International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 18

Modelling of Glass Fibre/Epoxy Composite Pipes

Under Multi-Axial Loadings Using Finite Element
Analysis
Z.S. Naziraha, M.S. Abdul Majida*, N.A.M. Amina, A.G. Gibsonb
a
School of Mechatronic Engineering, Universiti Malaysia Perlis, Pauh Putra Campus,
Arau, 02600, Perlis, Malaysia
b
School of Mechanical and Systems Engineering, Newcastle University, Newcastle upon Tyne NE1 7RU, UK
*Corresponding Author: Phone. +612-736 7500, Fax. +604-9885167
Email: [email protected]

Abstract-- The glass fibre reinforced epoxy (GRE) pipes under GRE pipes are considered to be thin-walled cylinders
multi-axial loadings were studied through its performance at in which the radius-to-wall thickness ratio is greater than 10
various temperatures and different failure criteria. Owing to the
orthotropic nature of GRE pipes, it is hard to analyse the stresses
( > 10) [4]. In this research, the radius and thickness of the
generated in them. Therefore, using finite element software, an GRE pipes were 200 mm and 6 mm, respectively. Such GRE
analysis was conducted to determine the first ply failure of pipes are classified as an orthotropic material, where the
composite pipes, which were subjected to five different stress material properties vary in perpendicular directions rather than
ratios, ranging from pure hoop to pure axial loadings, at room
the primary axes of an established coordinate system [5]. In
temperature, 65 °C and 95 °C. The Tsai-Wu, Hashin and Puck
failure criteria were used to predict the failure strength of addition, owing to their orthotropic properties, filament wound
composite pipes. The results were validated with experimental casings consisting of composite materials can be used to
data obtained from previously reported studies. Subsequently, reduce the degree of expansion of pipes, without increasing
the limits associated with the axial and hoop stress were the pipe thickness [6]. Based on previous netting analyses, this
expressed by failure envelopes graph. There were differences in investigation focuses on the optimum winding angle of pipes,
the results obtained for the various criteria; however, they which is ±55°. In particular, previous studies have shown that
showed relatively similar trends. During failure analysis, Hashin an angle of ±55° is optimal for piping systems in which the
criterion results in superior predictions for determining the FPF applied hoop-to-axial stress loading ratio is 2:1 [7].
of GRE pipes yielding the smallest % error between the
experimental data compared to the other failure criteria. Initial Fibre reinforcements are the main contributors of
failure stress found decreased at elevated temperatures, except at strength and stiffness of a composite structure. They work
a 2:1 stress loading, where the initial failure stress increased.
more efficiently when they are in bulk form because most of
the fibre material becomes stronger and stiffer in this form.
Index Term-- Finite element modelling; first ply failure; failure Common materials used to make these fibre reinforcements
criteria; multiaxial stress ratio; glass/ epoxy composite pipes.
are a brittle material such as glass, carbon and ceramic.
Composite with fibre reinforcement has excellent tensile
1. INTRODUCTION strength and stiffness in longitudinal or fibre direction but low
GRE pipes are being increasingly used and are mechanical properties in the transverse direction and
becoming a crucial class of engineering material with a broad longitudinal compression strength [8].
range of applications. These include high-pressure containers
for chemical plants and the aerospace industry, as well as gas Aziz Onder [9] proposed that mechanical behaviour
and liquid transfer pipes for the gas, oil and nuclear industries. can be characterised by a set of equivalent or effective module
This is owed to their high strength-to-weight ratio and and strength properties. Lamina properties were determined by
excellent corrosion resistance [1]. The present work concerns a phenomenological approach. An analytical method, finite
applications in the offshore oil and gas industries, particularly element method and experimental method were applied to
composite pipelines for aqueous liquids. GRE pipes are determine first failure pressure of composite pressure vessel.
usually designed to withstand high pressures. Their A glass epoxy composite layer is used and is oriented
lightweight, relatively thin-walled structures facilitate symmetrically or anti-symmetrically. The Tsai-Wu criterion is
handling and transportation, which results in reduced used to compute first failure pressure of composite layers.
installation costs. The majority of pipes for such applications The finite element analysis (FEA) method, originally
are produced by a filament winding process, and they are introduced by Turner (1956), is a powerful computational
usually subjected to a combination of internal pressure and technique used to obtain approximate solutions for a variety of
axial loading [2-3]. Figure 1 shows a schematic of a filament ‘real world’ engineering problems that have complex domains
winding machine. subjected to general boundary conditions. FEA has become an

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 19

essential tool for the design or modelling of physical continuous and progressive decline in stiffness after the
phenomena in various engineering disciplines [10]. initiation of matrix cracks. This is later used to calculate the
elastic constant of the lamina.
Therefore, in this research, numerical analyses were
performed using a finite element method (FEM) approach to Prior to investigating the failure modes of GRE pipes
evaluate the performance of thin cylindrical GRE pipes under subjected to a range of multiaxial load ratios, it is important
multi-axial loadings. Xia et al. studied multi-layered filament that the micromechanics of GRE pipes is fully understood.
wound composite pipes under internal pressure, and presented Extensive work has been carried out by earlier researchers on
an exact solution [11]. Concerning the failure and composite the performance behaviour of GRE pipes [14] which is still
damage of composite tubes, Ellyin and Martens going on. The primary concerns are the degradation of the
experimentally investigated the biaxial fatigue behaviour of a mechanical properties of the pipes and their elastic response
multidirectional filament wound glass/fibre/epoxy pipe [12]. prior to final failure. The elastic properties of GRE pipe work
are often modelled using laminate theory.
Filament wound GRE pipes have been the subject of
theoretical and experimental investigations for many years. The performance of GRE pipes under multi-axial
The stress–strain relationships, failure envelopes and failure loadings will be analysed during this study. This investigation
mechanisms of GRE pipes under biaxial loadings have been has three objectives. The first is to develop an FE model for
extensively evaluated. The behaviour of GRE pipes, filament wound glass fibre/epoxy composite pipes under
particularly under combinations of axial tension and internal multi-axial stress loading. The second is to simulate the effect
pressure, has not been elucidated because of the scarcity of of temperature on the GRE pipes to evaluate their
modelling data and the lack of understanding of this failure performance. The final objective is to analyse and develop an
mode. FE model for GRE pipes with a winding angle of ±55° using
several types of failure criteria. The material details for the
In 1998, an initiative was taken by Soden et al. to
GRE pipe employed in this study are presented in Table 1.
conduct a systematic study to test and compare prominent
failure theories developed in the past decades. Known as the
World Wide Failure Exercise (WWFE), the research was 2. EXPERIMENTAL METHODS
conducted by comparing the failure criteria of fibre reinforced 2.1 Model development
plastics (FRP) systems subjected to biaxial stress ratios against
a standard set of experimental test data, so that consistent When a pipe is subjected to internal pressure, the stress
comparison could be facilitated. Surprisingly most theories components induced in each ply have to be measured. FEM is
differed significantly from experimental observations even for employed to conduct stress analysis and evaluate stress
a simple laminate analysis. It was concluded that, while there distribution in the structural layers of GRE pipes. The fibre is
are numbers of failure criteria available, these were only wound, producing a pattern, which is repeated for every two
successful with limited ranges of data. Some failure theories layers, until the desired laminate thickness is achieved (in this
showed a wider range of applications whereas some produced case 10 layers).
predictions of greater accuracy. [13-14].  First layer: +55° + resin
 Second layer: −55° + resin
The Tsai-Wu criterion was also one of few that was
highly ranked in the WWFE study [15-17] for predicting the The average wall thickness is 6 mm. The full dimensions
strength of unidirectional laminates under combined loads. of GRE pipes were illustrated in Figure 2.
The approach to the WWFE was to present material properties Figure 3 shows a flowchart that details the protocol
of various laminates to participating investigators initially and opted to predict the failure of GRE pipes during the modelling
have the participants predict failure and analysis. According to the chart, all the elements of the
In these criteria, all stress components interact and model, including the material properties, boundary conditions
contribute simultaneously toward the failure of the composite and internal pressure P, were established either before or
systems. Tsai-Hill [18] and Tsai-Wu [19] failure criteria are during the modelling process. During the process, the failure
the two most commonly used interactive failure criteria in models for the failed plies (elements) are identified, and
determining the failure of fibre reinforce polymer structures. stiffness degradation will be conducted. The structure will be
These criteria assume linear elastic material properties and evaluated to verify whether a failure has occurred. If failure
expect degradations in stiffness after the first ply failure. does not occur, the internal pressure is increased
incrementally. However, if a failure occurs, the mechanical
However, the Puck theory is a well-known non- properties are reduced using appropriate degradation rules.
interactive failure criterion was rated in the WWFE study. The obtained results are analysed and compared with
Developed by Puck and Schǘrmann [20-21], the theory is experimental data. If no failure is observed, the load is further
based on physical damage and failure mechanisms in the increased, and the process is repeated to perform stress
constituents. The theory takes into account the non-linear analysis, as shown in Fig 3. This protocol was repeated at
stress-strain relationship and makes allowances for a

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 20

elevated temperatures, and various failure criteria were and mechanical structures such as ship hulls, aircraft bodies,
considered to evaluate the performance of GRE pipes. and machine housings, as well as mechanical components
such as pistons, mechanical parts, and tools [23].
A progressive damage model was developed to
determine the internal pressure associated with functional 2.3 Loads and boundary conditions
failure, which is a non-structural failure wherein the pipe can
still sustain loading but can no longer function effectively. The GRE pipe analysis was performed using FEA.
Structural failure, on the other hand, is expected to occur at To achieve 2:1 loading, the pipe was fitted with end caps and
higher pressures and temperatures; therefore, it is of vital the internal pressure was applied within the pipe. This loading
importance to predict the functional failure strength of GRE condition is also known as a closed-end or pressure vessel
pipes to facilitate their design. Progressive damage models are condition. As shown in Fig. 7, the two ends of the pipe were
developed using four primary stages, namely model closed and subjected to internal pressure. Two principal
preparation, stress analysis, failure evaluation and material stresses developed: axial stress along the longitudinal
degradation. There will be different micro-mechanic direction and hoop stress along the circumferential direction.
regulations used for predicting the mechanical properties of The internal pressure loadings, ranging from pure
composite layers as the output of model preparation, and the axial to pure hoop, were estimated using the following
inputs for stress analysis are evaluated with respect to three relationships; where he hoop stress,  H , acting on the wall
different failure criteria. thickness is represented as
pD
ANSYS software was used during the study. Initially, H 
engineering data associated with the properties of the 2t
materials employed in the study are set as material input to the (1)
The axial stress,  A , is given by
software. These properties can be revised within the model
without having to create a new project design. Figure 4 shows
pD
the properties of the materials that are employed in this
research.
A 
4t
Figure 5 shows the entire model analysis. Many (2)
different conditions are available for selection in the ANSYS where ‘D’ is the internal diameter, ‘t’ is the wall thickness,
and ‘p’ is the fluid pressure within the pipe. The applied
Workbench and this system was used for the static structural
pressure load was increased incrementally. Upon failure, the
analysis in this study.
corresponding ply is degraded, and the resulting stress and
2.2 Finite element mesh generation strain distribution are calculated.
Figure 6 (a) shows a solid model of a pipe developed
using the ANSYS Workbench and Fig 6 (b) displays the 2.4 Stress analysis and failure criteria
meshed model. Meshing is necessary to ensure that the Failure criteria are used to predict whether a layer
properties are modelled accurately. Fine meshing generation (ply) has failed due to the applied loads. Temperature and
would yield a more accurate and produced efficient model other failure criteria can also be considered. There are
[22]. Analyses were performed for every meshed area, and the numerous failure criteria available for composite material
summation of all the areas represents the total property design [24]. Before the initial failure, strength analyses of
gradient of the model. One can control the meshing by fibre reinforced composite structures are significantly more
selecting different properties for each mesh, such as the size of complicated than those of isotropic materials. Standard failure
the meshed area, meshing style and mesh thickness. The first criteria do not exist for fibre reinforced composite structural
and most crucial step in FEA is discretisation (mesh elements and thus, failure predictions are relatively unreliable.
generation). The results of the analysis depend on the type of Following the application of boundary conditions and
element used. In this step, the component or part is divided forces, the next step is to perform structural analysis of the
into smaller divisions. In the discretisation process in this GRE pipe. Stress analysis is primarily carried out to determine
study, there were 9,595 and 1,730 nodes and elements formed, the failure behaviour of the pressure vessel. During structural
respectively. The purpose of discretisation is to perform analysis, the hoop stress, axial stress and failure criteria were
analysis on each small division separately. In addition to these primarily considered. The total hoop and axial stresses in the
parameters, to produce reliable results with reasonable composite pipes under multi-axial loading are shown in Figs.
computation periods, it is important to select appropriate 8 (a-b) for 2:1 loading using the Tsai-Wu criterion. As shown
element types, mesh sizes and interface elements during the in the figure, initial failure starts to occur at a pressure of 11.7
structural simulation. MPa and a hoop stress of 191.7 MPa.
Structural analysis is probably the most common
application of the finite element method. The term structural Figure 9 illustrates the stress distribution for the
(or structure) implies not only civil engineering structures composite pipe at first ply failure (FPF). Figure 10 shows the
such as bridges and buildings, but also marine, aeronautical, Tsai-Wu failure criterion values at FPF. This was determined
to be 1.02588 for the pipe with the lay-up configuration of

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 21

[RT/±55°10] at a stress ratio of 2:1 and an internal pressure of until initial failure occurs via the employed failure criteria; the
11.7 MPa. The red coloured portions indicate that high process will then cease.
stresses exist in that region, and the blue coloured portion
shows the existence of little stress. The linear static analysis Figure 12 presents a stress–strain curve at room
was performed to obtain the stress distribution in each ply. It temperature (RT) at a stress ratio of 2.1, using Tsai-Wu failure
comprised of principal stress in directions 1 and 2, as well as criterion. A point at which the onset of stress-strain
shear stress/strain components. The analysis is acceptable relationship deviate from linear to become non-linear
because the pipe will not experience substantial deformation, indicating a degradation in stiffness which suggests damage
and the deflection is negligible compared with the general has already taken place. To plot the curve, the hoop stress
dimensions of the pipe. values were determined from the results of ANSYS analysis
as given in Fig. 11.
2.5 Failure criterion Following the validation of the developed modelling
procedure, a parametric study is conducted to investigate the
For composites, most failure criteria are considered influence of temperature and selected types of failure criteria
to be macroscopic. It is important to assess the critical stress on the functional failure pressure. The failure envelopes for
state for static or dynamic loadings. The accuracy of the the different temperatures were compared, and the failure
failure prediction depends strongly on the criteria used. Many criteria revealed systematic trends. The effect of temperature
criteria are used to predict the failure of composite materials and the selected types of failure criteria significantly influence
[25-26]. Failure theories are referred on stresses in a material the performance of the GRE pipes and their functional failure
axes or local axes because a lamina is orthotropic in nature. pressures. Moreover, an increase in temperature will also
Failure theories are not based on principal normal stresses and affect the mechanical properties. As shown in Table 3, the
maximum shear stresses. Failure theories are based on first, Young’s modulus decreases as the temperature increases. The
finding the stresses in the local axes and using five strength transverse stress component, which primarily governs the
parameters of a unidirectional lamina to determine whether a functional failure behaviour, decreases as the temperature
lamina has failed [27]. increases while a greater stress is induced along the fibre
Tsai-Wu, Puck and Hashin criteria were used to direction.
complete the objectives of this research. This approach can The results in Table 4 indicate that the predicted FPF
represent the interaction between direct and shear stresses, and pressures are almost identical for both Puck’s and Hashin’s
can account for differences in tensile and compressive methods, namely no physical damage of the layers can be
strength. The Tsai-Wu criterion is a well-known polynomial observed. However, the most appropriate failure criteria must
criterion [28-29]. It is the most universally used criterion, and be selected to obtain correct and accurate results. This strategy
its equation can be particularised to represent all polynomials. clearly shows the immediate benefit of layered composite
This approach can account for different material strengths structures, which can accommodate loading despite the
under tensile and compressive loading. As with all polynomial occurrence of FPF. In all cases, it was observed that the
criteria, it does not take into account the different mechanical primary mode of failure for FPF is matrix cracking.
behaviours of the constituents or fibre–matrix interface
phenomena. The results show that the strength parameters
correlate with the changes that occur in the material properties
The Puck criterion is capable of predicting the risk of with the loading variations. The calculated values were
failure and the corresponding failure mode. It distinguishes compared with the given strength parameters, and the failure
between fibre failure and several modes of Matrix Dominated values were as expected. The failure of the composite
Failure (MDF), where delamination is one of these modes. elements leads to stress redistribution and severe stress
The delamination strengths, typically, are estimated by concentration on the neighbouring composite elements. The
slightly reducing the transverse ply strengths [30]. stiffness degradation took place after the initial failure of the
2.6 Degradation rules composite elements started. The effects of elevated
The stress components induced in each layer are temperatures significantly influence the mechanical behaviour
input into the failure criteria section of the programme; these of GRE pipes, as well as the functional failure pressures.
stress components are determined from the stress analysis. Tables 5–7 summarise the data obtained from the analysis,
The magnitudes of the mechanical properties are reduced, which was subsequently graphically represented by the failure
based on the failure mode, if a failure occurs in any layer [31]. envelope.
The multi-axial failure envelopes for the analysis of
the GRE pipes were illustrated by plotting the axial stress σA,
3. RESULTS AND DISCUSSION against the hoop stress σH, in the biaxial principal stress space.
Figure 11 shows the stress distribution of a GRE pipe Figure 12 shows the initial failure stress-based failure
at a stress ratio of 2:1 under an applied internal pressure. As envelopes for the GRE pipes that were analysed at RT, 65 °C
the internal pressure increases, the hoop stress also increases and 95 °C, respectively. Each of the points denotes the initial

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 22

failure points, whereas the lines indicate the stress ratios at strength increase under the 2:1 loading condition. The
which the tests were conducted. maximum temperature is represented by the low glass
transition temperature (Tg) of the GRE materials. When the
Overall, the failure envelopes generated from the temperature of the resin within the composite increases above
FEM analysis show a high dependence on the stress ratios and Tg, the mechanical properties of the material degrade. This
test temperatures. At RT, the hoop strength (~270 MPa) was will consequently affect the interfacial bonding between the
over four times greater than that of the pure axial strength glass and fibres and reduce the efficiency of the resin to
(~70 MPa). This is because the load is strongly matrix transfer the load to the fibres; hence, the mechanical properties
dominated under pure axial loading. Therefore, as the stress of the material degrade. The value of Tg for GRE pipes is
ratio increases, the FEM axial and hoop stresses and approximately 160 °C [33-34]. As the temperature reaches Tg,
experimental failure strength also increase because of the the material properties change.
greater loads that are tolerated by, the stronger fibre.
With increasing temperature it is expected for
The results showed that the optimal loading condition material to exhibit increasing in strain to failure with time at
for a ±55° GRE pipe, regarded as the optimal design, is a constantly applied loads. It is widely known that many types
hoop-to-axial stress ratio of 2:1. In the case of the FE model, of fibre reinforced are sensitive to temperature increase.
the initial failure stress point was observed at a hoop stress of Elevated temperatures soften the polymer matrix so that the
193.85 MPa. However, the greatest first failure stress point for behaviour of the composite can be significantly different from
the FE model was recorded to be at a hoop stress of 338.51 the response if the same load was applied under ambient
MPa under a hoop-to-axial loading ratio of 4:1, although the temperatures [35-36].
axial failure strength was slightly lower than that obtained
with the 2:1 loading. It can be observed in all the failure A model that provides the most accurate predictions
envelope graphs that the FE and experimental failure strengths against experimental results is desired. Referring to Tables 5,
are in close agreement with those predicted. For the 4:1 stress 6 and 7, it can be observed that the initial failure stresses vary
ratio, the axial strength starts to decrease in the hoop by the failure criteria used in the FEA; however, similar trends
dominated loading region until a pure hoop loading is are produced for each failure criterion (Refer to Figure 13).
achieved. From the calculated percentage error between the modelling
and experimental data, the results obtained from the Hashin
Meanwhile, with the exception of the 2:1 loading criterion show the lowest percentage error at RT; an error of
condition, the initial failure stress is reduced at elevated 0.89% was obtained at a stress ratio of 2:1. This is followed by
temperatures. Highly noticeable stress reductions were the Tsai-Wu and Puck criteria, with errors of 2.03% and
observed for the pure axial and pure hoop loading conditions, 12.11%, respectively. For the stress ratio of 2:1, the lowest
where the failure mode is matrix dominated. Under pure axial error percentages were recorded with the Hashin and Puck
loading, the initial failure stress values decreased from 70 criteria at 65 °C and 95 °C, respectively. At 65 °C and 95 °C,
MPa to 60.3 MPa to just 58.6 MPa at RT, 65 °C and 95 °C, the corresponding error percentages were 4.94% and 3.14%
respectively, which represents a reduction of almost one-third under the Hashin criterion, and 6.72% and 3.14% under the
in axial strength. For the FE model, the initial failure stress Puck criterion. Conversely, under the Tsai-Wu criterion, larger
associated with the hoop stress also substantially declined, errors of 17.16% and 4.23% were obtained at 65 °C and 95
with a decrease of nearly 42 % from 296.5 MPa at RT to °C, respectively. It can be summarised that small error
171.7 MPa at 95 °C. This finding is in agreement with past percentage between the Hashin and Puck criteria was observed
work by Hale, who suggested that the resin matrix softens at since both were categorised under phenomenological criteria.
high temperatures, which significantly reduces its strength Owing to its polynomial nature, the higher percentage error
[32]. Under all the loading conditions, with the exception of was seen under the Tsai-Wu criterion compared to those by
2:1, the stress failures are matrix dominated; therefore, this the Hashin and Puck criteria. It can be concluded that the
results in significantly reduced initial failure stress resistance, Hashin criterion is optimal to determine the FPF of GRE
especially under pure hoop and pure axial loadings. At a hoop- pipes. This may be because the Hashin criterion considers the
to-axial loading ratio of 2:1, the first failure stress showed an effect of intermediate stress during the failure analysis, which
increase on one occasion; the strength rose from 174 at RT to occurs in reality. The Hashin criterion is advantageous
201 MPa at temperatures of both 65 °C and 95 °C. This is because all the strength values can be obtained from standard
because a loading of 2:1 is considered the optimal loading material tests. Under both the Hashin and Puck criteria, failure
condition for a pipe with a winding angle of ±55°, where the is predicted when the matrix failure or fibre failure is equal to
majority of the loads are withstood by the glass fibres rather the unit value.
than the resin matrix.
As shown in Figs. 13 various failure criteria, such as
It is believed that matrix systems become more the Tsai-Wu, Puck and Hashin criteria, were programmed and
malleable at higher temperatures. This temperature implemented in the ANSYS software to determine the best
dependence results in the shrinkage of the failure envelopes performing failure criterion with the greatest accuracy for
where it becomes slightly narrower to accommodate the predicting the performance of GRE pipes. Figure 13 (a–c),

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 23

shows the modelled and experimental based biaxial failure elevated temperatures. Highly noticeable stress reductions
envelopes for ±55° GRE pipes tested at RT, 65 °C and 95 °C. were observed for the pure axial and pure hoop loading
Given that the FPF point is defined as the point at which conditions, where the failure mode is matrix dominated. Tsai-
permanent damage is considered to start, it also can be Wu criterion also gives good prediction at RT and 65 °C,
referred to as the initial failure stress. In general, the failure where the functional failure stresses were recorded at 193.85
envelopes generated from the FPF point show a strong MPa and almost 200 MPa, respectively. The criterion found to
dependence on stress ratio and test temperatures. For 2:1 test closely predict the ultimate failure compared to experimental
at RT, the envelope also shows that pure hoop FPF strength results. Overall, Hashin failure criteria give good performance
(~296MPa) is over four times greater than axial UEWS to predict the damage initiation of the glass fibre/epoxy
strength (~70MPa). This is mainly because in pure axial composite layers, which includes four failure modes; the fibre
loading the load is strongly matrix dominated. As the stress tension and compression, the matrix tension and compression
ratio increases, the axial and hoop FPF failure strengths also at elevated temperatures. The finding well agrees with WWFE
increased due to the greater loads now being taken up by the study which suggested the superiority of Hashin’s over Tsai-
stronger fibre. Wu particularly on the failure modes identification [39].
For the effects of elevated temperatures, noticeable ACKNOWLEDGEMENTS
reductions were observed for the pure axial and pure hoop The authors would like to thank the Ministry of
loading conditions, where the failure mode is matrix Science and Technology Innovation (MOSTI), Malaysia, for
dominated. The failure envelope generated from FE simulated its financial funding under the e-Science Fund, research grant
FPF correlated well with previously published experimental number 03-01-15-SF0202. The authors would also like to
results [37]. Furthermore, the envelopes in Fig. 13 show that thank the School of Mechatronic Engineering, Universiti
the largest decline in the pipe stiffness was found under pure Malaysia Perlis for providing space, lab facilities and financial
axial loading, whereas less degradation was observed in pure assistance throughout the study.
hoop loading. For examples pipes with Tsai-Wu criterion
exhibited high performances under pure axial loading whereas REFERENCES
[1] Bai.J, Seeleuthner.P and Bompard.P, Compos. Sci. Technol, 1997,
those with Hashin criterion showed the highest ability in the 57 (2), 141-153.
case of hoop loading at elevated temperatures. According to [2] Gibson. A.G, Abdul Majid. M.S, Assaleh T.A, Fahrer. A, Rookus.
Hashin criteria, the longitudinal modulus reduced to zero C.A.P and Hekman.M, Plastic Rubber and Composites, 2011,
when fibre breakage occurred. If matrix failure occurred, the 40(2), 80-84.
[3] Mahdi.E, Hamouda. A.M.S, Sahari B.B and Khalid Y.A, Journal
transverse and shear moduli reduced to zero. The transverse of Materials Processing Technology, 2003, 132(1-3), 49-57.
stress found to increase as pressure increases, whereas [4] Griffiths. J and Gudimetla. P, Asian International Journal of
reducing the longitudinal strength. The three failure criteria Science and Technology in Production and Manufacturing, 2008,
evaluated in this study produced comparable results 1(2), 41-50.
[5] Nowak A.J, International Scientific Journal, 2013, 64(2), 198-204.
concerning the performance of GRE pipes albeit difference in [6] Farooq U and Gregory K, ARPN Journal of Engineering and
accuracy. Additionally, at elevated temperatures the material Applied Sciences, 2010, 5(4), 75-85.
is an oxidising environment, which means the failure mode [7] Krishnan P, Abdul Majid M.S, Afendi M, Gibson A.G and
may become a function of the microstructure instabilities [38]. Marzuki H.F.A, Journal of Material and Design, 2015, 88,196-206.
[8] Ahmad Zaidi A.M, Abdul Hamid H, Ghazali M.I, Abdul Rahman
Hence for high temperatures any reliable life prediction I, Mahzan S and Yusof M.S, International Journal of Integrated
criteria must consider the effects of multi-axially as well as Engineering, 2009, 1(1), 77-208.
that of the material microstructure. Therefore, an [9] Bhavya S, Ravi Kumar P, Abdul Karim S.D, IOSR Journal of
understanding of the failure modes and damage mechanics of Mechanical and Civil Engineering 9IOSR-JMCE), 2012, 3(3), 01-
07.
structural materials under creep-fatigue loadings at elevated [10] Ananthasagaram S and Gopinath V, International Journal of
temperatures becomes essential for developing reliable failure Modern Engineering Research (IJMER), 2014, 2(1), 127-133.
criteria. [11] Xia M, Takayanagi H and Kemmochi K, Composite. Structure,
2001, 53(4), 483-491.
4. CONCLUSIONS [12] Ellyin F and Martens M, Composite Science and Technology,
2001, 61(4), 491-502.
The effects of different failure criteria, such as the
[13] Azzi, V.D. and S.W. Tsai, Experimental Mechanics, 1965, 5(9),
Tsai-Wu, Hashin and Puck criteria, on the damage initiation 283-288.
and failure strength of composite vessels were comparatively [14] Abdul Majid M.S, Gibson A.G, Hekman M, Afendi M, Amin
studied. There were differences in the results obtained for the N.A.M, Plastics Rubber and Composites.2014, 43(9), 1-11.
[15] Soden, P.D., A.S. Kaddour, and M.J. Hinton, Composites Science
various criteria; however, they showed relatively similar
and Technology, 2004. 64(3-4), 589-604.
trends. During failure analysis, Hashin criterion results in [16] Hinton, M.J., P.D. Soden, and A.S. Kaddour, Strength of
superior predictions for determining the FPF of GRE pipes composite laminates under biaxial loads. Applied Composite
due to its consideration of the effect of the intermediate stress Materials, 1996. 3(3), 151-162.
[17] Hinton. M.J, Kaddour A.S and Soden P.D. Failure Criteria in Fibre
that actually occurs. Hashin’s criteria yielded the smallest %
Rein-forced Polymer Composites: The World-Wide Failure
error to experimental data compared to Puck and Tsai-Wu Exercise, Elsevier, 2004.
studied. Meanwhile, with the exception of the 2:1 loading [18] Azzi, V.D. and S.W. Tsai, Experimental Mechanics, 1965. 5(9),
condition, the initial failure stress is simulated to reduce at 283-288.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 24
[19] Tsai, S.W. and E.M. Wu, A Journal of Composite Materials, 1971. [30] Wimmer G, Schuecker C, Pettermann H.E, Wimmer et al 2006,
5, 58-80. 27-29.
[20] Rotem, A., Prediction of laminate failure with the Rotem failure [31] Laš V, Zem R, Kroupa T and Kottner R, Bull Appl. Mech, 2008,
criterion. Composites Science and Technology, 1998. 58(7): p. 4(14), 81-87.
1083-1094. [32] Blanco N, Des. Anal. Compos with Finite Element, 2010, 1-19.
[21] Puck, A. and H. Schurmann, Failure analysis of FRP laminates by [33] Bogert P, Satyanarayana A and Chunchu P, Struct.Adapt. Struct.
means of physically based phenomenological models. Composite Conf, 2006.
Science Technology, 1998. 58(7), 1045. [34] Barrère-Tricca C, Halary J.L and Dal Maso F,Oil & Gas Science
[22] Wang Y, Sun M, Zheng Z and Zhu S, International Conference on and Technology, 2002, 57(2), 169-175.
Educational and Network Technology, 2010, 6(122), 259- [35] Knox E.M, Cowling M.J and Hashim S.A, Composites: Part A,
262. 2000, 31, 583-590.
[23] Ananthasagaram S, Gopinath V, International Journal of Modern [36] Vijayaraghavan G.K and Sundaravalli S, Flash Thermography,
Engineering Research (IJMER), 2011, 2(1), 127-133. 2005, 52(9), 481-487.
[24] Rafiee R and Amini A, Computational Material Science, 2015, 96, [37] Abdul Majid M.S, Assaleh T.A, Gibson. A.G, Hale J.M, Fahrer A,
579-588. Rookus C.A.P and Hekman M, Compos. Part A Appl. Sci. Manuf,
[25] Ajit K.P, IOSR Journal of Mechanical and Civil Engineering, 2011, 42(10), 1500-1508.
2013, 10(1), 47-51. [38] Ming. D, Robert J.A study of failure prediction and design criteria
[26] Madhavi M, Rao K.V.J and Narayana Rao K, Defence Science for fiber composites under fire degradation, Composites Applied
Journal.2009, 59(1), 73-81. science and Manufacturing, 1999, 30, 123-131.
[27] Shanmugavel M, Velmurugan, IOSR Journal of Mechanical and [39] Hashin Z. Failure criteria for unidirectional fiber composites, J
Civil Engineering (IOSR-JMCE), 11-16. Appl Mech, 1980, 47.329-334.
[28] Kreculj D, Rašuo B, Tehnički vjesnik, 2013, 20(3), 485-495.
[29] Tsai, S.W. and E.M. Wu, Journal of Composite Materials, 1971,
(5) 58-80.

List of Figures

Fig. 1. Schematic of the filament winding process [3].

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 25

Fig. 2. Dimensional characteristics of pipe modelling

Fig. 3. Flow chart for failure prediction modelling of composite laminates.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 26

Fig. 4. Engineering data used for GRE pipe model.

Fig. 5. Composite pipe analysis of the model

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 27

(a) (b)
Fig. 6. Model development using ANSYS software; (a) solid model of GRE pipes; (b) meshed model of GRE pipes.

Fig. 7. The load and boundary conditions applied for the GRE pipe analysis.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 28

(a) (b)
Fig. 8. Structural analysis. (a) axial stress; (b) hoop stress at a stress ratio of 2:1 using Tsai-Wu criterion.

(a) (b)

(c)
Fig. 9. Stress deformation: (a) principal stress in direction 1 (s1), (b) principal stress in direction 2 (s2), (c) in-plane shear stress (s12) at a 2:1 stress ratio using
Tsai-Wu criterion.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 29

Fig. 10. Failure distribution at maximum load for a stress ratio of 2:1 using Tsai-Wu failure criterion.

Fig. 11. Stress distribution of GRE pipe at a stress ratio of 2:1.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 30

Fig. 12. Stress–strain curve at a stress ratio of 2:1 at room temperature.

(a)

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 31

(b)

(c)
Fig. 13. Combination of failure criteria in failure envelope for GRE pipe at (a) RT, (b) 65 °C and (c) 95 °C.

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 32

List of Tables

Table I
Material description [14].
Title Description
Fibre E-Glass
Type of fibres Long and continuous
Type of composite Laminate
Matrix Epoxy
Orientation ±55°
No. of ply 10

Table II
Mechanical property of E-glass fibre and Epoxy resin [14]
Property E-glass Epoxy resin
Elastic modulus E (MPa) 73000 2800
Shear modulus G (MPa) 30400 1000
Poisson’s ratio v (-) 0.2 0.4
Density ρ kgm-3 2600 1200

Table III
Physical and mechanical properties of GRE pipes at various temperatures [14].

Physical and mechanical properties 25 °C (RT) 65 °C 95 °C

Internal diameter (mm) 200 200 200
Average wall thickness (mm) 6 6 6
Length of pipes (mm) 2,000 2,000 2,000
No. of ply 10 10 10
-3
Density, ρ kgm 2,000 2,000 2,000
Matrix Young’s modulus Em, MPa 2,800 2,240 1,680
Fibre Young’s modulus Eg, MPa 73,000 58,400 43,800
Axial Young’s modulus Ea, MPa 11,700 9,000 7,000
Hoop Young’s modulus Eh, MPa 20,500 16,300 12,000
Ply stiffness, fibre direction E1, MPa 44,900 36,000 27,000
Ply stiffness, transverse fibre direction E2,
MPa 11,400 9,100 6,800
Poison’s ratio, v12 0.28 0.28 0.28
Poison’s ratio v21 0.071 0.071 0.071
Shear modulus G12, MPa 4,600 3,600 2,800

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:16 No:04 33
Table IV
Predicted pressure for first ply failure (FPF) using Tsai-Wu, Puck and Hashin failure criteria at a stress ratio of 2:1
FPF Pressure (MPa)
Temperature (°C)
Tsai-Wu Puck Hashin
RT 10.4 9.1 10.4

65 10.5 7.5 7.5

95 10.4 6 6

Table V
Failure envelop for a pipe with a winding angle of ±55° using the Tsai-Wu failure criterion; (a) RT; (b) 65 °C; (c) 95 °C.
Experimental Tsai-Wu
Stress RT 65°C 95°C RT 65°C 95°C
Ratio % % %
σhp σhp σhp σhp Error σhp Error σhp Error
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
60(axial 53(axial 35(axial 70(axial 60(axial 59(axial
0:1 stress) stress) stress) stress) 17 stress) 14 stress) 67
1:1 90 82 76 97 8 95 16 82 8
2:1 190 210 210 194 2 174 17 201 4
4:1 340 295 280 339 0 294 0 283 1
1:0 270 215 170 297 10 200 7 172 1

Table VI
Failure envelop for a pipe with a winding angle of ±55° using the Puck failure criterion; (a) RT; (b) 65 °C; (c) 95 °C.
Experimental Puck's
Stress RT 65°C 95°C RT 65°C 95°C
% % %
Ratio σhp σhp σhp σhp σhp σhp
Error Error Error
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
60(axial 53(axial 35(axial 67(axial 62(axial 54(axial
0:1 stress) stress) stress) stress) 12 stress) 17 stress) 54
1:1 90 82 76 101 12 85 3 77 1
2:1 190 210 210 213 12 196 7 217 3
4:1 340 295 280 360 6 293 1 271 3
1:0 270 215 170 298 10 201 6 179 5

Table VII
Failure envelop for a pipe with a winding angle of ±55° using the Hashin failure criterion; (a) RT; (b) 65 °C; (c) 95 °C.
Experimental Hashin
Stress RT 65°C 95°C RT 65°C 95°C
Ratio % % %
σhp σhp σhp σhp Error σhp Error σhp Error
(MPa) (MPa) (MPa) (MPa) (MPa) (MPa)
60(axial 53(axial 35(axial 67(axial 62(axial 54(axial
0:1 stress) stress) stress) stress) 12 stress) 16.77 stress) 54
1:1 90 82 76 101 12 85 3.35 89 17
2:1 190 210 210 192 1 220 4.94 217 3
4:1 340 295 280 294 13 295 0.05 271 3
1:0 270 215 170 269 1 221 2.70 179 5

161504-8989-IJMME-IJENS © August 2016 IJENS

IJENS