Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

Composites Part B 155 (2018) 414–424

Contents lists available at ScienceDirect

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

Effect of fire-retardant ceram powder on the properties of phenolic-based T


GFRP composites
Wahid Ferdousa,∗∗, Tuan Duc Ngoa,∗, Kate T.Q. Nguyena, Abdallah Ghazlana, Priyan Mendisa,
Allan Manalob
a
The University of Melbourne, Department of Infrastructure Engineering, Parkville, VIC 3010, Australia
b
University of Southern Queensland, Centre for Future Materials (CFM), School of Civil Engineering and Surveying, Toowoomba, QLD 4350, Australia

A R T I C LE I N FO A B S T R A C T

Keywords: This paper investigated the effect of ceram powder on the properties of composite laminates based on glass fibres
Laminated composites and phenolic resin. The amount of ceram in the polymer matrix was varied between 30% and 50% of the weight
Phenolic resin of resin. The density, void ratio, tensile strength, interlaminar shear strength, bond strength, bending modulus
Ceram powder and glass transition temperature were studied, and the effect of ceram on these properties was assessed. A
Properties
systemic decision-making strategy is applied to evaluate the optimal amount of ceram in the polymer matrix.
Optimal design
Results showed that while the increase of ceram decreased the strength properties of the composite laminates,
the bulk density and bending modulus increased. Moreover, the glass transition temperature increased by 32 °C
with the addition of 50% (by weight of resin) ceram powder. The strategic decision-making approach suggested
that a good balance of physical, mechanical, and thermos-mechanical properties can be achieved when ceram is
added at a fraction of 50% of the weight of resin, and this amount is considered as optimal for designing
laminated fibre composites.

1. Introduction composite skins need to be modified to meet the fire requirements of


building codes before their widespread application in modular con-
Composite sandwich panels have been the subject of active research struction.
in replacing conventional materials (such as concrete, aluminium and One approach in improving fire performance of polymer composites
steel) for prefabricated building façades due to their high strength-to- is by incorporating fire retardant filler as a constituent to the matrix
weight ratio, excellent durability, corrosion free properties, good in- [12,13]. Several fire-retardant fillers are available in the market and the
sulation, strong bonding and rapid offsite construction facilities [1–6]. most common types are Alumina Trihydrate (Al(OH)3), also known as
The sandwich systems are fabricated by attaching two thin laminated ATH, and Magnesium Hydroxide (Mg(OH)2), known as MDH [13–16].
composite skins to a thick lightweight core that recently attracted at- Instead of traditional fire retardant fillers, this study used Ceram
tention in civil construction [7,8]. When applied as a building façade, Polymerik powder because of its greater design flexibility, lower smoke
the skins are exposed to the environment including fire and elevated development, higher resistance against moisture, better thermal and
temperature, which limit the use of organic materials in construction electrical stabilities and cost effectiveness compared to existing systems
[9]. Recent incidents and consequences of fire in London Grenfell [17,18]. Phenolic resin [19] was also chosen due to its superior flame
Tower, Dubai Torch Tower, Lacrosse Building at Melbourne and others resistance and low cost [9] over traditional resin systems such as
are becoming a matter of great concern for design engineers. A report polyesters [20], vinyl ester [21] and epoxy [22]. Recent studies by
has been published that linked the rapid spread of fire to combustible Manalo et al. [23] suggested that sandwich beams with glass FRP
construction materials [10]. Traditional fibre reinforced polymer (FRP) (GFRP) skins and phenolic core can retain more than 80% of their
composites have relatively poor fire performance [11]. Therefore, the mechanical properties at 80 °C. While some studies showed that the


Corresponding author. ARC Training Centre for Advanced Manufacturing of Prefabricated Housing, Department of Infrastructure Engineering, The University of
Melbourne, Parkville, VIC 3010, Australia.
∗∗
Corresponding author. Centre for Future Materials (CFM), School of Civil Engineering and Surveying, University of Southern Queensland, Toowoomba, QLD
4350, Australia.
E-mail addresses: wahid.ferdous@unimelb.edu.au (W. Ferdous), dtngo@unimelb.edu.au (T.D. Ngo), kate.nguyen@unimelb.edu.au (K.T.Q. Nguyen),
abdallah.ghazlan@unimelb.edu.au (A. Ghazlan), pamendis@unimelb.edu.au (P. Mendis), allan.manalo@usq.edu.au (A. Manalo).

https://doi.org/10.1016/j.compositesb.2018.09.032
Received 28 April 2018; Received in revised form 15 September 2018; Accepted 15 September 2018
Available online 17 September 2018
1359-8368/ © 2018 Elsevier Ltd. All rights reserved.
W. Ferdous et al. Composites Part B 155 (2018) 414–424

mechanical properties of composites may degrade, the addition of fire reinforced composites once activated by high temperature (from 350 °C
retardant filler can greatly improve their fire resistance [12,24]. For to 800 °C or higher) under fire conditions [26]. The off-white colour
example, the bending strength of an epoxy-based polymer matrix was ceram powder has a specific gravity of 1.80 and is also effective in
reduced with the addition of ATH fillers but increased its glass transi- reducing overall weight, resisting moisture, and controlling smoke de-
tion temperature by 5 °C [25]. For composite laminates, the mechanical velopment. This material was supplied by Ceram Polymerik Pty Ltd,
properties cannot be sacrificed significantly for meeting fire require- Australia.
ments. Therefore, the present study focused on investigating the prop-
erties of phenolic-based laminated GFRP composites with the addition 2.2. Specimen preparation
of ceram powder and determining the optimal volume of ceram powder
to achieve a balance in mechanical properties. The laminated composite specimens were fabricated by three layers
Different percentages of ceram powder were added to the matrix of of fabrics with a stacking sequence of [0/-45/+45/+45/-45/90/
the laminated composites during the fabrication. The mechanical +45/-45/0], which were bonded with a phenolic polymer matrix. The
properties of these composites were then evaluated including tensile, longitudinal (0°), transverse (90°) and diagonal ( ± 45°) fibres provide
interlaminar shear, bond and bending properties. The effect of ceram a balance of strength and stiffness in all directions. Polymer matrices
powder on the glass transition temperature of laminates was also in- were produced with different mixing ratios of resin, hardener, urepac
vestigated. Finally, the optimal volume of ceram was determined using and ceram. The liquid resin, hardener and urepac were mixed together
a structured technique based on the experimental results. The outcome before adding the ceram powder. In all mixtures, the fractions of the
of this study will contribute scientific knowledge for designing lami- hardener and urepac were 20% and 10% of the weight of phenolic
nated composites for with optimal mechanical and fire resistance per- resin, respectively, while the ceram was added at 0%, 30%, 40% and
formance. 50% of the resin weight, thereby producing four different matrices
named C0, C30, C40 and C50, respectively. These percentages of filler
2. Experimental program were based on the previous study by Ferdous et al. [25] wherein they
found that epoxy-based polymer resin with 30%–50% of fillers will
2.1. Materials produce a matrix with reasonable properties. Furthermore, the addition
of ceram over 50% by weight did not produce a workable polymer
The materials employed in this investigation were fabrics, resin and matrix. The laminated composite without Ceram (C0) is considered as a
filler. reference specimen for comparing the performance of the C30, C40 and
C50 samples.
2.1.1. GFRP fabrics After hand laminating the GFRP laminates (nominal dimensions of
The glass fibre reinforced fabrics were composed of three plies of 375 mm × 375 mm × 2.5 mm), they were placed in an oven for heat
reinforcement in a stacking sequence of [0/-45/+45]. The bonded curing. In the oven, the laminates were positioned between two plain
fabrics strictly control fibre alignment, allowing for stronger, stiffer and aluminium sheets and non-stick hot plate liner to facilitate a smooth
lightweight laminates; and improve surface quality while minimising non-stick surface. Each specimen was then loaded with a 20 kg plate to
the formation of resin-rich pockets. The stacking sequence allows compress it during heating. The oven was preheated at 95 °C and each
flexibility in engineering laminated composites for specific strength specimen was subjected to three different temperatures, namely 95 °C,
requirements using different fibre orientations. 115 °C and 165 °C. The total curing time was approximately 100 min,
which consisted of the following phases: starting with an initial tem-
2.1.2. Phenolic resin perature of 95 °C for 20 min; increasing the temperature from 95 °C to
In this study, GP 486G34 Phenolic impregnating resin was used as a 115 °C over 10 min; settling at a temperature of 115 °C for 20 min; again
binder of fibreglass mat. The liquid amber resin was formulated by increasing the temperature from 115 °C to 165 °C over 20 min; and fi-
phenol and formaldehyde, which have a specific gravity of 1.24. To nally heating the sample at 165 °C for the final 30 min. This curing
enhance the polymerisation reaction within a practical time period, the regime was implemented to minimise pitting on the composites surface
resin was mixed with hardener. The liquid brown hardener (GP 4826C caused by gas release. When heated at 165 °C, the surface colour of the
Hardener) with a specific gravity of 1.41 was formulated by phosphoric, specimen C0 turned into light orange while the other samples with
arylsulphonic and sulphuric acid. UrePac+1008 with specific gravity of ceram (C30, C40 and C50) turned into light yellow as shown in Fig. 1.
1.14 was also introduced in the resin system to maintain the pH and
improve the workability of the polymer matrix. The resin and hardener 2.3. Mechanical testing
were supplied by Pacific Urethanes Pty Ltd, Australia.
The tensile, interlaminar shear, bond and bending properties of la-
2.1.3. Fire-retardant filler minated composites with different ceram volumes were evaluated fol-
The filler material used in this study is Ceram Polymerik FM-3H, lowing the relevant ASTM standards. The tests were conducted at room
which is a development grade of ceramifying additives in powder form. temperature under displacement controlled conditions at a rate of
This powder, when added to the resin, creates a ceramic shell in fibre 1–2 mm/min [27] using a 100 kN MTS testing machine. Five specimens

Fig. 1. Changes of colour with the increase of ceram in the laminated composites. (For interpretation of the references to colour in this figure legend, the reader is
referred to the Web version of this article.)

415
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Table 1 5.3%, 7.4% and 11.9% for C0, C30, C40 and C50 samples, respectively.
Specimen details and test standards. This range of voids is similar to that of other laminated fabric compo-
Type of test Nominal dimensions (L × W × T) No. Of Test Ref. sites, e.g. between 5.09% and 11.2% according to the findings of
samples standard Jeong's [33] study. The increase of ceram reduced the flowability of the
matrix, thereby increasing the void content. This finding is further
Tensile 250 mm × 25 mm × 2.5 mm 5 ASTM [27]
supported by the microscopic images as presented in Fig. 2(b) and 2(e).
D3039
Short beam 15 mm × 5 mm × 2.5 mm 5 ASTM [28]
The optical microscope captured the voids at the surface and it can be
shear D2344 seen that more and larger voids were formed with the increase of ceram
Lap shear Each leg 100 × 25 × 2.5 mm 5 ASTM [29] powder.
D5868
Bending 80 mm × 25 mm × 2.5 mm 5 ASTM [30]
3.2. Tensile properties
D790
DMA 55 mm × 10 mm × 2.5 mm 2 ASTM [31]
D4065 The failure modes of the tensile specimen in the longitudinal and
transverse directions are shown in Fig. 3(a) and (b), respectively. In the
longitudinal direction, all of the specimens failed due to tensile failure
were tested for each specimen type to account for statistical variation of the glass fibres within the gauge length with no observed slip or
and to achieve reliable results. The load and corresponding displace- failure in the anchorage zone. However, delamination was observed for
ments were directly recorded by the instruments and the failure beha- C50 specimen as shown in Fig. 3(a). This delamination can be due to
viour was observed during the testing. The specimens’ length (L), width the existence of large voids in this specimen. The delamination ob-
(W) and thickness (T) and the test standards followed are provided in served for C50 in the longitudinal direction is also reflected in the
Table 1. stress-strain behaviour shown in Fig. 3(c). As can be seen from the
figure, the stress in the laminates dropped when delamination occurred
2.4. Dynamic mechanical analysis (DMA) at a strain of around 1% and stress of 80 MPa but continued to carry
load until rupture of the longitudinal fibres. Similarly, a slight decrease
The thermomechanical behavior of the composite laminates was in stiffness was observed at the same level of strain for all longitudinal
investigated under the dynamic mechanical analysis (DMA) with a (Fig. 3(c)) and transverse (Fig. 3(d)) specimens. This decrease in stiff-
NETZCH 242 E Artemis according to the ASTM D4065 [31] testing ness was due to the formation of cracks in the matrix. The nonlinearity
standard. Two samples from each laminate type were prepared with a could also be from the progressive interfacial bond failure of ± 45° and
nominal thickness of 2.5 mm, width of 10 mm, and length of 55 mm. 90° GFRP layers as was also suggested by Manalo et al. [34] these fibres
These laminates were then loaded in a dual-cantilever setup at a fre- are not continuous throughout the length of the specimens. The non-
quency of 1 Hz from room temperature to 180 °C at a heating rate of linear stress-strain behaviour became more obvious with the increase of
3 °C/min. The storage modulus (E’) was recorded and plotted against ceram due to the increase in voids as highlighted in the previous sec-
temperature. The glass transition temperature (Tg) of the laminates was tion.
then measured based on the onset of the storage modulus curve. Fig. 3(c) and (d) indicates that the increase of ceram reduced the
tensile strength of the laminates. The ultimate tensile strength of C0,
3. Results and discussion C30, C40 and C50 specimens are 267, 247, 245 and 225 MPa in the
longitudinal direction or a reduction of strength by 7.5%, 8.2% and
3.1. Density and void content 15.7%, respectively. On the other hand, the tensile strength in the
transverse direction is 168, 142, 132 and 125 MPa for C0, C30, C40 and
The bulk density and void content of the laminated composites C50 specimens, respectively or a decrease of 15.5%, 21.4% and 25.6%
containing ceram powder was measured prior as they can have influ- for the samples with a ceram content of 30%, 40% and 50%, respec-
ence on the performance of composites. Fig. 2(a) shows the theoretical tively. This reduction of strength was due to the increased amount and
and measured density with the percentage of voids. The bulk density size of voids in the laminates. These defects tend to open in tension
was measured by the weight per unit volume and calculated theoreti- loading and continue to grow significantly with increased loading
cally using Eq. (1). The volume of void content is measured by Eq. (2). therefore giving a lower stiffness and overall strength. This behaviour
was also observed by Haj-Ali and Kilic [35] for their composite mate-
⎛w w w w wf ⎞ rials.
γt = 100/ ⎜ r + h + u + c +
γr γh γu γc γf ⎟ (1)
⎝ ⎠ 3.3. Interlaminar shear strength
v = 100(γt − γm)/ γt (2)
The interlaminar shear strength is an important characteristic to
In Eq. (1) and Eq. (2): γt and γm are the theoretical and measured evaluate the integrity of the fiber-resin interface in laminated compo-
density of the laminates; wr , wh , wu , wc and wf are the percentage sites [36]. When the transverse shear stress of a laminated composite
weights of resin, hardener, urepac, ceram and fibre in the laminates exceeds the interlaminar shear strength (ILSS), delamination failure
respectively; γr , γh , γu , γc and γf are the densities of these constituents; will occur between the fibre layers. From the experimental results, it
and v is the percentage void content. It is interesting to observe that the was observed that all specimens failed along the length at the mid-plane
theoretical density is increasing with an increase of ceram while the (Fig. 4(a)). Fig. 4(b) illustrates the load displacement behaviour of
measured bulk density is decreasing. The theoretical density obtained GFRP laminates with different ceram content. It can be seen that the
for C0, C30, C40 and C50 laminates are 1.777, 1.790, 1.792 and load displacement in ILSS are almost same for all specimens up to a load
1.795 g/cm3, respectively, while the corresponding measured densities of 50kN. After this load level, the load-displacement curve for laminates
are 1.705, 1.695, 1.659 and 1.581 g/cm3. This opposite trend between with ceram powder began to behave differently. This was due to the
theoretical and measured densities can be attributed to the formation of initiation of failure in these laminates, resulting in lower ILSS compared
voids in the polymer matrix. The formation of voids is a common to C0. The ILSS is determined from the ultimate load (P) and cross-
phenomenon that is unavoidable in composite materials, which is sectional area (A) relationship (ILSS = 0.75P/A) according to ASTM
caused by the mixing and consolidation of two or more discrete mate- D2344 [28]. Fig. 4(c) illustrates the ILSS for samples with different
rial components [32]. Fig. 2(a) shows that the void percentage is at 4%, percentages of ceram. The ILSS decreased from 14 MPa to 10.7, 9.4 and

416
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Fig. 2. Effect of ceram on density.

5.7 MPa with the increase of ceram by 30%, 40% and 50%, respec- the fracture surface for C0 (Fig. 5(a)) was rougher than other laminates
tively. The decrease of ILSS with the increase of ceram can be attributed (Fig. 5b and d) and became smoother with the increase percentage with
to the decrease of resin content and the formation of voids on which the ceram. Fig. 5(d) showed C50 had the smoothest fracture surface among
bond properties of the matrix are primarily dependant. Moreover, the the laminates, which explained the lowest ILSS for this sample.
decreasing ILSS for laminates with increasing percentage of ceram can
be explained by the failure behaviour. Fig. 5 showed the fracture sur-
face observed under a microscope. As can be noticed from the figures,

417
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Fig. 3. Effect of ceram on tensile properties.

3.4. Bond strength in Fig. 6(b) indicates that the increase of ceram in the matrix reduces
the ultimate failure load. The bond strength (Fig. 6(c)) obtained for C0,
The lap shear test was performed to further understand the effec- C30, C40 and C50 specimens are 10.1, 7.8, 7.2 and 5.1 MPa indicating a
tiveness of the ceram powder in bonding fibres. Fig. 6(a) shows the reduction of bond strength by 23%, 29% and 50% when ceram is added
bond failure under the lap shear test. In all specimens, failure was ob- at 30%, 40% and 50%, respectively. This is mainly due to the decrease
served due to adhesion. The load-displacement relationship illustrated in adhesion of the matrix with the increase of ceram. Compared to the

Fig. 4. Effect of ceram on interlaminar shear properties.

418
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Fig. 5. Failure of laminates in interlaminar shear.

short beam test, the measured bond strength from the lap shear test was 3.5. Bending properties
39%, 37%, 31% and 12% lower for C0, C30, C40 and C50 specimens,
respectively. This is because of the different loading mechanisms Failure of the specimen under three-point bending is shown in
wherein in the lap shear method, the failure plane was loaded in shear Fig. 7(a). It was observed that the specimens with ceram under bending
simultaneously with tension while the short beam method measures the load failed prematurely due to delamination. This can be explained by
almost pure shear of the laminates. the reduced ILSS shear strength of laminates as discussed in the pre-
vious section. Fig. 7(b) illustrates the load-displacement relationship for
C0, C30, C40 and C50 specimens under three-point bending while

Fig. 6. Effect of ceram on bond properties.

419
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Fig. 7. Effect of ceram on bending properties.

Fig. 7(c) illustrates a plot of the bending modulus with respect to dif- 3.6. Glass transition temperature
ferent ceram contents. The bending modulus (E = PL3/48δI) was de-
termined from the initial slope of the load-displacement curve (P/δ). It Fig. 8 shows the composite laminates after the test as well as the
can be seen from Fig. 7(b) that the bending strength of the laminates DMA plots. Fig. 8(b) shows an almost constant storage modulus (E′) was
decreased with the increase of ceram while the bending modulus observed for sample C0 from 30 °C up to around 75 °C. This indicates
(slope) increased with the addition of ceram. As mentioned previously, that the molecular chain mobility of the phenolic resin did not change,
the specimens failed due to delamination and the failure load is not since the temperature was below Tg. On the other hand, the samples
representative of the ultimate bending capacity for ceram-based lami- C30, C40 and C50 retained most of their storage modulus up to a
nates. The bending modulus on the other hand obtained for C0, C30, temperature of 90 °C. This result suggests that the addition of the ceram
C40 and C50 specimens are 18.6, 21.9, 24.2 and 25.5 GPa, respectively, powder made the mechanical properties of composite laminates stable
which indicated that the bending modulus increases by 18%, 30% and at elevated temperature. A rapid decrease in E’ was then observed at a
37% with the addition of 30%, 40% and 50% of ceram, respectively. temperature higher than 90 °C. This behavior indicates the transition of
The increase of bending modulus with the increase of ceram can be the matrix from a glassy to a leathery state. Under these conditions, the
attributed to the partial replacement of low modulus resin by high breakdown of secondary molecular bonds started leading to a decrease
modulus ceram in the polymer matrix. in mechanical properties.
At the onset of the storage modulus curves, Tg values of almost
86 °C, 97 °C, 103 °C, and 118 °C were recorded for the samples C0, C30,
C40, and C50, respectively. For composite laminates without ceram

Fig. 8. DMA test of laminated composites.

420
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Table 2 4.1. Analytic hierarchy process (AHP)


Tg of composite laminates with ceram filler measured from different plots.
Specimen Glass transition temperature, Tg (oC) The Analytic Hierarchy Process (AHP) is a strategic decision-making
method based on mathematics and psychology. This method derives the
Storage modulus Tan delta ratio scales from paired comparisons and the relative importance of
each attribute can be expressed by the AHP fundamental scale proposed
C0 86 101
C30 97 109 by Saaty and Vargas [40]. One advantage of this method is its ability to
C40 103 116 check and reduce inconsistencies in judgement, with a magnitude up to
C50 118 131 10% generally acceptable. The fundamental scale 1, 3, 5, 7 and 9 sig-
nifies the verbal judgments as ‘equal importance’, ‘moderate im-
portance’, ‘strong importance’, ‘very strong importance’ and ‘extreme
powder (sample C0), an almost similar Tg values was measured by importance’, respectively, while 2, 4, 6 and 8 represent intermediate
Manalo et al. [23] for the phenolic based composite laminates. Simi- behaviours. A relative importance matrix can be formulated for n at-
larly, the Tg values measured from the peak of the tan delta curve, tributes, where the relative importance of attribute i with respect to
which is approximately 10–15% higher than the Tg measured from attribute j is presented by aij. All the diagonal entries are equal to 1 in
storage modulus plot [37], are reported in Table 2, based on standard the n -order square matrix as a strong attribute always needs to be
practices and recommendations in ASTM D4065 [31]. The laminates compared with weight scale 1 for a weak attribute. Eq. (3) expresses a
with ceram powder consistently showed a higher Tg than that without typical relative importance matrix A = [aij]n × n which has reciprocal
ceram (C0). The increase in Tg of composite laminates with ceram properties where aji = 1/ aij and ann = 1.
powder can be due to the high temperature resistance of this material
a ⋯ a1n
which helped stabilizes the properties of the phenolic based resin. ⎡ 11 ⎤
Garrido et al. [38] suggested a maximum usable temperature (service A=⎢ ⋮ ⋱ ⋮ ⎥

⎣1/ a1n ⋯ ann ⎥
⎦ (3)
temperature) of 20 °C lower than the Tg of the materials can be specified
for civil engineering applications, which corresponds to more than The relative weights of the attributes or the priority matrix is the
80 °C for all laminates. Thus, the results from this study suggested that normalised Eigenvector of matrix A . The Eigenvector of matrix A can
the phenolic-based composite laminates can fulfill the requirements for be determined either manually or by Matlab and then be normalised by
civil infrastructure exposed to an in-service elevated temperature of at dividing the sum of all its elements that can be expressed by Eq. (4).
least 65 °C. With the addition of ceram powder, phenolic-based com-
aij
posite laminates can be used in applications subjected to an elevated aij∗ = n
temperature for up to 95 °C. ∑i = 1 aij (4)

In Eq. (4), aij is an element of the Eigenvector.


The Consistency Ratio (CR) needs to be checked to ensure consistent
4. Design of optimal ceram volume
relative weights among the attributes. The values are generally con-
sidered inconsistent in a pair-wise comparison when CR ≥ 0.1 and need
An optimal volume of ceram is the amount added to the matrix
to be revised. The CR can be expressed by Eq. (5).
which produces composite laminates with physical and mechanical
properties that can satisfy the performance requirements for a parti- CI
CR =
cular application. The results of the experimental investigations re- RI (5)
vealed that the addition of ceram powder from 30% to 50% can satisfy
In Eq. (5), CI is the Consistency Index that can be expressed by Eq.
the minimum performance requirements of building façades of at least
(6). The average values of Random Index (RI) with respect to n are
100 MPa tensile strength (obtained 225–247 MPa), 10 GPa bending
given in Table 3 [40].
modulus (obtained 21.9–25.5 GPa), 1.35 g/cm3 density (obtained
1.581–1.695 g/cm3) and 65 °C glass transition temperature (obtained λmax − n
CI =
97–118 °C) [39]. However, it is noted that the increase of ceram powder n−1 (6)
decreased some important properties whilst improving others. The
In Eq. (6), λmax is the Eigenvalue, and n is the number of compar-
addition of ceram reduced tensile, interlaminar shear and bond
isons.
strengths, and produced larger voids in the matrix. On the other hand,
The hierarchy of the analysis is divided into criteria and alter-
the bending modulus, glass transition temperature and density im-
natives. The main drivers of criteria are density, void content, tensile
proved with the increase of ceram. The lower bending stiffness of
strength, ILSS, bond strength, bending modulus, glass transition tem-
composites is a major problem compared with traditional construction
perature and cost. Each of these criteria has three alternatives (C30,
materials such as timber, concrete and steel. The addition of ceram
C40 and C50).
powder increase the glass transition temperature that help the lami-
nated composites to sustain higher temperature. The reduction of
density with the increase of ceram improves composites by lowering the 4.1.1. Relative intensities of criteria
structural weight. Therefore, higher bending modulus, higher glass The main drivers for selecting the optimal mix is the criteria because
transition temperature and lower density have been considered as po- a small change in intensity in this stage has a significant effect on the
sitive attributes. Moreover, resin is the most expensive component in final output. The design requirements depend on the priorities of the
polymer matrix composites. Therefore, the partial replacement of resin properties. Three main challenges of composites are their low fire
by the low cost ceram will reduce the overall cost of composites. This performance, high cost and low stiffness properties. Therefore, this
gives rise to the following question: what percentage of ceram content study provided high importance on glass transition temperature, cost
(out of 30%, 40% and 50%) is the most suitable whilst considering the
Table 3
positive and negative attributes compared with others? To solve this
Average values of RI.
complex problem and to determine the optimal amount of ceram, this
study applied the analytic hierarchy process. n 1 2 3 4 5 6 7 8 9 10

RI 0 0 0.52 0.89 1.11 1.25 1.35 1.40 1.45 1.49

421
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Table 4
Pairwise comparison of criteria.
Criteria Importance Intensity Criteria Importance Intensity

A B A B

Tensile Modulus B 3 ILSS Density A 4


Tensile ILSS B 1 ILSS Void A 4
Tensile Bond B 1 ILSS Tg B 5
Tensile Density A 4 ILSS Cost B 3
Tensile Void A 4 Bond Density A 4
Tensile Tg B 5 Bond Void A 4
Tensile Cost B 3 Bond Tg B 5
Modulus ILSS A 3 Bond Cost B 3
Modulus Bond A 3 Density Void B 1
Modulus Density A 6 Density Tg B 8
Modulus Void A 6 Density Cost B 6
Modulus Tg B 3 Void Tg B 8
Modulus Cost A 1 Void Cost B 6
ILSS Bond A 1 Tg Cost A 3

Table 5
Change of performance with respect to C0 and normalised relative intensities.
Properties % Improvement (+ve) or reduction (-ve) Normalised relative intensities

C30 vs C40 C30 vs C50 C40 vs C50

C30 C40 C50 C30 C40 C30 C50 C40 C50

Density 1.2 2.9 7.6 1 2.50 1 6.50 1 2.60


Void −32.5 −85.0 −197.5 2.62 1 6.08 1 2.32 1
Tensile strength −7.5 −8.2 −15.7 1.10 1 2.10 1 1.91 1
Bond strength −22.9 −28.9 −49.6 1.26 1 2.16 1 1.72 1
ILSS −23.6 −32.9 −59.3 1.39 1 2.52 1 1.80 1
Bending modulus 17.7 30.1 37.1 1 1.70 1 2.09 1 1.23
Cost 25.0 36.0 43.0 1 1.44 1 1.72 1 1.19
Tg 12.8 19.8 37.2 1.00 1.55 1.00 2.91 1.00 1.88

Fig. 9. Local priorities of the laminates.

Table 6
Global priorities of the laminates.
Samples Density Void Tensile Bond ILSS Modulus Cost Tg Total

C30 0.0026 0.0168 0.0319 0.0337 0.0360 0.0374 0.0429 0.0664 0.268
C40 0.0065 0.0064 0.0290 0.0268 0.0258 0.0633 0.0620 0.1028 0.323
C50 0.0169 0.0027 0.0152 0.0156 0.0143 0.0780 0.0739 0.1934 0.410

and bending modulus. The tensile strength, bond strength and inter- importance and intensities of the properties in criteria are provided in
laminar shear strength are given almost equal priority as they all re- Table 4.
present the strength of the composites and give moderate importance
over void content and density. Thus, the highest priority was given to 4.1.2. Relative intensities of alternatives
the glass transition temperature followed by the cost, bending modulus, The relative weightings of the alternative composite laminates (i.e.,
ILSS, bond strength, tensile strength, void and density. The relative C30, C40 and C50) are calculated based on their performances obtained

422
W. Ferdous et al. Composites Part B 155 (2018) 414–424

in this study. As previously mentioned, some of the properties were composite laminates can be used for civil infrastructure exposed to
found to be good for C30 while others were good for C50. The per- an in-service elevated temperature of up to 95 °C.
centage improvement or reduction of performance with respect to the • The amount of ceram in the polymer matrix is found to be optimal
control specimen (i.e., C0) is tabulated in Table 5. The increase of when it is added by 50% to the weight of resin. This amount of
ceram reduces the cost of the matrix and the percentage improvement is ceram can provide a good balance among the physical, mechanical
calculated based on the consideration that resin is 8 times more ex- and thermo-mechanical properties.
pensive than ceram. The intensity of the weak specimen is given 1 in
pairwise comparison for a particular property and increased pro- An in-depth investigation on the effect of fire-retardant ceram
portionally for the strong one. For example, when comparing normal- powder on the fire performance of laminated composites is currently
ised relative intensities between C30 and C40, the intensity of the being conducted and has shown promising results. The results of this
density attribute for C40 is 2.5 (i.e., 2.9/1.2) relative to 1.0 (i.e., 1.2/ work are currently being analysed and the significant findings will be
1.2) for C30. Using this concept, the normalised relative intensities of published in the near future.
the alternatives are determined and provided in Table 5.
Acknowledgements
4.1.3. Determination of priorities for laminates
The priorities of the alternatives with respect to the criteria and This project was funded through the ARC Training Centre for
those of the criteria with respect to the goal are calculated and pre- Advanced Manufacturing of Prefabricated Housing (ARC-CAMP.H) at
sented in Fig. 9. The global priority of GFRP laminates containing dif- the University of Melbourne. The technical assistance from Ms Jinghan
ferent amounts of ceram is determined by multiplying its corresponding Lu during specimen testing is highly acknowledged.
local priorities and summing them as expressed in Eq. (7).
References
Global priority = ∑ [Priority alternative × Priority criteria] (7)

Based on the comparative importance of each attribute with respect [1] Nguyen QT, Tran P, Ngo TD, Tran PA, Mendis P. Experimental and computational
investigations on fire resistance of GFRP composite for building façade. Compos B
to the goal, Table 6 presents the global priorities of each laminate and Eng 2014;62:218–29.
the optimal one out of the three alternatives are determined. It is ap- [2] Manalo A, Aravinthan T, Fam A, Benmokrane B. State-of-the-art review on FRP
parent from Table 6 that the highest global priority is obtained for C50 sandwich systems for lightweight civil infrastructure. J Compos Construct
2016;21(1):1–16.
specimen (global priority 41%) compared with C30 (26.8%) and C40
[3] Ferdous W, Manalo A, Aravinthan T. Effect of beam orientation on the static be-
(32.3%). This result suggests that the addition of 50% ceram in the haviour of phenolic core sandwich composites with different shear span-to-depth
polymer matrix can balance the attributes (i.e., density, void content, ratios. Compos Struct 2017;168:292–304.
tensile strength, ILSS, bond strength, bending modulus, cost and glass [4] Ferdous W, Almutairi AD, Huang Y, Bai Y. Short-term flexural behaviour of concrete
filled pultruded GFRP cellular and tubular sections with pin-eye connections for
transition temperature) of GFRP laminates in the best possible way. modular retaining wall construction. Compos Struct 2018;206:1–10.
[5] Ferdous W, Manalo A, Aravinthan T. Bond behaviour of composite sandwich panel
5. Conclusions and epoxy polymer matrix: Taguchi design of experiments and theoretical predic-
tions. Construct Build Mater 2017;145:76–87.
[6] Ferdous W, Bai Y, Almutairi AD, Satasivam S, Jeske J. Modular assembly of water-
This paper investigated the behaviour of GFRP laminated compo- retaining walls using GFRP hollow profiles: components and connection perfor-
sites with a fire retardant ceram powder employed as a filler in the mance. Compos Struct 2018;194:1–11.
[7] Ferdous W, Manalo A, Aravinthan T, Fam A. Flexural and shear behaviour of
polymer matrix. The effect of ceram powder on the density, void con- layered sandwich beams. Construct Build Mater 2018;173:429–42.
tent, tensile strength, interlaminar shear strength, bond strength, [8] Ferdous W, Manalo A, Van Erp G, Aravinthan T, Ghabraie K. Evaluation of an in-
bending modulus and glass transition temperature were examined. A novative composite railway sleeper for a narrow-gauge track under static load. J
Compos Construct 2018;22(2):1–13.
strategic decision-making method is applied to determine the optimal [9] Tyberg CS, Sankarapandian M, Bears K, Shih P, Loos AC, Dillard D, et al. Tough,
volume of ceram in the polymer matrix. The nature of the failure, load void-free, flame retardant phenolic matrix materials. Construct Build Mater
carrying capacity and stiffness behaviour were systematically in- 1999;13(6):343–53.
[10] Toscano N, Spooner R. Docklands apartment tower fire fuelled by material in
vestigated from which the following conclusions are drawn:
building's walls, says MFB. Australia: The Age Victoria; 2015.
[11] Nguyen QT, Ngo T, Tran P, Mendis P, Zobec M, Ayea L. Fire performance of pre-
• The addition of ceram powder in the polymer matrix resulted in a fabricated modular units using organoclay/glass fibre reinforced polymer compo-
site. Construct Build Mater 2016;129:204–15.
reduction in the bulk density of laminated composites. However,
[12] Bar M, Alagirusamy R, Das A. Flame retardant polymer composites. Fibers Polym
this led to the formation of more and bigger voids. 2015;16(4):705–17.
• The addition of ceram up to 50% by weight of the resin resulted in [13] Petersen MR, Chen A, Roll M, Jung SJ, Yossef M. Mechanical properties of fire-
retardant glass fiber-reinforced polymer materials with alumina tri-hydrate filler.
only a 15% and 25% reduction in tensile strength in the longitudinal
Compos B Eng 2015;78:109–21.
and transverse direction, respectively of the composite laminates. [14] Huber. Non-halogen fire retardant additives. China. 2014. p. 1–8.
This reduction in tensile strength was due to the delamination of the [15] Jeencham R, Suppakarn N, Jarukumjorn K. Effect of flame retardants on flame
composite layers before fibre rupture. retardant, mechanical, and thermal properties of sisal fiber/polypropylene com-

• The interlaminar shear strength and bond strength of composite


posites. Compos B Eng 2014;56:249–53.
[16] Suppakarn N, Jarukumjorn K. Mechanical properties and flammability of sisal/PP
laminates was significantly affected with the addition of ceram composites: effect of flame retardant type and content. Compos B Eng
powder. Up to 60% reduction in interlaminar shear strength and 2009;40(7):613–8.
[17] CeramPolymerik. Brochure on the Caeram Polymerik range of ceramifiable tech-
50% reduction in bond strength was observed for composite lami-
nology. 2016. p. 1–2. Australia.
nates with 50% ceram powder. Moreover, this resulted in a [18] Jeong S-H, Song J-B, Choi YH, Kim S-G, Go B-S, Park M, et al. Effect of micro-
smoother fracture surface compared to the composite laminates ceramic fillers in epoxy composites on thermal and electrical stabilities of GdBCO
coils. Compos B Eng 2016;94:190–6.
without the ceram powder.

[19] YukihiroKiuchi Iji M, Yanagisawa T, Shukichi T. Flame-retarding polylactic-acid
The addition of 50% ceram powder improved the bending modulus composite formed by dual use of aluminum hydroxide and phenol resin. Polym
by 37%. However, this led to the composite laminates failing pre- Degrad Stabil 2014;109:336–42.
maturely due to delamination. [20] Ferreira AJM, Tavares C, Ribeiro C. Flexural properties of polyester resin concretes.


J Polym Eng 2000;20(6):459–68.
The glass transition temperature of the phenolic-based composite [21] Lokuge W, Aravinthan T. Effect of fly ash on the behaviour of polymer concrete
laminates increased by 32 °C (from 86 °C to 118 °C) with the addi- with different types of resin. Mater Des 2013;51:175–81.
tion of 50% ceram powder. This suggests that the phenolic-based [22] Elalaoui O, Ghorbel E, Mignot V, Ouezdou MB. Mechanical and physical properties
of epoxy polymer concrete after exposure to temperatures up to 250 °C. Construct

423
W. Ferdous et al. Composites Part B 155 (2018) 414–424

Build Mater 2012;27(1):415–24. [32] Madsen B, Lilholt H. Physical and mechanical properties of unidirectional plant
[23] Manalo A, Surendar S, Van Erp G, Benmokrane B. Flexural behavior of an FRP fibre composites—an evaluation of the influence of porosity. Compos Sci Technol
sandwich system with glass-fiber skins and a phenolic core at elevated in-service 2003;63(9):1265–72.
temperature. Compos Struct 2016;152:96–105. [33] Jeong H. Effects of voids on the mechanical strength and ultrasonic attenuation of
[24] Liang JZ, Feng JQ, Tsui CP, Tang CY, Liu DF, Zhang SD, et al. Mechanical properties laminated composites. J Compos Mater 1997;31(3):276–92.
and flame-retardant of PP/MRP/Mg(OH)2/Al(OH)3 composites. Compos B Eng [34] Manalo A, Mutsuyoshi H, Matsui T. Testing and characterization of thick hybrid
2015;71:74–81. fibre composites laminates. Int J Mech Sci 2012;63(1):99–109.
[25] Ferdous W, Manalo A, Aravinthan T, Van Erp G. Properties of epoxy polymer [35] Haj-Ali R, Kilic H. Nonlinear behavior of pultruded FRP composites. Compos B Eng
concrete matrix: effect of resin-to-filler ratio and determination of optimal mix for 2002;33(3):173–91.
composite railway sleepers. Construct Build Mater 2016;124:287–300. [36] He Y, Makeev A. Nonlinear shear behavior and interlaminar shear strength of
[26] Ceram-Polymerik. Product data sheet - mixure of ceramifying additives (FM-3H). unidirectional polymer matrix composites: a numerical study. Int J Solid Struct
Ceram Polymerik Pty Ltd; 2013. 2014;51(6):1263–73.
[27] ASTM-D3039. Standard test method for tensile properties of polymer matrix com- [37] Shamsuddoha M, Islam MM, Aravinthan T, Manalo A, Lau K-t. Characterisation of
posite materials. USA: ASTM International; 2017. mechanical and thermal properties of epoxy grouts for composite repair of steel
[28] ASTM-D2344. Standard test method for short-beam strength of polymer matrix pipelines. Mater Des 2013;52:315–27.
composite materials and their laminates. USA: ASTM International; 2016. [38] Garrido M, Correia JR, Keller T. Effects of elevated temperature on the shear re-
[29] ASTM-D5868. Standard test method for lap shear adhesion for fiber reinforced sponse of PET and PUR foams used in composite sandwich panels. Construct Build
plastic (FRP) bonding. USA: ASTM International; 2014. Mater 2015;76:150–7.
[30] ASTM-D790. Standard test methods for flexural properties of unreinforced and re- [39] Abet-Laminati. Building façades technical manual. Material Exterior Grade; 2016.
inforced plastics and electrical insulating materials. USA: ASTM International; [40] Saaty TL, Vargas LG. Models, methods, concepts & applications of the analytic
2017. hierarchy process. second ed. New York, USA: Springer Science & Business Media;
[31] ASTM-D4065. Standard practice for plastics: dynamic mechanical properties: de- 2012.
termination and report of procedures. USA: ASTM International; 2012.

424

You might also like