Composite Structures 244 (2020) 112307

Contents lists available at ScienceDirect

Composite Structures
journal homepage: www.elsevier.com/locate/compstruct

Flax fabric-reinforced epoxy pipes subjected to lateral compression T

a,⁎ a b a
Saeed Eyvazinejad Firouzsalari , Dmytro Dizhur , Krishnan Jayaraman , Nawawi Chouw ,
Jason M. Inghama
a
Department of Civil and Environmental Engineering, The University of Auckland, New Zealand
b
Department of Mechanical Engineering, The University of Auckland, New Zealand

ARTICLE INFO ABSTRACT

Keywords: Application of flax fibre as reinforcement in polymer composites used in the construction and manufacturing
Natural fibre industries is a step towards achieving more sustainable construction practices. Flax fabric-reinforced epoxy (FFRE)
Flax fabric-reinforced epoxy pipe was used to manufacture forty-two pipes with a variety of internal diameters and flax fabric layers, all with a
Tube length-to-internal diameter ratio of 1.5:1. Lateral compression loading was performed on the FFRE pipes to es-
FRP
tablish the effect of internal diameter and number of fabric layers on the strength, energy absorption capability,
Lateral compression
strains, and failure mechanism. When compared to previously reported test results for pipes manufactured from
Parallel-plate loading
Circumferential bending different natural fibre composites, the FFRE pipes manufactured in this study and subjected to lateral compression
Crushing showed higher strength, flexibility, and specific energy absorbed while demonstrating comparable strength, spe-
Specific energy absorbed cific strength, and specific energy absorbed with synthetic and hybrid fibre composite pipes.

1. Introduction tensile strength between FFRE and glass fibre reinforced epoxy (GFRE)
composites [18], indicating the substantial potential for FFRE composites to
Natural fibres are readily available in many countries, and there is be used in structural components, especially in developing countries where
considerable potential for their application as a sustainable construction flax fibres are often available in abundance [16,17,20].
material [1,2] due to increasing environmental concern [3,4]. In order to GFRP composites used in pipes, tanks and corrosion-resistant ap-
achieve a more sustainable construction industry, the United States (US) plications represent the third largest group of glass fibre applications in
Department of Agriculture and the US Department of Energy have set goals Europe [2,21], such that this industry sector provides many potential
of having at least 10% of all basic chemical building blocks to be created opportunities for the application of flax fibres in composite pipe man-
from renewable and plant-based sources by 2020, increasing to 50% by 2050 ufacture. FFRE offers many advantages for manufacturing pipes when
[5,6]. Additionally, the International Council for Research and Innovation in compared to conventional pipe materials such as polyvinyl chloride
Building and Construction (CIB) published its Agenda 21 on Sustainable (PVC), glass fibre reinforced plastic (GRP), steel, and concrete. FFRE
Construction (CIB Report Publication 237) in 1999, placing emphasis on the resists corrosion [13,22] and no protective coating is required on the
use of agricultural waste products and other biological materials as building internal and external pipe surfaces when compared to their steel
products for further sustainable construction practices [7]. counterparts. PVC and GRP pipes are commonly used to address the
The substitution of synthetic fibres for natural fibres has recently corrosion issue, and these alternative pipe systems are manufactured
become a focus in composite research [8–15]. For example, Dittenber from non-sustainable resources. The application of synthetic fibres for
and Ganga Rao [12] compared the physical and mechanical properties the manufacture of water pipes may be a health hazard if any synthetic
of glass fibre to more than 20 commonly used natural fibres such as fibre debris causes contamination of the water passing along the pipe,
sisal, ramie, kenaf, jute, hemp, flax, coir, and cotton, and found that of whereas natural fibre debris may result in such health hazards being
the various natural fibres tested, flax fibre offered the best potential avoided [23]. Additionally, FFRE composites are approximately 16%
combination of low mass, low cost, high strength, and high stiffness. the density of steel and 50% the density of concrete [24], such that
The behaviour of structural members of various configurations made or FFRE pipes are expected to be more convenient for transportation and
strengthened with flax fabric-reinforced epoxy (FFRE) laminates has been installation when compared to their steel and concrete counterparts.
widely studied previously [16–19]. Such research shows close similarity in Furthermore, the reduced mass of FFRE results in lower inertia forces

Corresponding author.
⁎

E-mail addresses: seyv943@aucklanduni.ac.nz, s.eyvazinejad@gmail.com (S. Eyvazinejad Firouzsalari), ddiz001@aucklanduni.ac.nz (D. Dizhur),
k.jayaraman@auckland.ac.nz (K. Jayaraman), n.chouw@auckland.ac.nz (N. Chouw), j.ingham@auckland.ac.nz (J.M. Ingham).

https://doi.org/10.1016/j.compstruct.2020.112307
Received 8 December 2019; Received in revised form 18 March 2020; Accepted 27 March 2020
Available online 30 March 2020
0263-8223/ © 2020 Elsevier Ltd. All rights reserved.
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

being attracted during dynamic loadings such as earthquake excitations were the main pipe damage mechanisms. El-Sobkhy and Singace [44]
when compared to their steel and concrete counterparts. experimentally investigated the deformation of double-skin profiled
Although FFRE composites have the potential to be used in pipe polyethylene pipes subjected to lateral compression and concluded that
manufacture, several challenges for practical application of FFRE pipes increasing the pipe diameter improved the pipe energy absorbed and pipe
still exist. One major concern is the durability of FFRE composites specific energy absorbed. Niknejad et al. [50], Moeinifard et al. [51], and
[25,26], with the shortage of existing data related to the durability of Rouzegar et al. [52] performed experimental studies to investigate the
FFRE being one major challenge that needs to be overcome prior to effect of pipe geometric properties on structural behaviour of E-glass fibre
widespread acceptance of FFRE for pipe manufacture. FFRE composites composite tubes when subjected to lateral compression, with the results
have relatively poor moisture resistance because of the presence of hy- showing that the total energy absorption capacity of the larger specimens
droxyl and other polar groups in the fibres when compared to glass or is more than for smaller tubes.
carbon FRP composites [27,28], and flax fibres with a high moisture
uptake have weak fibre/matrix interfacial bonding which may compro- 2. Experimental procedures
mise the mechanical properties of the FFRE composites [28,29]. It is
therefore necessary to enhance the hydrophobicity of the natural fibres 2.1. Materials
by treating them with suitable coupling agents or by coating the fibres
with an appropriate resin matrix to generate FFRE composites having Commercial balanced plain-woven (bidirectional) flax fabric with a
better mechanical properties and superior durability performance mass per unit area of 550 g/m2 manufactured by Lineo located in
[13,30]. A more comprehensive analysis should involve potential frac- France [53], see Fig. 1a, and 105 West System epoxy and its 209 extra
ture mechanics and the fatigue behaviour of FFRE pipes. Other issues to slow hardener with the mixture ratio being 3.5:1 by mass, respectively
be considered are that the void ratio and void behaviour may be different [54], were used to manufacture pipe specimens. The moisture content
for gravity flow when compared to low, moderate and high-pressure pipe of the flax fabric before composite manufacture was measured using
flow, and that the life span of FFRE pipes will merit attention and oven-drying tests on 10 flax fabric samples according to ASTM D2495
comparison against the performance of existing pipe materials. Also, the [55], with the tests showing an average moisture content of 6.25% with
pressure requirements for different pipe applications will vary and the a standard deviation of 0.87. Tensile properties of flax yarn were de-
design requirements will be dependent on the application. By properly termined using a universal Instron 5567 at a constant temperature of
defining the required specification, the performance of the FFRE com- 23 °C, with the gauge length set to 100 mm and the cross-head dis-
posite pipes can be properly benchmarked and assessed. placement rate set to 2 mm/min. The mean diameter of the sample was
Lateral compression can arise within buried pipelines during their calculated as the average of 5 measurements along the fibre using an
operation from the weight of the soil above the pipe (dead load) or by optical microscope. An axial pre-tension equal to 25 MPa was applied to
vehicles traversing the pipeline on the ground surface [31,32]. Tubes are the yarn samples before starting the tensile test to alleviate the effect of
also commonly used as energy-absorbing devices that are subjected to yarn tightness on the measured strain. Twenty tests were conducted,
lateral compression during impact events. For example, tubes are used to and the average of strength, failure strain, and Young’s modulus are
improve the crashworthiness of vehicles such as cars, lifts, aircraft, and given in Table 1. A typical stress–strain curve for flax yarn is given in
ships [33,34], crash barriers [35], road bridges, offshore structures, and oil Fig. 2a, with the curve showing that the stress–strain response can be
tankers [34], or tubes are used to confine concrete columns or soil, with divided into three phases: (i) a non-linear phase extending from 0 to
the application of that being associated with reinforcement of concrete 0.015 strain; (ii) a linear phase extending from 0.015 to 0.02 strain, and
columns or soil in infrastructure applications [16,17]. In this study the (iii) a linear phase extending from 0.02 strain until rupture of the yarn.
term ‘tube’ is used to refer to cylindrical hollow sections used for structural The mechanical properties of flax fibres obtained from tensile tests
purposes such as energy-absorbing devices or concrete filled steel/com- on dry flax yarn may differ from the equivalent fibre properties when
posite tubes used in infrastructure applications, whereas the term ‘pipe’ is used in composites. Recent studies on fibre tensile properties showed
used to refer to a hollow tubular section that transmits fluids or gasses. that impregnated composite yarn samples more accurately represented
The aim of the study reported herein was to address the feasibility of the fibre tensile properties within a composite when compared to dry
FFRE pipes to replace pipes manufactured from non-sustainable re- yarns [56,57]. Fifty flax yarns were cut from the flax fabric and fully
sources, such as GRP and PVC pipes, from structural performance per- immersed in the Gurit PRIME™ 27 epoxy resin and its PRIME™ slow
spective. If so, FFRE pipes would have a substantial potential to be used hardener with the mixture ratio being 100:28 by mass, respectively
especially in developing countries where flax fibres are available in [58], at an environment temperature of 21 °C, degassed in 2 mbar for
abundance. In this study, the response of FFRE pipes with varying dia- 10 min, and then the excess resin was removed from the yarn external
meters and fabric layers when subjected to lateral compression loading surface and the yarns were elongated and cured at 21 °C for 5 days. The
was investigated. A complete explanation of the pipe fabrication and tensile properties of the impregnated flax yarns were determined ac-
testing methodology is provided, and pipe crown vertical deformation, cording to ASTM D3039 [59] at a constant temperature of 23 °C and
pipe springline horizontal deformation, strength, circumferential strain with the cross-head displacement rate set to 2 mm/min. The stiffness
at different locations, and pipe failure mechanisms are evaluated. and strength of the flax yarn were back-calculated from the composite
Additionally, energy absorption and specific energy absorbed (SEA: en- properties via the rule of mixtures. The fibre volume fraction Vf was
ergy absorbed per unit mass) characteristics of FFRE pipes presented calculated using the following equation [56]:
herein are compared to energy-absorbing tubes with different materials.
mf
Parallel-plate loading is a standardised test for ensuring that the stiff-
f
ness and strength of a pipe subjected to lateral compression meets speci- Vf = mf mc mf × 100
fied levels of performance [36,37]. There are many studies in the literature +
f m (1)
which focus on steel pipe/tube lateral compression, with the main con-
cerns of these studies being the effect of different geometric parameters on where mf is the mass of flax fibres, ρf is the density of the flax fibres, mc
the energy absorption properties of the tubes [38–43]. Research on lateral is the mass of FFRE composite, and ρm is the density of the matrix. From
compression of composite pipes/tubes has been previously undertaken by the measured composite properties of the impregnated fibre yarn, the
many researchers [44–52]. Faria [47], Sebaey and Mahdi [48], Park et al. flax fibre properties were calculated using the following formulae [56]:
[49], Gupta and Abbas [45], and Abdewi et al. [46] investigated the lat- Ec Em × (1 Vf )
eral compression behaviour of composite pipes manufactured from glass- Ef =
Vf (2)
fibre resins, and showed that inter-layer delamination and fibre breakage

2
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 1. Micrograph pictures.

Table 1
Mechanical properties of flax yarn and epoxy.
Material Diameter (mm) Density (g/cm3) Tensile Modulus (GPa) Ultimate tensile strength (MPa) Tensile strain at break (%)

Flax fibre (dry yarn) 0.63 1.45 16.40 (4.65) 152.67 (28.70) 3.78 (1.23)
Flax fibre (based on impregnated yarn) 0.63 1.45 3.80 (0.20) 224.58 (105.47) 1.75 (0.55)
West system epoxy N/A 1.18 3.09 (0.45) 58.26 (6.07) 5.27 (0.92)
Gurit epoxy N/A 1.08 3.58 (0.846) 50.67 (8.52) 4.65 (1.33)

Note: Values inside parenthesis show standard deviation.

Fig. 2. Example of typical stress–strain curves of a flax yarn and an epoxy sample.

Table 2
Physical and mechanical properties of FFRE composites.
No. of flax Thickness Tensile Tensile Tensile strain Flexural Flexural Strain at Void content Fibre volume
fabric layers (mm) strength modulus (GPa) at failure strength (MPa) modulus (GPa) flexural fraction
(MPa) strength Buoyancy Microscopic

2-layer 2.96 34.51 3.38 0.013 41.43 (5.87) 1.83 (0.311) 0.073 (0.009) 0.052 0.055 39%
3-layer 4.49 43.87 3.98 0.016 50.80 (2.53) 2.11 (0.139) 0.061 (0.004) 0.056 0.053 41%
4-layer 5.80 49.77 3.51 0.021 58.02 (4.20) 2.81 (0.0403) 0.046 (0.006) 0.055 0.056 43%

Note 1: Values inside parenthesis show standard deviation.

Note 2: ‘N/A’ indicates ‘not applicable’.

c
'
m × (1 Vf ) ultimate tensile strength of the fibre, and σ′m is the stress in the matrix
u, f = at the failure strain, which can be calculated assuming linear elastic
Vf (3)
deformation of the matrix up to the instant of fibre (and hence com-
posite) failure.
where Ef is the tensile modulus of the fibre, Ec is the tensile modulus of
FFRE composite, Em is the tensile modulus of the matrix, σu,f is the

3
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 3. Manufacturing process of FFRE pipe specimens.

Table 3
FFRE pipe geometric properties.
Specimen type Avg. internal diameter Nom. of Avg. thickness Avg. length Pipe mass Void content Fibre mass Fibre volume
(mm) layers (mm) (mm) (g) fraction fraction
Buoyancy Microscopic

D38-N2-S1~S3 38.1 2 3.0 68 28.3 0.05 0.06 0.42 0.37

D38-N3-S1~S3 3 4.3 74 47.0 0.06 0.05 0.40 0.38
D38-N4-S1~S3 4 5.5 77 54.1 0.06 0.06 0.42 0.40
D60-N2-S1~S3 60.4 2 3.2 103 73.8 0.05 0.05 0.41 0.38
D60-N3-S1~S3 3 4.9 110 131.9 0.05 0.06 0.42 0.38
D60-N4-S1~S3 4 6.1 100 146.0 0.06 0.06 0.43 0.41
D100-N2-S1~S3 100.1 2 3.3 167 205.5 0.05 0.06 0.43 0.37
D100-N3-S1~S3 3 4.8 174 303.5 0.06 0.06 0.44 0.39
D100-N4-S1~S3 4 6.4 174 433.2 0.06 0.07 0.44 0.42
D160-N4-S1~S3 160.5 4 5.4 229 752.0 0.06 0.06 0.43 0.42
D205-N4-S1~S3 208.3 4 5.6 320 1414.3 0.05 0.06 0.43 0.41
D60-N2-S1~S3 (VI) 60.3 2 2.2 90 59.5 0.04 0.05 0.55 0.48
D60-N3-S1~S3 (VI) 3 3.2 100 93.2 0.04 0.06 0.53 0.46
D60-N4-S1~S3 (VI) 4 4.1 100 115.5 0.04 0.06 0.53 0.46

Note: ‘VI’ indicates pipes fabricated using vacuum infusion technique.

according to ASTM D638 [60] by conducting tensile tests on 10 samples

of both group of resins at a constant temperature of 23 °C and with the
cross-head displacement rate set to 2 mm/min, with the average results
given in Fig. 2b and Table 1. The Young’s modulus of impregnated flax
fibre was less than that for dry flax yarn (see Table 1), with the reason
for this behaviour being attributed to physical and geometrical im-
perfections of testing samples concluded from sample preparation
combined with different testing methodologies. Tensile properties of
FFRE composite with 2, 3, and 4 fabric layers were determined ac-
cording to ASTM D3039 [59] and the flexural properties of the FFRE
were determined according to ASTM D790 [61] at a constant tem-
perature of 23 °C. Fifteen coupon specimens were tested for each fabric-
layer configuration for both tensile and flexural tests, with the average
mechanical properties of these tests listed in Table 2.
The void contents of the FFRE pipes were determined using a
buoyancy (Archimedes) technique, according to ASTM D2734 [62], and
also using an optical digital microscopy technique combined with
image analysis software. Fifteen samples for each specific pipe group
Fig. 4. Details of lateral (transverse) compression test set-up. were tested using the buoyancy technique. For the microscopic analysis
of FFRE, fifteen micro-sections were made using a cold mounting
' method by embedding FFRE pieces from various parts of the pipes in an
m = Em × u, c (4) epoxy resin, followed by grinding and polishing according to ASTM E3-
11 [63]. A micrograph of an FFRE composite with 4-layers is given in
in which εu,c is the ultimate strain of the fibre (and hence composite).
Fig. 1b. The average values of the void contents for each specific pipe
The average results of elastic modulus, ultimate tensile strength and
group and for both techniques are given in Table 2.
tensile strain at break are given in Table 1.The mechanical properties of
After obtaining the density and void content for FFRE composites
Gurit PRIME™ 27 epoxy resin and its PRIME™ slow hardener and 105
with a specific number of fabric layers, the fibre volume fraction for the
West System epoxy and its 209 extra slow hardener were determined

4
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 5. A schematic view of strain gauge locations on a pipe cross-section.

Fig. 6. Averaged load-vertical displacement response of FFRE pipes subjected to monotonic and cyclic lateral compression.

composite was derived from the fibre-to-epoxy resin weight ratio and mould release combined with the plastic sheet greatly facilitated re-
the densities of both fibre and epoxy resin matrix [64]. The fibre vo- moval of the FFRE pipes from the mandrel. The impregnated flax fabric
lume fraction Vf was calculated using the following equation: was wrapped around the mandrel until the targeted number of fabric
layers had been achieved (Fig. 3b). The warp orientation relative to the
1
Vf = 1 Vf
Vv pipe axial axis was θ = 90° and the weft orientation was θ = 0°
1+ Vm (5) (Fig. 3a). The specimens were consolidated for 10 days by curing the
specimens at a constant temperature of 29 °C, after which the pipes
where Vv is the void content of composite and Vm is the volume of epoxy were easily demoulded (Fig. 3c).
resin. The calculated fibre volume fractions of the FFRE composites Thirty-three pipe specimens were produced, with three specimens
with different fabric layers are listed in Table 2. manufactured for each specific configuration of fabric layers and in-
ternal diameter. The pipes had different internal diameters of Dint = 38,
2.2. Specimens manufacture 60, 100, 160, and 205 mm, which are within the typical range for pipe
applications [65]. The number of fabric layers was 2, 3, and 4, while the
Steel tubes were used as mandrels to manufacture the FFRE pipes specimen lengths were equal to 1.5 times the external diameter
(Fig. 3), with the steel tube surface rubbed with mould release, and (L = 1.5Dext). Pipes with Dint = 160 and 205 mm were fabricated with
plastic sheet then taped tightly around the tube. The application of only four fabric layers because for 2 and 3 fabric layers the pipes would

5
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 7. Load-dispalcement response of pipes with different diameters and number of fabric layers.

6
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 8. Typical load-vertical displacement of an FFRE pipe when subjected to lateral compression and photographs of deformation at start and end of each de-
formation phase.

have an impractical D/t ratio. Specimen geometric configurations are 2.3. Test set-up and instrumentation
given in Table 3, with an identifier ‘Dx-Ny-Sz’ assigning for each spe-
cimen, in which the designation ‘Dx’ indicates the internal diameter Lateral compression testing was conducted with a constant loading
(38, 60, 100, 160, and 205), ‘Ny’ indicates the number of fabric layers rate of 2 mm/min at a constant temperature of 23 °C, with the pipe
(2 to 4), and ‘Sz’ indicates the number of the specimen in each group (1 specimen placed between parallel rigid plates at the pipe top and
to 3). For example, D60-N4-S2 indicates the second specimen of the bottom, according to ASTM D 2412 [36] and BS EN 1226 [67]. The
three manufactured pipes having Dint = 60 mm and composed of 4 width of the top and bottom rigid plates was larger than the half-
fabric layers. The length of each pipe specimen was measured at six perimeter of the pipe cross-sections to ensure that full rigid plate con-
locations, the thickness was measured at 12 locations, and the internal tact was maintained throughout the displacement process [68–71]. The
diameter was measured at 3 locations, according to AS 3572.10 [66], experimental configuration designed, as shown in Fig. 4, to specifically
with the averaged values given in Table 3. replace the mechanical model and failure mechanism of the pipe sub-
jected to lateral compression, as clearly illustrated in Fig. 5. Details of
the test set-up are provided in Fig. 4. For each specimen, eight

7
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Table 4
Elastic limit load and peak pipe load of pipes and their associated normalised displacement.
Specimen Elastic limit load (kN.m−1) δv/Dint at elastic limit load Peak pipe load (kN.m−1) δv/Dint at peak pipe load Peak pipe load (Eq. (6), kN.m−1) Error

D38-N2-S1 7.92 0.058 15.13 0.54 14.92 0.05

D38-N2-S2 5.60 0.096 15.30 0.55
D38-N2-S3 6.09 0.058 16.61 0.53
Average 6.53 0.071 15.68 0.54
D38-N3-S1 16.69 0.074 28.21 0.40 38.08 0.19
D38-N3-S2 17.63 0.073 32.96 0.42
D38-N3-S3 15.72 0.057 30.82 0.40
Average 16.68 0.068 30.66 0.406
D38-N4-S1 23.02 0.046 49.22 0.44 68.11 0.29
D38-N4-S2 26.20 0.052 48.93 0.38
D38-N4-S3 25 0.060 47.43 0.47
Average 24.74 0.052 48.52 0.43
D60-N2-S1 5.71 0.070 12.91 0.64 10.88 0.12
D60-N2-S2 4.89 0.069 12.44 0.56
D60-N2-S3 5.09 0.058 11.59 0.60
Average 5.23 0.065 12.31 0.60
D60-N3-S1 17.64 0.066 29.16 0.28 31.48 0.08
D60-N3-S2 17.59 0.058 26.18 0.17
D60-N3-S3 17.46 0.058 31.54 0.18
Average 17.56 0.061 28.96 0.21
D60-N4-S1 19.11 0.038 36.11 0.17 55.74 0.34
D60-N4-S2 21.05 0.050 37.40 0.33
D60-N4-S3 17.86 0.037 36.21 0.19
Average 19.34 0.042 36.57 0.23
D100-N2-S1 2.32 0.076 8.39 0.65 7.18 0.11
D100-N2-S2 2.44 0.076 8.04 0.64
D100-N2-S3 2.63 0.076 7.78 0.66
Average 2.46 0.076 8.07 0.65
D100-N3-S1 5.605 0.054 15.90 0.57 19.27 0.18
D100-N3-S2 5.87 0.055 16.03 0.51
D100-N3-S3 5.94 0.054 15.77 0.49
Average 5.80 0.054 15.90 0.52
D100-N4-S1 15.49 0.047 27.48 0.40 37.94 0.27
D100-N4-S2 11.57 0.048 28.75 0.40
D100-N4-S3 11.82 0.047 26.32 0.40
Average 12.96 0.047 27.51 0.40
D160-N4-S1 4.27 0.082 12.51 0.60 17.24 0.29
D160-N4-S2 4.25 0.082 11.86 0.55
D160-N4-S3 4.26 0.081 12.20 0.57
Average 4.26 0.082 12.19 0.57
D205-N4-S1 2.91 0.089 10.745 0.68 14.39 0.26
D205-N4-S2 2.96 0.089 10.23 0.67
D205-N4-S3 2.79 0.089 10.86 0.69
Average 2.88 0.089 10.61 0.67
D60-N2-S1 (VI) 3.52 0.057 10.65 0.28 − −
D60-N2-S2 (VI) 3.87 0.059 10.48 0.31 − −
D60-N2-S3 (VI) 3.79 0.058 9.93 0.29 − −
Average 3.73 0.058 10.35 0.29 − −
D60-N3-S1 (VI) 7.29 0.047 16.52 0.20 − −
D60-N3-S2 (VI) 7.32 0.054 16.15 0.19 − −
D60-N3-S3 (VI) 7.67 0.055 15.71 0.18 − −
Average 7.43 0.052 16.13 0.19 − −
D60-N4-S1 (VI) 12.06 0.035 26.52 0.13 − −
D60-N4-S2 (VI) 9.69 0.037 24.26 0.14 − −
D60-N4-S3 (VI) 11.77 0.042 25.94 0.15 − −
Average 11.17 0.038 25.57 0.14 − −

circumferential strain gauges were used (Fig. 5), with the strain gauges 3. Results and discussion
placed on the internal surfaces of the pipes at the crown (S-C-I, in which
the letter ‘S’ indicates a strain gauge, ‘C’ shows crown, ‘I’ indicates the 3.1. Monotonic versus cyclic characteristics
internal surface) and invert (S-I-I, in which the first I indicates invert)
and internal and external surfaces of the east springline (S-ES-I and S- Lateral compression testing was conducted using deformation-con-
ES-E, in which ‘ES’ shows east springline and the letter ‘E’ indicates trolled loading, and to amalgamate the three lateral compression
external surface) and west springline (S-WS-I and S-WS-E, in which ‘WS’ testing conducted for each specific pipe group, the average lateral
shows west springline), and internal and external surfaces of west compression load-vertical displacement relationship was calculated.
haunch (S-H-I and S-H-E, in which ‘H indicates haunch). Vertical dis- The average lateral compression load–displacement relationships for
placement of the specimens at the load location was measured by the the D60 pipes with 2, 3, and 4 fabric layers when subjected to mono-
Instron machine while the pipe horizontal displacement in the spring- tonic and cyclic loadings were compared in Fig. 6. The vertical dis-
line was measured manually using digital Vernier callipers with an placement normalised by pipe internal diameter (δv/Dint). The load-
accuracy of +/−0.02 mm. vertical displacement relationships show that the monotonic results
provided an approximate envelope to the cyclic response, and therefore

8
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 9. Comparison of theoretical and experimental vertical and horizontal displacement of pipes at crown and springline.

Fig. 10. Experimental and theoretical pipe stiffness.

monotonic tests were conducted to investigate the lateral compression PR3

= 0.149
response of FFRE pipes. v
EI (6)
and the springline horizontal displacement being equal to
3.2. Deformation characteristics
PR3
= 0.137
Load–vertical displacement relationships of the pipes with the same
h
EI (7)
number of fabric layers and internal diameters along with the asso- where E is the modulus of elasticity of the pipe wall material (MPa) and
ciated average load-vertical displacement are given in Fig. 7, with the I is the moment of inertia of a unit length of pipe which is equal to t3/12
results showing that there was a high level of consistency between the where t is the wall thickness, and R is the pipe nominal radius (half
pipe responses and the average response in each particular group. mean internal diameter plus half mean wall thickness). The results from
The displacement response of the pipes can be classified by three Eqs. (6) and (7) were compared with the average response of specimens
phases i.e. elastic displacement, inelastic displacement, and failure D160-N4 (i.e., D160-N4-S1~S3) as shown in Fig. 9a, with the equations
phases (Fig. 8a): providing a close estimation of the experimental results in the elastic
range (δ/Dint ≤ 0.08). The vertical and horizontal displacements up to
3.2.1. Elastic displacement δv/Dint = δh/Dint = 0.06 were the same, beyond which the change in
The vertical displacement is elastic up to the “elastic limit load” (i.e. the vertical diameter was greater than the change in the horizontal
the load at which large inelastic deformation is initiated, Fig. 8b-Nom. diameter. This behaviour reflects that beyond the elastic displacement,
2). Table 4 gives elastic limit load data for the pipes and their corre- the vertical stiffness was lower than the horizontal stiffness due to pipe
sponding normalised displacements. At a given cross-section, the basic ovalisation, which produced a larger section modulus with respect to
deformation was circumferential bending (Fig. 8b-Nom. 2), and there- the vertical axis than when compared to response about the horizontal
fore, as listed in Table 4, the elastic limit load for the pipes increased axis. Average normalised vertical versus horizontal displacement of
with an increasing number of fabric layers or a decrease in the dia- specimens D160-N4 and D205-N4 are compared in Fig. 9b, with the
meter. The pipe vertical displacement at the elastic limit load decreased figure showing that response of both pipes was similar. The diameter
with increasing the number of fabric layers or decreasing the pipe changes were linear up the δv/Dint = 0.08, beyond which a nonlinear
diameter. response occurred. The vertical displacement was approximately 1.10
Elastic vertical and horizontal displacement of a thin-walled pipe at times the horizontal displacement at the start of loading, which closely
the crown and the springline when subjected to lateral compression is a correlates with the value calculated by Eqs. (6) and (7) [72] in the
classic engineering problem that has been solved in Refs. [72,73], with elastic range, i.e. δv/δh = 1.08, and was 1.70 times the horizontal
the vertical displacement being equal to: displacement at the limited crush load (Fig. 8a). For δv/Dint = 0 to 0.15,

9
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 11. Effect of number of fabric layers (a, b, and c) and diameter (d, e, and f) on load–displacement relationships of FFRE pipes.

Fig. 12. Deformation of FFRE pipes within failure phase.

the ratio between vertical displacement to horizontal displacement was PS can also be found by conducting lateral compression testing on a
δv/δh = 1.20~1.28, which is consistent with the displacement pattern section of pipe at δv/Dint = 0.05 using the formula below, as defined in
of SRHDPE pipes [74] (Fig. 9b), flexible metal (e.g. corrugated steel and ASTM D2412 [36] and ASTM D3262 [65]:
ductile iron) pipes [75], and HDPE pipes [76].
Pipe stiffness (PS) and stiffness factor (SF) can be calculated by a
PS = Q/ v in kPa (9)
closed-form formula for structural analysis of the pipes [77,78]:
The averaged values of PS of each specific pipe group were mea-
EI sured and compared with Eq. (9) as shown in Fig. 10. The experimental
PS = values were slightly greater than the theoretical values except for D38-
0.149R3 (8)
N3 and D38-N4 pipes, for which the theoretical values over-predicted

10
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig 13. Effect of pipe manufacturing process on (a, b, and c) load–displacement relationships and (d, e, and f) specific load–displacement relationships of pipes.

the PS, with the reason for this response being attributed to the rela- Peak pipe load is compared with Eq. (10) in Table 4, with results
tively poor quality of FFRE in the D38 pipes due to the manufacturing showing that the predictions deviated from the experiments by a
difficulties associated with small diameter pipes. maximum value of 34%. For each specific pipe group, the deviation
increased by increasing the number of fabric layers, attributed to the
3.2.2. Inelastic displacement development of debonding between the layers which reduced the peak
Resin micro-cracking appeared on the internal surface of the pipes pipe load (Fig. 12a).
at the crown and the invert and shortly afterwards appeared on the FFRE pipes started yielding within the range of δv/Dint =
external surface at the springline. The vertical displacement increased 0.033~0.07. This range is close to that obtained for steel-reinforced
at a greater rate when compared to the elastic phase due to the de- high-density polyethylene pipes [74,75], but smaller than for corru-
velopment of additional micro-cracks, thus decreasing the FFRE secant gated steel pipes (CSP) [77] and SRHDPE pipes [74], which were within
modulus of elasticity and reducing the pipe section modulus due to δv/Dint = 0.06. The FFRE pipe displacement at peak pipe load (δv/
cross-sectional ovalisation (Fig. 8b-Nom. 3) with the pipe reaching the Dint = 0.21~0.67, Table 4) was much larger than for the CSP [77] and
peak pipe load at the end of this phase. The load-vertical displacement SRHDPE pipes [74], which were within δv/Dint = 0.10, and for GFRE
relationships in Fig. 11 show that by decreasing the diameter or by tubes [48] which were within δv/Dint = 0.06.
increasing the number of fabric layers the pipe strength increased. Peak Although there is no sharp criterion for a flexible pipe definition, the
pipe load and the associated displacement were measured and are given value of δv/Dint = 0.02 is often noted in the literature as the boundary
in Table 4, with the measurements showing that by increasing the between flexible and rigid pipes [75,79]; i.e. a flexible pipe should be
number of fabric layers or by decreasing the pipe internal diameter the capable of withstanding a vertical displacement of δv/Dint = 0.02
peak pipe load increased while the displacement associated with the without significant structural damage, such as cracking. Visual ob-
peak pipe load decreased. The rate of increase of peak pipe load with servation of the tested pipes (made with the unaided eye, according to
pipe fabric layers was greater for tubes of smaller diameter. ASTM D2412 [36] and ASTM D3262 [65]) showed that any pipe struc-
By assuming that the failure of the pipes started by developing tural damage occurred after δv/Dint > 0.02, and thus FFRE pipes can be
fracture lines at the springline, crown, or invert (Fig. 8b-Nom. 3), the classified as flexible pipes. Design standards for flexible pipes allow
peak pipe load can be calculated by vertical pipe displacements to be as large as δv/Dint = 0.05~0.075
[80,81]. Regarding the displacements of FFRE pipes associated with
ut
2
elastic limit load and peak pipe loads given in Table 4, similar vertical
Qpeak =
R (10) pipe displacements can be considered for FFRE pipe design.

11
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 14. Comparison of FFRE pipe load–displacement relationships with tubes made of different materials.

Table 5
Geometric properties of tubes with different materials.
Composite Internal diameter (mm) Wall thickness (mm) Nom. of layers Fibre orientation Reference

FFRE 38.11, 60.98, 100.08, 160.52, and 208.33 2.98~6.37 2, 3, and 4 Woven fabric (0° and 90°) Current study
FFRE 64.00 and 86.00 N/P 2, 4, and 6 Woven fabric (0° and 90°) [15]
KFRP 50.00 N/P 1, 2, and 3 0o, 5o, 10o [83]
JFRE 110.00 2.15 and 3.94 2, 4, and 6 ± 45o [86]
CTFRE 100.00 1.30 Woven roving fibre [88]
GFRE 110.00 2.46 and 4.13 2, 4, and 6 ± 45o [86]
GFRE 160.00 3.70 6 Woven roving fibre [46]
GTFRE 100.00 1.30 N/P Woven roving fibre [88]
GFRE 39.60, 50.00, and 38 3.20, 3.50, and 1.00 7, 7, and 3 Random orientation [45]
CFRE 100.00 1.30 N/P Woven roving fibre [88]
EGFRV 40.00, 50.00, and 60.00 N/P 8 ± 30o [52]
EGFRV 40.00, 51.00, 60.00, and 80.00 N/P 6 and 7 ± 30o [50]
EGFRP 40.00, 50.00, and 60.00 N/P 4 ± 30o [52]
J-GFRE 110.00 6.41 2, 4, and 6 ± 45o [86]
CT-CFRE 100.00 1.30 N/P Woven roving fibre [88]
CT-GTFRE 100.00 1.30 N/P Woven roving fibre [88]
GT-CFRE 100.00 1.30 N/P Woven roving fibre [88]
GT-CT-CFRE 100.00 1.30 N/P Woven roving fibre [88]
Aluminium 30.30~149.00 1.48~3.34 N/A N/A [89]
Brazen 35.00 1.00 N/A N/A [52]

Note: ‘N/P’ indicates ‘not provided’.

3.2.3. Failure phase delamination occurred between the fabric-layers (Fig. 12a) and the pipe
Four fracture lines were developed at the crown, invert, and strength dropped. Further delamination and cross-sectional ovalisation
springline, with the phase angle α ≈ 90° (Fig. 8b-Nom. 3, four-fracture occurred with no significant change in loading. Beyond peak pipe load
line cross-sectional model). These fracture lines functioned as hinges, and the subsequent drop in pipe strength, the lateral compression load
allowing the damaged pipes to deform further. In some specimens, began to increase as the points of contact between the plates and the

12
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 15. Comparison of specific load–displacement relationships for FFRE pipes and tubes made of different materials.

Fig. 16. Developed strains in D160-N4-S2 and D205-N4-S1 pipes.

pipe continuously relocated away from the crown and invert (Fig. 12b). the crown and invert were in contact, which resulted in a sharp rise in
This behaviour resulted in a shorter moment arm, which therefore re- the load magnitude; this condition is called limited crush load [82]
quired greater loading to continue the deformation process. The in- (Fig. 8b-Nom. 4). The pipe was unloaded, with some recovery of de-
crease in pipe strength continued until the inner surfaces of the pipe at formation taking place (Fig. 8b-Nom. 5).

13
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 17. Experimental and theoretic strain distribution on pipe internal (a and b) and external surfaces (c and d).

Progressive fracture is distinguished by several strength drops in the 3.2.4. Comparison to pipes manufactured using vacuum infusion technique
load–displacement relationship during the failure (Fig. 11c). It is as- Gurit PRIME™ 27 epoxy resin and its PRIME™ slow hardener with
sumed that there was considerable variation in the strain across the the mixture ratio being 100:28 by mass, respectively [58], were used to
wall thickness of 3 and 4-layer pipes, leading to delamination between manufacture pipe specimens using the infusion vacuum technique. The
the fabric layers (Fig. 12a), which prevented sudden fracture across the flax fabric was heated in an oven at 96 °C for 12 h before composite
thickness. The fracture initiated from the most internal fabric layer manufacture. Nine pipe specimens were manufactured, with all pipes
(tensile surface) at the crown or invert and continued towards the ex- having Dint = 60 mm and fabric layers of 2, 3, and 4, with three spe-
ternal layer, while at the springline the fracture developed from the cimens manufactured for each specific configuration of fabric layers.
most external layer towards the internal layer. The length of the specimens was equal to 1.5 times the external dia-
In the case of GFRE tubes [45] and FFRE tubes reported by Yan et al. meter (L = 1.5Dext). The void content and fibre volume fraction of the
[15], the phase angle between the fracture lines depended upon the D/t pipes was determined and is given in Table 3.
ratio. For D/t > 10 the phase angle was 90°, while for tubes with D/ The load–displacement and specific load–displacement relation-
t < 10, two fracture lines were developed in the contact regions with ships of series D60 FFRE pipes manufactured using the hand layup and
the parallel plates and the two other fracture lines were developed at a vacuum infusion techniques are compared in Fig. 13. The elastic limit
phase angle of 60° from the crown and invert in an anticlockwise di- and peak pipe loads and the associated displacement are calculated and
rection. For E-glass fibre reinforced epoxy (EGFRE) tubes [51], only two given in Table 4. The response of the FFRE pipes manufactured using
plastic hinges were produced at the springline. the vacuum infusion technique when subjected to lateral compression is

14
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 18. Effect of number of layers on strain-vertical pipe displacement relationships.

Fig. 19. Effect of number of fabric layers on energy absorbed and specific energy absorbed as a relationship of pipe displacement.

characterised by several sudden drops in the load–displacement re- 3.2.5. Comparison to tubes with different materials
lationship, with the elastic limit and peak pipe loads occurring at The load–displacement relationships of FFRE pipes are compared
smaller displacements when compared to the FFRE pipes manufactured with FFRE [15], kenaf fibre reinforced polyester resin (KFRP) [83],
using the hand layup technique. The reason for this response was the GFRE [45,46], E-glass fabric-reinforced vinylester (EGFRV), and E-glass
smaller thickness of the vacuum infusion pipes, which produced smaller fabric-reinforced polyester (EGFRP) [52] tubes in Fig. 14. The geo-
circumferential bending strengths when compared to their hand layup metric specifications of comparable tubes are listed in Table 5. The
counterparts, which had larger thicknesses (see Table 3). compared tubes either had similar internal diameters and fabric layers
(Fig. 14a~d) or similar internal diameters and wall thicknesses
(Fig. 14e). FFRE tube response to lateral compression reported in Ref.

15
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 20. Effect of diameter on energy absorbed and specific energy absorbed as a relationship of pipe displacement.

[15] is characterised by several sudden drops in the load–displacement 3.3. Strain characteristics
relationship, with the peak pipe load occurring at smaller displace-
ments when compared to the tested FFRE pipes reported in the current Closed-form formulae for structural analyses of a pipe have been
research, and for the case of 4-layer tubes, the strength was significantly used to calculate the distribution of internal forces (Nθ) and moments
lower than the measured strength of FFRE pipes reported in the current (Mθ) developed by opposite linear distributed load at crown and invert
research (Fig. 14a). Although KFRP tubes [83] had a smaller diameter (Fig. 5) [72,84]:
than FFRE pipes, the strength of FFRE pipes was considerably higher Q
than measured strength of the KFRP tubes reported in Ref. [83], with N = 2
cos 2 2
the peak pipe load of FFRE pipes occurring at larger displacements N =
Q
cos
3
2 2 2 (11)
(Fig. 14b). The EGFRV tubes [52] had more fabric layers (N = 6 and 8,
Fig. 14a and c) and the tube strength was less than for D60-N4 and and
comparable with for D38-N4. The EGFRP tube [52] had four fabric
QR 2
layers (Fig. 14c) and the tube strength was considerably less than for M = 2 (
cos ) 2 2
D38-N4, but the displacement associated with the peak pipe load was M =
QR 2
+ cos )
3

slightly larger than D38-N4. FFRE pipes showed comparable peak pipe 2 ( 2 2 (12)
load to GFRE tubes [45,46], with the displacement associated with peak where Q is the distributed load and θ is the orientation angle defined in
pipe load being larger than that for GFRE tubes [45,46] (Fig. 14d, e). Fig. 5. From the circumferential bending moment given in Eq. (12) and
The specific load–displacement relationship of the FFRE pipes tested using EI, the curvature radius of the pipe wall is obtained as shown:
in the current research by taking into account pipe mass are compared
EI
with the specific load properties of FFRE [15], GFRE [45,46], E-glass =
fabric-reinforced vinylester (EGFRV), and E-glass fabric-reinforced
M (13)
polyester (EGFRP) [52] tubes in Fig. 15. The specific load of the FFRE where the extreme fibre strain on the pipe external or internal surface is
pipes tested in the current research were greater than for the FFRE calculated from the curvature radius by [85]:
tubes reported in Ref. [15] and for GFRE [45,46] and EGFRP tubes
t
[52], while the EGFRV tubes [52] had the specific load that was com- =
2 (14)
parable to that obtained for the FFRE pipes tested in the current re-
search. Pipe strain vs. load per unit length relationships for D160-N4-S1 and
D160-N4-S2 pipes at different locations are given in Fig. 16a and 16b.
Positive values reflect tensile strains while negative values show

16
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 21. Comparison of EA-displacement relationships for FFRE pipes and GFRE tubes.

compressive strains. S-ES-E, S-WS-E, S-C-I, S-I-I, and S-H-E were sub- Dint = 0.03, 0.05, and 0.07, with the results showing that strains can be
jected to tensile strains while S-ES-I, S-WS-I, and S-H-I recorded com- effectively estimated up to the elastic limit load.
pressive strains. These recorded strains are consistent with the classic Strain-pipe vertical displacement recorded by S-I-I for the pipes with
pipe ovalisation pattern reported for tested FFRE pipes in Section 3.2.1, the same diameter but a different number of fabric layers are compared
and with flexible pipes e.g. corrugated steel and iron pipes [76]. Gen- in Fig. 18, with the results showing greater strain development for the
erally, maximum strain ranged from 0.015 to 0.025. Beyond same vertical displacement with increasing fabric layers due to the
Q = 6 kN.m−1 in specimen D160-N4-S1 and Q = 4 kN.m−1 in spe- greater distance of the extreme fibre from the neutral axis of the wall
cimen D205-N4-S2 the circumferential strains increased with greater thickness for the pipes with more fabric layers.
gradient due to a steady decrease in the secant modulus of the pipes.
The maximum strain on the pipe external surface occurred at the 3.4. Energy absorbed and specific energy absorbed characteristics
springline while the maximum strain on the pipe internal surface oc-
curred at the invert, followed by the crown, and was likely attributable Pipe energy absorbed (EA) at the instant of each particular dis-
to stress concentrated at the crown and invert which were in direct placement was measured by the area under the load–displacement
contact with the loading plates. Strains measured at the haunch were curve from δv = 0 to the displacement considered. EA and SEA-nor-
substantially lower than the strains measured at the crown, invert, and malised vertical displacement relationships for pipes with the same
springline, indicating a conventional hourglass strain distribution, be- diameter but a different number of fabric layers are given in Fig. 19,
cause the circumferential bending moments changed sign between the and corresponding relationships for pipes with the same number of
springline and the invert at an angle of θ = 230°, calculated from Eq. fabric layers but different diameters are given in Fig. 20. The vertical
(12) (Fig. 5). displacement normalised by pipe nominal diameter (mean internal
Fig. 16c and d show that peak extreme fibre strains occurred at the diameter plus one time the mean wall thickness, δv/Dnom). An increase
crown and invert, and then occurred at the springline, such that it is not in the number of fabric layers or decrease in diameter resulted in a
necessary to monitor FFRE pipes in the vicinity of the haunches up to significant increase in EA and SEA, with the gradient of the curves
the normalised displacement of δv/Dint 0.65. Beyond δv/Dint = 0.65, the decreasing beyond the peak pipe load. This observation is contrary to
peak moment might move toward the haunch regions for more flexible that reported by Yan et al. [15] who noted that an increase in the
pipes such as the 2-layer specimens. diameter of FFRE tubes produced larger SEA. The increase in the EA
Strains on the internal surface of the pipes were calculated using Eq. with increasing fabric layers was due to the greater amount of material
(14) and in Fig. 17 are compared with the experiments for δv/ across the section of the tube, which effectively absorbed the energy

17
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 22. SEA comparison of FFRE pipes with hollow tubes of different materials.

18
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 23. Experimental and analytical specific energy absorbed-displacement relationships.

and required greater energy to deform. EA-displacement relationships SEA-normalised vertical displacement of FFRE pipes, and by increasing
for FFRE and GFRE pipes with similar diameters and wall thicknesses the diameter and decreasing the number of fabric layers the consistency
are compared in Fig. 21, with the results showing a larger EA for FFRE of experimental results and the prediction increased. The Gupta et al.
pipes having an equal displacement. relationship (Eq. (15)) [42] is an empirical equation, which was ob-
Different materials are used to fabricate energy absorbing hollow tained from the lateral compression loading on mild steel and alumi-
tubes for applications such as vehicle collision protection or ships impact nium tubes, with the discrepancy between the experiments and the
protection [33,34], with the geometric specification of comparable prediction by Eq. (15) being attributed to the difference in the material
hollow tubes listed in Table 5. In Fig. 22a, FFRE pipe SEA is compared to properties of mild steel or aluminium and FFRE composite.
that for hollow tubes made of already available materials with similar Mean pipe crushing load (Qavg) is the average of the lateral com-
nominal diameters and fabric layers and in Fig. 22b FFRE pipes SEA is pression load through complete deformation. This parameter is useful
compared with the hollow tubes with similar nominal diameters and wall for measurement of the performance of the energy absorbers and plays
thicknesses. The comparison of SEA values shows that the SEA of FFRE an important role in their design. It can be calculated as [90]:
pipes is superior to that for natural composite hollow tubes i.e., KFRP
Etot
tubes [83], FFRE hollow tubes reported in Ref. [15], and jute FRE hollow Qavg =
Lc (16)
tubes [86] and comparable to that for synthetic composite hollow tubes
i.e., GFRE [87], glass tissue FRE (GTFRE) [88], EGFRV [52], aluminium where Etot is the total absorbed energy and Lc is the crush length, which
hollow tubes [89], and hybrid composite i.e., jute-GTFRE (JGTFRE), is equal to pipe vertical displacement at the limited crush load (Fig. 8a
cotton-carbon FRE (CT-CFRE), cotton-glass tissue FRE (CT-GTFRE), glass and b. Nom-4). Stroke efficiency (Se) is considered as a good indicator
tissue-carbon FRE (GT-CFRE), and glass tissue-cotton-carbon FRE (GT- for describing the amount of material that can be used during the en-
CT-CFRE) hollow tubes [88]. These conclusion are applicable specifically ergy absorber failure. This indicator is very useful in the applications
for hollow FFRE pipes and composite tubes as found in the literature. which have restrictions on the energy absorber space. Stroke efficiency
SEA is also defined as energy absorbed per unit volume of the tube for a pipe subjected to lateral compression loading is calculated as
[42], and herein to differentiate it with SEA per unit mass, SEA* (SEA [82,90]:
asterisk) is used as an alternative abbreviation to define SEA per unit
Lc
volume of the tube. Gupta et al. [42] proposed an expression to predict Se =
the SEA* of mild steel and aluminium tubes subjected to lateral com- Dext (17)
pression: The energy efficiency indicator is given by [82]:
1.22
Dnom E
SEA = 0.0106e 4.0633 v / Dnom . Ee =
t (15) Qpeak . Lo (18)
3
in which e is Napier's constant and equals to 2.718 and SEA is in J/mm . where Lo is the initial length of the pipe. The work effectiveness is a
In Fig. 23 the Gupta et al. equation (Eq. (15)) [42] is compared with the combination of the specific energy absorbed capacity with the stroke

19
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

Fig. 24. Energy absorption characteristics of FFRE pipes. Note: Error bars indicate standard deviation.

20
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

efficiency indictor and is defined as follows [82]: Acknowledgement

Weff = SEA × Se (19) The authors gratefully acknowledge Adhesive Technologies NZ Ltd
for providing the epoxy resin and hardener used in this research.
The work effectiveness indicator is very useful indictor for appli-
cations in which there are restrictions on both weight and space of the
Author statement
energy absorbers [82].

• Study conception and design:

The variations in mean pipe crushing load, stroke efficiency, energy
efficiency, and work effectiveness for FFRE pipes with different dia-
meters and fabric layers are shown in Fig. 24. It can be inferred from
Eyvazinejad Firouzsalari, Chouw, and Jayaraman
this figure that by increasing the fabric layers and decreasing the dia-

• Acquisition of data:
meter mean pipe crushing load and work efficiency increased.

4. Conclusions Eyvazinejad Fioruzsalari

Lateral compression of flax fabric-reinforced epoxy (FFRE) pipes • Analysis and interpretation of data:
with different flax fabric layers and internal diameters were in-
vestigated. Pipe invert vertical deformation, pipe springline horizontal Eyvazinejad Firouzsalari
deformation, strength, energy absorbed, specific energy absorbed and
failure mechanism were investigated, alongside circumferential strain • Drafting of manuscript:
measured on the internal surface of the crown and invert, as well as
strain measured on both internal and external surfaces of the springline Eyvazinejad Firouzsalari
and haunch. Conclusions drawn are listed as follows:
• Critical revision:
• The non-elastic phase of deformation was extended when the pipe
diameter increased and the fabric layer decreased. Dizhur, Jayaraman, Ingham
• FFRE pipes showed an ability to withstand a vertical pipe dis-
placement-to-internal diameter ratio greater than 0.02 without Appendix A. Supplementary data
structural damage and hence can be classified as flexible pipes. The
vertical pipe displacement-to-internal diameter ratio at elastic limit Supplementary data to this article can be found online at https://
load was within 0.042–0.089, while at failure was within 0.21–0.67, doi.org/10.1016/j.compstruct.2020.112307.
depending on pipe diameter and the number of fabric layers.
• Pipe elastic limit load and peak pipe load and the associated dis- References
placements, energy absorbed, and specific energy absorbed increased
with a decrease in the diameter or an increase in fabric layers. [1] Pervaiz M, Sain MM. Carbon storage potential in natural fiber composites. Resour

• Failure of the FFRE pipes was accompanied by formation of long- Conserv Recycl 2003;39:325–40. https://doi.org/10.1016/S0921-3449(02)
00173-8.
itudinal fracture lines at the crown, invert and springline. Fractures [2] Ticoalu A, Aravinthan T, Cardona F. A review of current development in natural
initiated under the loading plates at the invert. The phase angle be- fiber composites for structural and infrastructure applications. Proc. South. Reg.
tween the fracture lines was 90°. Pipes with smaller diameter-to- Eng. Conf. (SREC 2010), Engineers Australia. 2010. p. 113–7.
[3] Crocker J. Natural materials. Mater Technol 2008;23:174–8.
thickness ratios were more susceptible to failure through delamination.
•
[4] Singh B, Gupta M, Verma A. Mechanical behaviour of particulate hybrid composite
FFRE pipes with 3 and 4 fabric layers experienced progressive laminates as potential building materials. Constr Build Mater 1995;9:39–44.
fractures due to considerable variation in the strain across the wall https://doi.org/10.1016/0950-0618(95)92859-F.
[5] Mohanty AK, Misra M, Drzal LT. Natural fibers, biopolymers, and biocomposites.
thickness, leading to delamination between the fabric-layers which CRC Press; 2005.
prevented sudden fracture across the thickness. [6] Mohanty Amar K, Misra M, Drzal LT. Natural fiber polymer composites. Illustrate.
• FFRE pipes showed the potential to be used as energy absorbing CRC Press; 2005.
[7] Du Plessis C. Agenda 21 for sustainable construction in developing countries. CSIR
devices. The specific energy absorbed of FFRE pipes was superior to
Rep BOU E. 2002. p. 204.
that of natural fibre composite hollow tubes i.e., kenaf FRP and jute [8] Bismarck A, Mishra S, Lampke T. Plant fibers as reinforcement for green composites.
FRE based on available literature and was comparable to that for Boca Raton, FL: Taylor & Francis; 2005.
glass fibre reinforced epoxy hollow tubes, hybrid composite hollow [9] Symington MC, Banks WM, Opukuro DW, Pethrick RA. Tensile testing of cellulose
based natural fibers for structural composite applications. J Compos Mater
tubes of synthetic and natural fibres, and aluminium hollow tubes as 2009;43:1083–108. https://doi.org/10.1177/0021998308097740.
found in the literature. [10] Mohanty AK, Misra M, Hinrichsen G. Biofibres, biodegradable polymers and bio-

• Maximum strain occurred on the internal surface of the crown and composites: an overview. Macromol Mater Eng 2000;276–277:1–24. https://doi.
org/10.1002/(SICI)1439-2054(20000301)276:1<1::AID-MAME1>3.0.CO;2-W.
the invert and then at both internal and external surfaces of [11] Malkapuram R, Kumar V, Negi YS. Yuvraj Singh Negi. Recent development in
springlines. Strains measured at the haunch were minimal because natural fiber reinforced polypropylene composites. J Reinf Plast Compos
the circumferential bending moments changed sign around the 2008;28:1169–89. https://doi.org/10.1177/0731684407087759.
[12] Dittenber DB, GangaRao HVS. Critical review of recent publications on use of
haunch.
•
natural composites in infrastructure. Compos Part A Appl Sci Manuf
Classic analytical models for the calculation of pipe circumferential 2012;43:1419–29. https://doi.org/10.1016/j.compositesa.2011.11.019.
bending, peak pipe load, and circumferential strains were compared [13] Yan L, Chouw N, Jayaraman K. Flax fibre and its composites – a review. Compos
Part B Eng 2014;56:296–317. https://doi.org/10.1016/J.COMPOSITESB.2013.08.
with the experiments, with the results showing a close correlation
014.
between the calculations and experiments. [14] Yan L, Chouw N. Crashworthiness characteristics of flax fibre reinforced epoxy
tubes for energy absorption application. Mater Des 2013;51:629–40. https://doi.
org/10.1016/j.matdes.2013.04.014.
The behaviour of buried pipes might be different from that observed
[15] Yan L, Chouw N, Jayaraman K. Lateral crushing of empty and polyurethane-foam
in this study, and further research is required to evaluate the effect of filled natural flax fabric reinforced epoxy composite tubes. Compos Part B Eng
soil on the lateral compression response of FFRE pipes. Further in- 2014;63:15–26. https://doi.org/10.1016/j.compositesb.2014.03.013.
vestigations are required to explore the response of FFRE pipes sub- [16] Xia Y, Xian G, Kafodya I, Li H. Compression behavior of concrete cylinders ex-
ternally confined by flax fiber reinforced polymer sheets. Adv Struct Eng
jected to a variety of pipeline loadings.

21
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

2014;17:1825–33. https://doi.org/10.1260/1369-4332.17.12.1825. tubes. Int J Impact Eng 2016;92:18–25. https://doi.org/10.1016/j.ijimpeng.2015.

[17] Xia Y, Xian G, Wang Z, Li H. Static and cyclic compressive properties of self-com- 11.007.
pacting concrete-filled flax fiber–reinforced polymer tubes. J Compos Constr [49] Park J-S, Hong W-H, Lee W, Park J-H, Yoon S-J. Pipe stiffness prediction of buried
2016;20:4016046. GFRP flexible pipe. Polym Polym Compos 2014;22:17.
[18] Assarar M, Scida D, El Mahi A, Poilâne C, Ayad R. Influence of water ageing on [50] Niknejad A, Assaee H, Elahi SA, Golriz A. Flattening process of empty and poly-
mechanical properties and damage events of two reinforced composite materials: urethane foam-filled E-glass/vinylester composite tubes – an experimental study.
flax–fibres and glass–fibres. Mater Des 2011;32:788–95. https://doi.org/10.1016/j. Compos Struct 2013;100:479–92. https://doi.org/10.1016/j.compstruct.2013.01.
matdes.2010.07.024. 009.
[19] Huang L, Yan B, Yan L, Xu Q, Tan H, Kasal B. Reinforced concrete beams [51] Moeinifard M, Liaghat G, Rahimi G, Talezadehlari A, Hadavinia H. Experimental
strengthened with externally bonded natural flax FRP plates. Compos Part B Eng investigation on the energy absorption and contact force of unstiffened and grid-
2016;91:569–78. https://doi.org/10.1016/J.COMPOSITESB.2016.02.014. stiffened composite cylindrical shells under lateral compression. Compos Struct
[20] Pham D-C, Su Z, Sridhar N, Qian X, Huang Z, Sobey AJ, et al. Experimental and 2016;152:626–36. https://doi.org/10.1016/J.COMPSTRUCT.2016.05.067.
numerical studies of large-scaled filament wound T700/X4201 composite risers [52] Rouzegar J, Assaee H, Niknejad A, Elahi SA. Geometrical discontinuities effects on
under bending. Eur. Conf. Compos. Mater. 2016. lateral crushing and energy absorption of tubular structures. Mater Des
[21] Marsh E. Composites in infrastructure-building new markets. First Oxford: Elsevier; 2015;65:343–59. https://doi.org/10.1016/J.MATDES.2014.09.041.
2000. [53] Lineo. Technical data sheet-flaxply 2015. http://www.lineo.eu/download/2015-
[22] Saheb DN, Jog JP, Nabi Saheb D, Jog JP. Natural fiber polymer composites: a re- TDS-FlaxPly.pdf.
view. Adv Polym Technol 1999;18:351–63. https://doi.org/10.1002/(SICI)1098- [54] Technical data sheet. 105 Epoxy Resin/209 Extra Slow Hardener n.d. https://www.
2329(199924)18. westsystem.com/wp-content/uploads/105-209-Epoxy-Resin-1.pdf.
[23] Yu HN, Kim SS, Hwang IU. Application of natural fiber reinforced composites to [55] ASTM D2495—07, Standard test method for moisture in cotton by oven-drying
trenchless rehabilitation of underground pipes. Compos Struct 2008;86:285–90. 2012.
[24] Yan L, Chouw N. Behavior and analytical modeling of natural flax fibre-reinforced [56] Bensadoun F, Verpoest I, Baets J, Müssig J, Graupner N, Davies P, et al.
polymer tube confined plain concrete and coir fibre-reinforced concrete. J Compos Impregnated fibre bundle test for natural fibres used in composites. J Reinf Plast
Mater 2013;47:2133–48. Compos 2017;36:942–57. https://doi.org/10.1177/0731684417695461.
[25] Moudood A, Rahman A, Khanlou HM, Hall W, Öchsner A, Francucci G. [57] Depuydt DEC, Sweygers N, Appels L, Ivens J, van Vuure AW. Bamboo fibres sourced
Environmental effects on the durability and the mechanical performance of flax from three global locations: a microstructural, mechanical and chemical composi-
fiber/bio-epoxy composites. Compos Part B Eng 2019;171:284–93. https://doi.org/ tion study. J Reinf Plast Compos 2019;38:397–412. https://doi.org/10.1177/
10.1016/J.COMPOSITESB.2019.05.032. 0731684419828532.
[26] Jeannin T, Berges M, Gabrion X, Léger R, Person V, Corn S, et al. Influence of [58] Technical data sheet. PRIMETM 27/PRIMETM Slow Hardener n.d. https://www.
hydrothermal ageing on the fatigue behaviour of a unidirectional flax-epoxy la- gurit.com/sitecore/content/Old-Product-Pages/Other-Products/Laminating-
minate. Compos Part B Eng 2019;174:107056https://doi.org/10.1016/J. Infusion-Systems/PRIME-27.
COMPOSITESB.2019.107056. [59] ASTM D3039/D3039M: standard test method for tensile properties of polymer
[27] Pickering KL, Efendy MGA, Le TM. A review of recent developments in natural fibre matrix composite materials. Annu B ASTM Stand 2014:1–13. https://doi.org/10.
composites and their mechanical performance. Compos Part A Appl Sci Manuf 1520/D3039.
2016;83:98–112. https://doi.org/10.1016/J.COMPOSITESA.2015.08.038. [60] ASTM D638-14. Standard test method for tensile properties of plastics 2014:17.
[28] Azwa ZN, Yousif BF, Manalo AC, Karunasena W. A review on the degradability of [61] ASTM D790 - 82 Standard Test Methods for Flexural Properties of Unreinforced and
polymeric composites based on natural fibres. Mater Des 2013;47:424–42. Reinforced Plastics and Electrical Insulating Materials (supersedes ASTM D790 -
[29] Al-Maharma AY, Al-Huniti N. Critical review of the parameters affecting the ef- 81). Am Soc Test Mater 1981.
fectiveness of moisture absorption treatments used for natural composites. J [62] ASTM International. Standard test methods for void content of reinforced plastics.
Compos Sci 2019;3:27. ASTM D 2734-16 2009. https://doi.org/10.1520/D2734-09.2.
[30] Bledzki AK, Letman-Sakiewicz M, Murr M. Influence of static and cyclic climate [63] ASTM International. ASTM E3-11. Standard guide for preparation of metallographic
condition on bending properties of wood plastic composites (WPC). Express Polym specimens. Am Soc Test Mater 2001:1–17. https://doi.org/10.1520/E0003-
Lett 2010;4:364–72. 11R17.1.
[31] Moore ID, Selig ET, Haggag A. Elastic buckling strength of buried flexible culverts. [64] Heslehurst RB. Composite structures engineering design vs. fabrication require-
Transp Res Rec 1988. ments. ACUN-5 Int. Compos. Conf. Dev. Compos. Adv. infrastructure, Nat. nano-
[32] Whidden WR. Buried flexible steel pipe: design and structural analysis. American composites, UNSW Sydney, Australia; 2006.
Society of Civil Engineers; 2009. [65] ASTM D3262-16. Standard specification for “fiberglass” (Glass-Fiber-Reinforced
[33] Johnson W, Walton AC. An experimental investigation of the energy dissipation of a Thermosetting-Resin) sewer pipe. ASTM Int 2016:1–9. https://doi.org/10.1520/
number of car bumpers under quasi-static lateral loads. Int J Impact Eng D3262-16.2.
1983;1:301–8. https://doi.org/10.1016/0734-743X(83)90024-6. [66] AS 3572.10. Plastics-Glass filament reinforced plastics (GRP)-Methods of test.
[34] Alghamdi AA. Collapsible impact energy absorbers: an overview. Thin-Walled Method 10: Determination of the initial ring stiffness of glass filament reinforced
Struct 2001;39:189–213. https://doi.org/10.1016/S0263-8231(00)00048-3. plastics pipes. Aust Stand 1989.
[35] Reid JD, Sicking DL. Design and simulation of a sequential kinking guardrail [67] Plastics piping systems, Glass-reinforced thermosetting plastics (GRP) pipes, Test
terminal. Int J Impact Eng 1998;21:761–72. https://doi.org/10.1016/S0734-743X method to prove the resistance to initial ring definition, British Standard BS EN
(98)00029-3. 1226: 1996, BS 2782 : Part 12 : Method 1214E : 1996 1226 1996, BS 2782 : Part
[36] ASTM D 2412. Standard test method for determination of external loading char- 12 : Method. London, UK: 1996.
acteristics of plastic pipe by parallel-plate loading. ASTM Int 2008;08(04):1–3. [68] Firouzsalari SE, Showkati H. Investigation of free-spanned pipeline behavior due to
https://doi.org/10.1520/D2412-11.2. axial forces and local loads. J Constr Steel Res 2013;86:128–39. https://doi.org/10.
[37] Methods of test for plastics pipes and fittings. Method 23: Method for determination 1016/j.jcsr.2013.03.018.
of ring flexibility. Aust Zeal Stand 2006;AS/NZS 146. [69] Eyvazinejad Firouzsalari S, Showkati H. Behavior of pre-compressed tubes sub-
[38] DeRuntz JA, Hodge PG. Crushing of a tube between rigid plates. J Appl Mech jected to local loads. Ocean Eng 2013;65:19–31. https://doi.org/10.1016/j.
1963;30:391. https://doi.org/10.1115/1.3636567. oceaneng.2013.03.003.
[39] Burton R, Craig J. An investigation into the energy absorbing properties of metal [70] Firouzsalari SE, Showkati H. Thorough investigation of continuously supported
tubes loaded in the transverse direction. University of Bristol; 1963. pipelines under combined pre-compression and denting loads. Int J Press Vessel Pip
[40] Reddy TY, Reid SR. Phenomena associated with the crushing of metal tubes be- 2013;104:83–95. https://doi.org/10.1016/j.ijpvp.2013.01.008.
tween rigid plates. Int J Solids Struct 1980;16:545–62. [71] Eyvazinejad Firouzsalari S, Showkati H, Ingham JM. Free-spanning and base-sup-
[41] Redwood RG. Discussion: “Crushing of a tube between rigid plates” (DeRuntz, Jr., ported tubes subjected to combined axial compression and indentation loads. J
John A., and Hodge, Jr., PG, 1963, ASME J. Appl. Mech., 30, pp. 391–395). J Appl Constr Steel Res 2019;161:341–54. https://doi.org/10.1016/j.jcsr.2019.07.001.
Mech 1964;31:357–8. [72] Timoshenko SP. Strenght of materials. Courier Corporation; 1941.
[42] Gupta NK, Sekhon GS, Gupta PK. Study of lateral compression of round metallic [73] Young OC, Trott JJ. Buried rigid pipes: structural design of pipelins. Elsevier Appl
tubes. Thin-Walled Struct 2005;43:895–922. Sci London, (SW/1940), 1984, 234 1984.
[43] Karamanos SA, Charis E. Collapse of pressurized elastoplastic tubular members [74] Khatri DK, Han J, Parsons RL, Young B, Brennan JJ, Corey R. Laboratory evaluation
under lateral loads. Int J Mech Sci 2004;46:35–56. https://doi.org/10.1016/J. of deformations of steel-reinforced high-density polyethylene pipes under static
IJMECSCI.2004.02.015. loads. J Mater Civ Eng 2013;25:1964–9.
[44] El-Sobky H, Singace AA. Profiled polymer pipes as re-usable energy absorption [75] Moser AP, Folkman SL. Buried pipe design. New York: McGraw-Hill; 2001.
elements. Int J Mech Sci 1999;41:1385–400. [76] Munro SM, Moore ID, Brachman RWI. Laboratory testing to examine deformations
[45] Gupta NK, Abbas H. Lateral collapse of composite cylindrical tubes between flat and moments in fiber-reinforced cement pipe. J Geotech Geoenvironmental Eng
platens. Int J Impact Eng 2000;24:329–46. https://doi.org/10.1016/S0734-743X 2009;135:1722–31.
(99)00173-6. [77] Schluter J, Shade J. Flexibility factor or pipe stiffness: significant stiffness con-
[46] Abdewi EF, Sulaiman S, Hamouda AMS, Mahdi E. Quasi-static axial and lateral siderations. Transp Res Rec J Transp Res Board; 1999. p. 45–50.
crushing of radial corrugated composite tubes. Thin-Walled Struct 2008;46:320–32. [78] Howard AK. Method for prediction of flexible pipe deflection. US Department of the
https://doi.org/10.1016/j.tws.2007.07.018. Interior, Bureau of Reclamation; 1997.
[47] Faria H. Realistic modeling of the failure of composite pipes under ring deflection [79] Talesnick M, Baker R. Failure of a flexible pipe with a concrete liner. Fail. Anal.
and internal pressure loading conditions. ASME 2011 Press. Vessel. Pip. Conf., Case Stud. II. Elsevier; 2001. p. 31–43.
American Society of Mechanical Engineers; 2011, p. 1229–35. [80] ASTM F679-16. Standard Specification for Poly(Vinyl Chloride) (PVC) Large-
[48] Sebaey TA, Mahdi E. Crashworthiness of pre-impacted glass/epoxy composite Diameter Plastic Gravity Sewer Pipe and Fittings. Am Soc Test Mater 2016. https://

22
S. Eyvazinejad Firouzsalari, et al. Composite Structures 244 (2020) 112307

doi.org/10.1520/F0679-16. [86] Khalid AA. Behaviour of hybrid jute-glass/epoxy composite tubes subjected to lat-
[81] AWWA MANUAL MI 1. Steel pipe: a guide for design and installation, (M11). 4th eral loading. IOP Conf Ser Mater Sci Eng 2015;100:12068. https://doi.org/10.
edition. American Water Works Association; 2011. 1088/1757-899X/100/1/012068.
[82] Baroutaji A, Olabi A-G. Lateral collapse of short-length sandwich tubes compressed [87] Sisodia R, Waghmare N, Dixit A. An experimental and numerical crashworthiness
by different indenters and exposed to external constraints: Laterales Versagen von study of composite tubes. ASME early career Tech. Conf., Atlanta, Georegia, USA: n.
kurzen Verbundrohren unter Druckbelastung mittels verschiedener Eindringkörper d., p. 49–56.
und äußerer Auflagerbedingungen. Materwiss Werksttech; 2014. p. 45. [88] Abosbaia AS, Mahdi E, Hamouda AMS, Sahari BB, Mokhtar AS. Energy absorption
[83] Ismail AE, Sahrom MF. Lateral crushing energy absorption of cylindrical kenaf fiber capability of laterally loaded segmented composite tubes. Compos Struct
reinforced composites. Int J Appl Eng Res 2015;10:19277–88. 2005;70:356–73. https://doi.org/10.1016/j.compstruct.2004.08.039.
[84] Fahimi A, Evans TS, Farrow J, Jesson DA, Mulheron MJ, Smith PA. On the residual [89] Shen J, Lu G, Ruan D, Chiang Seah C. Lateral plastic collapse of sandwich tubes with
strength of aging cast iron trunk mains: physically-based models for asset failure. metal foam core. Int J Mech Sci 2015;91:99–109. https://doi.org/10.1016/j.
Mater Sci Eng A 2016;663:204–12. ijmecsci.2013.11.016.
[85] Frieze PA, Cho SR. Impact damage and assessment of offshore tubulars. Offshore [90] Eyvazian A, Akbarzadeh I, Shakeri M. Experimental study of corrugated tubes under
Technol. Conf., Offshore Technology Conference. 1993. lateral loading. Proc Inst Mech Eng Part L J Mater Des Appl 2012;226:109–18.

23