Professional Documents
Culture Documents
Effect of Surface Treatment On Creep of Flax FRC Under Hygrotherm Aging Conditions
Effect of Surface Treatment On Creep of Flax FRC Under Hygrotherm Aging Conditions
Effect of Surface Treatment On Creep of Flax FRC Under Hygrotherm Aging Conditions
h i g h l i g h t s
a r t i c l e i n f o a b s t r a c t
Article history: The application of natural FRP (Fiber Reinforced Polymer) is being targeted in various fields due to both
Received 7 November 2018 environmental and economic benefits. FFRP (Flax Fiber Reinforced Polymer) is one of the major natural
Received in revised form 28 February 2019 FRPs. The durability and long-term performance of FFRP have been proven a key to its practical engineer-
Accepted 1 March 2019
ing application. Some experimental works have been conducted to investigate the creep properties of
FFRP. However, fewer efforts have been made to improve its creep performance thus far. In this paper,
the effect of surface treatment on the creep behavior of FFRP under hygrothermal aging conditions is
Keywords:
studied, and a fractional-order creep model is established to predict its creep behavior.
Flax fiber
Composite
Ó 2019 Elsevier Ltd. All rights reserved.
Creep
Surface treatment
Hygrothermal
Fractional calculus
https://doi.org/10.1016/j.conbuildmat.2019.03.001
0950-0618/Ó 2019 Elsevier Ltd. All rights reserved.
X. Wang et al. / Construction and Building Materials 208 (2019) 220–227 221
untreated flax fiber, the surface of alkali-treated fiber is cleaner and In the presence of moisture, silane is hydrolyzed to form silanol
rougher because of the hemicelluloses, lignin and surface impuri- (Scheme 3(a)). The silanol then reacts with the hydroxyl group of
ties (such as waxes and oils) are taken away [9,10]. The treated fiber cell to form a strong covalent bond between fiber and silane
fiber surface facilitates both mechanical interlocking and bonding (Scheme 3(b)). The other end of silanol forms a bond with matrix
reaction. Therefore, the performance of fiber/matrix bond interface functional group [21]. This co-reactivity provides molecular conti-
is improved. The reaction of sodium hydroxide with fiber cell is nuity across the interface of the composite. Therefore, the perfor-
[15] (Scheme 1). mance of fiber/matrix interface is improved.
Besides, alkali-treatment reduces the hydrophilic hydroxyl Although some experimental works have been reported on the
groups in the fiber and improves the moisture resistance [16]. creep of FFRP, fewer analytical efforts have been made to model
Bledzki et al. [11] reported that the tensile strength of FFRP and predict its creep behavior. Creep of FFRP is influenced by the
increases by 20–35% after acetylation treatment, and the water mechanical properties of flax fiber, matrix and the interface
uptake of the specimen is significantly reduced. And they found between them. To get accurate fitting results, Marklund et al.
the surface morphology of the treated fiber is smother because [22] and Varna et al. [4] introduced Schapery’s model [23], which
the wax and cuticle are taken away. Acetyl group (CH3CO) reacts contains three stress-dependent functions to characterize the vis-
with the hydrophilic hydroxyl groups (OH) of the fiber (as shown coelastic behavior of FFRP. Wong and Shanks [24] compared Max-
in Scheme 2 [18]). The hydrophilic of the modified fiber is reduced well model, Kelvin-Voigt model and Burgers model, and Burgers
and the dimensional stability is improved [17]. model is found to produce a better creep description of modified
Georgiopoulos et al. [14] investigated the effect of silane treat- FFRP with different additives. Gemant et al. [25] found some limi-
ment on the mechanical properties of FFRP. They found that the tations of traditional integer order calculus-based viscoelastic
specimen treated with 1 wt% silane solution shows an optimum models and introduced fractional differential into the basic differ-
increment of 11% in flexural modulus, and the specimen treated ential equation to model the viscoelastic behavior of materials.
with 2 wt% silane solution shows an optimum increment of 18% Some researches [26,27] show that, compared with traditional vis-
in flexural strength. Compared with the SEM pictures of the coelastic models, the fractional calculus-based ones have proved to
untreated specimens, there are more matrix attached to the trea- be powerful tools to characterize the creep behaviors of polymers
ted fiber, which indicates the adhesion between the flax fibers and other materials with less parameters. However, the fractional
and the matrix can be improved by silane treatment. They also calculus approach to creep of FFRP is rarely reported up to now.
noted that silane treatment of flax fiber slows the process of creep In this work, the effect of surface treatment on the creep behav-
in FFRP. Firstly, the silane coupling agent creates a chemical bond ior of flax fiber reinforced composites under hygrothermal aging
between the fiber and matrix. Secondly, silane coupling agent condition is studied, and a fractional derivative creep model is
forms a hydrophobic organic silicate protective layer on the fiber adopted to describe the creep response of the FFRP.
surface, which can effectively reduce water abortion [19]. The reac-
tion between fiber cell and the silane coupling agent is shown in
Scheme 3 [20]. 2. Test methods
O O O O
O OCH3 O OCH3
OCH3 OCH3
O OCH3 O OCH3
OCH3 OCH3
Scheme 3. Reaction between silane coupling agent and fiber cell. Fig. 1. Flax fabric.
222 X. Wang et al. / Construction and Building Materials 208 (2019) 220–227
Synthetic Material Co. were used as the matrix. The FFRP specimen
was made of approximately 30% fiber in volume. The dimension of
FFRP specimen in rectangular form is 50 mm 10 mm 1.35 mm.
After curing (25 °C for 2 weeks), the specimens were first dried
in an oven at 60 °C for 24 h, and then soaked in distilled water at
60 °C in an environmental test chamber. Both the water uptake
and creep properties of the specimens were measured.
(1) Water uptake tests. Dry specimens were weighed after oven
drying. The soaked specimens in test chamber were period-
ically taken out and dried by blotting paper and weighed to
assess the water uptake.
(2) Creep tests. The creep tests scheme is shown in Table 1.
Creep tests were carried out with DMS (Dynamic Mechanical
Spectrometer) 6100 (Fig. 2), at a three-point bending mode, Specimen DMS
with a constant stress level of 20 MPa for 30 min. All exper-
imental values were obtained by an average value of three
specimens. After the creep test was finished, a fractional-
Fig. 2. Creep test of FFRP.
order creep model was adopted, and the test data were used
to determine the parameters of the model.
Table 1
Creep test scheme.
10
Untreated (2) Set n as a counter. The one-dimensional search
Untreated-Fick's model space is based on dynamic changes of xðn;i1Þ , e.g.
8
ðk;1Þ ðn;i1Þ
Alkalization xi 2 d1 x ; d2 xðn;i1Þ
, i ¼ 1; 2; c. Let xðn;0Þ ¼ xðn1Þ , for
6 Alkalization-Fick's model i ¼ 1; 2; c, the one-dimensional search starts from the initial
M t (%)
4
Silanization
point xðn;i1Þ , along the direction of lðn;iÞ , to find wi that makes
Silanization-Fick's model
2 Acetylation
Acetylation-Fick's model f ðxðn;i1Þ þ wi lðn;iÞ Þ ¼ minf ðxðn;i1Þ þ wi lðn;iÞ Þ; ð7Þ
0
0 5 10 15 20
t0.5 (hour 0.5) Then update x ðn;iÞ
¼x þ wi l ; i ¼ 1; 2; c:
ðn;i1Þ ðn;iÞ
Where
0 ðn;0Þ 1 0 1 0 1
x1 0 0
Fig. 4. Relative water uptake of FFRP. B 0 C B xð2n;0Þ C B 0 C
lðn;1Þ ¼ B C B C
B .. C lðn;2Þ ¼ B .. C lðn;cÞ ¼ B @ ... A,
C wi is
@ . A @ . A
0 0
ðn;0Þ
xc
xi 0; i ¼ 1; c 1
s:t: restricted as wi 2 ½d1 ; d2 . To ensure that the fractional derivative
0 xc 1
order parameter 0 xc 1, wc should satisfy
ðn;i1Þ
where Jðx; ti Þ is the creep compliance calculated by Eq. (5) at time t i , wc min d2 ; 1=xc 1 .
r i is the creep compliance obtained from the test. x ¼ ðx1 ; xc ÞT is (3) Let xðnÞ ¼ xðn1;cÞ , set a permissible error e > 0, if the conver-
the column vector of material parameters. c is the number of gent condition (as shown in Eq. (8)) is met, end the computation, or
parameters. The parameter represents the order of fractional calcu- n ¼ n þ 1, go to (1).
lus is set as xc . m is the number of test points, and m > c is required.
As the derivative of the Mittag-Leffler function is not obtainable.
xðnÞ xðn1Þ
The optimization problem in Eq. (6) can be solved by replacing the i
i
< e i ¼ 1; 2; c: ð8Þ
Mittag-Leffler function by an approximate differentiable function, xðn1Þ
i
or using other optimization methods without calculating deriva-
tives. The second approach is adopted. A direct search strategy The search speed of this algorithm is related to its initial value,
Powell method [29] is applied. The search process is executed by search range and termination criteria. In order to obtain better fit-
round, and each round is carried out following a group of feasible ting results and minimize the amount of computation, xð0Þ should
orthogonal directions in turn as follows. assigned appropriately. The initial values of FPT model parameters
(1) Set the initial value of xð0Þ , the linearly dependent search can be determined according to the characteristics of Mittag-Leffler
0 ð0Þ 1 0 1 function. When t ¼ 0, Ea;1 ð0Þ ¼ 1, the initial creep compliance is
x1 0 h
i
B 0 C B xð20Þ C J ð0Þ ¼ E11 \* MERGEFORMAT. When t ! þ1, lim Ea;1 st
a
¼ 0\*
B C B C
directions are defined as lð1Þ ¼ B . C lð2Þ ¼ B . C lðcÞ ¼ t!þ1
@ .. A @ .. A MERGEFORMAT, the long term creep compliance (or equilibrium
0 0 creep compliance) is lim J ðtÞ ¼ E11 þ E1 . The creep compliances at
0 1 t!þ1
0
B 0 C the beginning and the end of the test are denoted as J t0 and J tmax .
B . C. J t0 and J tmax can be approximately equal to the initial and
@ .. A
ð0Þ
xc
Table 2
Maximum relative water uptake M m and diffusion coefficient D of FFRP.
1 day
4 days
9 days
16 days
0.5
0.125 U-1
0.12 U-0
0.45 Al-1
0.115 Al-0
S-1
0.11 S-0 0.4 Ac-1
0.105 Ac-0
0.1 U-1-F
U-0-F 0.35
0.095
Al-0-F Al-1-F
0.09
0.085 S-0-F 0.3 S-1-F
0.08 Ac-0-F Ac-1-F
0.075 0.25
0 10 20 30 40 0 10 20 30 40
Time/min Time/min
(a) 0 day (b) 1 days
0.6
0.6
U-4 U-9
Creep compliance J/GPa -1
S-4-F S-9-F
0.35 0.35
Ac-9-F
Ac-4-F
0.3 0.3
0 10 20 30 40
0 10 20 30 40
Time/min
Time/min
(c) 4 days (d) 9 days
0.75
U-16
Creep compliance J/GPa -1
0.7
0.65 Al-16
0.6 S-16
0.55 Ac-16
0.5 U-16-F
0.45 Al-16-F
0.4
S-16-F
0.35
Ac-16-F
0.3
0 10 20 30 40
Time/min
(e) 16 days
Fig. 6. Creep curves of FFRP from experiment and FPT model (Note: F stands for the fitting curve of FPT model).
a
long-term creep compliances as J t0 ¼ E11 \* MERGEFORMAT, an approximation is adopted as Ea ðza Þ ’ ez . It is assumed that
J tmax ¼ E11 þ E1 \* MERGEFORMAT, then the initial values of E1 and there is one point t m ; Jtm on the creep compliance test curve, which
E can be approximately considered as: satisfies t m ¼ ms, 0 < m
1. Combining the approximation with
Eq. (5), Eq. (9), and Eq. (10) leads to the initial value of a as
ð0Þ 1
E1 ¼ ; ð9Þ J tmax J m
J t0 að0Þ ¼ logm ln : ð12Þ
J tmax Jt0
1
Eð0Þ ¼ : ð10Þ
J tmax J t0
4. Results and discussion
when b ¼ 1, 0 a 1, t=s ¼ 0:7, the value of Mittag-Leffler func-
tion is within the range of Ea 0:7a 2 ð0:474; 0:5. In this paper, 4.1. Water uptake results
an approximate is made as: Ea 0:7a ¼ 0:5. The corresponding
time when Jt0:5 ¼ 0:5ðJt0 þ J tmax Þ is denoted as t0:5 . Then the initial The relative water uptake of FFRP is assessed according to:
value of s is
Wt W0
Mt ¼ 100% ð13Þ
sð0Þ ¼ t0:5 =0:7: ð11Þ W0
when 0 < z
1, the Mittag-Leffler function is approximated by where W 0 and W t is the weight of the dry specimen and the wet
[30]: Ea ðza Þ ’ eðz =Cð1þaÞÞ . When 0 a 1, Cð1 þ aÞ 2 ð0:885; 1,
a
specimen, respectively. Generally, the moisture uptake of FFRP
X. Wang et al. / Construction and Building Materials 208 (2019) 220–227 225
Table 3
Parameters of FPT model.
22
7 1.5
2 4 6 8 Al-test
S-test 19
τ (s)
3 16 S-test
Ac-test
1 13
0 2 4 6 8 Ac-test
M t (%)
10
0 2 4 6 8
Fig. 7. Evolution of E1. Mt (%)
Fig. 9. Evolution of s.
120
10
U-test
100
8 1.00
U-test
80 6 0.95
Al-test
E * (MPa)
60 0.90 Al-test
4
2 4 6 8 0.85
40 S-test
S-test
0.80
20
Ac-test 0.75 Ac-test
0 0.70
0 2 4 6 8 0 2 4 6 8
M t (%) M t (%)
4.2. Creep test results A four-parameter FPT model is proposed to describe the creep
behavior of FFRP, and the results show that the fractional-order
After the creep test data was obtained, the algorithm in creep model agrees well with test data. Although this model is
Section 3 and software Mathematica were used to calculate the established for the FFRP in this paper, it may have potential appli-
parameters of FPT model by fitting the test results with Eq. (5). cations in creep analysis of other natural fiber reinforced compos-
The curve-fitting results in Fig. 6 show good agreement with test ite materials, and influencing factors such as creep stress,
data for both untreated and treated specimens. The parameters temperature, et al. can be introduced by parameter modification.
of the FPT model are presented in Table 3. The influence of mois-
ture content on FPT model parameters is depicted in Figs. 7–10.
The creep compliance increases with the hygrothermal aging Conflict of interest
time, as shown in Fig. 6. For example, the creep compliance of
the untreated group at the end of the test increases by 316.03%, None.
366.28%, 398.61%, 494.80%, after 1 day, 4 days, 9 days and 16 days
of aging, respectively. Compared with the untreated specimens Acknowledgement
under the same aging time, the creep compliances of the treated
specimens are significantly reduced. For example, after 16 days of This work is partially supported by the Ministry of Education,
aging, compared with the untreated group, the creep compliance Youth and Sports of the Czech Republic and the European Union
at the end of the test decreases by 31.54%, 20.78%, 25.73% after (European Structural and Investment Funds – Operational Pro-
alkalization, silanization and acetylation treatment, respectively. gramme Research, Development and Education) in the frames of
The elastic modulus of the Hook spring in series part E1 deter- the project ‘‘Modular platform for autonomous chassis of special-
mines the initial strain of FPT model: e0 ¼ r=E1 . As illustrated in ized electric vehicles for freight and equipment transportation”,
Table 3 and Fig. 7, the value of E1 decreases with hygrothermal Reg. No. CZ.02.1.01/0.0/0.0/16_025/0007293.
aging time and moisture content, resulting in the increase of e0 .
For example, e0 of the untreated group increases by 255.24%,
291.01%, 310.55%, 371.66% after 1 day, 4 days, 9 days and 16 days References
of aging, respectively. Compared with the untreated specimens
[1] M.P.M. Dicker, P.F. Duckworth, A.B. Baker, G. Francois, M.K. Hazzard, P.M.
under the same aging time, e0 of the treated group is smaller. For Weaver, Green composites: A review of material attributes and
example, after hygrothermal aging of 16 days, e0 decreases by complementary applications, Compos. A 56 (1) (2014) 280–289.
23.96%, 25.54%, 31.08% after alkalization, silanization, acetylation [2] L. Yan, N. Chouw, K. Jayaraman, Flax fibre and its composites - a review,
Compos. B 56 (1) (2014) 296–317.
treatment, respectively. [3] A. Shahzad, Hemp fiber and its composites - a review, J. Compos. Mater. 46 (8)
When t ! þ1, the final strain of FPT model is (2012) 973–986.
e1 ¼ r=E1 þ r=E . The elastic modulus of Hook spring in the paral- [4] J. Varna, E. Spārnin ß š, R. Joffe, K. Nättinen, J. Lampinen, Time-dependent
behavior of flax/starch composites, Mech. Time-Dependent Mater. 16 (1)
lel part E determines the final creep strain ec1 ¼ r=E . It can be (2012) 47–70.
observed from Table 3 and Fig. 8 that the value of E decreases with [5] K. Cheour, M. Assarar, D. Scida, R. Ayad, X.L. Gong, Effect of water ageing on the
the hygrothermal aging time and relative water uptake, leading to mechanical and damping properties of flax-fibre reinforced composite
materials, Compos. Struct. 152 (2016) 259–266.
the increase of ec1 and e1 . Compared with the untreated specimens [6] D. Scida, M. Assarar, C. Poilâne, R. Ayad, Influence of hygrothermal ageing on
under the same aging time, ec1 and e1 of the treated group are the damage mechanisms of flax-fibre reinforced epoxy composite, Compos. B
smaller. For example, after hygrothermal aging of 16 days, e1 48 (10) (2013) 51–58.
[7] A. Thuault, S. Eve, D. Blond, J. Breard, M. Gomina, Effects of the hygrothermal
decreases by 18.62%, 34.72% and 42.06%, and ec1 decreases by environment on the mechanical properties of flax fibres, J. Compos. Mater. 48
5.07%, 58.02% and 69.92%, after alkalization, silanization, acetyla- (14) (2014) 1699–1707.
tion treatment, respectively. [8] Y. Du, N. Yan, M.T. Kortschot, An experimental study of creep behavior of
lightweight natural fiber-reinforced polymer composite/honeycombcore
As shown in Table 3 and Fig. 9, hygrothermal aging does not sandwich panels, Compos. Struct. 106 (106) (2013) 160–166.
have a noticeable effect on s. As illustrated in Table 3 and [9] L. Yan, N. Chow, X. Yuan, Improving the mechanical properties of natural fibre
Fig. 10, a increases with hygrothermal aging time and moisture fabric reinforced epoxy composites by alkali treatment, J. Reinf. Plast. Compos.
31 (6) (2012) 425–437.
content. For example, after 1 day, 4 days, 9 days and 16 days of [10] A. Amiri, C. Ulven, S. Huo, Effect of chemical treatment of flax fiber and resin
aging, a of the untreated group increases by 8.81%, 12.20%, manipulation on service life of their composites using time-temperature
16.91%, 19.15%, respectively. Compared with the untreated speci- superposition, Polymers 7 (10) (2015) 1965–1978.
[11] A.K. Bledzki, A.A. Mamun, M. Lucka-Gabor, V.S. Gutowski, The effects of
mens, after 16 days of aging, a decreases by 3.79%, 7.68%, 11.86% acetylation on properties of flax fibre and its polypropylene composites,
after alkalization, silanization and acetylation treatment, respec- eXPRESS Polym. Lett. 2 (6) (2008) 413–422.
tively. According to the definition of the fractional derivative, the [12] A.K. Bledzki, M. Lucka, A.A. Mamun, J. Michalski, Biological and electrical
resistance of acetylated flax fibre reinforced polypropylene composites,
close of a to 1, the nearer the material to the ideal fluid. The trend
Bioresources 4 (1) (2008) 111–125.
of a indicates that more viscous character after the hygrothermal [13] Y. Xie, C.A.S. Hill, Z. Xiao, H. Militz, C. Mai, Silane coupling agents used for
aging. natural fiber/polymer composites: A review, Compos. A 41 (7) (2010) 806–
819.
[14] P. Georgiopoulosa, E. Kontoua, G. Georgousisa, Effect of silane treatment
loading on the flexural properties of PLA/flax unidirectional composites,
5. Conclusions Compos. Commun. 10 (2018) 6–10.
[15] L.Y. Mwaikambo, M.P. Ansell, Chemical modification of hemp, sisal, jute, and
kapok fibers by alkalization, J. Appl. Polym. Sci. 84 (12) (2002) 2222–2234.
The experimental study carried out in this paper aimed at [16] M.J. John, R.D. Anandjiwala, Recent developments in chemical modification
developing a better understanding of the creep behavior of flax and characterization of natural fiber-reinforced composites, Polym. Compos.
29 (2) (2010) 187–207.
fiber reinforced composite. The results show that the performance
[17] M.S. Sreekala, M.G. Kumaran, S. Joseph, M. Jacob, S. Thomas, Oil palm fibre
of FFRP is improved by alkalization, silanization and acetylation reinforced phenol formaldehyde composites: influence of fibre surface
treatment. The water uptake and diffusion coefficients of the trea- modifications on the mechanical performance, Appl. Compos. Mater. 7 (5)
ted specimens are lower than untreated group. The effect of the (2000) 295–329.
[18] L.Y. Mwaikambo, M.P. Ansell, The effect of chemical treatment on the
chemical treatment is further confirmed by the lower creep defor- properties of hemp, sisal, jute and kapok for composite reinforcement,
mation of the treated specimens. Macromol. Mater. Eng. 272 (1) (1999) 108–116.
X. Wang et al. / Construction and Building Materials 208 (2019) 220–227 227
[19] S. Amidi, J. Wang, Engineering FRP-to-concrete bonded joints for better [27] A. Sapora, P. Cornetti, A. Carpinteri, O. Baglieri, E. Santagata, The use of
durability, Proceedings of the American Society for Composites, University of fractional calculus to model the experimental creep-recovery behavior of
California San Diego, La Jolla, CA USA, 2014. modified bituminous binders, Mater. Struct. 49 (1) (2016) 1–11.
[20] D.K.F. Anna, A. Balan, H. Bin, G. Xian, S. Thomas, Effect of surface modification [28] Z. Xu, W. Chen, A fractional-order model on new experiments of linear
of jute fiber on the mechanical properties and durability of jute fiber- viscoelastic creep of Hami Melon, Comput. Math. Appl. 66 (5) (2013) 677–681.
reinforced epoxy composites, Polym. Compos. 2018 (2018) 1–10. [29] M.J.D. Powell, An efficient method for finding the minimum of a function of
[21] E. Boix, F. Georgi, P. Navard, Influence of alkali and Si-based treatments on the several variables without calculating derivatives, Comput. J. 7 (1964) 155–162.
physical and chemical characteristics of miscanthus stem fragments, Ind. [30] F. Mainardi, M. Raberto, R. Gorenflo, E. Scalas, Fractional calculus and
Crops Prod. 91 (2016) 6–14. continuous-time finance II: the waiting-time distribution, Physica A 287 (3)
[22] E. Marklund, J. Varna, L. WallströM, Nonlinear viscoelasticity and (2000) 468–481.
viscoplasticity of flax/polypropylene composites, J. Eng. Mater. Technol. 128 [31] E.H. Saidane, D. Scida, M. Assarar, H. Sabhi, R. Ayad, Hybridisation effect on
(4) (2006) 527–536. diffusion kinetic and tensile mechanical behaviour of epoxy based flax–glass
[23] R.A. Schapery, Nonlinear viscoelastic and viscoplastic constitutive equations composites, Compos. A Appl. Sci. Manuf. 87 (2016) 153–160.
based on thermodynamics, Mech. Time-Dependent Mater. 1 (2) (1997) 209– [32] C.-H. Shen, G.S. Springer, Moisture absorption and desorption of composite
240. materials, J. Compos. Mater. 10 (1) (1976) 2–20.
[24] S. Wong, R. Shanks, Creep behaviour of biopolymers and modified flax fibre [33] P. Georgiopoulos, A. Christopoulos, S. Koutsoumpis, E. Kontou, The effect of
composites, Compos. Interfaces 15 (2–3) (2008) 131–145. surface treatment on the performance of flax/biodegradable composites,
[25] A. Gemant, A method of analyzing experimental results obtained from elasto- Compos. B Eng. 106 (2016) 88–98.
viscous bodies, Physics 7 (8) (1936) 311–317. [34] D.K.F. Anna, A. Balan, H. Bin, G. Xian, S. Thomas, Effect of surface modification
[26] M.D. Paola, A. Pirrotta, A. Valenza, Visco-elastic behavior through fractional of jute fiber on the mechanical properties and durability of jute fiber-
calculus: An easier method for best fitting experimental results, Mech. Mater. reinforced epoxy composites, Polym. Compos. (2018) 1–10.
43 (12) (2011) 799–806.