Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

https://doi.org/10.1007/s40722-020-00184-6

RESEARCH ARTICLE

Fatigue damage analysis of GFRP composites using digital image

correlation
Hassan Gonabadi1 · Adrian Oila1 · Arti Yadav1 · Steve Bull1

Received: 11 August 2020 / Accepted: 31 December 2020 / Published online: 18 February 2021
© The Author(s) 2021

Abstract
Glass fibre reinforced polymer composites are frequently used in marine applications where the combined effects of cyclic
loads and the seawater environment limit their fatigue life. This paper aims to demonstrate the degradation that seawater
causes to the stiffness of the composites. Three-point bending fatigue properties of cross-ply woven glass fibre composites
commonly used to manufacture tidal turbine blades are reported for both wet and dry conditions. Failure analysis based on the
Digital Image Correlation method was performed to identify damaged zones on the test coupon surface and to follow failure
progression during the fatigue tests. To characterize the damage in the composite, stiffness degradation has been monitored
during the entire fatigue history. Scanning electron microscopy was used to identify multiple failure mechanisms on the
specimen fracture surface. In addition, for further verification of microscopy results, X-Ray Micro-computed tomography,
was used to characterize the internal damage such as delamination. From the full-field strain measurement technique and
microscopic examination of failed samples, it was found that distributed localized strains are evidence of the number of resin
cracks and de-bonded areas. SEM examination shows a degraded fibre/matrix interface region due to the action of seawater.

Keywords Composites · Tidal turbine blades · Digital image correlation · Fatigue

List of symbols Nf Number of cycles to failure

F Force D Damage
L Distance between supports in 3 PB test set up E Elastic modulus
b Width of test specimen Ei Initial Elastic modulus
h Thickness of test specimen Ef Final elastic modulus
d Vertical deflection lin
X10 Point of line corresponding to 10% of consumed
σf Flexural stress fatigue life
ef Flexural strain lin
X90 Point of line corresponding to 90% of consumed
emax, i Max initial strain fatigue life
exx Longitudinal normal strain ΔX% Index of irreversible composite degradation
eyy Transvers normal strain
exy Shear strain
K and A Constants in power law equation 1 Introduction

* Hassan Gonabadi One of the major challenges in the marine energy industry is
hassan.izadi‑[email protected] the ability to predict fatigue failure of composite structures.
Adrian Oila High-cycle fatigue occurs in many marine energy conversion
[email protected] devices; in particular tidal turbine blades that are subject to
Arti Yadav an environmentally affected fatigue load, and therefore the
[email protected] corrosion fatigue capability is a life-limiting factor. Glass
Steve Bull fibre reinforced epoxy resin composites are frequently used
[email protected] in the design of tidal turbine blades and the primary fatigue
failure modes are delamination and fibre breakage (Bois-
1
School of Engineering, Newcastle University, seau et al. 2013). In this application, although there is a
Newcastle upon Tyne, UK

13
Vol.:(0123456789)
26 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

layer of gel-coat to protect the GFRP material; under some fluid diffusion accelerates damage induced by mechanical
circumstances due to high static and cyclic loads the gel-coat stresses (Suri and Perreux 1995). In the literature, It has
could fail, leaving the GFRP material directly exposed to the been shown that the mechanisms of environmental stress
seawater environment. Therefore, it is important to under- corrosion cracking (ESCC) in FRP composites (Renaud
stand the amount of degradation caused by seawater and its and Greenwood 2005) are fibre and fibre/matrix interface
effect on the mechanical properties and failure mechanisms degradation which is due to the water penetration into the
of these materials. network of resin cracks and de-bonded fibres (Schmitz and
Determining the resistance of composites to the com- Metcalfe 1966, Metcalfe and Schmitz 1972, Vauthier et al.
plex state of fluctuating stress is a fundamental problem in 1995, Izadi Gonabadi 2019). The main causes of degradation
many engineering applications. Composite materials exhibit are either corrosive species (i.e. chloride) or the mechanisms
very complex failure phenomena including multiple dam- of ionic exchange between metallic cations ( Na+ and K+ )
age mechanisms and their interactions. This is due to the and hydrogen ions ( H+) from the attacking environment. In
anisotropic characteristics of the mechanical properties of the latter mechanisms, separation of metallic ions from the
laminates as well as the presence of interfaces causing stress fibre surface can develop nucleation sites for micro-cracks
redistribution. Unlike isotropic materials where the predomi- (Charles 1958b, a; Charles 1958b, a; Izadi Gonabadi 2019).
nant failure mechanism is the initiation and propagation of The other factor causing the degradation of polymer-
a single crack, fatigue failure in fibre reinforced composites based composite tidal turbine blades is tribological phe-
is due to general degradation throughout the specimen vol- nomena such as sediment and solid particle erosion as well
ume. Fatigue failure modes in composites are resin cracking, as cavitation erosion. Tribological mechanisms in GFRP
interfacial de-bonding between fibre and polymer matrix, composites have been investigated by considering the effects
delamination and fibre breakage (Vassilopoulos and Kel- of erosion parameters, contributing to understand the effect
ler 2011). The different damage mechanisms coupled with of seawater environment on the degradation of composite
complex strain fields significantly limit understanding of the (Ahamed et al. 2016, Rasool et al. 2016, Rasool and Stack
true nature of fatigue failure. 2019).
The most important obstacle to predict the fatigue life Digital image correlation (DIC) is non-contact optical
of composites is the lack of information regarding com- technique for the measurement of mechanical displacement
posite material properties that could be used as a basis for and strain. Numerous investigations of damage development
the development of the fatigue failure criteria (Schaff and using DIC have been reported in the literature (Chu et al.
Davidson 1997, Quaresimin et al. 2010). In order to estab- 1985, Sun et al. 1997, Zhang et al. 2006, Godara and Raabe
lish acceptable failure criteria, fatigue failure can be defined 2007, Canal et al. 2012, He et al. 2012, Laurin et al. 2012,
as stiffness degradation which is a suitable failure criterion Crupi et al. 2013). Inter-laminar shear and tensile properties
for many composites. Changes in stiffness as an index of of composites have been investigated using the DIC method
composite microscopic damage can be easily measured where full-field strain maps show inhomogeneous strain dis-
and interpreted. In fatigue bending, the degradation rate is tribution in the areas on the sample surface where delamina-
related to the stress within the composite material. Initially, a tion occurs (Chu et al. 1985, Godara and Raabe 2007, Canal
high load is required to deflect the test coupon. As the cyclic et al. 2012, He et al. 2012, Laurin et al. 2012, Crupi et al.
loading continues the load required for the same deflection 2013), however, little knowledge is available about apply-
becomes smaller. This reduction in load (and stress) within ing DIC analysis to environmentally assisted fatigue loads.
the specimen signals the degradation. Therefore, in the pro- In the past, the investigations of the fatigue behaviour
cess of damage evolution of a constant bending amplitude of GFRP composites have been focused on tensile fatigue,
fatigue test, at the beginning of life (first stage), damage with less attention being given to the fatigue behaviour under
initiation and propagation causes a drop in the composite bending modes especially in the case where the material is
stiffness; in the second stage, a gradual material degradation undergoing fluctuating loads in a fluid. The present inves-
which is a function of cyclic load levels and environmental tigation provides data for fatigue tests of specimens during
conditions is observed. Finally, the third stage follows with seawater exposure. A large number of coupons were tested
a catastrophic reduction of material properties and a large under a three-point bending (3 PB) fatigue load configu-
loss in stiffness in the damaged zones (Shokrieh and Lessard ration at various constant displacements in both seawater
1997, Giancane et al. 2010). and air. Using the DIC technique a better knowledge of
In terms of the environmentally affected fatigue perfor- fatigue behaviour of a standard epoxy/glass fibre composite
mance of composite materials it must be noted that envi- was acquired by measuring surface strain maps during the
ronmental exposure and mechanical forces can act together fatigue test. The DIC was also used to determine the change
to speed up the degradation process in the composite mate- in elastic modulus (modulus decay) which is associated with
rial i.e. mechanical forces accelerate fluid penetration and fatigue damage. Modulus decay gives clear evidence about

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 27

Fig. 1  Micrographs of the

GFRP laminate, a SEM reveals
the cross section of micro-
structure with biaxial fibre
orientation. b 3D reconstructed
volumetric model obtained by
µ-CT

Table 1  GFRP composite specifications

Weave style Symmetric laminate with plain

woven fabric as shown in
Fig. 2
Layups (0◦/90°)
Thickness of each lamina 0.2 mm

Fig. 3  Experimental set up for a 3 PB test equipped with DIC system

fabric (orthogonally oriented). 25 plies of fabric were lami-

nated in a stacking sequence of (0º/90º) layup. By the matrix
burn-off method the glass fibre volume fraction of the GFRP
composites was determined as approximately 35 ± 2%. To
examine the quality of GFRP laminates, scanning electron
microscopy was used to observe a polished cross section
(Fig. 1). The glass transition temperature, Tg, of the epoxy
resin matrix was measured by Differential Scanning Calo-
Fig. 2  Schematic of plain-woven fabric, (ply thickness: 0.2 mm)
rimetry (DSC) to be 122 ºC. Table 1 lists the composite
laminate specification.
damage localization and gradual degradation of the mechan-
ical properties of composites during their fatigue history. 2.2 Quasi‑static three point bending test (3PBT)

The quasi-static tests have been conducted by means of a

2 Experimental universal test machine with a 50KN load cell in the three-
point bending (3 PB) mode. This test produces a tensile
2.1 Material stress in the convex side and compressive stress on the con-
cave side. This creates a region of shear stress along the
The material under study is woven E-glass fibre reinforced midline; in order to ensure that this is minimized the sam-
epoxy resin composite produced by a commercial supplier ple thickness must be kept to less than one-sixteenth of the
to the G10 standard (Johnson and Johnson 1987). The as- distance between the lower supports (span). The machine
received materials were 1.2 m ×1.2 m × 0.005 m plates. was also equipped with a video gauge (DIC measurement
Plates were manufactured using a vacuum-assisted resin set up) to record the strain during the test (Fig. 3). Bar spec-
infusion process. Structurally, the reinforcement material imens of dimensions as shown in Table 2 were cut from
consists of longitudinal and transverse bundles of glass plates according to the standard test method of BS EN ISO
fibres as shown in Fig. 1. The fibres are in a plain-woven 14,125 (STANDARD and ISO 1998). During the test, a

13
28 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

Table 2  Geometrical dimensions of specimens

Values Length (mm) Width (mm) Thickness (mm)

Nominal 150 15 5
Actual 150 ± 0.8 15 ± 0.3 5 ± 0.05

load cell and the video gauge record the load and displace-
ment, respectively. Force–displacement was converted to
stress–strain to standardize the curves for all samples, this
was done by applying standard bending equations (devel-
oped for a homogeneous material by assuming a constant
sample curvature), where F is the force applied by the load-
ing roller, L is the distance between supports, h and b are
thickness and width of the test coupon, respectively, and D
is the vertical deflection of the sample at the centre.

𝜎f = 3FL∕2bh2
Fig. 4  Exploded-view schematic of 3 PB fatigue rig

ef = 6Dh∕L2 (1)
Surface related damage during fatigue was observed and
failure mechanisms were analysed using reflected light and
Scanning Electron Microscopy (SEM). The fracture surfaces
2.3 3 PB fatigue test were sputter coated with gold prior to SEM examination.

Fatigue tests were performed in the 3 PB mode at the loading 2.4 Digital image correlation
frequency of 1 Hz in two different environments, i.e. natural
North Seawater from the Dove Marine Laboratory (3.5% DIC is a full-field strain measurement technique which can
dissolved salts, predominantly NaCl (2.4%), M ­ gSO4 (0.5%) be used to build the strain maps in composite materials, pro-
and ­MgCl2 (0.3%) and laboratory air. The specimens were viding information about strain localization (Izadi Gonabadi
cut from the same panel to minimise sample-to-sample vari- 2019). The technique is based on comparing digital images
ation. A total of 60 samples were obtained; 30 samples were before and after deformation and tracking blocks of pixels in
tested under fatigue loading while exposed to air, whilst the a target area. This process is viable when the object surface
other 30 specimens were tested under fatigue loading while has a random speckle pattern, which depending on the sur-
immersed in seawater with the same configuration at room face morphologies of the samples, may need to be created
temperature. 3 PB fatigue tests involved cycling specimens first. In this study, the pattern was created by alternate spray-
under constant deflection using a specially designed test ing white and black paint on the sample surface. The DIC
machine (load cell of 25 KN) until failure. In order to avoid technique can also be used to monitor the evolution of strain
specimen motion, impact on the sample and the variation of during the mechanical testing of materials (Metwalli et al.
the loaded point on test coupon during the fatigue test, the 1987; Ukyo and Masuda 2004), to investigate the crack tip
strain ratio was kept at R = 𝜀 min = 0.1. The configuration for
𝜀
max
and crack propagation (Abanto-Bueno and Lambros 2002,
the 3 PB fatigue test (Fig. 4) was the same as used for quasi- Lee et al. 2009, Richter‐Trummer et al. 2011), and to detect
static 3 PB test in terms of test coupon dimensions, geom- damage development in composites (Izadi Gonabadi 2019).
etry, span supports and loading roller positions. According However, as the spatial resolution in DIC method depends
to the standard test method of ES BN ISO13003 (ISO 2003) on the optical resolution of the experimental set up and the
for bending fatigue behaviour of composites, five coupons quality of speckle pattern, studies on the above phenomena,
were tested at each of different strain levels between 50 and in particular crack propagation, are challenging. Although
90% of their dry maximum static strain. in many cases large crack opening can be identified using
Fatigue tests were interrupted at regular time intervals the naked eye, smaller cracks that are not visible to the
to perform quasi-static 3 PB tests to measure elastic modu- naked eye can be identified using combined digital image
lus decay due to the cyclic loads. Post static bending tests correlation and dye penetrant on the surface of the sam-
on fatigue loaded samples also included a DIC measure- ple. By using DIC as a map of localized deformation/strain,
ment to obtain a full-field strain map on the sample surface. with one image taken before cracking occurs and one taken

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 29

afterwards, the full extent of crack opening and crack propa- maximum power of 10 W. The highest achievable resolu-
gation paths can be examined (McCormick and Lord 2010). tion is 0.9 µm. A spot size of < 5 µm is obtained at 4 W. A
The CCD camera (IMETRUM) with a 300 mm work- precision object manipulator with two translations and one
ing distance lens was placed in front of the specimen so rotation facilitates rotating the sample for the acquisition
that it could be viewed edge on during bending. The cam- of tomographic data, raising/lowering the sample to select
era features a spatial resolution of 0.0055 mm/pixel and a region of interest, and translation along the optical axis
was equipped with lenses of 0.193 magnification. Digital to adjust the magnification. The X-ray detector consists of
images of the specimen surface were recorded with a fre- an X-ray scintillator screen which is fiber optically coupled
quency of 25 Hz. After testing, the series of digital images (at an image ratio of 3.7:1) to a high resolution (1024/1024
were processed using the VIC 2D software (from Correlated pixel) cooled 2 k × 2 k 16bit CCD x-ray camera.
Solutions, Inc). A field size (5 mm × 60 mm) was defined
on each initial image of the fatigued samples (Fig. 5). The
most important parameters related to calculations using DIC
software are facet size, step size and interpolation type. The 3 Results and discussion
results here were obtained with facet size of 30 pixels cor-
responding to 165 m, step size to 10 pixels which is equal In this study, using a data logger connected to the load cell in
to a spatial resolution of 55 m and bi-cubic interpolation. the fatigue rig, the load-time response during fatigue tests is
recorded (Fig. 6a). When the first load drop occurs (indicat-
2.5 X‑ray 3D micro‑computed tomography ing failure in one of the samples), the fatigue test was inter-
rupted and the samples were checked under a light micro-
X-ray (µ-CT) measurements were carried out using an X scope to identify the failed sample which was removed. The
Radia/Zeiss Versa XRM 410. The system consists of a fatigue test is then carried on until the next load drop occurs.
micro-focus X-ray source operating at 150 kV, within a The same procedure was used for the rest of the samples
until the all have failed. Although fatigue tests have been
conducted for multiple samples with the above procedure, a
fatigue test was repeated using a single sample for each strain
level to check consistency; each test is therefore repeated 5
times (Fig. 6b). As the bending fatigue tests are conducted
under strain control mode, the applied load, required to keep
the given deflection constant, decreases with the number of
cycles. In other words, stiffness degradation of composites
during fatigue tests can be confirmed directly via the output
from the fatigue test rig.

3.1 Strain‑life curves

Fig. 5  a An image of spray-coated specimen showing the stochastic Various curves can be drawn for the presentation of fatigue
speckle pattern and the region of interest for the strain analysis. Sche-
test results and life prediction. In this study a power-law rela-
matic drawing illustrating the fundamentals of the DIC approach indi-
cating b initial image (before deformation), c distorted image (after tionship between maximum initial strain (𝜀max,i) and fatigue
loading) life was used.

Fig. 6  Load versus time during

the fatigue test of a multiple
specimens b single speci-
men captured via data logger
(applied strain level is 1.2%)

13
30 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

(
emax,i = K logNf
)−A
(2) The power-law coefficients determined by fitting the
experimental data using the least-squares method are shown
where Nf is the number of cycles to failure, K and A are in Fig. 7. For the prediction of the fatigue life of the com-
constants. The constants can be determined by plotting the posites under strain levels outside the testing range, Eq. 2 is
strain versus number of cycles to failure and fitting Eq. 2 used. The regression fitting parameter (A) is 0.846 for GFRP
to the data. In order to determine the lifetime diagram, five in air and 0.88 in seawater indicating the strength loss per
specimens were tested at a minimum of seven separate lev- decade of cycles of the material. The loss in strength per
els of imposed strain. Also for measuring the monotonic decade is higher for GFRP composites when tests are per-
strength and its corresponding deflection/strain level for the formed in seawater.
selected test method, five specimens were initially tested
under 3 PB to failure. The fatigue failure criterion was 3.2 Fatigue failure analysis using DIC
defined as the onset of a DIC-detectable delamination. The
strain-life fatigue test results for GFRP composite along Damage analysis based on DIC has been conducted to locate
with the corresponding fits to Eq. 2 are shown in Fig. 7, in the damaged zones on the specimen surface and to follow
which each data point indicates one valid flexural fatigue failure evolution during fatigue life. Figure 8 shows an
test. Comparing ε-N curves obtained in the air with seawater example of the random speckle pattern created using black
environment provides an indication of the durability of these and white spray paints on a sample. The figure also shows
materials and clearly reveals the influence of the environ- a typical failure in one of the fatigue bending specimens
ment on reducing fatigue life. after 5.9 million cycles. The failure mode is a delamination
As can be seen in Fig. 7, fatigue data points can be clas- starting beneath the loading roller (upper support) on the
sified into two distinctive regimes, a regime of stress-envi- tension side of the sample. All bending specimens exhibited
ronment dependence, and a regime of stress-dependence and the same behaviour.
environment-independence. This is shown by a transition Figure 9 shows the results of strain analysis at the very
line and the two arrows in Fig. 7. For samples fatigue loaded beginning of cycling and at the peak cyclic strain equal to
at high strain levels (strain level higher than 1.1%), fatigue 30% of the ultimate static strain. The displacements meas-
data points are mostly overlapping, so low cycle fatigue of ured in this method were numerically differentiated using
composite can be categorized as in a stress-dependent and the strain computation algorithm built in the VIC-2D soft-
environment-independent regime. A substantial difference ware to compute the Lagrange surface strain tensor com-
in fatigue data points can be observed for those of specimens ponents. With reference to the Fig. 9, distribution of axial
tested at higher cycles (strain levels lower than 1%). Speci- strain through the thickness direction at various length coor-
mens tested in air survived to longer lives approximately 5 dinates are shown in Fig. 10. Applying deflection of 2 mm
× ­106 cycles, however, specimens tested in seawater failed at (equivalent to 30% of maximum deflection in static 3 PB
less than ~ 106 cycles, consequently high cycle fatigue can be test) at the centre of 3 PB test specimen, the flexural strain
classified as a stress-dependent and environment-dependent using Eq. 1 is calculated to be 0.6% indicating that DIC cal-
regime. culated strains agree well with the strain values calculated
from Eq. 1.
Figures 11 and 12 illustrate typical DIC analysis results
for the surface strain components, loaded in the x–z princi-
pal material plane, at the peak cyclic strain equal to 30% of
ultimate static strain in both seawater and air environments,
respectively. Figures 13 and 14 show the typical flexural
normal engineering strain histories from the starting point
at the different peak flexural strain, as functions of cycles

Fig. 7  Strain –Number of cycles to failure plots of GFRP composites

under different strain level in wet and dry conditions. The standard Fig. 8  E-glass/epoxy three-point bend fatigue test specimen failure at
uncertainty in the maximum applied strain is estimated as 0.1% 5,9 million cycles

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 31

during fatigue tests. The last point in the fatigue history

matches to the onset of a DIC-detectable delamination. A
64–116% increase in the localized flexural normal strain
can be observed during the first 90% of fatigue life in both
seawater and air conditions, and a rapid and large increase
of the flexural normal strain occurs shortly before the onset
of delamination. A possible reason for this response is the
increase in the non-visible damage accumulation in the
epoxy resin prior to detectable cracking.
By comparing Figs. 11 and 12 and according to the
microscopic features of GFRP composites it can be found
that the damaged surface of the fatigued specimen in the air
is more uneven (i.e. more localized strains can be seen) with
much more de-bonding failure than samples tested in sea-
water. A possible reason for the higher number of localized
strains is a large number of resin cracks. Indeed microscopic
Fig. 9  Lagrange strain tensor components for a 3 PB fatigue test at
observation of fatigue loaded samples reveals that for those
peak strain equal to 30% of ultimate static strain at the beginning of
life cycle (a), longitudinal normal strain, exx (b), transverse normal specimens tested in the air many resin cracks are created
strain, eyy and (c), shear strain, exy prior to failure, while only a few resin cracks are found in

Fig. 10  DIC calculated distribu-

tion of axial strain through the
thickness direction at the length
coordinates of a x = 0 (centre),
b x = 12.5 mm, c x = 25 mm, d
x = 37.5 mm, e x = 50 mm (at
support)

13
32 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

concentrated portion of the composite specimen as failure

occurs just in the central weaker section. The fatigue loaded
specimens in air fail in the tension side while the samples
tested in seawater fail in both tension and compression sides
suggesting that there may be an environmentally affected
fatigue mechanism acting (Price and Hull 1983).

3.3 Fatigue failure mechanisms

Fatigue damage evolution is evaluated by the measurement

of changes in flexural elastic modulus and investigation of
the microstructural features of the sample surface during
the fatigue cycling. To reveal information about damage
development and understand how fatigue cycling affects the
composites, fatigue tests have been interrupted after differ-
ent proportions of the fatigue life in both high and low cycle
Fig. 11  Lagrange strain tensor components for a 3 PB fatigue test
fatigue. Then a quasi-static 3 PB test was performed to fail-
(seawater environment) at peak cyclic strain equal to 30% of ultimate
static strain and 1.300 million cycles. (a), longitudinal normal strain, ure to determine residual flexural modulus as illustrated in
exx (b), transverse normal strain, eyy and (c), shear strain, exy Fig. indicating stiffness reduction as a result of cyclic loads.
It has been well documented that variation in the flexural
elastic modulus is an indication of the failure state for a
composite material (Stinchcomb and Reifsnider 1979; Hahn
et al. 1980; Konur and Matthews 1989, Agarwal et al. 2017).
Such a bending modulus degradation curve is characterized
by a drop within the first 10% of life, followed by a relatively
flat region, representing a slower damage development rate,
to about 90% of life and then termination by a sudden drop
during the last stages of life. In Sect. 3.5 and 3.6 it has been
shown that the modulus decay and fatigue damage accu-
mulation processes in a GFRP woven-fabric composite are
divided into three such stages.
During the initial stage of fatigue life, the initiation of
fatigue failure is revealed by the appearance of transverse
cracks in the fill direction of the weave followed by longi-
tudinal inter-laminar cracks. Transverse matrix cracks usu-
ally initiate from the edge of specimen, presumably due to
the effects of microdamage during sample cutting. As the
Fig. 12  Lagrange strain tensor components for a 3 PB fatigue test
cyclic load continues, the crack density increases and this
(air environment) at peak cyclic strain equal to 30% of ultimate static
strain and 5.900 million cycles. (a), longitudinal normal strain, exx is followed by crack coupling and interfacial de-bonding
(b), transverse normal strain, eyy and (c), shear strain, exy (Fig. 15). De-bonding between transversely oriented fibers
and the matrix may occur as these transverse resin cracks
propagate. A rapid decay of modulus during this stage is
specimens tested in seawater. This indicates that failure fol- mainly caused by crack propagation and accumulation in
lows quickly after the first crack formation in the seawater several stress concentration locations.
environment. Consequently, resin cracking can be acceler- In the next stage of fatigue life, the main damage mecha-
ated and overall fatigue life is reduced. This description of nism is delamination between adjacent plies and shear fail-
resin cracking effects in environmental behavior of FRP ure. Generally, failure modes such as interfacial de-bonding
composites is seen by other authors (Carswell and Roberts between fibres/matrix and delamination are associated with
1980, Hofer et al. 1987). the difference in interlayer stress which is due to the differ-
In Figs. 13 and 14, it can be observed that the damage ence in the load among the layers in the thickness direction
phenomena are concentrated in the central zone near the of the FRP specimens. Due to the stress concentration asso-
tension and compression sides of the specimen (starting ciated with the difference in interlayer stress, inter-laminar
from the beginning of cycling) and gradually evolve in a cracks propagate into the epoxy resin adjacent to the fibers.

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 33

Fig. 13  bending engineering

strain history for a 3 PB fatigue
test (seawater environment) at
peak cyclic strain equal to 30%
of the ultimate static strain and
1.300 million cycles

Fig. 14  bending engineering

strain history for a 3 PB fatigue
test (air environment) at peak
cyclic strain equal to 30% of the
ultimate static strain and 5.900
million cycles

Fig. 15  a Scanning electron

micrograph shows interfacial
de-bonding between fibres and
matrix, b inter-laminar cracks
with delamination

13
34 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

This leads to the creation of fiber-matrix interfacial de-bond- destructive effect of seawater on the fiber/matrix adhesion.
ing and the crack growth in the fibers. Similar behavior of degraded fiber surface due to the wet
In the final stage of fatigue life, when the residual strength conditions has been also reported in (Watanabe and Tsai
of the composites is overcome all failure mechanisms are 1979, Sekine et al. 1988). Consequently, the process of envi-
activated and catastrophic failure occurs with the evidence ronmentally affected fatigue of GFRP composites, involves
of fiber breakage. Fiber failure happens in two modes; ten- combined interaction between the degradation of E-glass/
sile and compressive (Fig. 16). Epoxy resin interfaces and resin cracking. After crack initia-
tion and propagation in the epoxy resin, access of fluid into
3.4 The effect of fiber/matrix interface the composite occurs at an accelerated rate, which further
speeds up the degradation processes in the interphase region.
The effect of fiber/matrix interface on the environmentally Therefore the glass fibers are exposed to corrosion at a faster
affected fatigue performance of fiber-reinforced composites rate. From the fatigue data, fiber strength decay cannot be
has been investigated in a number of studies (Fried 1967, quantified, however, microscopic investigations of the failed
Hofer Jr et al. 1976, Watanabe and Tsai 1979, Hofer et al. specimens can establish a strong indication of fiber strength
1987, Shih and Ebert 1987). The destruction of the fibre/ deterioration.
matrix interface region by corrosive fluid leakage facilitates
matrix crack propagation because fibres no longer sustain 3.5 Elastic modulus evolution
their reinforcing roles. This process is further aided by
the fact that the failure strain of the matrix (epoxy resin) To assess different test results with various strain levels and
is diminished upon exposure to seawater. In this study fatigue life, all measurements of modulus decay are normal-
SEM examination of the failed surfaces revealed the nature ized with respect to the maximum values of elastic modu-
of the surface of fibers. For specimens tested in dry con- lus and the number of cycles to failure. Given the stiffness
ditions more matrix material appears attached to the fiber degradation normalized curves for all fatigue tests (Fig. 18),
surface. This relatively rough surface of fibers can be seen two slope transition points can be observed; they exist in
in Fig. 17a, while “smoother” and cleaner fiber surfaces with correspondence with a specified percentage of total fatigue
less matrix bonded are observed from specimens failed under life and three different zones can be identified. The values
fatigue in seawater (Fig. 17b). This difference indicates the of these slope transition points for stiffness decay in terms

Fig. 16  Scanning electron

micrograph shows fibre break-
age as a result of flexural fatigue
loads, a tensile failure of fibre
in the bottom (tension) side of
GFRP specimen, b fibre buck-
ling/kicking in the upper side

Fig. 17  Scanning electron

micrographs of glass fibre
surface for specimen failed by
cycling in a air and b seawater

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 35

Fig. 18  Modulus decay in

woven fabric composites dur-
ing fatigue life in a dry b wet
condition, the relative standard
uncertainty in the normalized
modulus data is estimated as
2.5%

of the number of cycles are positioned at around 10–90% Ei − E

of total life. Given the modulus decay graphs, the initial D= (3)
Ei − Ef
phase is dominated by a reduction in stiffness in the order of
5–10%, until an equilibrium stage is reached. This behaviour where D covers a range of values between 0 and 1 and
is influenced by internal fibre-matrix viscoelastic interac- depends on the initial Modulus, Ei , and final Modulus,Ef .
tions and crack accumulation in several stress concentra- Figure 19 shows the evolution of damage resulting from
tion locations. The second stage occupies 90% of consumed analysis of modulus decay data using the above expression.
fatigue life and is in the same life percentage range for all These figures and the stiffness degradation curves show that
samples. In addition, the trend in this second phase is quite the stiffness degradation and damage state criterion (D) are
linear and represents a similar slope regardless of the strain strictly correlated in fatigue for composite materials and
level and environment. The final stage is characterized by both can express the progressive fatigue damage.
coalescence of defects and a huge stiffness reduction can be
observed. 3.7 Assessment of residual fatigue life

3.6 Fatigue damage estimation As shown earlier, the stiffness degradation is well-correlated

to the damage state of composite test coupon and both of
To determine the damage (D) in GFRP test coupon in terms them can be adopted as a potential damage measure. This
of residual life to failure, a knowledge of elastic modulus means that a method can be suggested to predict the residual
reduction is useful, provided that constant displacement fatigue life of GFRP components. A representative example
or strain level is applied. Elastic modulus decay has been is presented in Fig. 20; drawing a line that fits the second
used to quantify the damage state of composite materials stage of stiffness degradation, it is possible to identify two
for a long time (Kuo and Pon 1997; Whitworth 1997; Van points: X10
lin
: The point of line corresponding to 10% of con-
Paepegem and Degrieck 2002, Tang et al. 2004). In this sumed fatigue life and X90 lin
: The point of line corresponding
work, for the calculation of damage (D) an expression is to 90% of consumed fatigue life on modulus decay diagram.
suggested which is a function of Elastic modulus variation; Thus the following expression can be defined as

Fig. 19  Damage accumulation

in woven fabric composites dur-
ing fatigue life in a dry, b wet
condition, the relative standard
uncertainty in the normalized
modulus data is estimated as
2.5%

13
36 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

Fig. 20  Elastic modulus evolution and line fitting the second stage
(40% of ultimate Flexural deflection)

Fig. 21  Micrograph of a typical delamination pattern. The micro-

lin
X10 lin
− X90 graph is taken from the edge of composite test specimens near mid
ΔX% = lin (4) span after fatigue
X10

The quantity (ΔX ) is defined as an index of irreversible com-

posite degradation. By measuring this quantity for all strain Therefore, the process of environmentally affected fatigue
levels in fatigue tests of GFRP composites, it was found behavior for the GFRP composite involves resin cracking
that the modulus decay rate of the fatigue loaded samples in and fiber/matrix interface degradation. Although the above
seawater is higher than that in the air. mechanisms occur in high cycle regimes (low strain levels)
of fatigue in Fig. 7 where the difference between fatigue life
in the air and seawater is significant, in the low cycle regime
3.8 Environmental effects on fatigue crack growth (high strain levels), the amount of time required for seawater
diffusion is insufficient, therefore no significant difference
With reference to Figs. 13 and 14, it is seen that many cracks between fatigue life in seawater and the air is observed. Sim-
in the epoxy resin are created prior to final failure for sam- ilar results have been also reported in the literature (Carswell
ples tested in air, however, a only a few cracks were found in and Roberts 1980, Hofer et al. 1987).
specimens tested in seawater. This indicates that the major
damage follows immediately after the development of the 3.9 Fabric and delamination analysis using µ‑CT
first cracks in seawater. During the environmental fatigue
testing, seawater diffuses into the network of resin cracks The progression of bending fatigue damage in 0º/90º ori-
and de-bonded fibre/matrix affecting the mechanical prop- ented glass fiber-reinforced epoxy was previously studied
erties of the matrix (epoxy resin) and fibre/matrix inter- (Sect. 3.3), and delamination within the specimen was
face region of the plies and extends rapidly beyond the last observed (Fig. 21). In this figure, characterization of the
cracked ply. In Sect. 3.4, evidence of the destructive effect of geometry of delamination required slicing of consecutive
seawater on the fibre/matrix interface region was presented. sections through the specimen and preparation of each
The degraded fibre/matrix interface region is extended by section for optical and/or electron microscopy which is a
seawater diffusion and can facilitate crack propagation in the laborious and destructive process. To determine subsurface
matrix since the fibres cannot continue their reinforcing roles damage non-destructively specimens from this study were
in the composite. In addition, once the seawater penetrates examined using µ-CT, and an example is presented here.
the epoxy resin matrix, it can stimulate cracking and craz- The three-dimensional geometry of the delamination is
ing in the matrix (Arnold 1996; Wright 1996). As a result, revealed in the volumetric reconstruction in Fig. 22a. The
access of seawater into the composite occurs at a higher rate, µ- CT results provide a method to measure the delamina-
which further accelerates the degradation in the interface tion area, without the need for successive sectioning of
region, finally leading to the significant reduction in fatigue the sample and the potential for introduction of additional
life of the composites when fatigue testing is carried out in damage during sectioning. Figure 22b and c presents a slice
seawater. This agrees with the findings in Sect. 3.7, as it is (19.6 µm voxels) revealing the delamination. The location
shown that the rate of modulus decay of the samples tested in of this cross-section is identified by the cutting plane line in
seawater is higher than the samples tested in the air (Fig. 20). the central region of a fatigued specimen where macroscopic

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 37

Fig. 22  E-glass/epoxy fatigue

loaded specimen with delamina-
tion, a volumetric reconstruc-
tion of the delamination in the
micro-CT specimen, b Micro-
CT reconstructed cross-section
of the specimen in x–z plane, c
Micro-CT reconstructed cross-
section of the specimen in x–y
plane

failure was observed. The delamination is resolved based mechanical loading, thus providing a more realistic estima-
on the difference in X-ray absorption between the glass/ tion of the environmentally affected fatigue behaviour of the
epoxy material and air. The delaminated area had a lower composite and the degradation mechanisms of composite for
absorption coefficient than the composite plies and appears tidal turbine applications.
darker in the image. This clearly matches the delamination Although in the literature (Kennedy et al. 2012, Li et al.
geometry observed at the surface by optical and electron 2014, Suzuki 2017, Jaksic et al. 2018) models for fatigue life
microscopy (comparing Figs. 21 and 22b). Therefore, the prediction of FRP composite tidal turbine blades have been
µ-CT can provide an excellent characterization of the 3D proposed, the fatigue testing technique to provide the input
damage geometry. test data (S–N curves) for the majority of these models ignore
either the effects of environmental conditions or the bending
3.10 State of the art mode on the composites leading to an overestimation of life.
In the present study, in order replicate the bending behav-
Tidal turbine blades are fatigue critical components and ior of the blade, associated stresses and failure modes, bend-
degradation of blade material (fibre reinforced composites) ing fatigue testing is employed to characterize the strain-life
under combined effects of cyclic loads and seawater environ- behavior of composites specimens under more realistic marine
ment are the major considerations in the design of the blades conditions. The results show that high cycle fatigue of GFRP
(Davies et al. 2013, Jaksic et al. 2018, Izadi Gonabadi 2019). composites is a stress and environment-dependent regime,
Although in recent years, novel techniques for manufacturing meaning that the combined effect of stress and environment
fibre-reinforced composites tidal turbine blades with improved is significant and therefore a higher safety factor should be
properties have been investigated (Robert et al. 2019; Murray considered by tidal turbine designers. In addition, the better
et al. 2020), accounting for the environmental degradation of understanding of microscopic failure mechanisms of compos-
the composite is an essential element for multi-scale Finite ite constituents under combined effect of seawater condition
Element modelling of composite blades and it requires reli- and cyclic loading obtained in the present study should help
able fatigue testing data input. The approach to account for the designer to consider the magnitude of the extra safety fac-
the environmental degradation of composite materials, based tors necessary when selecting these materials for tidal turbine
on testing samples in the seawater condition is developed in applications.
this study involving interaction between seawater diffusion and

13
38 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

4 Conclusions Compliance with ethical standards

For the specimens cyclically loaded in the low cycle regime Conflicts of interest There are no conflicts of interest associated with
the submission of this manuscript.
above 60% of the maximum flexural strain, no substantial
difference in fatigue life was recorded in air or seawater. Ethics approval No formal ethical approval was required to carry out
However, for the specimens cyclically loaded in the high this research. The work described has not been published before.
cycle regime below 60% of the dry maximum flexural strain,
Consent to participate Each author has participated in the work. HIG
all specimens tested in air survived beyond 1­ 06 cycles, while carried out the mechanical and fatigue test experiments with the assis-
all those tested in the seawater environment did not. Conse- tance of AY who also took an important role in the microscopy. AO and
quently, it is found that the long-term environmental fatigue SJB planned and directed the research and approved the publication.
performance of woven GFRP composites is controlled by a
Consent for publication All authors give consent for publication.
combination of both load and the fluid environment.
Microscopic observation of fatigue failed samples reveals
that for those specimens tested in the air many resin cracks Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
are created prior to failure, while only a few resin cracks are tion, distribution and reproduction in any medium or format, as long
found in specimens fatigued in the seawater. The higher num- as you give appropriate credit to the original author(s) and the source,
ber of resin cracks can be seen by distributed localized strain provide a link to the Creative Commons licence, and indicate if changes
in DIC maps and this technique clearly confirms the existence were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated
of more resin cracks in the fatigue failed samples in the air otherwise in a credit line to the material. If material is not included in
than in seawater. In addition, the greater accumulated dam- the article’s Creative Commons licence and your intended use is not
age in samples tested in seawater shows that failure follows permitted by statutory regulation or exceeds the permitted use, you will
quickly after the first crack formation. This is due to the fact need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/.
that seawater enters into resin cracks and is quickly distributed
through the sample. This is the reason why the DIC strain map
for seawater affected fatigue samples shows damage concen-
tration only in limited zones near the top and bottom sides, References
however, in the case of the air environment, the DIC strain
map is more uneven due to the larger number of resin cracks. Abanto-Bueno J, Lambros J (2002) Investigation of crack growth in
From the investigation of stiffness degradation curves functionally graded materials using digital image correlation. Eng
Fract Mech 69(14–16):1695–1711
(modulus decay) during the fatigue testing, it was found
Agarwal BD, Broutman LJ, Chandrashekhara K (2017) Analysis and
that 80% of fatigue life is taken in the second stage where performance of fiber composites. Wiley
delamination is the main failure mode, therefore control- Ahamed RAR et al (2016) Impact angle effects on erosion maps of
ling the rate of delamination plays an important role in GFRP: applications to tidal turbines. J Bio- Tribo-Corr 2(2):14
Arnold J (1996) Environmental stress crack initiation in glassy poly-
increasing the fatigue life of composite structures.
mers. Trends Poly Sci 12(4):403–408
Microscopic examination of the fatigued test coupons Boisseau A et al (2013) Fatigue behaviour of glass fibre reinforced
reveals a degraded fibre/matrix interface region in those composites for ocean energy conversion systems. Appl Compos
samples where environmental exposure caused the pre- Mater 20(2):145–155
STANDARD B, ISO B (1998) Fibre-reinforced plastic composites—
mature failure. It is believed that fibre/matrix interfacial
Determination of flexural properties
degradation is the controlling factor in the environmentally Canal LP et al (2012) Application of digital image correlation at the
affected fatigue behavior of GFRP composites. microscale in fiber-reinforced composites. Compos A Appl Sci
The µ- CT results provide a method to measure the Manuf 43(10):1630–1638
Carswell WS, Roberts RC (1980) Environmental fatigue stress failure
delamination area, without the need for successive sec-
mechanism for glass fibre mat reinforced polyester. Composites
tioning of the sample and the introduction of additional 11(2):95–99
damage during sectioning. Charles R (1958a) Static fatigue of glass. I. J Appl Phys
29(11):1549–1553
Acknowledgements We thank our colleagues from the REC-ASM Charles R (1958b) Static fatigue of glass. II. J Appl Phys
and NECEM projects who provided insight and expertise that greatly 29(11):1554–1560
assisted the research. Chu TC et al (1985) Applications of digital-image-correlation tech-
niques to experimental mechanics. Exp Mech 25(3):232–244
Crupi V, Guglielmino E, Risitano G, Tavilla F (2013) Application
Funding This research was supported by EPSRC through the ReC-
of Digital Image Correlation for the effect of glass fibres on
ASM (EP/K013319/1) and NECEM (EP/R021503/1) projects.
the strength and strain to failure of polyamide plastics. Gruppo
Italiano Frattura. Convegno Nazionale IGF. Acta Fracturae, 22,
Data availability The data associated with this paper is available and 287
can be provided.

13
Journal of Ocean Engineering and Marine Energy (2021) 7:25–40 39

Davies P et al (2013) Evaluation of the durability of composite tidal Price JN, Hull D (1983) Propagation of stress corrosion cracks
turbine blades. Philosoph Transactions Royal Soc A Math Phys in aligned glass fibre composite materials. J Mater Sci
Eng Sci 371(1985):20120187 18(9):2798–2810
Fried N (1967) Degradation of composite materials: the effect of Quaresimin M et al (2010) Fatigue behaviour and life assessment of
water on glass-reinforced plastics. Pergamon Press, Pensylvania, composite laminates under multiaxial loadings. Int J Fatigue
Mechanics of Composite Materials 32(1):2–16
Giancane S et al (2010) Characterization of fatigue damage in long Rasool G et al. (2016) Tribology of tidal turbine blades: impact angle
fiber epoxy composite laminates. Int J Fatigue 32(1):46–53 effects on erosion of polymeric coatings in sea water conditions,
Godara A, Raabe D (2007) Influence of fiber orientation on global Asian Wave and Tidal Energy Conference (AWTEC)
mechanical behavior and mesoscale strain localization in a short Rasool G, Stack MM (2019) Some views on the mapping of erosion
glass-fiber-reinforced epoxy polymer composite during tensile of coated composites in tidal turbine simulated conditions. Tribol
deformation investigated using digital image correlation. Compos Trans 62(3):512–523
Sci Technol 67(11):2417–2427 Rasool G et al (2016) Mapping synergy of erosion mechanisms of
Hahn HT, HT H., TT C (1980) Long-term behavior of composite mate- tidal turbine composite materials in sea water conditions. J Bio
rials, vol 1; pp 584–596; BIBL. 17 REF.; ISBN 0-08-026717-3 Tribo-Corr 2(2):13
He Y et al (2012) Characterization of nonlinear shear properties for Renaud CM, Greenwood ME (2005) Effect of glass fibres and environ-
composite materials using digital image correlation and finite ele- ments on long-term durability of GFRP composites. In: Proceed-
ment analysis. Compos Sci Technol 73:64–71 ings of 9 EFUC Meeting, Wroclaw, Poland
Hofer Jr KE, et al. (1976). Effect of various fiber surface treatments on Richter-Trummer V et al (2011) Analysis of crack growth behavior
the fatigue behavior of glass fabric composites in high humidity in a double cantilever beam adhesive fracture test by different
environment. Proceeding of the 31st Annual Technical Conference digital image processing techniques. Materialwiss Werkstofftech
of SPI, Washington (DC) 42(5):452–459
Hofer KE et al (1987) Effect of moisture on fatigue and residual Robert C, Pecur T, McCarthy ED, Brádaigh C (2019) Tidal turbine
strength losses for various composites. J Reinf Plast Compos blade composites using basalt fibre reinforced powder epoxy.
6(1):53–65 Engineers Australia, Barton, Australia, p 3687. https:​ //www.resea​
ISO B (2003) 13003. Fibre-reinforced plastics–determination of fatigue rchga​te.net/publi​catio​n/33972​5966
properties under cyclic loading conditions. British Standards Schaff JR, Davidson BD (1997) Life prediction methodology for com-
International posite structures. Part I—Constant amplitude and two-stress level
Izadi Gonabadi H (2019) Performance of low cost composites for tidal fatigue. J Compos Mater 31(2):128–157
turbine applications. PhD diss., Newcastle University Schmitz G, Metcalfe A (1966) Stress corrosion of E-glass fibers. Indust
Jaksic V et al (2018) Influence of composite fatigue properties on Eng Chem Product Res Develop 5(1):1–8
marine tidal turbine blade design. Durab Comp Mar Env 2:195– Sekine H et al (1988) Strength deterioration and degradation mecha-
223 (Springer) nism of glass chopped reinforced plastics in water environment.
Johnson KL, Johnson KL (1987) Contact mechanics. Cambridge uni- JSME Int J Ser Solid Mech Strength Mater 31(3):619–626
versity press Shih GC, Ebert LJ (1987) The effect of the fiber/matrix interface on
Kennedy CR et al (2012) A preliminary design methodology for fatigue the flexural fatigue performance of unidirectional fiberglass com-
life prediction of polymer composites for tidal turbine blades. Pro- posites. Compos Sci Technol 28(2):137–161
ceed Instit Mech Eng Part L J Mater Des Appl 226(3):203–218 Shokrieh MM, Lessard LB (1997) Multiaxial fatigue behaviour of
Konur O, Matthews FL (1989) Effect of the properties of the constitu- unidirectional plies based on uniaxial fatigue experiments—II.
ents on the fatigue performance of composites: a review. Com- Experimental evaluation. Int J Fatigue 19(3):209–217
posites 20(4):317–328 Stinchcomb WW, Reifsnider KL (1979) Fatigue damage mechanisms in
Kuo W-S, Pon B-J (1997) Elastic moduli and damage evolu- composite materials: a review. Fatigue mechanisms. https​://www.
tion of three-axis woven fabric composites. J Mater Sci astm.org/DIGIT​AL_LIBRA​RY/STP/PAGES​/STP35​914S.htm
32(20):5445–5455 Sun Z et al (1997) Measuring microscopic deformations with digital
Laurin F et al (2012) Determination of the properties of composite image correlation. Opt Lasers Eng 27(4):409–428
materials thanks to digital image correlation measurements. Pro- Suri C, Perreux D (1995) The effects of mechanical damage in a
cedia IUTAM 4:106–115 glass fibre/epoxy composite on the absorption rate. Compos Eng
Lee D et al (2009) Experimental study of dynamic crack growth in 5(4):415–424
unidirectional graphite/epoxy composites using digital image cor- Suzuki T (2017) Development of a comprehensive design methodology
relation method and high-speed photography. J Compos Mater and fatigue life prediction of composite turbine blades under ran-
43(19):2081–2108 dom ocean current loading (Doctoral dissertation, Florida Atlan-
Li H et al (2014) Reliability-based fatigue life investigation for a tic University). https:​ //search​ .proque​ st.com/docvie​ w/201552​ 7971​
medium-scale composite hydrokinetic turbine blade. Ocean Eng ?pq-origs​ite=gscho​lar&fromo​penvi​ew=true
89:230–242 Tang R et al (2004) An appropriate stiffness degradation parameter
Mamalis D et al (2019) Novel carbon-fibre powder-epoxy composites: to monitor fatigue damage evolution in composites. Int J Fatigue
interface phenomena and interlaminar fracture behaviour. Compos 26(4):421–427
B Eng 174:107012 Ukyo S, Masuda M (2004) Investigation of the true stress-strain rela-
McCormick N, Lord J (2010) Digital image correlation. Mater Today tion in shear using the digital image correlation method. J Jpn
13(12):52–54 Wood Res Soc (Japan). https​://agris​.fao.org/agris​-searc​h/searc​
Metcalfe A, Schmitz G (1972) Mechanism of stress-corrosion in e glass h.do?recor​dID=JP200​40076​30
filaments. Glass Technol 13(1):5–000 Van Paepegem W, Degrieck J (2002) A new coupled approach of resid-
Metwalli S et al (1987) Determination of plastic stress-strain behavior ual stiffness and strength for fatigue of fibre-reinforced compos-
by digital-image-processing techniques. Exp Mech 27(4):414–422 ites. Int J Fatigue 24(7):747–762
Murray JJ et al (2020) Manufacturing of unidirectional stitched glass Vassilopoulos AP, Keller T (2011) Fatigue of fiber-reinforced com-
fabric reinforced polyamide 6 by thermoplastic resin transfer posites. Springer Science & Business Media. https ​ : //doi.
moulding. Mater Des 189:108512 org/10.1007/978-1-84996​-181-3

13
40 Journal of Ocean Engineering and Marine Energy (2021) 7:25–40

Vauthier E, et al. (1995). Hygrothermal aging and durability of uni- Zhang D et al (2006) Displacement/strain measurements using an opti-
directional glass-epoxy composites. Tenth international confer- cal microscope and digital image correlation. Optical Eng 45:3
ence on composite materials. VI. Microstructure, Degradation, (033605-033605-033609)
and Design
Watanabe M and TW Tsai (1979) Effect of water environment on Publisher’s Note Springer Nature remains neutral with regard to
fatigue behavior of fiberglass reinforced plastics. Composite mate- jurisdictional claims in published maps and institutional affiliations.
rials: testing and design (Fifth conference), ASTM STP
Whitworth HA (1997) A stiffness degradation model for composite
laminates under fatigue loading. Compos Struct 40(2):95–101
Wright DC (1996) Environmental stress cracking of plastics. iSmithers
Rapra Publishing

13