ACI MATERIALS JOURNAL

TECHNICAL PAPER

Title no. 102-M03

Pultruded Fabric-Cement Composites by Alva Peled and Barzin Mobasher The use of reinforcement in thin cement-based elements is essential to improve the tensile and flexural performance. The reinforcements can be either short fibers or continuous reinforcements, in a fabric form. Practical use of fabric-cement composites requires an industrial, cost-effective production process. The objective of this study was to develop the pultrusion technique as an industrial, cost-effective production method of prefabricated thin-sheet fabric-reinforced cement composites. Woven fabrics made from low-modulus polyethylene and glass meshes were used to produce the pultruded cement composites. The influence of fabric type, cell opening, application of pressure during casting, and cement-based matrix modification were examined. The tensile strength and ductility of the pultruded fabric-cement components were found to be relatively high, exhibiting strain hardening behavior even for fabrics with low modulus of elasticity. The best performance was achieved from glass fabric composites with a high content of fly ash. The mechanical properties were significantly affected by the matrix formulation, rheology of the matrix, and the intensity of the pressure applied after the pultrusion process. The promising combination of fabric reinforcement in cement composite products using the pultrusion process is expected to lead to a new class of highperformance fabric-cement composite materials. Keywords: cement; composite; cracking; process; tensile strength.

INTRODUCTION The past decade has seen an increased use of prefabricated cement-bonded fiberboard around the world. Such elements are used for wall panels, exterior siding, pressure pipes, and roofing and flooring tiles. The use of reinforcement in these elements is essential to improve the tensile and flexural performance. The reinforcements can be either short fibers or continuous reinforcements in a fabric form. Recently, several researchers reported very promising results of cement-based products reinforced with fabrics.1-7 In addition to ease of manufacturing, fabrics provide benefits such as excellent anchorage and bond development.4 Peled, Bentur, and Yankelevsky3 found that the flexural strength of cement-based composite products with low-modulus polyethylene (PE) fabrics is almost two times higher than the strength of composites reinforced with straight continuous polyethylene yarns. In addition to the improved strength, these fabric-reinforced composites exhibited strain-hardening behavior even though the reinforcing yarns had low modulus of elasticity. Practical use of fabric-cement composites requires an industrial, cost-effective production process, which has not yet been developed. The pultrusion process is a natural candidate for this purpose, as it is based on a relatively simple setup using low-cost equipment while assuring uniform production. Pultrusion has been examined to produce cement composites with continuous filaments (filament winding technique) by several researchers, exhibiting significantly improved performance. Cement composites containing 5% alkali resistance (AR) unidirectional glass ACI Materials Journal/January-February 2005

fibers achieved tensile strengths of 50 MPa,8 compared with an average tensile strength of approximately 6 to 10 MPa of conventional glass fiber-reinforced cement (GFRC) composites. Pultrusion products reinforced with polyacrylonitrile (PAN)-based carbon continuous filaments achieve superior flexural strength of approximately 600 MPa with 16% content by volume9 and 800 MPa with 23% content by volume.10 The manufacturing process could significantly affect the properties of the composite even when the same matrix and fibers are used.11-14 Igarashi, Bentur, and Mindess12 found that increasing the processing time of the fresh mixture influences the fiber-matrix bond strength due to changes in the interfacial microstructure, when using the same processing method, material, and fiber. Delvasto, Naaman, and Throne11 investigated the effects of applied pressure after casting on the flexural response of cement composites. They found that the performance of the composite depended on the applied pressure. Pressed composites showed increase in flexural strength but reduction in the postcracking response. Peled and Shah13 compared the properties of cast and extruded composites having similar matrixes and fibers. They also found significant effects of the processing method. The objective of this study was to evaluate the pultrusion technique for production of prefabricated high-performance thin-sheet fabric-reinforced cement composites. Woven fabrics made from low-modulus polyethylene as well as glass meshes were used to produce the pultruded cement composites in this study. The influence of the opening of the fabric as well as the applying pressure during the pultrusion process was studied by tensile tests. Also, composites made from various mixtures containing high-range water-reducing admixture, silica fume, and fly ash as replacements for cement were examined to find the appropriate rheology for the pultrusion process. Attention was given to the crack formation during the tensile tests. A microstructural analysis was conducted and correlated with the mechanical performance of the composite. RESEARCH SIGNIFICANCE The challenge in developing an industrial production method to produce thin-sheet cement elements with fabrics is not only technological, but also scientific. It has been demonstrated in numerous circumstances that the production method can have substantial impact on the properties of the final product. In this research, many parameters that affect the overall performance of cement-based composite systems using fabrics are discussed, and the information is presented ACI Materials Journal, V. 102, No. 1, January-February 2005. MS No. 03-328 received August 14, 2003, and reviewed under Institute publication policies. Copyright © 2005, American Concrete Institute. All rights reserved, including the making of copies unless permission is obtained from the copyright proprietors. Pertinent discussion including authors’ closure, if any, will be published in the NovemberDecember 2005 ACI Materials Journal if the discussion is received by August 1, 2005.

15

ACI member Alva Peled is a senior lecturer at the Structural Engineering Department, Ben Gurion University, Israel. She received her BTech from Shenkar College, Israel in textile technology; MS from the Hebrew University, Israel, in polymers and textiles chemistry; and DSc from the Technion, Israel Institute of Technology, in civil engineering. Her research interests include cement-based composite materials, processing of composites, and microstructure of cement materials. ACI member Barzin Mobasher is a professor of civil and environmental engineering at Arizona State University, Tempe, Ariz. He is a member of ACI Committees 446, Fracture Mechanics; 544, Fiber Reinforced Concrete; and 549, Thin Reinforced Cementitious Products and Ferrocement. His research interests include fiber-reinforced concrete, toughening mechanisms, and modeling of durability.

Fig. 1—Schematic description of pultrusion process. Table 1—Ingredients of pultruded mixtures Volume fraction, % Matrix Cement

1 42

2 42

3 40

4 19

Silica fume

5

5

10

5

Fly ash

—

—

—

24

High-range waterreducing admixture

0.1

0.2

0.4

0.05

to encourage further development of this area for new classes of products applicable to the construction and infrastructure repair. EXPERIMENTAL PROGRAM Fabric types Two types of fabrics were used for this study: bonded and woven (plain weave). The fabric structures differ by the way the yarns are combined together. In bonded fabrics, a perpendicular set of yarns (warp and weft) are glued together at the junction points. In these fabrics, the reinforcing yarns are in a straight form. In woven fabrics, the warp and the fill (weft) yarns pass over and under each other, resulting in a crimped geometry of the reinforcing yarns. These differences in fabric and yarn geometry can effect the mechanical performance of the fabric-cement composites.5 Two different bonded fabrics were examined in this study made from AR and E-glass fibers with 2 and 8 yarns/cm, respectively, in both directions of the fabric. This was to study the effects of fabric spacing on composite performance. The woven fabric was made from monofilament PE with 22 yarns/cm in the reinforcing direction (warp yarns) and 5 yarns/cm in the perpendicular direction (fill yarns). The AR glass fibers were with tensile strength of 1270 to 2450 MPa, elasticity modulus of 78,000 MPa, bundle diameter of 800 microns and filament diameter of 13.5 microns. The PE fibers had a tensile strength of 260 MPa, 1760 MPa modulus of elasticity, 16

and 0.25 mm diameter. The E-glass meshes were commercial meshes used for a variety of applications with a bundle having the diameter of 400 microns. Both AR and E-glass were coated with epoxy during fabric production. Pultrusion process The pultrusion process was used to produce the fabriccement specimens. In this process as shown in Fig. 1, the fabrics were passed through a slurry infiltration chamber, and then pulled through a set of rollers to squeeze the paste between the fabric openings while removing excessive paste. The fabric-cement composite laminate sheets were then formed on a plate-shaped mandrel14 resulting in samples with width of 20 cm, length of 33 cm (continuous length) and thickness of approximately 1 cm. Each cement board was made with seven layers of fabrics, resulting in a reinforcement content of approximately 4.5, 7, and 9.5% by volume of AR glass fabric, E-glass fabric, and PE fabric, respectively. After forming the samples, pressure was applied on top of the fabric cement laminates to improve penetration of the matrix between the opening of the fabrics. A constant pressure of 15 KPa due to a 900 N load applied on the surface of the fabric-cement sheet was used for all specimens. Most of this pressure was reduced within 1 h after the pultrusion process to a level corresponding to 100 N load (1.7 KPa). This pressure was maintained up to 24 h from the pultrusion process. In the case of the AR glass fabric, loads at two levels of 450 and 100 N (7.5 and 1.7 KPa) were applied to examine the effect of the static pressure on the mechanical performance of the composite. Also, the load of 450 N was reduced within 1 h after the pultrusion process to a level corresponding to 100 N load (1.7 KPa). The intensity of the applied pressure is limited because the fresh matrix cannot support elevated pressures, and 900 N (15 KPa) was the highest pressure examined. The duration of the applied pressure was also examined for the E-glass fabrics by maintaining a 900 N (15 KPa) load for 20 h and comparing the results with those samples subjected to only 1 h of pressure application. Matrix modification Controlling the rheological properties of the cement mixture is an important factor during the pultrusion process. The mixture should be sufficiently fluid to enable the fabric to transfer through the cement slurry but dense enough so that it will remain on the fabric when it leaves the cement bath. To develop a mixture with optimal rheology for the pultrusion process various amounts of silica fume, fly ash (Class F) as well as high-range water-reducing admixture were used. The matrix mixture proportions used in this study are presented in Table 1. In all cases, the water-binder ratio by weight was 0.4. The fly ash addition was studied with the AR glass fabric only. The various silica fume contents were applied for the PE and AR glass fabrics. The effect of the high-range water-reducing admixture was examined for the glass fabrics. Table 2 summarizes the experimental program for all tested systems. Cutting and curing After 24 h from the pultrusion process, the applied loads were removed and then the fabric-cement laminates were cut to at least four specimens from each system, with a length of 250 mm, thickness of 13 mm, and width of 25 mm. The specimens were cured for 3 days in an oven at 80 °C, 100% relative humidity and then stored in room environment for ACI Materials Journal/January-February 2005

Table 2—Experimental program of different composite systems Applied load Curing Matrix after type processing, N Accelerated 28 days

Specimen name

Fabric material

ARG-P100 ARG-P450

AR glass AR glass

1 1

100 450

+ +

— —

ARG-P900 ARG-SP0.2

AR glass AR glass

1 2

900 900

+ +

+ —

ARG-SF10 ARG-FA

AR glass AR glass

3 4

900 100

+ +

— +

EG-SP0.1 EG-1H

E-glass E-glass

1 2

900 900

+ +

— —

EG-20H

E-glass

2

900*

+

—

PE-SF5 PE-SF10

Polyethylene Polyethylene

2 3

900 900

+ +

+ —

*

Applied load removed 20 h after pultrusion process. Note: ARG = AR glass fiber; EG = E-glass fiber; PE = polyethylene fiber; P = pressure; SP = high-range water-reducing admixture (superplasticizer); SF = silica fume; FA = fly ash; and H = hour.

another 3 days, until testing in tension at 7 days after the pultrusion process. TESTING AND ANALYSIS METHODS Tensile tests Mechanical performance of the pultruded laminates was studied using direct tensile tests with closed loop control on a MTS testing machine. The rate of cross head displacement was set at 0.5 mm/min. The composite laminates were examined along the pultrusion direction also referred to as the machine direction (MD). Metal plates with dimension of 25 x 30 mm and 1 mm thick were glued on the gripping edges of the specimen to minimize localized damage and allow better load transfer from the hydraulic grips. Tensile load and deformation was recorded and converted to stress and strain plots during the data analysis phase using the nominal dimensions of the specimens. The reported results reflect the average values of four tested samples per each composite category. Typical stress-strain curves representing the tensile behavior of individual composites were chosen for comparison. Image analysis of crack formation Throughout the tensile testing, formation of parallel distributed cracks was observed as presented in Fig. 2(a). The cracking pattern was digitally recorded using a highresolution digital camera and stored in a computer. Images with a resolution of 480 x 640 pixels were taken every 15 s throughout the loading history. A monochromatic light source was used to illuminate the surface of the specimens. The formed cracks at each stage of loading were traced as shown in Fig. 2(b). By using an image-processing algorithm, the spacing between the parallel cracks was quantitatively measured. The average crack spacing for each strain level was correlated with the stress-strain response. As an example, average crack spacing versus strain is presented in Fig. 2(c) for composite with AR glass fabric and fly ash addition. Composite bond Using the computed average crack spacing, the average bond between the fabric and the cement matrix was calcuACI Materials Journal/January-February 2005

Fig. 2—(a) Formation of crack during loading of specimen (with fly ash) at certain stage of loading; (b) traced cracks based on above image; and (c) crack spacing versus strain during loading calculated based on successive images. lated by ACK model.16 This model is presented in the following equation V m σ mu r τ fu =  --------------- V f  2x

(1)

where τfu = bond strength; Vm and Vf = volume fraction of matrix and fiber (reinforcing yarns), respectively; σmu = matrix tensile strength; r = yarn radius; and x = average distance between the cracks (the values of all parameters are presented in Table 3). Rheology measurements The viscosity of the fresh mixture was measured by shear rheometery to study the effects of the fluidity of the fresh mixture on the overall performance of the pultruded composite laminates. The rheology was measured using a viscometer with external control. The test was done at room temperature with a shear rate of 0.02 1/s, 10 min after mixing. This time lag represents the average time required to produce the pultruded sheet. Forty-nine g was the mass of the mixture used for each measurement. The shear stress was recorded for each mixture. The rheology properties were correlated with the mechanical performance of the composites. Composite density The density of the glass fabric pultruded specimens was measured and calculated based on ASTM C 948-81. In this method, the immersed mass, saturated surface-dry mass, and 17

Table 3—Composite bond strength calculated based on ACK model including parameters required for calculation Specimen name

Fig. 3—Comparison of tensile behavior of composites with different fabric types made by pultrusion process and conventional glass fiber-reinforced cement. oven-dry mass are measured. Based on these measurements, the apparent porosity was calculated as the differences of the saturated surface-dry mass minus oven-dried mass divided by the differences of saturated surface-dry mass minus immersed mass. This was only studied for the specimens with the various applied pressures to examine differences in porosity due to processing. While this methodology does not differentiate between the fabric and cement density, it provides a basis for comparison of various specimens containing a similar volume fraction of fabrics. Microstructure characteristics The microstructure of the different composites were characterized and correlated with their mechanical properties. For these observations, fragments of specimens obtained after tensile tests were dried at 60 °C and gold-coated. Microstructural features such as matrix penetration in between the opening of the fabrics, size and distribution of flaws, porosity, and yarn damage were evaluated using a scanning electron microscopy (SEM). RESULTS AND DISCUSSION Mechanical performance Influences of processing technology and fabric types— Different fabrics with a range of openings were used in this study resulting in various fabric modules (PE with low modulus and glass with high modulus). A preliminary evaluation demonstrated that it was possible to pultrude all fabric types to form composite laminates. A comparison of the tensile behavior of the various fabric types produced by the pultrusion method with the conventional GFRC produced by the spray process containing short glass fibers is presented in Fig. 3. A strain-hardening behavior is observed for both cases. The benefit of using the pultruded fabric laminates, however, is obvious in this figure. A significant superior tensile behavior in both strength and toughness is seen for the AR glass pultruded specimen compared with much poorer response of the conventional GFRC. The improvement in tensile strain (at peak) of the AR glass pultruded composite is greater than fourfold. The improvement in tensile strength is almost twice as that of the conventional GFRC, indicating the 18

* †

Crack * * spacing, mm Vf , % Vm, %

Matrix strength,† MPa

Bond strength, MPa

ARG-P100 ARG-P450

16.8 12.4

4.7 4.4

95.3 95.6

4.4 4.4

0.63 0.93

ARG-P900 ARG-SP0.2

10.2 8.7

4.4 4.7

95.6 95.3

4.4 4.4

1.13 1.23

ARG-FA ARG-SF10

7.7 15.0

6.4 4.5

93.6 95.5

2.8 2.5

0.63 0.43

EG-20H EG-1H

9.1 9.1

8.2 7.0

91.8 93.0

4.4 4.4

0.49 0.58

PE-SF5 PE-SF10

4.2 6.9

9.5 7.7

90.5 92.3

4.4 2.5

0.63 0.28

Vf , Vm = fabric and matrix volume fractions, respectively. Calculated from bending tests.

advantages of using fabrics with the pultrusion process for the production of structural elements. When comparing the tensile behavior of all fabric systems (Fig. 3), a strain-hardening behavior is clearly seen for all cases even though the polyethylene fabrics are made from low-modulus fibers. The greatest tensile strength is observed for the AR glass fabric at 18 MPa compared with only 11 MPa for the E-glass and the 7 MPa for the PE fabric. The increased tensile performance of the AR glass fabric composites can be attributed to the high modulus of elasticity of this glass fabric (78 GPa). The low performance of the E-glass fabric might be attributed to its smaller diameter and reduction in fabric properties during heating (curing), leading to fabric breakage at low loads and poor postcracking behavior. The ductile behavior of the PE fabric composite is clearly observed in this figure. The relatively low tensile performance of the PE fabric composites and the E-glass fabrics compared with the improved performance of the AR-glass fabric composites is also summarized in Table 4, showing the average tensile properties of these composites. Effects of pressure—After forming the composite with the pultrusion process, additional pressure was applied on top of the laminates to improve penetration of the matrix between the opening of the fabrics. To better understand the pressure effects, three different levels were examined for AR glass fabric composite at 100, 450, and 900 N (1.7, 7.5, and 15 KPa). Figure 4 clearly shows that the intensity of the pressure significantly influences the tensile response of the composite. When the processing pressure increases, the tensile strength of the composite is enhanced. Increasing the pressure from 100 to 900 N improves the tensile strength of the composite by approximately 40% (Table 4). The ductility of the low-pressure composite (100 N), however, is much greater than that of the high-pressure composite (900 N). The tensile properties of the pultruded composite were hardly affected by the duration of the applied pressure, as shown in Table 4 for the E-glass fabric composites. A slight improvement in strength is observed for the specimen with the long duration of pressure (20 h, EG-20H) compared with the short duration of pressure (1 h, EG-1H). A reduction in the strain at peak of the long duration composite, EG-20H compared with EG-1H, is also observed (εultimate in Table 4). ACI Materials Journal/January-February 2005

Fig. 4—Effect of pressure applied after pultrusion process on tensile response of composites with AR glass fabrics. Matrix modification—Controlling the rheological properties of the cement mixture is an important factor during the pultrusion process. To develop a mixture with optimal rheology, various contents of high-range water-reducing admixture as well as replacement of cement with fly ash and silica fume were examined (Table 1). In the first case, 60% by volume of cement was replaced by fly ash. The results as presented in Fig. 5 clearly indicate an improvement in the mechanical behavior of the cement composite compared to similar composites without fly ash. The improvement due to the addition of fly ash is seen either when comparing with a control composite pressed with 900 N loads (Fig. 5(a)) or with a control composite pressed with 100 N loads (Fig. 5(b)) (ARG-P900 and ARG-P100, respectively, in Table 2). Note that the composite with the fly ash was pressed with 100 N loads only since higher pressures resulted that most of the paste was flow away from the fabric. When comparing the fly ash composite with a control specimen (without fly ash), which was pressed by only 100 N loads, the improvement by the addition of fly ash is pronounced mainly in tensile strength (Fig. 5(b)), whereas in Fig. 5(a), the improvement in ductility is more significant. This improvement by the fly ash suggests that the use of fly ash as a replacement for cement can be uniquely beneficial for the pultrusion process when glass fabrics are used. A tensile strength of approximately 23 MPa at a strain capacity of approximately 5% is observed, representing a tensile strength of more than eight times, and a strain capacity of more than 400 times the plain cement-based materials. Compared with GFRC, these materials are as much as five times stronger with approximately six times more strain capacity (Fig. 3). The influence of silica fume on the tensile behavior of the composite is presented in Fig. 6. In this case, the composites with the lower content of silica fume, 5% by volume, were preformed significantly better than that with the higher content of silica fume (10% by volume). This improvement is seen for both fabric types: AR glass fibers (Fig. 6(a)) and PE fibers (Fig. 6(b)). The improvement of the composites with the low content of silica fume compared with the high content composites is above 50% for both fabric systems (Table 4, PE-SF10 compares with PE-SF5, and ARG-SF10 compares with ARG-SP0.2). It should be noted that the matrixes with the different content of silica are differ also by ACI Materials Journal/January-February 2005

Fig. 5—Effect of fly ash addition on tensile behavior of various composites pressed with: (a) 900 N; and (b) 100 N. Table 4—Average tensile properties of different pultruded specimens exposed to accelerated curing

ARG-P100

σBOP , MPa 1.94

σultimate, MPa 13.02

εultimate, % 5.86

Toughness, N⋅mm 18,874

ARG-P450 ARG-P900

2.09 2.73

18.05 18.09

2.92 2.43

14,655 10,319

ARG-SP0.2 ARG-FA

2.52 1.68

21.06 22.68

3.11 5.56

13,406 25,942

ARG-SF10 EG-20H

1.34 2.19

12.09 9.23

4.15 1.71

13,909 3206

EG-1H PE-SF5

2.16 2.39

8.33 7.56

2.09 5.62

3699 10,553

PE-SF10

1.55

5.49

2.69

4428

Specimen name

the content of high-range water-reducing admixture (Table 1, Mixtures 2 versus 3). AR glass fabric composites and E-glass fabric composites with 0.1 and 0.2% high-range water-reducing admixture by volume were also produced by the pultrusion process. The tensile results indicated that the AR glass fabric composite with the increased content of high-range water-reducing admixture performed better than similar composites with the lower content of high-range water-reducing admixture; the tensile strength of the specimen with the high-content highrange water-reducing admixture was 21.06 MPa (ARGSP0.2, Table 4) compared with 18.09 MPa of the low-content high-range water-reducing admixture (ARG-P900, Table 4). For composites with the E-glass fabrics, only specimens with 19

Fig. 7—Rheology properties of various fresh cement mixtures versus tensile strength of hardened fabric-cement composites.

Fig. 6—Effect of silica fume content on tensile behavior of composites with: (a) AR glass fabrics; and (b) polyethylene fabrics. 0.2 high-range water-reducing admixture content were tested in tensile, as specimens with the lower high-range water-reducing admixture content broke in the grips before the start of testing, suggesting weak specimens. Rheology properties Better fluidity of the fresh matrix can help with the pultrusion process and leads to efficient penetration of the cement matrix between the fabric opening, leading to improved mechanical behavior of the composite. The viscous properties of fresh matrix mixtures were characterized by shear rheometry. The shear stresses of the fresh mixtures versus the tensile strength of the hardened composites are presented in Fig. 7. A good correlation is observed between the rheology properties of the fresh mixture and the tensile strength of the hardened composite, indicating that when the shear yield strengths are reduced, the tensile strength is increased. The lower fluidity and high tensile strength were recorded for the mixture with the high fly ash content (60% by volume), resulting in shear stresses of 586 D/cm2 (Dyns/cm2) and tensile strength of 22.68 MPa. The highest shear strength of 6972 D/cm2 was measured for the mixture with 10% by volume of silica fume, and shows the lowest mechanical properties—12.09 MPa (Table 4). Image analysis of crack formation During the tensile tests, successive cracks developed along the width of the specimen, as presented in Fig. 8. The development of the cracks was recorded by taking pictures 20

using a digital frame-grabber at regular time intervals of 15 s throughout the tensile loading cycle. The calculated crack spacing based on these images was correlated with the applied strains and plotted together with the stress-strain response for several tested systems. The plots presented in Fig. 8 compare the influences of the type of fabric, PE, and AR glass (Fig. 8(a)), the pressure applied after the pultrusion process (Fig. 8(b)), and silica fume content (Fig. 8(c)). In all the compared systems (Fig. 8), the spacing between the formed cracks decreases during loading up to a certain strain level. This strain level varies depending on the type of the tested system. A further increase in the applied load beyond this point does not lead to a significant reduction in crack spacing. While the spacing between the cracks is approximately maintained constant, indicating that no new cracks are forming, a further increase in the strain beyond this level mainly causes a widening of existing cracks by fabric pullout due to debonding at the fabric-matrix interface. Based on the previous discussion, one can differentiate three regions of response. The first portion of the curve represents the behavior of the whole composite as a linear material, whereas in the second portion, the crack formation and distribution in the matrix is the primary response. In the third region, crack widening is the dominant mechanism, representing mainly the behavior of the fabric and the interface. Comparing the mechanical behavior of the composites with the PE and AR glass fabrics, the authors observe that the primary mechanism of the PE composite is crack widening by fabric pullout, whereas with the glass fabric composites crack widening is not the governing mechanism up to approximately 75% of its strength (Fig. 8(a)). The glass fabric is well bonded to the matrix, and the tensile behavior is well representative of the entire composite. Only toward the end of testing (at approximately 15 MPa) did the fabric begin to pull out from the matrix, where no new cracks developed. Note that the volume fraction of the PE fabric is 9.5% and that of the glass fabric is only 4.4% (Table 3). Crack widening is also the control mechanism of the lowpressure composites and the composite with the high-content silica fume (Fig. 8(b) and (c), respectively), suggesting relatively poor bond of these systems. Thus, in these composites, a majority of the stress-strain response represents the fabric properties. On the other hand, the high pressed composite and the composite with the low silica fume content demonstrate an enhanced interaction between the fabric and the cement matrix which is indicated in the tensile response (Fig. 8(b) ACI Materials Journal/January-February 2005

Fig. 8—Crack spacing versus strain of various composites comparing effects of: (a) fabric type; (b) pressure applied after pultrusion process for AR glass fabric; and (c) silica fume content, AR glass fabrics. and (c), respectively). This is perhaps due to a better bond between the fabric and the cement matrix. In general, when comparing the different composites, lower crack spacing for the entire test suggests higher bond strength between the fabric and the cement matrix for systems having similar fabric content. Such a difference in crack spacing is clearly observed in Fig. 8(b), which compares the effects of the pressure applied after the pultrusion process. The crack spacing throughout all the strain ranges with the increased pressure (900 N) is significantly lower than that of the composite with the low pressure (100 N), suggesting better bonding and improved mechanical behavior of the high-pressured composite. A similar trend is also observed when comparing the composites with the different silica fume content (Fig. 8(c)). The composite with the high content of silica fume (10% by volume) developed fewer cracks across the specimen length with larger crack spacing, suggesting poorer bond in this case. On the other hand, the composite with the lower content of silica fume shows smaller crack spacing, which indicates improvement in the bond, leading to better tensile behavior. Bonding In view of the previous section, the bond strengths of the different composites were calculated using the ACK model (Eq. (1)15). The volume fraction of all fabric types was calculated taking into account only the yarns along the tensile loading. Because the glass yarns were impregnated in epoxy prior to fabric preparation, their volume fraction as reported in Table 3 was calculated based on the bundle diameter, assuming no penetration of the cement matrix between the filaments of the bundle. The PE fabric was made from a monofilament yarn. Other parameters used for this calculation are presented in Table 3 and were used with the final crack spacing to calculate bond strengths. In general, Table 3 points out that the AR glass fabrics develop the highest bonding with the cement matrix compared with the E-glass fabric and the PE fabric with similar matrix. Bond strengths of 1.13, 0.58, and 0.63 MPa were calculated for AR glass fabric, E-glass fabric, and PE glass fabric, respectively. Poor bonding is seen for the composites with the high content of silica fume, 10% by volume (Table 3). This is the case for the composites with the AR glass fabric (ARG-SF10) as well as with the PE fabric (PE-SF10), having bond strength values of 0.43 and 0.28 MPa, respectively, as compared with 1.13 MPa (ARG-P900) and 0.63 MPa (PE-SF5) of similar ACI Materials Journal/January-February 2005

Fig. 9—Bond strength versus apparent porosity for composites with various pressures applied after pultrusion process. composites with only 5% by volume silica fume (Table 3). Such low bond values for a high-content silica fume system can explain the comprehensive crack widening observed in these composites as discussed previously (Fig. 8(c)), leading to low mechanical properties (Fig. 6, Table 4). Values of 0.63, 0.93, and 1.13 MPa were calculated for composite exposed to applied pressures of 100, 450, and 900 N, respectively (ARG-P100, ARG-P450, and ARG-P900, Table 3), indicating that the bond strength between the matrix and the fabric is highly affected by the pressure applied during sample preparation. Higher pressure results in denser composite leading to improvement in the bond strength, as demonstrated in Fig. 9. This figure presents the bond strengths versus apparent porosity of the AR glass composites with the various applied pressures. It should be noted that the apparent porosity is measured for the whole specimen and indirectly addresses the fiber-matrix interface porosity. The image analysis of the crack formation indicated that the dominant crack widening due to the poor bond can lead to the low tensile strength of this composite with increased ductility. The superior bond of the high-pressed composite leads to its improved mechanical performance (Table 4, Fig. 4 and 8(b)). Microstructure Figure 10 presents SEM micrographs of the different fabrics embedded in the cement matrix, indicating that the pultrusion process can successfully be used for even relatively dense fabrics. A sufficient penetration of the cement matrix is observed for all fabric types, even for the PE fabric with the finest opening (less than 1 mm), as presented in Fig. 10. 21

Fig. 12—Effect of silica fume on: (a) matrix with high-content silica fume; and (b) surface of polyethylene yarn at junction of fabric in composite with high content of silica fume.

Fig. 10—Scanning electron microscopy micrographs of various fabrics embedded in cement matrix.

Fig. 11—Effect of pressure on fabric-cement interface: (a) 100 N; and (b) 900 N. Note that in all fabric systems, the applied pressure after the pultrusion process was 100 N (1.7 KPa). Figure 11 presents the fabric-matrix interface of composites with AR glass fabric exposed to different pressures during sample preparation (100 N in Fig.11(a) and 900 N in Fig.11(b)). The observations were focused on the junction of the mesh, where orthogonal yarns are connected together to form the fabric. Note that both specimens were prepared 22

exactly the same, except the difference in the applied pressure, using the pultrusion process. Also, both specimens were exposed to the same curing procedure. A larger gap between the fabric and the matrix is observed for the composite with the lower pressure, indicating poorer bonding of this laminate composite (Fig. 11(a)). A much smaller gap between the matrix and the fabric is seen for the high-intensity pressure composite (Fig. 11(b)), indicating improved bonding with this high-pressure system. These observations correlate with the bond strengths calculated for these composites (Table 3), and the crack-widening mechanism observed with the lowpressed fabric (Fig. 8(b)). A higher bond is developed at the interface when the pressure applied during sample preparation is increased, thus improving mechanical performance (Fig. 4 and Table 4). The low mechanical performance of the composites with the high-content silica fume (ARG-SF10, PE-SF10, Table 4) can be explained based on the SEM observations presented in Fig. 12. The fresh mixture with the high content of silica fume was very stiff, showing shear stresses of 6972 D/cm2 (Fig. 7); therefore, a relatively high content of high-range water-reducing admixture was added to this mixture to enable using it with the pultrusion process (Table 1, matrix No. 4). This high content of high-range water-reducing admixture formed air pockets in the matrix as observed in Fig. 12(a). These air pockets can lead to a reduction in the properties of the matrix as seen in Table 3 for ARG-SF10 and PE-SF10 systems. When these foams are at the fabricmatrix interface, bond strength suffers (Table 3). The high viscosity of the fresh mixture combined with the low matrix performance and poor bonding can lead to the low mechanical performance of the composite obtained in the present study (Table 4, Fig. 6). Moreover, during manufacturing, the high stiffness of the fresh mixture with the high silica fume content (Fig. 7) caused some damage to yarn at the fabric junction, as observed in Fig.12(b). Such damage was not observed for the other systems with a lower content of silica fume. This damage may additionally explain the reduction in the mechanical performance of the composite (Fig. 7). CONCLUSIONS Based on the research presented, the following conclusions can be made: 1. Several fabric types were successfully used in this study with the pultrusion process. Results indicate that an effective manufacturing tool for production of cement-based composites ACI Materials Journal/January-February 2005

with relatively dense fabrics has been developed. This may open the use of fabrics for a wide range of products and structures; 2. The benefit of using pultruded fabric cement laminates was clearly demonstrated in this study. Superior tensile behavior in both strength and toughness was found with the pultruded composites as compared with the conventional GFRC (with short fibers). The improvement in strain at peak of the pultruded AR glass fabric composite was greater than fourfold. In addition to superior tensile strength, strain-hardening behavior was observed even for fabrics with low modulus of elasticity. The best performance was achieved for glass fabric composites with 60% by volume of fly ash in its matrix. This observation suggests that the use of fly ash as replacement for cement can be uniquely beneficial for the pultrusion process when glass fabrics are used; 3. The mechanical performances of the pultruded composites were found to be significantly affected by the rheology properties of the fresh matrix. Fresh mixtures having high viscosity caused a reduction in the mechanical performance of the composite. Such reduction could occur due to low penetration of the matrix in the fabric opening. Moreover, when the fresh matrix is relatively stiff, the friction effects during forming may damage the fabric, suggesting that rheological parameters are a controlling factor in the pultrusion process; and 4. The intensity of the static pressure applied after casting affects the mechanical behavior of the pultruded composites. Increasing the pressure improves the tensile strength due to improvement in bond strength. ACKNOWLEDGMENTS The authors would like to thank Nippon Electric Glass Co., Ltd. for their cooperation and for providing the AR glass fabrics used in this study. The National Science Foundation, program 0324669-03, and the BSF (United States Israel Binational Science Foundation) are acknowledged for the financial support in this research.

REFERENCES 1. Swamy, R. N., and Hussin, M. W., “Continuous Woven Polypropylene Mat Reinforced Cement Composites for Applications in Building Construction,” Textile Composites in Building Construction, P. Hamelin and G. Verchery, eds., Part 1, 1990, pp. 57-67. 2. Perez-Pena, M.; Mobasher, B.; and Alfrejd, M. A., “Influence of Pozzolans on the Tensile Behavior of Reinforced Lightweight Concrete,” Innovations in the Development and Characterization of Materials for Infrastructure, Materials Research Society, Boston, Mass., Dec. 1991.

ACI Materials Journal/January-February 2005

3. Peled, A.; Bentur, A.; and Yankelevsky, D., “Flexural Performance of Cementitious Composites Reinforced by Woven Fabrics,” Materials in Civil Engineering, ASCE, Nov. 1999, pp. 325-330. 4. Peled, A.; Bentur, A.; and Yankelevsky, D., “Effects of Woven Fabrics Geometry on the Bonding Performance of Cementitious Composites: Mechanical Performance,” Advanced Cement Based Materials Journal, V. 7, 1998, pp. 20-27. 5. Peled, A., and Bentur, A., “Geometrical Characteristics and Efficiency of Textile Fabrics for Reinforcing Composites,” Cement and Concrete Research, V. 30, 2000, pp. 781-790. 6. Kruger, M.; Ozbolt, J.; and Reinhardt, H. W., “A New 3D Discrete Bond Model to Study the Influence of Bond on Structural Performance of Thin Reinforced and Prestressed Concrete Plates,” Proceedings of the Fourth International RILEM Workshop on High Performance Fiber Reinforced Cement Composites (HPFRCC4), A. E. Naaman and H. W. Reinhardt, eds. Ann Arbor, Mich., 2003, pp. 49-63. 7. Meyer, C., and Vilkner, G., “Glass Concrete Thin Sheets Prestressed with Aramid Fiber Mesh,” Proceedings of the Fourth International RILEM Workshop on High Performance Fiber Reinforced Cement Composites (HPFRCC4), A. E. Naaman and H. W. Reinhardt, eds., Ann Arbor, Mich., 2003, pp. 325-336. 8. Mobasher, B.; Pivacek, A.; and Haupt, G. J., “Cement Based CrossPly Laminates,” Journal of Advanced Cement Based Materials, No. 6, 1997, pp. 144-152. 9. Kazuhisa, S.; Noayoshi, K.; and Yasuo, K., “Development of Carbon Fiber Reinforced Cement,” Advanced Materials: The Big Payoff National SAMPE Technical Conference, SAMPE, Covina, Calf., V. 21, 1998, pp. 789-802. 10. Nishigaki, T.; Suzuki, K.; Matuhashi, T.; and Sasaki, H., “High Strength Continuous Carbon Fiber Reinforced Cement Composite (CFRC),” Proceedings of the Third International Symposium on Brittle Matrix Composites, A. M. Brandt and I. H. Marshall, eds., Elsevier Applied Science, Warsaw, Poland, 1991, pp. 344-355. 11. Delvasto S.; Naaman A. E.; and Throne J. L., “Effect of Pressure After Casting on High Strength Fiber Reinforced Mortar,” International Journal of Cement Composites and Lightweight Concrete, V. 8, No. 3, 1986, pp. 181-190. 12. Igarashi, S.; Bentur, A.; and Mindess, S., “The Effect of Processing on the Bond and Interfaces in Steel Fiber Reinforced Cement Composites,” Cement and Concrete Composites, V. 18, 1996, pp. 313-322. 13. Peled, A. and Shah, S. P., “Processing Effects in Cementitious Composites: Extrusion and Casting,” Journal of Materials and Civil Engineering, ASCE, Mar.-Apr. 2003, pp. 192-199. 14. Pivacek, A., and Mobasher, B., “A Filament Winding Technique for Manufacturing Cement Based Cross-Ply Laminates,” Innovations Forum, Journal of Materials Engineering, ASCE, May 1997, pp 55-58. 15. Avenston A.; Cooper, G. A.; and Kelley, A., “Single and Multiple Fracture in the Properties of Fiber Composites,” Proceedings, National Physical Laboratories, IPC, Science and Technology Press, UK, 1971, pp. 15-24.

23