Materials Science and Engineering A 527 (2010) 5507–5513
Contents lists available at ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Fatigue behavior of sisal fiber reinforced cement composites Flávio de Andrade Silva a , Barzin Mobasher b,∗ , Romildo Dias Toledo Filho c a
Institute of Construction Materials, TU Dresden, 01062 Dresden, Germany Department of Civil and Environmental Engineering, Arizona State University, Tempe, AZ, United States c Department of Civil Engineering, COPPE/Universidade Federal do Rio de Janeiro, P.O. Box 68506, CEP 21941-972, Rio de Janeiro – RJ, Brazil b
a r t i c l e
i n f o
Article history: Received 5 March 2010 Received in revised form 29 April 2010 Accepted 4 May 2010
Keywords: Cement composites Sisal fibres Fatigue behavior Microcracking Stress–strain
a b s t r a c t The tensile fatigue behavior of long aligned sisal fiber reinforced cement composites was investigated. The fatigue behavior was examined in terms of the stress versus cycles and stress–strain hysteresis behavior of the composites. Composites were tested at stress levels ranging between 4 and 9.8 MPa which represent approximately 30–80% of the monotonic ultimate tensile strength. The composites did not fail in fatigue below a maximum fatigue level of 6 MPa up to 106 cycles. Monotonic tensile testing was performed for composites that survived 106 tests to determine the residual strength. Crack spacing was measured by image analysis technique. There was no observed loss in strength, but a decrease in Young’s modulus and an increase in first crack strength was observed with increasing fatigue stress. Fluorescent optical microscopy was used to investigate the micro-crack formation in composites subject to fatigue loading. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Fiber reinforced cement composites made with long aligned sisal fibers have been investigated under tension and bending tests in previous studies [1–4]. These composites are a new class of sustainable construction materials with superior tensile strength and ductility which can be used as load bearing structural members in different applications such as structural panels, impact and blast resistance, repair and retrofit, construction products, earthquake remediation, and strengthening of unreinforced masonry walls. While static strength data such as ultimate tensile strength (UTS), strain capacity, ductility, and modulus of rupture (MOR) were obtained, no information regarding the fatigue resistance of these composites is known. Fatigue type damage causes permanent, localized, and progressive structural change [5]. The cyclic strains are created by loads smaller than the UTS of the material in a static test. In practice, fatigue affects all of the known engineering materials subjected to repetitive cyclic loads. Examples of such cyclic loads include machine vibration, marine structures, wind action, and automobile traffic [6], and can result in structural failure when loads lower than design load are applied for a large number of stress cycles [7]. The exposure to repeated loading results in a steady decrease in
∗ Corresponding author at: Arizona State University, Civil, Environmental, and Sustainability Engineering, P.O. Box 875306, Tempe, AZ 85287-5306, United States. Tel.: +1 480 965 0141; fax: +1 480 965 0557. E-mail address:
[email protected] (B. Mobasher). 0921-5093/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2010.05.007
the stiffness of the structure, which may eventually lead to fatigue failure [6]. The interest in the fatigue of concrete started with the development of concrete railroad bridges which were exposed to millions of cycles during their entire life [8]. Hsu [8] summarized the ranges of interest of fatigue in a spectrum as shown in Table 1. There is however a lack in understanding of the fatigue behavior of concrete materials and that is even more pronounced for fiber reinforced concrete (FRC). While natural fibers such as cotton [9], wood pulp [10] and sisal [11] have been tested under fatigue load, cement composites reinforced with those fibers have not yet been investigated. Most of the fatigue tests in concrete or FRC have been performed under bending loads [12–17]. Naaman and Hammoud [14] found that fiber reinforced concrete mixtures containing 2% of hooked steel fibers can sustain bending fatigue stresses more than twice that of plain concrete. Pre-cracked FRC specimens presented average fatigue lives of the order of 10 cycles for loads within 10–90% of static strength, 8000 cycles for a load range of between 10% and 80%, and more than 2.7 × l06 cycles for load range of 10–70% [14]. Parant et al. [13] tested multi-scale steel fiber cement composite (MSCC) under bending fatigue and observed that below a loading ratio of 0.88 (maximum fatigue stress ranging from 35.9 to 40.8 MPa for a MOR of 61.5 MPa), specimens survived up to 2 million load cycles. Only a few publications address fatigue in uniaxial compression [18,19] and tension loads [20–22]. In this paper we investigate the stress versus number of fatigue cycles for sisal fiber reinforced composites. The composites were subjected to tensile fatigue load with maximum stresses ranging from 4 to 9.6 MPa at a frequency of 2 Hz, representing a range
5508
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
Table 1 Relationship between ranges of fatigue cycles and types of applications [8]. Classification
Low-cycle
High-cycle
Super high-cycle
Application
Structures subjected to earthquake
Airport pavements and bridge
Highway and highway bridges, highway pavements and concrete railroad ties
Mass rapid transfer structures
Sea structures
Number of cycles
0 to 103
103 to 105
105 to 107
107 to 5 × 107
5 × 107 to 5 × 108
Fig. 1. Sisal fiber morphology. The diverse geometry may result in different fiber–matrix bond adhesion.
of approximately 30–80% of the UTS. Similar to the conditions of monotonic tensile tests performed earlier [3], the fatigue tests were intended for a complete failure of the composite, but tests were stopped at 106 cycles. Composites that survived 106 cycles were tested under monotonic tension to establish their residual strength. Optical and fluorescent microscopies were used to investigate the microstructure after fatigue tests. 2. Experimental program 2.1. Materials and processing Continuous sisal fibers were obtained from an agricultural farm located in the city of Valente, state of Bahia – Brazil. Mechanical properties of bulk fibers defined in terms of Young’s modulus and tensile strength of 19 GPa and 400 MPa, respectively were reported by Silva et al. [23]. The sisal fibers are extracted from the plant leaf which is a functionally graded composite structure reinforced by three types of fibers: structural, arch, and xylem fibers [23]. The structural fibers are located in the periphery of the leaf providing resistance to tensile loads. The arch and xylem fibers are located in the middle of the leaf and present secondary reinforcement as well as a path for nutrients. These fibers present different geometries as shown in Fig. 1 which may result in different fiber–matrix bond strengths. After receiving the sisal fibers they were washed, cut to 400 mm long sections, weighed and separated into five different layers in order to result in a total volume fraction of 10%. These layers were aligned in the load direction. The sisal fibers were stitched by three cotton fibers to make a homogeneous spacing between them so as to facilitate the molding process.
Wollastonite fiber, a naturally occurring white, non-metallic mineral with an acicular morphology, obtained from Energyarc (JG class, CaSiO3 ), with an average equivalent diameter of 40 m and an aspect ratio of 15 was used as a micro-reinforcement in the composite production (Vf = 5%). The role of the wollastonite fiber was to increase the matrix first crack strength and to improve the matrix rheology. The matrix was produced using the Portland cement CPII F32, Metakaolin (MK), calcined waste crushed clay brick (CWCCB) from an industry located in Itaborai – RJ, Brazil, burned at 850 ◦ C, and river sand with maximum diameter of 1.18 mm and density of 2.67 g/cm3 . A naphthalene based superplasticizer with content of solids of 44% was used to achieve the desired rheology properties. The mortar matrix used a mix design ratio of 1:1:0.4 (cementitious material:sand:water by weight). The Portland cement was replaced by 30% of MK and 20% of CWCCB following recommendations of previous work [1] to increase the durability of the fibers within the alkaline cementitious system. The matrix was produced using a bench-mounted mechanical mixer of 20 l capacity. The cementitious materials were dry mixed during 30 s (for homogenization) with the subsequent addition of sand and then a volume fraction of 5% wollastonite. The powder material was mixed for an additional 30 s before the superplasticizer and the remainder of water were added for an additional 3 min of mixing. Laminate production was achieved by placing the mortar mix in a steel mould one layer at a time, followed by a layer of unidirectional aligned fibers (up to 5 layers). Additional consolidation was achieved by vibration, resulting in a sisal fiber volume fraction of 10% (see Fig. 2). All test specimens were prepared with a geometry of 400 mm × 50 mm × 12 mm
Fig. 2. Molding procedure for long aligned sisal fiber reinforced composites.
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
5509
(length × width × thickness) and reinforced by 10% of 400 mm sisal fibers distributed in layers. The final composite layered structure presented 5 layers of fibers (∼1.2 mm each) and 6 matrix layers (∼1.0 mm each). The vibrating table was used at a frequency of 65 Hz. After casting, composites were compressed at 3 MPa for 5 min, followed by curing in the mold for 24 h. After demolding, samples were fog-cured for 28 days in a curing chamber with 100% RH and 23 ± 1 ◦ C. 2.2. Fatigue tests Composites were tested under tensile fatigue loading at a stress ratio (R ratio = min / max ) of 0.2 and frequency of 2 Hz. Testing was conducted on a MTS 810 testing system under force control. Fig. 3 shows a typical sinusoidal waveform of force versus time resulting in a maximum stress level of 4 MPa. The experiment was conducted on samples with a 300 mm gage length at five different stress levels: 4, 4.8, 6, 7.2, and 9.6 MPa in consideration of an average static UTS of 12 MPa [3]. These stress levels represent approximately 30%, 40%, 50%, 60% and 80% of the mean static UTS, respectively. Three replicate samples (400 mm × 50 mm × 12 mm – length × width × thickness) were tested for each stress level. The method of tensile specimen preparation is presented in an earlier work [3]. The tests were stopped after 106 cycles or failure, whichever occurred first. Specimens that survived 106 cycles were tested under monotonic tensile load using the same testing system (MTS 810). Monotonic tests were controlled by the cross-head displacement at a rate of 0.1 mm/min. Crack spacing was measured during the
Fig. 3. Load control of the fatigue test performed using MTS 810 at stress level of 4 MPa (30% of UTS). A high accuracy is obtained by the system at the applied stress level.
monotonic tests following procedures developed during earlier work [3]. 2.3. Microstructural investigation Microstructural analysis was performed on vacuum impregnated specimens that had failed in fatigue. Samples that were tested up to 106 cycles of fatigue were pre-loaded to 0.4% strain and then
Fig. 4. Stress versus cycles fatigue curve for composites subjected to maximum stress levels ranging from 4 MPa to 9.8 MPa (30–80% of UTS) at constant R ratio of 0.2. Fatigue runout was taken at 106 cycles. The maximum stress was normalized by the ultimate tensile stress (UTS) of the composites in (b).
Fig. 5. Monotonic tensile behavior of composites that have survived 106 cycles: (a) stress–strain curves of composites subjected to maximum fatigue stresses of 4 MPa, 4.8 MPa and 6 MPa (30%, 40% and 50% of UTS) and (b) effect of the cycles on modulus of elasticity and first crack strength.
16.70 (2.33) 6.31 (2.34) 2.12 (0.82) – – 16.78 (2.61) 9.92 (3.45) 2.55 (0.92) – – 16.80 (2.62) 15.52 (3.00) 5.60 (1.23) 2.08 (0.87) –
16.76 (2.84) 10.2 (2.89) 3.21 (1.00) – – 16.81 (2.42) 16.30 (3.21) 8.70 (2.70) 3.11 (1.12) 2.62 (1.20)
glued with steel pieces and epoxy so the cracks remained open. These samples were then embedded in a polymer mixed with fluorescent dye for analysis using a Nikon Inverted microscope. This microscope is outfitted with two cameras; a low light level CCD camera (Quantix) and also a standard CCD camera that allows real time image capture. A fluorescent light (including both phase and DIC contrast enhancement capabilities) was used. Images were captured and processed by Inovision’s hardware/software interface on a SGI O2 R5000 computer.
Percentage of average monotonic ultimate tensile strength.
11.22 (1.43) 3.47 (0.84) 2.26 (0.73) – – 5.22 (0.32) 6.87 (0.44) 7.59 (0.66) – – 10.39 (0.84) 9.51 (1.12) 10.72 (1.40) – – 4 (30%) 4.8 (40%)a 6 (50%)a 7.2 (60%)a 9.8 (80%)a
a
Cycle 105 Cycle 104 Cycle 103 Cycle 102
Fig. 6. Effect of the cycles on the crack spacing of composites tested to monotonic tensile load after being subjected to maximum fatigue stress of 6 MPa (50% of UTS) for 106 cycles.
3. Discussion and analysis
a
Fatigue modulus (GPa)
Cycle 1 Young’s modulus (GPa) First crack strength (MPa) UTS (MPa)
Post-fatigue monotonic tensile test Max. fatigue stress (MPa)
Table 2 Summary of post-fatigue monotonic tensile and fatigue tests. Mean values and standard deviation in parentheses.
16.69 (2.28) 5.13 (1.87) 1.96 (0.78) – –
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
Cycle 106
5510
Fig. 4 shows the stress versus number of cycles behavior of the sisal reinforced cement composite tested at various maximum stresses (4–9.8 MPa). It can be seen that the composites can survive 106 cycles up to 6 MPa, representing 50% of the UTS. The stress level of 6 MPa can be considered as the threshold limit where composites may present fatigue failure at cycles close to 106 . Beyond 6 MPa all the composites failed below 103 cycles. It was observed that for high stress levels (i.e. >6 MPa) all the cracks are formed during the first few cycles. Examination of failed samples indicates that the number of cracks (12) in fatigue specimens were the same as the ones observed in monotonic tensile tests [3]. After their formation, cracks started to grow under fatigue load. The cycles at these high stress levels caused a degradation of the fiber–matrix interface which increased the rate of crack opening, ultimately leading the complete composite failure at low cycles (i.e. <103 ). Composites that survived 106 cycles were tested under monotonic tensile load and the results are presented in Fig. 5 and Table 2. When comparing the UTS values between monotonic and postfatigue tensile tests, a slight decrease was observed. Nevertheless, this decrease is within the standard deviation of the monotonic tests and no significant variation among the post-fatigue UTS for Table 3 Crack spacing versus strain equation for monotonic tension (reference) and postfatigue monotonic tension. Tests
Reference 106 cycles at 6 MPa
S1 + S0∗ exp–˛(εi –εmu ) S1
S0
˛
εmu (mm/mm)
25.2 30.02
321.8 60.76
673.5 854.7
0.00155 0.00066
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
5511
Fig. 7. (a) Monotonic tensile behavior of composites cycled at maximum stress level of 6 MPa (50% of UTS). The fatigue tests were stopped at 4.5 × 104 , 2 × 105 and 106 cycles. (b) Effect of the cycles on the modulus of elasticity and first crack strength.
different stress levels was observed. Stiffness degradation was observed when calculating the modulus for the post-fatigue tensile tests (refer to Table 2). Fig. 5(b) shows that samples subjected to a maximum fatigue stress level of 4 MPa (30% of UTS) maintained their initial modulus in the same range as those samples reported for the monotonic tensile tests. Above the 4 MPa stress level, the modulus decreased from approximately 11.5–2 GPa. It is important to note that these modulus values were calculated from cross-head displacement data and may include spurious deformations. First crack strength values in the post-fatigue tensile tests increase with fatigue stress levels (see Fig. 5(b)). This may be attributed to the formation of early stage cracks during the fatigue cycles. New cracks will then initiate at a higher strain level and strength.
The crack spacing was measured during a post-fatigue monotonic tensile test on the sample that survived 106 cycles at a stress level of 6 MPa (50% of UTS). Fig. 6 shows a significant difference between the crack spacing behavior of a monotonic sample compared to the post-fatigue samples. This difference can be addressed using a phenomenological representation of the decrease in crack spacing with applied strain as a function of three parameters S0 , S1 , ˛, and εmu (Eq. (1)). S(εi ) = S1 + S0 e−˛(εi −εmu )
εi > εmu
(1)
where S(ei ) = crack spacing as a function of strain, εmu = average strain at the first cracking level, or where the first set of measurements were obtained, εi = independent parameter, S0 and
Fig. 8. Hysteresis stress–strain behavior of composites subjected to 106 cycles. Maximum stress levels of (a) 4 MPa (30% of UTS), (b) 4.8 MPa (40% of UTS) and (c) 6 MPa (50% of UTS).
5512
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
Fig. 9. Effect of the cycles on the fatigue modulus and maximum strain of composites subjected to maximum stresses of (a) 4.8 MPa (30% of UTS) and (b) 6 MPa (50% of UTS).
˛ = constants representing the initial length of the specimen and rate of crack formation as a function of strain, and S1 = final crack spacing length. A decrease in S0 and increase in ˛ parameter was observed for the cycled specimen (see Table 3). This can be explained due to the fact that almost all the cracks in the specimen subjected to fatigue load were formed during the first cycles. Therefore, when the specimen previously subjected to fatigue was tested under monotonic load, many cracks appeared in the initial loading period test (i.e. at low strain levels). Nevertheless, the composites presented the same final crack spacing, approximately 20 mm. In the schedule for post-fatigue tests, specimens were initially subjected to a maximum fatigue stress level of 6 MPa (50% of UTS) up to 45 000, 2 00 000 and 106 cycles. These specimens were then tested under monotonic tensile load up to complete failure. The results are shown in Fig. 7. A slight decrease in the UTS is observed for the post-fatigue specimens when compared to control samples.
This decrease is in the error range of the monotonic tested samples discussed elsewhere [3]. A decrease in the modulus from 12 to 2 GPa is observed (see Fig. 7(b)), which is associated with an increase in the first crack strength for sample tested under monotonic load compared to those tested under fatigue up to 45 000. No change in first crack strength is observed after 45 000 cycles (see Fig. 7(b)). When comparing Figs. 5 and 7 it is noted that the modulus of elasticity decreases with an increasing fatigue stress level as well as an increasing number of cycles for any particular fatigue level. First crack strength measured in specimens pre-tested under fatigue increased when increasing the fatigue level due to previously formed cracks but did not change after cycling at a particular fatigue level (see Figs. 5(b) and 7(b)). The evolution of damage was addressed from stress–strain hysteresis measurements of composites that survived 106 cycles at maximum stress levels of 4, 4.8 and 6 MPa (30%, 40% and 50% of UTS). Data are presented in Fig. 8. The Young’s modulus was com-
Fig. 10. Cracks on composites subjected to 106 cycles at maximum fatigue level of 6 MPa (50% of UTS). Fluorescent and conventional optical microscopy. Note the higher contrast in the fluorescent optical microscopy that allows the visualization of small cracks (<20 m).
F. de Andrade Silva et al. / Materials Science and Engineering A 527 (2010) 5507–5513
puted from the unloading cycle at several stages (refer to Fig. 8). The area bounded by the individual hysteresis loops can be considered as a measure of inelastic damage or energy during a given cycle. At the maximum stress level of 4 MPa, this area was not significant, did not change with cycles, and no stiffness degradation was observed. By increasing the maximum fatigue stress to 4.8 and 6 MPa a different behavior was observed. An increase in the area within the hysteresis loops as a function of cycles was observed for both levels. This behavior can be explained due to the formation of several cracks in the first cycles followed by their widening in follow up cycles due to the degradation in the fiber matrix interface. It can be seen in the hysteresis loops that after 105 cycles, the loop changes to an “s” shape due to a continuous degradation process in the fiber–matrix interface. Stiffness degradation and incremental strain increases are presented in Fig. 9. A higher degradation was observed for the composites cycled at a maximum stress of 6 MPa as shown in Fig. 9. At 106 cycles, the maximum strain was 0.8% and Young’s modulus of approximately 2 GPa. For the 4.8 MPa stress level, the maximum strain was 0.23% and Young’s modulus of approximately 5 GPa at 106 cycles. Based on these results, four stages in the fatigue cycles can be observed and classified: (i) Maximum fatigue stress (MFS) ≤ 4 MPa (30% of UTS) – characterized by no fatigue and no damage to the material; (ii) 4 MPa (30% of UTS) < MFS < 6 MPa (50% of UTS) – characterized by no fatigue with moderate damage to the material; (iii) MFS = 6 MPa (50% of UTS) – characterized by fatigue at high cycles (close to 106 ) or no fatigue with high damage to the material; (iv) MFS > 6 MPa (50% of UTS) – characterized by fatigue at low cycles (i.e. <1000 cycles). Samples that survived 106 cycles at a stress level of 6 MPa (50% of UTS) were investigated using fluorescent optical microscopy. Fig. 10 shows the capacity of the fibers to arrest and bridge the cracks formed during fatigue cycles in lateral (Fig. 10(a), (c) and (d)) and transverse (Fig. 10(b)) cross-section views. This behavior supports the high efficiency in the fiber matrix bond adhesion of the composite system even when subjected to 106 cycles at a maximum stress of 6 MPa (50% of UTS). Two ranges of crack widths were observed in the micrographs: (i) from 1 to 20 m and (ii) from 150 to 200 m at a deformation of 0.4%. Impregnation by means of fluorescent dye was effective in visualization of micro-cracks less than 20 m in width that were not otherwise observed in conventional optical microscopy. The enhancement in contrast was also achieved with the fluorescent microscopy. Only the cracks, voids, and fibers are exhibiting fluorescence while the matrix is shown in the background. 4. Conclusions Long aligned sisal fiber reinforced cement composites were tested under tensile fatigue loading and the main findings are described below:
5513
• Composites did not fatigue up to 106 cycles when subjected to maximum stress level below 6 MPa (50% of UTS). Above this stress, the composites presented fatigue below 103 cycles. • Composites that survived 106 cycles and were tested under monotonic tension did not exhibit a significant reduction in UTS but were accompanied with a decrease in Young’s modulus. The first crack strength increased when increasing the fatigue stress levels. • Hysteresis stress–strain curves showed no signs of degradation for maximum stress level of 4 MPa (30% of UTS). At the maximum stress levels of 4.8 and 6 MPa (40% and 50% of UTS) an increase in the hysteresis area and decrease in the Young’s modulus was observed. • High contrast images of the micro-cracks were obtained with the fluorescent microscopy. The sisal fibers were able to arrest and bridge the cracks even when subjected to 106 cycles at 6 MPa (50% of UTS) of maximum stress. Acknowledgements The authors acknowledge the W.M. Keck Bioimaging Laboratory at ASU for the use of the Nikon inverted microscope. The first author acknowledges financial support from CNPq (Brazilian National Science Foundation). References [1] R.D. Toledo Filho, F.A. Silva, E.M.R. Fairbairn, J.A. Melo Filho, Construct. Build. Mater. 23 (2009) 2409–2420. [2] F.A. Silva, R.D. Toledo Filho, J.A. Melo Filho, E.M.R. Fairbairn, Construct. Build. Mater. 24 (2010) 777–785. [3] F.A. Silva, B. Mobasher, R.D. Toledo Filho, Cem. Concr. Compos. 31 (2009) 721–730. [4] F.A. Silva, D. Zhu, B. Mobasher, C. Soranadom, R.D. Toledo Filho, Mater. Sci. Eng. A 527 (2010) 544–552. [5] D. Revuelta, A. Miravete, Int. Appl. Mech. Int. 38 (2002) 121–134. [6] M.K. Lee, B.I.G. Barr, Cem. Concr. Compos. 26 (2004) 299–305. [7] H. Li, M. Zhang, J. Ou, Int. J. Fatigue 29 (2007) 1292–1301. [8] T.C.C. Hsu, ACI J. 78 (1981) 292–304. [9] J.W.S. Hearle, J.T. Sparrow, Text. Res. J. 49 (1979) 242–243. [10] W.Y. Hamad, Cellulose 4 (1997) 51–56. [11] F.A. Silva, N. Chawla, R.D. Toledo Filho, Mater. Sci. Eng. A 516 (2009) 90– 95. [12] F. Herández-Olivares, G. Barluenga, B. Parga-Landa, M. Bollati, B. Witoszek, Construct. Build. Mater. 21 (2007) 1918–1927. [13] E. Parant, P. Rossi, C. Boulay, Cem. Concr. Res. 37 (2007) 264–269. [14] A.E. Naaman, H. Hammoud, Cem. Concr. Compos. 20 (1998) 353–363. [15] C.D. Johnston, R.W. Zemp, ACI Mater. J. 88 (1991) 374–383. [16] J. Zhang, H. Stang, ACI Mater. J. 95 (1) (1998) 58–67. [17] V. Ramakrishnan, B.J. Lokvik, in: H.W. Reinhardt, A.E. Naaman (Eds.), High Performance Fiber Reinforced Cement Composites: Proceedings of the International RILEM/ACI workshop, E&FN SPON, London, 1992, pp. 271– 287. [18] L. Gao, T.C.C. Hsu, ACI Mater. J. 95 (1988) 575–581. [19] A.S. Rafeez, A. Gupta, S. Krishnamoorthy, J. Mater. Civil Eng. 12 (2000) 172– 179. [20] M. Xianhong, S. Yupu, J. Wuhan Univ. Technol.-Mater. Sci. Ed. 22 (2007) 564–568. [21] J. Zhang, H. Stang, V.C. Li, J. Mater. Civil Eng. 12 (2000) 66–73. [22] M. Saito, Cem. Concr. Res. 17 (1987) 211–218. [23] F.A. Silva, N. Chawla, R.D. Toledo Filho, Comp. Sci. Technol. 68 (2008) 3438–3443.