was less sensitive to stress concentration, as
compared with neat PEEK. Due to fiber ends effect, a higher per-
centage of fatigue fracture life was spent on crack growth in short
fiber PEEK, compared with neat PEEK. Master curves were obtained
by plotting Kt times either of maximum nominal stress where a
macro-crack does not appear, or threshold nominal stress where
a non-propagating crack exists, versus reciprocal of notch root
radius.
Mortazavian and Fatemi [74] studied effects of a notch on
R = 1 and R = 0.1 axial load-controlled fatigue behavior of short
glass fiber reinforced polybutylene terephthalate. A plate-type
specimen geometry with a gage section width of 10 mm was
machined parallel and perpendicular to the mold flow direction.
Fig. 19. Effects of hole diameter on net stress-based S–N curves of a short glass fiber A central circular hole with diameter of 2 mm was used with
reinforced polyamide-6.6 [22]. Kt = 2.5. Effect of the circular notch on nominal fatigue strength
312 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Fig. 20. Effect of the notch stress concentration on fully-reversed axial fatigue strength of short glass fiber reinforced polyamide-6.6 [59]. k values on the graph are reciprocals
of the slope of S–N lines.

was found to be significant, with several orders of magnitude life composite shows more resistant to the subsequent low stress. At
reduction in HCF. Lower notch sensitivity was observed in LCF than the low stress level, cracking occurs in matrix with no plastic defor-
in HCF. Tensile mean stress decreased the fatigue strength of mation and crack propagates by the subsequent high stress which
notched specimens and this effect was more pronounced in LCF. results in shorter fatigue life.
Neuber’s rule provided reasonably accurate predictions of notched Zago et al. [78] evaluated variable amplitude fatigue behavior of
fatigue strength. smooth specimens made of short glass fiber reinforced co-polyam-
Meneghetti and Quaresimin [75,76] evaluated notched axial ide. Specimens were used with their axis in 0°, 45°, and 90° with
fatigue behavior of short glass fiber reinforced polyamide-6.6 using respect to the mold flow direction. Two different blocks of variable
load-controlled fatigue tests under R = 0.1 condition. A rectangular amplitude loading were applied including blocks of constant stress
bar, in line with mold flow direction, with the width of 40 mm and amplitude with different levels and blocks of random stress ampli-
a central circular hole with a diameter of 10 mm or a central slit tude with mean stress. Rainflow cycle counting method was
notch with tip radius of 0.5 mm was used. The stress concentration employed to transfer the variable amplitude loading to blocks of
factor for these specimens was 2.84 and 7.82, respectively. Nomi- constant amplitude. Fatigue lives were correlated using the gener-
nal fatigue strength decreased significantly with increasing the alized Palmgren–Miner’s rule:
severity of the notch. A model was proposed on the basis of the
X ni b
first law of thermodynamics over a control volume of material. ¼1 ð17Þ
By assuming a steady state heat transfer condition and under a i
Ni
condition with no mechanical work, this model results in a specific
where ni and Ni are the number of applied cycles and the number of
heat loss (Q) equation given by:
cycles to failure, respectively at ith stress level, and b is a material
qc @T constant found to be 0.92. Zago and Springer [79] also investigated
Q ¼ ð16Þ
f @t fatigue behavior of 10 mm wide plate specimens with a central cir-
cular hole of 2 mm diameter under various blocks of variable ampli-
where q is material density, c is material specific heat, T is temper-
tude loading. Fatigue lives were estimated based on the generalized
ature, t is time, and f is frequency. oT/ot was measured from the
Palmgren–Miner’s rule and estimated lives were close to the
cooling curve. The specific heat loss correlated the fatigue strength
experiments.
of smooth and notched specimens reasonably well. This method
Sonsino and Moosbrugger [59] studied variable amplitude fati-
was not able to predict the fatigue strength of slit notched speci-
gue behavior of short glass fiber reinforced polyamide-6.6 using
mens due to small and unreliable temperature rise at the tip of
blocks of Gaussian random load sequences under either fully-
the notch.
reversed or R = 0 condition. Palmgren–Miner’s rule was used to esti-
mate the fatigue life under either of these mean stress conditions,
10. Variable amplitude loading expressed as:
X n
Variable amplitude loading is typical of most service load histo- D¼ ¼1 ð18Þ
ries. Hoppel et al. [77] studied damage accumulation of short glass
N i

fiber reinforced styrene-maleic anhydride with specimen axis in An estimation of fatigue life with the theoretical damage sum
the mold flow direction. Step loading fatigue tests with stress ratio Dth = 1 was found to be nonconservative for R = 0, but conservative
of 0.1 were performed in a high stress level followed by a low stress for R = 1. Recommendations about allowable damage sums were
level (HL) and in a low stress level followed by a high stress level also provided. Nonconservative calculated life under a positive
(LH). Linear damage rule over-estimated LH fatigue test results, mean stress was attributed to the influence of cyclic creep which
while it under-estimated HL fatigue test results. SEM of fractured was not considered by the linear damage rule.
surfaces indicates extensive matrix crazing in HL tests, compared Dreißig et al. [80] conducted pulsating Gaussian variable ampli-
with LH tests. At the high stress, significant plastic deformation tude load testing on un-notched specimens of short glass fiber
occurs in matrix and clamping force on fiber increases, therefore, polyamide-6.6 machined in 0°, 30°, and 90° directions with respect
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 313

to the mold flow direction. Linear damage accumulation rule (I) also existed. Under pure torsion cracks started at the center of
was applied and damage sums were 0.2 for 0° and 30° direction specimen on maximum shear stress plane and grew along this
tests and about 0.35 for 90° direction tests, as shown in Fig. 21. plane until the friction force between the failure surfaces over-
Generalized Palmgren–Miner’s rule mentioned earlier as Eq. (17) comes the driving force in shear. Then the crack propagated in
with (II) and without (IV) load dependent fitting coefficient was an angle between 30° and 45° relative to the specimen axis. For
also applied and results are shown in Fig. 21. A three parameter combined loading with biaxiality ratio of one, failure patterns were
nonlinear damage accumulation (III) and a nonlinear stress-time similar to pure torsion, indicating shear stress was dominant.
history based cumulative model (V) were also applied. As seen in Klimkeit et al. [82] studied multiaxial fatigue behavior of two
Fig. 21, the nonlinear cumulative damage model (II) calculates short glass fiber composites, one with a mix of polybutylene tere-
the fatigue life well; however, this model requires the use of a fit- phthalate and polyethylene terephthalate, and the other with poly-
ting parameter. amide-6.6. A flat dog-bone shaped specimen and a double V-
shaped notched specimen in 0°, ±45°, and 90° directions relative
to the mold flow direction were used to investigate axial and pure
11. Multiaxial stresses shear fatigue behaviors, respectively. A tubular specimen with
fiber orientation in specimen axis direction was also used to inves-
Multiaxial stresses exist in many components and structures. tigate combined axial-torsion fatigue behavior. Fig. 23 shows the
Multiaxial stresses can even be present under uniaxial loading experimental results plotted in terms of nominal stress amplitudes
due to presence of notch or other stress concentrations. De Monte ra,n or sa,n. Under axial loading the tubular specimens resulted in
et al. [81] performed an experimental study on multiaxial fatigue higher fatigue strength, compared to the flat specimens. They
behavior of short glass fiber reinforced polyamide-6.6. A hollow explained this effect to be related to triaxiality of stress state in
tubular specimen geometry was fabricated in the mold flow direc- tubular specimens, formation of a core layer with 90° fibers orien-
tion and tested under combined in-phase and 90° out-of-phase tation with respect to the mold flow direction in flat specimens, as
tension and torsion loading. The effect of biaxiality ratio, phase well as the uniformity of stress across the flat specimen thickness,
angle, and load ratio at room temperature and 130 °C were dis- but not the tubular specimen. A higher shear stress resulted in a
cussed. The results at room temperature plotted based on material lower fatigue strength, as seen in Fig. 23(a). This effect was due
normal stress in fiber direction are shown in Fig. 22. At room tem- to turning of the principal stress direction, depending on the ratio
perature, significant strength reduction is observed under com- of torsion and tension stresses. Under fully-reversed pure shear
bined loading compared to pure axial or pure torsion loading. At loading condition, 0° and 90° direction specimens indicated identi-
130 °C, the combined loading condition was dominated by torsion cal fatigue strength, and ±45° direction specimens displayed higher
mode and fatigue lives from combined and torsion loadings fell fatigue strength than 0° and 90° direction specimens, as shown in
within a single scatter band. At room temperature, no effect of Fig. 23(b). Under the R = 0.1 pure shear loading condition, 45°
phase angle is observed under the fully-reversed loading condition, direction indicated higher fatigue strength than both 0° and 90°
whereas under the R = 0 condition, fatigue strength increased with directions, but + 45° direction displayed lower fatigue strength
shifting the phase, as can be seen from Fig. 22. At 130 °C, no influ- than 0° and 90° directions, as can be seen in Fig. 23(b). This is
ence of phase shifting was observed under either R = 1 or R = 0 because under the effect of mean stress, fibers are under compres-
condition. Biaxiality ratio affected the fatigue behavior at both sion in the +45° specimen, but under tension in the 45° specimen.
temperatures so that the fatigue strength reduced with increasing A local compressive stress on fibers in +45° specimens resulted in a
the ratio of torsion loading to axial loading. Stress ratio displayed detrimental effect on fatigue life. Crack path was depended on both
much more effect on axial loading than in torsion loading and it the loading mode and the microstructure of the material. In tubu-
even had no influence on torsional fatigue strength at 130 °C. Dif- lar specimens, under pure tension loading cracks initiated on the
ferent failure paths were observed depending on the loading mode plane perpendicular to the specimen axis and propagated in 45°
and level. At lower stress levels usually one main crack was direction to the specimen axis. Under pure torsion, cracks initiated
observed, whereas at higher stress levels many secondary cracks and grew in orientation of fibers (i.e. longitudinal cracks) which

Fig. 21. Comparison of different damage accumulation methods. (I) Linear damage rule, (II) nonlinear damage accumulation with one fitting parameter, (III) nonlinear
damage accumulation with three fitting parameter, (IV) nonlinear damage accumulation based on load depended fitting coefficient, and (V) nonlinear damage accumulation
based on the stress-time history [80].
314 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

mal and shear stresses, out-of-phase loading results in a smaller


equivalent or von Mises stress, as compared to in-phase loading.
The influence of phase shifting in notched specimens was more
pronounced, as compared with smooth specimens, due to locally
different deformation evolutions.
Launay et al. [84,85] applied the constitutive nonlinear model
mentioned earlier in [9] to multiaxial fatigue analysis of short glass
fiber reinforced polyamide-6.6 in [81,82]. The model parameters
identified based on tensile samples accurately predicted mechani-
cal response on tubular specimens under multiaxial stress state.
Criteria including von Mises stress, Hill stress, equivalent shear
stress and hydrostatic stress, modified version of Tsai–Hill, princi-
pal stress, principal strain, equivalent inelastic strain, and dissi-
pated energy density were employed to estimate the multiaxial
fatigue lifetime. Physical quantities involved in each of these crite-
ria (e.g. the shear stress amplitude and the hydrostatic stress) were
obtained based on the mechanical response at 20 cycles and then
material parameters were calibrated performing least square
method on the entire data set. Among these, dissipated energy
density estimated the fatigue life reasonably well. This criterion
considered the influence of microstructure and load ratio in multi-
axial fatigue life estimation.

12. Environmental effects

Components made of SFRPCs are often exposed to different


temperatures and humidity levels in their operating environment.
Therefore, temperature and moisture effects are two primary envi-
ronmental considerations with regards to fatigue behavior of
SFRPCs. If the temperature is cyclic in addition to the cyclic load,
thermo-mechanical fatigue is also an important aspect to consider.

12.1. Moisture effect

Hoppel [47] investigated axial load-controlled fatigue behavior


of short glass fiber reinforced styrene-maleic anhydride specimens
prepared in the mold flow direction and tested under R = 0.1 con-
dition. Dry tests were performed at temperature of 25 °C and 50%
relative humidity. Wet test samples were immersed in water for
48 hours prior to and throughout the test. Fatigue life in the wet
Fig. 22. Wohler curves in terms of material normal stress in fiber direction obtained condition was much shorter than the dry condition and this effect
from tubular specimens made of short glass fiber reinforced polyamide 6.6. Data for
was more apparent for materials with no silane coupling agent.
pure torsional loading are plotted in terms of shear stress amplitude [81].
Silane agent makes a strong bonding between fibers and matrix
and improves fatigue strength of the composite. Throughout the
was along the maximum shear stress plane. Under combined axial- test, the stiffness of the wet composite decayed at a higher rate
torsion loading, cracks propagated in 35° direction relative to the as compared to the dry composite.
specimen axis, which was close to the maximum shear stress Barbouchi et al. [86] evaluated the effect of moisture on bend-
plane. Three fatigue failure criteria including principal stress, von ing fatigue behavior of short glass fiber reinforced polyamide-6.6
Mises stress, and the energy criterion were applied to estimate at room temperature. Fully-reversed strain-controlled tests were
the fatigue life. Principal stress and von Mises criteria were not conducted at frequency of 10 Hz with applied stress in the mold
accurate in estimating the fatigue life. The energy criterion esti- flow direction. Two sets of samples with 0.2 wt% and 3.3 wt% water
mated fatigue life more accurately and it also included a mean contents were studied. The glass transition temperature were 70 °C
stress correction parameter. and 11 °C for samples with water content of 0.2 and 3.3 wt%,
Moosbrugger et al. [83] investigated multiaxial fatigue behavior respectively. This deviation is due to the plasticizing effect of water
of short glass fiber reinforced polyamide-6.6 using a tubular spec- which increases the mobility of polymer chains. In specimens with
imen with or without a circumferential V-shaped notch with tip higher percentage of water, testing temperatures were above the
radius of 0.2 mm. Axial and torsional stress concentrations for glass transition temperature from the beginning of the test. Tem-
the notched specimens were calculated by FEA to be 4.9 and 2.4, perature progressively increased to about 130–150 °C, then failure
respectively. Combined in-phase and out-of-phase axial-torsion occurred. This effect was evident at both high and low applied
loading were applied under stress ratios of 0 and 1. When com- strains. In specimens with lower percentage of water, the evolution
pared based on the same applied nominal axial and shear stresses, of temperature rise was dependent on the level of applied strain. At
for unnotched specimens out-of-phase loading resulted in slightly the same strain amplitude, the fatigue life was longer for speci-
longer life than in-phase loading, while for notched specimens out- mens with a higher moisture content, because the induced stress
of-phase loading resulted in shorter life than in-phase loading. It was lower due to the plasticizing effect of water. Samples with
should, however, be noted that for the same applied nominal nor- 0.2 wt% water content indicated higher fatigue strength compared
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 315

Fig. 23. Combined axial-torsion and torsion fatigue behaviors of a short glass fiber composite with a blend of polybutylene terephthalate and polyethylene terephthalate (a)
using tubular specimens, and (b) pure shear double V-shaped notched specimens [82].

to the samples with 3.5 wt% water content, as can be seen in


Fig. 24. In experiments on dried samples, a sharp decrease in mod-
ulus was observed due to the transition of matrix from glassy state
to the rubbery state. The wet samples were in the rubbery state
from the beginning of the test and the decrease of modulus was
predominantly due to damage initiation.
Bernasconi and Kulin [68] evaluated axial load-controlled fati-
gue behavior of short glass fiber reinforced polyamide-6 specimens
machined in the mold flow direction under R = 0.1 condition. Prior
to testing, samples were immersed at room temperature water
under the relative humidity of 50% for 7 days. During the cyclic
loading, crack initiation and propagation were not observed and
a general degradation in the surface of specimen was detected
instead. The transverse white lines in the surface of the specimen
were characterized as the sites of plastically deformed matrix.
Fotouh and Wolodko [38] studied the effect of moisture on axial Fig. 24. Effect of moisture on axial strain-controlled fatigue behavior of a short
glass fiber polyamide-6.6 tested under R = 1 condition and a frequency of 10 Hz
load-controlled fatigue behavior of natural hemp fibers in high [86]. Test specimen results shown with diamonds contain 0.2 wt% water tested at
density polyethylene. Neat specimens and specimens reinforced 23 °C and 50% relative humidity and test specimen results shown with circles
with 20 wt% natural fibers prepared in the mold flow direction. contain 3.5 wt% water tested at 23 °C and 90% relative humidity.
316 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

Fatigue tests were performed in dry and wet conditions. In the wet Noda et al. [57] investigated the effect of temperature on stress-
condition 1.46 wt% water was absorbed after 35 days of immersion controlled tension–tension fatigue behavior of short glass fiber
and the fatigue strength decreased by about 10%, as compared to polyamide-6.6 specimens in the mold flow direction. Tests were
the dry condition. The moisture swells fibers and decreases their conducted at temperatures in the range of 0–150 °C, and the S–N
strength and stiffness. As a result, the shear stress increases lines and data are shown in Fig. 26(a). The slope and the intercept
between fibers and the matrix and fiber debonding occurs faster. of S–N lines indicated a linear relationship with temperature as
shown in Fig. 26(b). This relationship was different for tempera-
tures above and below Tg due to different failure mechanisms
12.2. Temperature effects observed.
De Monte et al. [28] evaluated effects of temperature on axial
Jia and Kagan [87] evaluated effects of aging time and temper- load-controlled fatigue behavior of short glass fiber reinforced
ature on axial load-controlled fatigue behavior of short glass fiber polyamide-6.6 at room temperature and 130 °C. S–N lines at room
polyamides-6 and short glass fiber polyamide-6.6. Specimens were temperature and 130 °C had about the same slopes. Guster et al.
injection molded in the mold flow direction and tension–tension [25] investigated the effect of temperature on axial stress-con-
tests were carried out under R = 0.1 condition at 40, 23, and trolled fatigue behavior of short glass fiber reinforced polyamide-
121 °C. The highest fatigue strength was achieved at 40 °C, and 6T/6I with specimens molded parallel and perpendicular to the
fatigue strength significantly reduced at elevated temperatures. flow direction. Increasing the temperature from 23 to 120 °C
At 40 °C, polyamide-6 composite indicated higher fatigue decreased the fatigue strength considerably. In each flow direction,
strength than polyamide-6.6 composite, whereas at 23 °C polyam- longitudinal or transverse, the slopes of S–N curves remained con-
ide-6.6 composite showed higher fatigue strength than polyamide- stant at different temperatures. In the study of Sonsino and Moosb-
6 composite. For 121 °C experiments, fatigue strengths of both rugger [59] mentioned earlier, effect of temperature on axial
materials were nearly the same. Fatigue behavior of both materials fatigue behavior of short glass fiber reinforced polyamide-6.6
aged up to 1000 h was also investigated. Aging at room tempera- was investigated at temperatures of 22, 80, and 130 °C. High tem-
ture increased fatigue strength of both materials slightly, but at peratures reduced the fatigue strength significantly and S–N line
121 °C, aging did not influence fatigue strengths, as can be seen slopes were nearly constant at different temperatures.
from Fig. 25. At different temperatures and for different aging con-
ditions, normalized fatigue curves of both short glass fiber rein- 12.3. Thermo-mechanical fatigue (TMF) behavior
forced composites with tensile strength were collapsed on a
single curve. Thermo-mechanical fatigue (TMF) includes the combined effect
In the study of Handa et al. [67] on axial stress-controlled fati- of mechanical and thermal cycling and can be an important life lim-
gue behavior of short glass fiber reinforced polyamide-6.6 men- iting factor. Many components and structures in high temperature
tioned earlier with specimens in the mold flow direction and
tested under R = 0 condition, the influence of testing temperature
was evaluated at 0–120 °C. Both the slope and the intercept of S–
N lines indicated a linear relationship with the reciprocal of tem-
perature. This relationship was divided into two regions, divided
by the glass transition temperature, Tg of the material at 57 °C. At
temperatures above Tg, the slope and the intercept did not show
any dependency on temperature. At temperatures below Tg, the
slope and the intercept increased with increasing the reciprocal
of temperature. Hence, the failure mechanisms are different at
temperatures above and below Tg.

Fig. 26. Effect of testing temperature on fatigue behavior (a) and variation of S–N
Fig. 25. Effects of aging time and temperature on tension–tension fatigue behavior line slope and intercept with temperature (b) for short glass fiber reinforced
of short glass fiber reinforced polyamide-6 composite [87]. polyamide-6.6 [57].
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 317

applications are subjected to this type of fatigue. TMF of short fiber ends and grew along fiber and matrix interface and resulted in
composites has been the subject of a few studies. Pierantoni et al. fiber pull out. Fiber fracture was not a dominant failure mecha-
[88] performed TMF tests in temperature ranges of 40 °C to nism. SEM of fractured surfaces of TMF sharply notched specimens
40 °C and 40 °C to 120 ° on smooth and sharply notched (Kt = 9.8) was similar to those at isothermal fatigue condition at
(Kt = 9.8) specimens of short glass fiber reinforced polybutylene 40 °C and, therefore, it was concluded that TMF occurs at respec-
terephthalate. Prior to temperature cycling, a pre-strain level of tive minimum temperature and independent from the maximum
10, 50, or 80% of the tensile strain at room temperature was applied, temperature. TMF data were also collapsed on isothermal 40 °C
as indicated in Fig. 27(a). A linear viscoelastic material model was fatigue data by averaging SED range over a control volume around
developed to produce stress profile under TMF, based on Prony ser- the notch, as seen in Fig. 27(c). This parameter was validated with
ies [89]. Stabilized maximum nominal stress during TMF is shown isothermal experiments with various stress concentrations
against cycles to failure in Fig. 27(b). As pre-strain value increased, (2.5 < Kt < 9.8). Effect of mean stress was also considered by utiliz-
stabilized maximum nominal stress increased and fatigue life ing a mean stress parameter developed in [92].
decreased. Due to increased relaxation at higher temperature, as
the maximum applied temperature increased, maximum nominal
stress decreased, while fatigue life decreased. A higher stress con- 13. Fatigue behavior of SFRPC components
centration also resulted in shorter TMF life [90,91].
Schaaf et al. [90,91] studied the mechanisms of damage under Some studies have evaluated fatigue behavior of components
TMF tests described in [88]. Damage typically initiated at fiber made of SFRPCs. This includes studies of vibration welding, which

Fig. 27. (a) Variation of nominal stress under TMF load, (b) maximum nominal stress normalized by tensile strength of the notched specimens tested under thermo-
mechanical cyclic conditions versus cycles to failure, and (c) master curve of SED range versus cycles to failure correlating TMF and isothermal fatigue data [91].
318 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

is a widely used technique to join thermoplastics components, and performed at different alternating compressive loads. Fiber length
the influence of weld line on fatigue performance. Weld line is had a significant effect on fatigue performance of the leaf spring.
formed when a melting polymer fills in the injection molding cav- Short fiber composite indicated significantly shorter fatigue life,
ity with two or more flows. as compared with long fiber composite and this effect was more
Tsang et al. [93] studied axial fatigue behavior of vibration- pronounced in LCF.
welded butt joints of neat polyamide-6 and 6.6 and short glass Hartman et al. [97] evaluated fatigue behavior of smooth and
fiber reinforced polyamide-6 and 6.6. Load-controlled fatigue tests notched specimens and a pressure vessel made of short glass fiber
with applied stress in the mold flow direction were carried out reinforced polyamide-6.6. By means of FEA, the highly stressed
under R = 0.1 condition. The applied vibration dissipates frictional material volumes of specimens and the component were related
and viscous energy, and causes melting at the contact surface. together. Guster et al. [25] developed a fatigue life assessment
Applying a pressure then causes the two parts to be integrated. method for an injection molded ring spanner made of short glass
The welding decreased fatigue strength of these materials signifi- fiber reinforced polyamide-6T/6I. A simulation was performed to
cantly. In the absence of thermal failure, the welded material take into account the fiber orientation and distribution. The aniso-
was found as the site of crack initiation. Low pressure weld dis- tropic material behavior obtained from mold flow simulation was
played a better fatigue resistant than high pressure weld. then implemented into a finite element analysis. A good agreement
Lockwood et al. [94] studied the effect of temperature ranging regarding the area of maximum damage was obtained between
between 24 °C and 120 °C on axial R = 0.1 fatigue strength of vibra- experimental testing of the component and the simulation results.
tion-welded and unwelded short glass fiber polyamide-6. As tem- However, a conservative fatigue life within a factor of about 3 dif-
perature increased, fatigue strength of both welded and unwelded ference with the tested component was estimated.
composites reduced. Welded samples indicated lower fatigue Bernasconi et al. [98] investigated the effect of reprocessing of
strength than unwelded samples, due to great majority of fibers injection molded short glass fiber reinforced polyamide-6.6 in a flat
in welded samples transversely oriented with respect to the load- specimen and a clutch pedal. Fiber shortening effect was observed
ing direction. after processing in both specimen and component geometry, with
Boukhili et al. [95] evaluated the effect of weld line on fatigue a higher degree in clutch pedal due to a more complex shape. Fatigue
resistance of mold flow oriented samples of short glass fiber poly- strength reduced after recycling, however, fatigue strength was
carbonate under R = 0.1 loading condition and at a frequency of directly related to the tensile strength. In LCF, fatigue strength nor-
10 Hz. Tensile strength of welded short glass fiber polycarbonate malized with tensile strength was comparable in specimen and
was even lower than that of un-welded neat polycarbonate, indi- clutch pedal, while in HCF the normalized fatigue data were consid-
cating the detrimental effect of welding. Fatigue life of welded erably lower for pedal which was attributed to presence of sharp
samples was about 20% of fatigue life of un-welded samples. An notches and different types of loading in the specimen and clutch.
equation was derived to predict the effect of weld line on fatigue Vervoort [99] estimated fatigue life of a component on the bot-
strength based on the slope of un-welded S–N curve and tensile tom of the spare wheel of a car made of short fiber reinforced poly-
strengths of un-welded and welded samples. In the study of Bru- propylene. The orientation of fibers was simulated by a flow
nbauer et al. [26] on R = 0.1 axial and bending fatigue behavior of simulation program. Poisson’s ratios and longitudinal and trans-
short glass fiber polyamide-6.6 mentioned earlier, brittle fracture verse Young’s moduli representing the orthotropic behavior of
surface was observed in samples with weld line. Zhou and Mallick the material were input to the linear elastic finite element simula-
[22] studied the effect of weld line on axial fatigue behavior of tion. Simulations were performed based on both isotropic and
short glass fiber reinforced polyamide-6.6 specimens with axis orthotropic material properties. Significant differences at critical
perpendicular to the mold flow direction. A detrimental effect of locations were found between the two analyses. A good agreement
the weld line on the fatigue strength is evident from Fig. 4. between the orthotropic fatigue simulation and component testing
Sonsino and Moosbrugger [59] tested a fuel rail component was reported.
made of short glass fiber reinforced polyamide-6.6 under constant Gaier et al. [100] developed a method to estimate fatigue life of
and variable amplitude loading. Transferability of a fatigue life esti- belt pulley and ring spanner components made of short glass fiber
mation method from notched specimens to component was evalu- polyamide-6T/6I. Fully-revered fatigue tests were performed on
ated. Fatigue curves of notched specimens were transferred from short glass fiber polyamide-6T/6I specimens machined parallel
nominal system into local system by taking into account the stress and perpendicular to the mold flow direction. Fatigue strength at
concentration factor. The effect of stress gradient was also taken various directions was linearly interpolated between these two
into account by considering the volume of material with stresses directions. Mold flow simulations were performed on the compo-
larger than 80% of the maximum local stress. Linear elastic FEA nents to obtain orientation distribution of fibers in a tensor and
model of the fuel rail component was utilized to calculate the local characterize principal directions of anisotropy as well as share of
maximum principal stress as well as the maximum stressed vol- fibers in principal directions. S–N curves were locally modified to
ume. Estimated fatigue life under variable amplitude loading was each node of finite element mesh and at each node a fatigue dam-
close to the experimental life, utilizing Palmgren–Miner’s linear age analysis was performed by applying liner Palmgren–Miner’s
damage rule. rule (Eq. (17)) with the same b as of Zago et al. [78] study. Due
Zago and Springer [79] estimated the fatigue life of an automo- to rotation of principal stress directions, a factor was introduced
tive gear shift link made of short glass fiber reinforced co-polyam- to correct normal stresses, expressed as:
ide under variable amplitude loadings. Fatigue life under variable  
amplitude loadings was estimated based on the generalized Palm- rf r3
f ¼1þ 1 ð19Þ
gren–Miner’s rule in conjunction with the results of mold flow sim- sf r1
ulation software for distribution of fibers and FEA for stress
distributions. Estimated lives were close to experimental fatigue where rf and sf are tensile and shear fatigue limits, respectively, and
lives of the component. r3 and r1 are minimum and maximum absolute principal stresses,
Subramanian and Senthilvelan [96] evaluated fatigue behavior respectively. Critical location and fatigue life of the belt pulley were
of leaf springs made of short glass fiber reinforced polypropylene. reasonably estimated. A factor of about 7 difference was reported
Leaf springs were fabricated with various average fiber lengths between estimated fatigue lives under anisotropic and isotropic
ranging between 0.44 and 1.25 mm. Bending fatigue tests were finite element analyses of the ring spanner.
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 319

14. Summary been modeled with linear as well as nonlinear cumulative damage
models. Although nonlinear damage models may provide more
Applications of short fiber reinforced polymer composites have accurate fatigue life estimations, they typically require the use of
been rapidly increasing due to their low cost and easy manufactur- fitting parameter(s) from experimental data. Several studies have
ing with a high production rate. Most of such components are investigated multiaxial fatigue of SFRPCs including the effects of
under cyclic loading and prone to fatigue failure. Therefore, fatigue normal to shear stress ratio and phase shift at different tempera-
behavior and modeling of SFRPCs are of particular interest in tures and stress ratios. Stress ratio was found to have much more
recent years. This literature review presented a broad review of effect in axial loading as compared to torsion loading. Different
many factors influencing fatigue behavior and damage develop- failure paths have been observed, depending on the material, load-
ment in SFRPCs, including microstructural related effects as well ing mode, and stress level. A variety of fatigue criteria including
as effects related to loading condition and their service environ- energy-based models have been used to correlate the multiaxial
ment. However, it should be noted that due to the large variations fatigue data.
of matrices and fibers used in their manufacture, broad generaliza- Presence of moisture can have a substantial degrading effect on
tion of their fatigue behavior characterization based on a small fatigue behavior of SFRPCs and this effect is more pronounced for
number of studies is difficult. materials with no silane coupling agent between fiber and matrix.
Due to continuous initiation and growth of damage in the The glass transition temperature (Tg) decreases with increasing
matrix and at the fiber–matrix interface or at fiber ends, cyclic soft- moisture content due to the plasticizing effect of water and
ening is typically observed. Viscous characteristics can also increased chain mobility. Temperature also has a very significant
increase with increasing cycling frequency, temperature, or with effect on fatigue behavior of these materials. Fatigue strength sig-
fiber weight content. In injected modeled short fiber composites nificantly reduces by increasing the temperature from below to
a shell-core morphology typically exists where fiber orientation above Tg. At temperatures above Tg, the slope of S–N lines for differ-
is mainly along the mold flow direction in the shell and transverse ent temperatures have been found to be nearly constant in several
to the mold flow direction in the core. Fatigue strength of SFRPCs studies. If the temperature is cyclic in addition to the cyclic load,
strongly depends on the number of fibers aligned with the load thermo-mechanical fatigue is also an important aspect to be con-
direction and has been found to be proportional to the tensile sidered for these materials.
strength in the corresponding direction in some studies. The presence of weld line resulting from joining processes in
As the fiber aspect ratio increases, fatigue strength also manufactured components made of SFRPCs significantly reduces
increases until it reaches a plateau at a particular fiber aspect ratio. their fatigue performance. The welded region can be the site of
The effect of fiber length is typically different and more pro- crack initiation and fibers can orient in an unfavorable direction
nounced in LCF, as compared with HCF, and in bending versus axial with respect to the loading direction in this region. Mold flow sim-
loading. The interface between fiber and matrix is an important ulation and FEA along with fatigue analysis techniques in conjunc-
factor influencing fatigue behavior. Addition of silane coupling tion with smooth specimen constant amplitude fatigue data have
agent can significantly improve fatigue strength of SFRPCs. been successfully used to estimate fatigue life of components sub-
Matrix rupture, fiber debonding and pull out, and crazing are jected to complex loading conditions. However, since accurate
the commonly observed fatigue failure mechanisms. Which failure cumulative damage models for variable amplitude and/or complex
mechanism is dominant depends on a variety of factors including multiaxial stress states are not yet available, experimental compo-
the stress amplitude level, strength of the fiber–matrix interface, nent validation tests are essential.
fiber orientation with respect to the load direction, glass transition
temperature, and the present shell-core morphology. Dissipative
Acknowledgement
response of SFRPCs under cyclic loading has been used to charac-
terize and model their fatigue damage behavior and accumulation.
Financial support for this project was provided by General
Changes in the size and shape or slope of hysteresis loops, stiffness
Motors.
degradation and its variation under tensile and compressive loads,
cyclic creep or damping, hysteretic energy, strain energy, and non-
linear viscoelasticity are among the parameters used for such char- References
acterization and modeling.
The effect of mean stress on fatigue life of SFRPCs is significant [1] Biron M. Thermoplastics and thermoplastic composites. 2nd
ed. Massachusetts: William Andrew; 2013.
at all temperatures and is often accompanied by cyclic creep or rat- [2] Mallick PK. Fiber-reinforced composites: materials, manufacturing, and
cheting. This effect is less pronounced in notched geometries due design. 3rd ed. Florida: CRC Press; 1993.
to the stress gradient effect. A variety of parameters have been [3] Thomas S, Visakh PM. Handbook of engineering and specialty thermoplastics:
polyethers and polyesters. New Jersey: Jon Wiley & Sons; 2011.
used to model the mean stress effect. Fatigue behavior is also sub-
[4] Harper CA. Modern plastics handbook. New York: McGraw-Hill; 2000.
stantially influenced by the testing frequency. Self-heating is the [5] Silane coupling agents: connecting across boundaries. Version 2.0.
primary consequence of increased cyclic frequency resulting in [6] Stephens RI, Fatemi A, Stephens RR, Fuchs HO. Metal fatigue in engineering.
shorter fatigue lives, depending on the material, loading mode, 2nd ed. New York: Jon Wiley & Sons; 2000.
[7] Wang SS, Chim EM. Fatigue damage and degradation in random short-fiber
and stress level. Energy-based models have been used to predict SMC composite. J Compos Mater 1983;17:114–34.
the level of self-heating. Due to their viscoelastic nature, strength- [8] Launay A, Marco Y, Maitournam MH, Raoult I, Szmytka F. Cyclic behavior of
ening effect of frequency may also be observed at low frequencies short glass fiber reinforced polyamide for fatigue life prediction of automotive
components. Procedia Eng 2010;2:901–10.
for some SFRPCs, prior to self-heating. [9] Launay A, Maitournam MH, Marco Y, Raoult I, Szmytka F. Cyclic behaviour of
Stress concentrations and their gradient have a significant influ- short glass fibre reinforced polyamide: experimental study and constitutive
ence on fatigue behavior of SFRPCs, with a decreased influence as equations. Int J Plast 2011;27:1267–93.
[10] Mielke A. A mathematical framework for generalized standard materials in
temperature increases. As most components manufactured from the rate-independent case. Berlin Heidelberg: Springer; 2006.
SFRPCs contain geometric contours or features resulting in stress [11] Rice JR. Inelastic constitutive relations for solids: an internal-variable theory
concentrations, experimental results with notched specimens are and its application to metal plasticity. J Mech Phys Solids 1971;19:433–55.
[12] Janzen W, Ehrenstein GW. Hysteresis measurements for characterizing the
of particular significance to the component level performance eval- cyclic strain and stress sensitivity of glass fiber reinforced PBT. In: Plastics
uation. Fatigue behavior under variable amplitude loading has create a world of difference; 1989. p. 659–62.
320 S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321

[13] Lee SW, Youn JR. Characterization of short glass fiber filled polystyrene by [42] Kabir MR, Lutz W, Zhu K, Schmauder S. Fatigue modeling of short fiber
fiber orientation and mechanical properties. Macromol Symp reinforced composites with ductile matrix under cyclic loading. Comput
1999;148:211–27. Mater Sci 2006;36:361–6.
[14] Shaharuddin SIS, Salit MS, Zainudin ES. A review of the effect of moulding [43] Wang SS, Chim EM, Suemasu H. Mechanics of fatigue damage and
parameters on the performance of polymeric composite injection moulding. degradation in random short-fiber composites. II: analysis of anisotropic
Turk J Eng Environ Sci 2006;30:23–34. property degradation. J Appl Mech 1986;53:347–53.
[15] Vélez-García GM. Experimental evaluation and simulations of fiber [44] Wang SS, Goetz DP, Corten HT. Shear fatigue degradation and fracture of
orientation in injection molding of polymers containing short glass random short-fiber SMC composite. J Compos Mater 1984;18:2–20.
fibers. Blacksburg (VA): Virginia Polytechnic Institute and State University; [45] Klimkeit B, Castagnet S, Nadot Y, Habib AE, Benoit G, Bergamo S, et al. Fatigue
2012. damage mechanisms in short fiber reinforced PBT+PET GF30. Mater Sci Eng: A
[16] Horst JJ, Spoormaker JL. Mechanisms of fatigue in short glass fiber reinforced 2011;528:1577–88.
polyamide 6. Polym Eng Sci 1996;36:2718–26. [46] Hour KY, Sehitoglu H. Damage development in a short fiber reinforced
[17] Tanaka K, Kitano T, Egami N. Effect of fiber orientation on fatigue crack composite. J Compos Mater 1993;27:782–805.
propagation in short-fiber reinforced plastics. Eng Fract Mech [47] Hoppel CP. The effects of tension-tension fatigue on the mechanical behavior
2014;123:44–58. of short fiber reinforced thermoplastics. In: Proceedings American Society for
[18] De Monte M, Moosbrugger E, Quaresimin M. Influence of temperature and Composites; 1992. p. 509–18.
thickness on the off-axis behaviour of short glass fibre reinforced polyamide [48] Esmaeillou B, Fitoussi J, Lucas A, Tcharkhtchi A. Multi-scale experimental
6.6 – quasi-static loading. Compos Part A: Appl Sci Manuf 2010;41: analysis of the tension-tension fatigue behavior of a short glass fiber
859–71. reinforced polyamide composite. Procedia Eng 2011;10:2117–22.
[19] Arif MF, Saintier N, Meraghni F, Fitoussi J, Chemisky Y, Robert GM. Multiscale [49] Lazzarin P, Zambardi R. A finite-volume-energy based approach to predict the
fatigue damage characterization in short glass fiber reinforced polyamide-66. static and fatigue behavior of components with sharp V-shaped notches. Int J
Compos Part B: Eng 2014;61:55–65. Fract 2001;112:275–98.
[20] Zago A, Springer GS. Constant amplitude fatigue of short glass and carbon [50] De Monte M, Quaresimin M, Lazzarin P. Modelling of fatigue strength data for
fiber reinforced thermoplastics. J Reinf Plast Compos 2001;20:564–95. a short fibre reinforced polyamide 6.6 based on local strain energy density. In:
[21] Yau SS, Chou TW. Flexural fatigue of short fiber reinforced PEI, PES, and PEEK Proceedings of 16th international conference on composite materials. Kyoto,
thermoplastics. In: Advancing technology in materials and processes, vol. 30. Japan; 2007.
1985. p. 406–13. [51] Moosbrugger E, Wieland R, Gumnior P, Gerharz JJ. Design and dimensioning
[22] Zhou Y, Mallick PK. Fatigue performance of an injection-molded short E-glass of high loaded plastic parts in engine compartments. Materialprufung
fiber-reinforced polyamide 6,6. I. Effects of orientation, holes, and weld line. 2005;47:445–9.
Polym Compos 2006;27:230–7. [52] Ladeveze P, LeDantec E. Damage modelling of the elementary ply for
[23] Cosmi F, Bernasconi A. Fatigue behaviour of short fibre reinforced polyamide: laminated composites. Compos Sci Tech 1992;43:257–67.
morphological and numerical analysis of fibre orientation effects. Forni di [53] Nouri H, Meraghni F, Lory P. Fatigue damage model for injection-molded
Sopra 2010:239–42. short glass fibre reinforced thermoplastics. Int J Fatigue 2009;31:934–42.
[24] Cosmi F, Dreossi D. Numerical and experimental structural analysis of [54] Nouri H, Czarnota C, Meraghni F, Lory P. Fatigue damage model for short glass
trabecular architectures. Mecc 2007;42:85–93. fibre reinforced thermoplastics (PA6-GF30).
[25] Guster C, Pinter G, Mösenbacher A, Eichlseder WC. Evaluation of a simulation [55] Meraghni F, Nouri H, Bourgeois N, Czarnota C, Lory P. Parameters
process for fatigue life calculation of short fibre reinforced plastic identification of fatigue damage model for short glass fiber reinforced
components. Procedia Eng 2011;10:2104–9. polyamide (PA6-GF30) using digital image correlation. Procedia Eng
[26] Brunbauer J, Mösenbacher A, Guster C, Pinter G. Fundamental influences on 2011;10:2110–6.
quasistatic and cyclic material behavior of short glass fiber reinforced [56] Avanzini A, Donzella G, Gallina D. Fatigue damage modelling of PEEK short
polyamide illustrated on microscopic scale. J Appl Polym Sci 2014. 10.1002/ fibre composites. Procedia Eng 2011;10:2052–7.
app.40842. [57] Noda K, Takahara A, Kajiyama T. Fatigue failure mechanisms of short glass-
[27] Wang ASD, Tung RW, Sanders BA. Size effect on strength and fatigue of a fiber reinforced nylon 66 based on nonlinear dynamic viscoelastic
short fiber composite material. In: Emerging technologies in aerospace measurement. Polym 2001;42:5803–11.
structures, design, structural dynamics and materials; 1980. p. 37–52. [58] Komatsu S, Takahara A, Kajiyama T. Effect of cyclic fatigue conditions on
[28] De Monte M, Moosbrugger E, Quaresimin M. Influence of temperature and nonlinear dynamic viscoelasticity and fatigue behaviors for short glass-fiber
thickness on the off-axis behaviour of short glass fibre reinforced polyamide reinforced Nylon6. Polym 2003;35:844–50.
6.6 – cyclic loading. Compos Part A: Appl Sci Manuf 2010;41:1368–79. [59] Sonsino CM, Moosbrugger E. Fatigue design of highly loaded short-glass-fibre
[29] Bernasconi A, Davoli P, Basile A, Filippi AA. Effect of fibre orientation on the reinforced polyamide parts in engine compartments. Int J Fatigue
fatigue behaviour of a short glass fibre reinforced polyamide-6. Int J Fatigue 2008;30:1279–88.
2007;29:199–208. [60] Avanzini A, Donzella G, Gallina D, Pandini S, Petrogalli C. Fatigue behavior and
[30] Agarwal BD, Broutman LJ, Chandrashekhara K. Analysis and performance of cyclic damage of peek short fiber reinforced composites. Compos Part B: Eng
fiber composites. 3rd ed. New Jersey: Jon Wiley & Sons; 2006. 2013;45:397–406.
[31] Grove D, Kim H. Fatigue behavior of long and short glass reinforced [61] Mallick PK, Zhou Y. Effect of mean stress on the stress-controlled fatigue of a
thermoplastics. Adv Automot Plast Compon Technol 1995:77–83. short E-glass fiber reinforced polyamide-6,6. Int J Fatigue 2004;26:941–6.
[32] Lavengood RE, Gulbransen L. The effect of aspect ratio on the fatigue life of [62] Sauer J, McMaster A, Morrow D. Fatigue behavior of polystyrene and effect of
short boron fiber reinforced composites. Polym Eng Sci 1969;9:365–9. mean stress. J Macromol Sci, Part B: Phys 1976;12:535–62.
[33] Meneghetti G, Ricotta M, Lucchetta G, Carmignato S. An hysteresis energy- [63] Oka H, Narita R, Akiniwa Y, Tanaka K. Effect of mean stress on fatigue strength
based synthesis of fully reversed axial fatigue behaviour of different of short glass fiber reinforced polybuthyleneterephthalate. Key Eng Mater
polypropylene composites. Compos Part B: Eng 2014. 10.1016/ 2007;340:537–42.
j.compositesb.2014.01.027. [64] Bellenger V, Tcharkhtchi A, Castaing P. Thermal and mechanical fatigue of a
[34] Subramanian C, Senthilvelan S. Effect of fiber length on hysteretic heating of PA66/glass fibers composite material. Int J Fatigue 2006;28:1348–52.
discontinuous fiber-reinforced polypropylene. Int J Polym Mater [65] Esmaeillou B, Ferreira P, Bellenger V, Tcharkhtchi A. Fatigue behavior of
2009;58:347–54. polyamide 66/glass fiber under various kinds of applied load. Polym Compos
[35] Takahara A, Magome T, Kajiyama T. Effect of glass fiber–matrix polymer 2012;33:540–7.
interaction on fatigue characteristics of short glass fiber-reinforced [66] Lang R, Manson J. Crack tip heating in short-fibre composites under fatigue
poly(butylene terephthalate) based on dynamic viscoelastic measurement loading conditions. J Mater Sci 1987;22:3576–80.
during the fatigue process. J Polym Sci Part B: Polym Phys 1994;32: [67] Handa K, Kato A, Narisawa I. Fatigue characteristics of a glass-fiber-reinforced
839–49. polyamide. J Appl Polym Sci 1999;72:1783–93.
[36] Yamashita A, Higashi S, Komatsu ST, Takahara A, Kajiyama T. Fatigue analysis [68] Bernasconi A, Kulin RM. Effect of frequency upon fatigue strength of a short
of short glass fibre-reinforced Nylon 6 on the basis of dynamic viscoelastic glass fiber reinforced polyamide 6: a superposition method based on cyclic
measurements under cyclic fatigue. Key Eng Mater 1997;137:147–54. creep parameters. Polym Compos 2009;30:154–61.
[37] Mandell JF, McGarry FJ, Huang DD, Li CG. Some effects of matrix and interface [69] Jegou L, Marco Y, Le Saux V, Calloch S. Fast prediction of the Wöhler curve
properties on the fatigue of short fiber-reinforced thermoplastics. Polym from heat build-up measurements on Short Fiber Reinforced Plastic. Int J
Compos 1983;4:32–9. Fatigue 2013;47:259–67.
[38] Fotouh A, Wolodko J. Fatigue behavior of natural fiber reinforced [70] Doudard C, Calloch S, Cugy P, Galtier A, Hild F. A probabilistic two-scale
thermoplastic composites in dry and wet environments. In: ASME model for high-cycle fatigue life prediction. Fatigue Fract Eng Mater Struct
international mechanical engineering congress and exposition; 2011. p. 71–7. 2005;28:279–88.
[39] Zhou J, D’Amore A, Yang Y, He T, Li B, Nicolais L. Flexural fatigue of short glass [71] Haldar AK, Senthilvelan S. Notch effect on discontinuous fiber reinforced
fiber reinforced a blend of polyphenylene ether ketone and polyphenylene thermoplastic composites. Key Eng Mater 2011;471:173–8.
sulfide. Appl Compos Mater 1994;1:183–95. [72] Yamamoto T, Hyakutake H. Fatigue damage and failure for notched plates of
[40] Casado JA, Gutiérrez-Solana F, Polanco JA, Carrascal I. The assessment of short glass fiber reinforced polypropylene. Sci Eng Compos Mater
fatigue damage on short-fiber-glass reinforced polyamides (PA) through the 2005;12:117–24.
surface roughness evolution. Polym Compos 2006;27:349–59. [73] Nisitani H, Noguchi H, Kim YH. Evaluation of fatigue strength of plain and
[41] Hitchen S, Ogin S. Damage accumulation during the fatigue of an injection notched specimens of short carbon-fiber reinforced polyetheretherketone in
moulded glass/nylon composite. Compos Sci Tech 1993;47:83–9. comparison with polyetheretherketone. Eng Fract Mech 1992;43:685–705.
S. Mortazavian, A. Fatemi / International Journal of Fatigue 70 (2015) 297–321 321

[74] Mortazavian S, Fatemi A. Notch effects on fatigue behavior of thermoplastics. [88] Pierantoni M, De Monte M, Papathanassiou D, De Rossi N, Quaresimin M.
Adv Mater Res 2014;891–892:1403–9. Viscoelastic material behaviour of PBT-GF30 under thermo-mechanical cyclic
[75] Meneghetti G, Quaresimin M. Fatigue strength assessment of a short fiber loading. Procedia Eng 2011;10:2141–6.
composite based on the specific heat dissipation. Compos Part B: Eng [89] Tzikang C. Determining a Prony series for a viscoelastic material from time
2011;42:217–25. varying strain data; NASA/TM-2000-210123 ARL-TR-2206, 2000.
[76] Meneghetti G, Quaresimin M, De Monte M. Fatigue strength assessment of a [90] Schaaf A, De Monte M, Hoffmann C, Vormwald M, Quaresimin M. Damage
short fibre-reinforced plastic based on the energy dissipation. In: Proceedings mechanisms in PBT-GF30 under thermo-mechanical cyclic loading. In: The
of 16th international conference on composite materials. Kyoto, Japan; 2007. 29th international conference of the polymer processing society, vol. 1593.
[77] Hoppel CP, Pangborn RN, Thomson RW. Damage accumulation during 2014. p. 600–5.
multiple stress level fatigue of short-glass-fiber reinforced styrene-maleic [91] Schaaf A, De Monte M, Moosbrugger E, Vormwald M, Quaresimin M. Life
anhydride. J Thermoplast Compos Mater 2001;14:84–94. estimation methodology for short fiber reinforced polymers under thermo-
[78] Zago A, Springer GS, Quaresimin M. Cumulative damage of short glass fiber mechanical loading in automotive applications. In: Proceeding of 4th
reinforced thermoplastics. J Reinf Plast Compos 2001;20:596–605. symposium on structural durability. Darmstadt, Germany; 2014.
[79] Zago A, Springer GS. Fatigue lives of short fiber reinforced thermoplastics [92] Lazzarin P, Sonsino CM, Zambardi R. A notch stress intensity approach to assess
parts. J Reinf Plast Compos 2001;20:606–20. the multiaxial fatigue strength of welded tube-to-flange joints subjected to
[80] Dreißig J, Jaschek K, Büter A, De Monte M. Fatigue life estimation for short combined loadings. Fatigue Fract Eng Mater Struct 2004;27:127–40.
fibre reinforced polyamide components under variable amplitude loading. In: [93] Tsang K, DuQuesnay D, Bates P. Fatigue properties of vibration-welded nylon
2nd International conference on material and component performance under 6 and nylon 66 reinforced with glass fibres. Compos Part B: Eng
variable amplitude loading. Darmstadt, Germany; 2009. 2008;39:396–404.
[81] De Monte M, Moosbrugger E, Jaschek K, Quaresimin M. Multiaxial fatigue of a [94] Lockwood KT, Zhang Y, Bates PJ, DuQuesnay DL. Effect of temperature on
short glass fibre reinforced polyamide 6.6 – fatigue and fracture behaviour. fatigue strength of vibration welded and unwelded glass reinforced nylon 6.
Int J Fatigue 2010;32:17–28. Int J Fatigue 2014;66:111–7.
[82] Klimkeit B, Nadot Y, Castagnet S, Nadot-Martin C, Dumas C, Bergamo S, et al. [95] Boukhili R, Gauvin R, Gosselin M. The effect of weld-lines on the fatigue
Multiaxial fatigue life assessment for reinforced polymers. Int J Fatigue resistance of short glass fiber reinforced polycarbonate. J Thermoplast
2011;33:766–80. Compos Mater 1989;2:78–90.
[83] Moosbrugger E, De Monte M, Jaschek K, Fleckenstein J, Büter A. Multiaxial [96] Subramanian C, Senthilvelan S. Fatigue performance of discontinuous fibre-
fatigue behaviour of a short-fibre reinforced polyamide – experiments and reinforced thermoplastic leaf spring. In: Proceedings of the institution of
calculations. Materialwissenschaft Werkstofftechnik 2011;42:950–7. mechanical engineers, Part L: Journal of materials design and applications,
[84] Launay A, Maitournam MH, Marco Y, Raoult I. Multiaxial fatigue models for vol. 224. 2010. p. 93–100.
short glass fiber reinforced polyamide – part I: nonlinear anisotropic [97] Hartmann J, Moosbrugger E, Büter A. Variable amplitude loading with
constitutive behavior for cyclic response. Int J Fatigue 2013;47:382–9. components made of short fiber reinforced polyamide 6.6. Procedia Eng
[85] Launay A, Maitournam MH, Marco Y, Raoult I. Multiaxial fatigue models for 2011;10:2009–15.
short glass fibre reinforced polyamide. Part II: fatigue life estimation. Int J [98] Bernasconi A, Davoli P, Armanni C. Fatigue strength of a clutch pedal made of
Fatigue 2013;47:390–406. reprocessed short glass fibre reinforced polyamide. Int J Fatigue
[86] Barbouchi S, Bellenger V, Tcharkhtchi A, Castaing P, Jollivet T. Effect of water 2010;32:100–7.
on the fatigue behaviour of a pa66/glass fibers composite material. J Mater Sci [99] Vervoort S. Fatigue analysis of fibre-reinforced polymers. In: Proceedings of
2007;42:2181–8. the European congress on computational methods in applied sciences and
[87] Jia N, Kagan VA. Effects of time and temperature on the tension-tension engineering. Vienna, Austria; 2012.
fatigue behavior of short fiber reinforced polyamides. Polym Compos [100] Gaier C, Dannbauer H, Werkhausen A, Wahlmueller R. Fatigue life prediction
1998;19:408–14. of short fiber reinforced plastic.

Potrebbero piacerti anche