Flexural behaviour of glass fibre-reinforced polymer and ultra-high-strength fibre-reinforced concrete composite beams subjected to elevated temperature

Abstract

Composite beams consisting of pultruded glass fibre-reinforced polymer (GFRP) I-beams and ultra-high-strength fibre-reinforced concrete (UFC) slabs have been developed for use in short-span bridges. Fibre-reinforced polymer bolts (fibre-reinforced polymer threaded rods) and epoxy adhesive were used to connect the UFC slab to the GFRP I-beam. The authors conducted material tests and large-scale static bending tests at room and elevated temperatures (less than 90°C) to investigate the flexural behaviour of GFRP-UFC composite beams subjected to elevated temperature. The test results demonstrated that the mechanical properties of the GFRP I-beams, fibre-reinforced polymer bolts and epoxy adhesive were significantly deteriorated at elevated temperatures due to the glass transition of their polymer resin matrices. As a result, the stiffness and ultimate flexural capacity of the GFRP-UFC composite beams under elevated temperatures were significantly reduced. More than 85% of the flexural capacity of the GFRP-UFC composite beams was retained up to 60°C but that was decreased to 50% at 90°C. Fibre model analysis results confirmed that the stiffness of the GFRP-UFC composite beams is not significantly affected by actual hot environments, where there is a moderate temperature gradient across the beam cross-section.

Keywords

composite beam elevated temperature fibre-reinforced polymer bolts flexural behaviour glass fibre-reinforced polymer short-span bridge ultra-high-strength fibre-reinforced concrete

Introduction

The high durability of fibre-reinforced polymer (FRP) composites makes them suitable for short-span bridges that are exposed to severe environmental conditions (Foster et al., 2000; Hayes et al., 2000). The superior features of FRP include high tensile strength, high corrosion resistance, low weight and high fatigue resistance. Benefitting from these advantages, the first FRP pedestrian bridge in Japan was constructed in 2000 (Uno and Kitayama, 2003). The studies carried out on this bridge concluded that the use of FRP reduces the life-cycle cost and also minimizes the carbon dioxide emissions as compared to ordinary prestressed concrete bridges (Nishizaki et al., 2006; Tanaka et al., 2006).

The structural design of FRP bridges is governed mostly by the deflection limitation due to the relatively low stiffness of the FRP composites. The American Association of State Highway and Transportation Officials (AASHTO) and the Japan Society of Civil Engineers (JSCE) suggest that the deflection limit for pedestrian bridges should be less than L/500, where L corresponds to the bridge span (AASHTO, 2008; JSCE, 2011). In the hybrid fibre-reinforced polymer (HFRP) I-beams (consisting of carbon and glass FRP) used for a pedestrian bridge in Hiroshima prefecture, Japan in 2011 (Figure 1), only 10% of their ultimate tensile strength was utilized at the design deflection limit (Hai et al., 2010). Moreover, Hai et al. (2010) reported that the tensile capacity of the HFRP I-beams subjected to monotonic flexural loading was not effectively utilized because of the premature delamination failure of the HFRP I-beam’s compression flange. Manalo et al. (2012) overcame this problem by strengthening the compression flange of the HFRP I-beam with a reinforced concrete slab. The use of ultra-high-strength fibre-reinforced concrete (UFC) as an alternative to the conventional reinforced concrete has advantages such as high compressive strength and low weight. Also, UFC is a high durable material (Watanabe et al., 2007). Nguyen et al. (2015) used a UFC slab instead of the reinforced concrete slab and improved the flexural performance and the durability of the HFRP-concrete composite beams. The UFC slab was connected to the HFRP I-beam using steel bolts and epoxy adhesive. The use of the HFRP-UFC composite beams may result in relatively high initial cost for pedestrian bridge construction compared to the cost of the glass fibre-reinforced polymer (GFRP)-UFC composite beams. The effectiveness of the GFRP-UFC composite beams in pedestrian bridges has been reported in the literature (Gonilha et al., 2014; Mendes et al., 2011). The authors conducted large-scale static bending tests on the GFRP-UFC composite beams with different types of bolt shear connectors and developed a high corrosion resistant composite beam for construction of short-span pedestrian bridges in severe corrosive environments (Wijayawardane et al., 2013, 2014). In this innovative GFRP-UFC composite beam, FRP threaded rods (hereafter ‘FRP bolts’) and epoxy adhesive were used to connect the UFC slab to the GFRP I-beam (Figure 2). The thickness of the epoxy adhesive layer is approximately 2 mm and it was uniformly applied on the GFRP I-beam’s compression flange.

Figure 1.

HFRP short-span pedestrian bridge in Hiroshima, Japan.

Figure 2.

Details of GFRP–UFC composite beam and details of the bending test set-up (dimensions in millimetre).

The physical and mechanical properties of the resin matrix in the FRP composites are influenced by temperature and they degrade at temperatures close to and above their glass transition temperatures (T_g) (Bisby et al., 2005; Foster and Bisby, 2008). When the temperature of the FRP materials approaches T_g, the polymer resin matrix changes from rigid to rubbery state (Bank, 2006) and that causes the degradation of the mechanical properties of the FRP materials. Generally, the T_g of the resin matrices used in GFRP reinforcing bars is between 100°C and 120°C (Karbhari and Wang, 2004). However, the glass transition temperature of some commercially available FRP systems typically varies from 60°C to 82°C (ACI, 2008; Foster and Bisby, 2008). Furthermore, Van Erp (2008) reported that the T_g of resin matrices used in the FRP materials is varied from warm temperature to moderately high temperature, 60°C and 110°C for epoxy resins; 60°C and 120°C for vinylester resins and 85°C and 125°C for polyester resins. The deterioration of strength and stiffness properties of conventional pultruded FRP profiles as a function of temperature for various resin types can be found elsewhere (Strongwell, 2002).

The temperature dependence of the FRP material properties makes it important to study the performance of the FRP composite beams subjected to elevated temperature conditions. Farhey (2005) and Keller et al. (2007) investigated the long-term performance of short-span FRP bridges and reported the acceptable serviceability and durability of those bridges. However, the maximum temperature taken into account was less than 40°C. Dai et al. (2012) reported that FRP and concrete structures can experience 50°C or higher temperatures when they are located in hot climates and industrial environments. According to the field study on a full-scale GFRP bridge conducted by Sirimanna et al. (2011), the maximum temperature experienced by the bridge deck can reach up to 60°C during summer where the ambient temperature is around 33°C. The hot temperature environments can degrade the stiffness and flexural capacity of the FRP composite beams and may result in large deformation of the FRP bridges. Thus, the investigation of the FRP composite beams subjected to elevated temperatures is significant. To the authors’ knowledge, there have been very limited studies pertaining to the influence of elevated temperatures on the flexural behaviour of FRP composite beams. Kwon et al. (2004) conducted experiments on the fatigue behaviour of the FRP composite bridge deck prototypes subjected to cyclic loading under elevated and freezing temperature conditions (50°C and −30°C) and the results showed a significant degradation in stiffness of the FRP composite bridge decks under elevated temperatures.

To fill in the knowledge gap, this article investigates the influence of elevated temperatures (less than 90°C) on the mechanical properties of the materials used in the GFRP-UFC composite beams and on the flexural behaviour of the GFRP-UFC beams. The aim of the investigation is to determine whether or not the degraded material properties (resulting from the elevated temperature exposure) affect the flexural capacity and stiffness of the GFRP-UFC composite beam.

Material properties at room and elevated temperatures

Details of GFRP-UFC composite beams

The GFRP I-beams used for GFRP-UFC composite beams were manufactured by the pultrusion process using glass fibres embedded in a vinylester resin. Figure 3 shows the glass fibre lay-up and stacking sequence of the flange and the web of the GFRP I-beam. The overall length, width and height of the GFRP I-beams are 3500, 95 and 250 mm, respectively. The flange thickness is 14 mm and that of the web is 9 mm (Figure 4(a)).

Figure 3.

Glass fibre lay-up and stacking sequence of GFRP I-beam.

Figure 4.

Cross-sectional details of the beams (dimensions in millimetre): (a) GFRP I-beam and (b) GFRP-UFC composite beam.

The UFC segments constituting the UFC slabs were precast and were 300 mm in length, 95 mm in width and 35 mm in height. The UFC is consisted of premixed cementitious powder, water, sand, high-strength steel fibres and water-reducing agent. The premixed cementitious powder consisted of ordinary Portland cement, silica fume and ettringite. The nominal diameter and the tensile strength of the steel fibres were 0.2 mm and 2000 MPa, respectively. The mix proportions of the materials used for the UFC are given in Table 1. The FRP bolts (without bolt-head) were embedded 30 mm into the UFC segments during casting. The manufacturing process and other details of the UFC are reported by Nguyen et al. (2014).

Table 1.

Mix proportions of UFC.

Unit quantity (kg/m³)					Air content (%)
Water	Premixed cement	Sand	Water-reducing admixture	Steel fibre	Air content (%)
205	1287	898	32.2	137.4	2.0

The GFRP-UFC composite beams were fabricated by connecting the UFC segments to the GFRP I-beam top flange using 16-mm-diameter FRP bolts (containing glass fibres) and epoxy adhesive. Figure 4(b) shows the cross-sectional details of the GFRP-UFC composite beam. The centre-to-centre spacing of the FRP bolts is 150 mm and the bolts were tightened to a torque of 20 N m. In order to prevent web buckling in the GFRP-UFC composite beams, nine GFRP box sections were bonded to both sides of the web using epoxy adhesive (Figure 2). The nominal dimensions of these stiffeners are 30 × 60 × 220 mm with a wall thickness of 4 mm.

In order to understand the influence of elevated temperature on the mechanical properties of the materials of the GFRP-UFC composite beam, a number of material tests were carried out at room and elevated temperatures. Table 2 shows the experimental variables for the material tests. The material testing procedures and the test results are described in the following sections.

Table 2.

Material tests.

Test	Materials	Test temperatures
Glass transition temperature	GFRP (vinylester resin)	–
	FRP bolt	–
	Epoxy adhesive	–
Coefficient of thermal expansion	GFRP I-beam (a 100-mm-long section)	Heated from 30°C to 85°C
Coefficient of thermal expansion	UFC cuboid	Heated from 30°C to 85°C
Tensile strength	GFRP flange coupons	20°C, 50°C, 70°C, 90°C
	GFRP web coupons	20°C, 50°C, 70°C, 90°C
	FRP bolts	20°C, 50°C, 70°C, 90°C
Compressive strength	GFRP flange coupons	20°C, 60°C, 90°C
	GFRP web coupons	20°C, 60°C, 90°C
	FRP bolts	20°C, 50°C, 70°C, 90°C
	UFC cylinders	30°C, 50°C, 70°C, 90°C
Shear strength	FRP bolts	20°C, 60°C, 90°C
Shear strength	Epoxy adhesive	20°C, 60°C, 90°C

GFRP: glass fibre-reinforced polymer; FRP: fibre-reinforced polymer; UFC: ultra-high-strength fibre-reinforced concrete.

Glass transition temperature of GFRP, FRP bolt and epoxy adhesive

The glass transition temperature tests were carried out on the GFRP, FRP bolt and epoxy adhesive according to the differential scanning calorimetry (DSC) method described in the Japanese Industrial Standard K-7121 (JIS, 1987). During the test, heat flow and temperature values of each material were measured separately using a differential scanning calorimeter. The T_g of the materials was determined from the graphs between the heat flow and temperature. The test results are given in Table 3 and the T_g of all the materials was in between 50°C and 60°C.

Table 3.

Glass transition temperatures of materials.

	GFRP (vinylester resin)	FRP bolt	Epoxy adhesive
Glass transition temperature (°C)	58	53	56

GFRP: glass fibre-reinforced polymer; FRP: fibre-reinforced polymer.

Coefficient of thermal expansion of GFRP I-beam and UFC slab

The longitudinal thermal expansion rates of the GFRP I-beam flanges, GFRP web and UFC slab were measured using a 100-mm-long GFRP I-beam section and an UFC cuboid (20 × 20 × 100 mm). Specimens were gradually heated from 30°C to 85°C over a period of 20.5 h and the axial strains were measured throughout the test. The longitudinal coefficients of thermal expansion (α) of the GFRP flanges, GFRP web and UFC were constant with temperature, as shown in Table 4. The α values of the GFRP compression flange and UFC are not significantly different, and hence, the bending effect in the GFRP-UFC composite beam (resulting from the mismatch of the α values of the two materials) at elevated temperatures would be very small.

Table 4.

Coefficient of thermal expansion.

	GFRP flanges	GFRP web	UFC
Coefficient of thermal expansion (α) (×10⁻⁶/°C)	9.6	13.7	11.6

GFRP: glass fibre-reinforced polymer; UFC: ultra-high-strength fibre-reinforced concrete.

Tensile strength of GFRP I-beam and FRP bolts

The tensile behaviour of the GFRP flanges, GFRP web and the FRP bolts at elevated temperatures was examined by coupon tests at 20°C, 50°C, 70°C and 90°C. The GFRP flange and web coupon specimens (hereafter ‘coupons’) were machined from the flanges and the web of the GFRP I-beam in the longitudinal direction. The cross-sectional dimensions of the flange and the web coupons were 14 × 10 mm and 10 × 9 mm, respectively. In the FRP bolt tensile test, 16 mm diameter (φ16) threaded rods were used. Figure 5 shows the details of the tensile test specimens of the GFRP flanges/web coupons and the FRP threaded bolts. Three specimens were tested at each temperature level. Specimen temperatures were monitored using a thermocouple fixed near the longitudinal strain gauge [Figure 5(b) and (c)]. Prior to applying the tensile force, the coupons were gradually heated up to the designated temperature and kept for 90 min under the same condition. The keeping time of the specimens was determined considering the time duration when the strain gauge readings become constant, after the specimens attained the test temperature.

Figure 5.

Details of tensile test specimens: (a) tensile test specimen, (b) GFRP flanges/web and (c) FRP bolt (threaded rod).

Figure 6 shows the relationship between the temperature and tensile strength of the GFRP flanges, GFRP web and FRP bolts. The tensile strength of the flange and web coupons reduced by more than 20% at temperatures beyond 70°C. The main reason for this strength degradation may be attributed to the glass transition of the vinylester resin (Table 3). However, both the GFRP flange and the GFRP web coupons also lost significant tensile strength when the temperature increased from 20°C to 50°C due to the softening of the vinylester resin.

Figure 6.

Tensile strength of GFRP flange, GFRP web and FRP bolts.

Unlike the GFRP coupons, the FRP bolts did not exhibit a significant reduction in tensile strength with temperature (Figure 6). This can be explained by the orientation of the glass fibres. The tensile strength of the GFRP materials is higher when they have a large number of glass fibres layers in the longitudinal direction (Huh et al., 2012). However, in the GFRP materials, there is a contribution to the tensile strength from the fibres oriented in the directions other than the longitudinal direction. As the GFRP material is heated, this contribution to the tensile strength is reduced because of the weakening of the bonding between the fibres and the polymer resin matrix. Therefore, the GFRP materials having a large number of fibre layers in the longitudinal direction become less temperature dependent. In the FRP bolts, all the glass fibres are oriented at 0° to the longitudinal direction, whereas in the GFRP flanges and in the GFRP web, there are 0°, 90°, ±45° and randomly oriented glass fibres (Figure 3).

The relationships between the temperature and Young’s modulus of the GFRP flange and the GFRP web are shown in Figure 7. In both cases, the Young’s modulus decreases as the temperature increases. However, the declination becomes steeper at temperatures beyond 50°C, attributed to the glass transition of vinylester resin. Correia et al. (2013) conducted GFRP coupon tensile tests at room and high temperatures (between 20°C and 220°C) and reported that the tensile strength of the GFRP decreases before and after the glass transition temperature, but the Young’s modulus of the GFRP was not significantly affected by T_g. Similar behaviour of the tensile strength and the Young’s modulus was observed in this study. In contrast to the GFRP tensile test coupons, the Young’s modulus of the FRP bolt specimens was not noticeably affected by the elevated temperature (Figure 7). Similar relationship between the tensile strength and temperature was observed for the FRP bolts. The Young’s modulus versus temperature relationship of the UFC is discussed in the next section.

Figure 7.

Young’s modulus of GFRP flange, GFRP web, FRP bolts and UFC.

Compressive strength of GFRP I-beam, FRP bolts and UFC slab

The compression tests for the GFRP flanges and web were carried out at 20°C, 60°C and 90°C. The details of the compressive test coupons were similar to the tensile test coupons shown in Figure 5(a) and (b), except for the effective length (l_e = 45 mm) and the length of the steel pipe (l_s = 50 mm). FRP bolt compression tests were carried out at temperatures of 20°C, 50°C, 70°C and 90°C. The test configurations for FRP bolt compressive specimens were similar to the tensile test specimens (Figure 5(a) and (c)), except the l_e (40 mm) and l_s (50 mm).

All the compressive test specimens were gradually heated up to the designated temperature and kept constant for 90 min prior to loading. The failure patterns of the GFRP flange and the GFRP web compressive test coupons were almost identical. At 20°C and 60°C, both the flange and the web coupons failed due to delamination and fibre crushing, whereas the coupons at 90°C failed by fibre kinking and crushing (Figure 8(a)). The fibre kinking of the GFRP coupons at 90°C is a sign of softening of the vinylester resin due to the glass transition. On the other hand, the FRP bolts failed by fibre crushing at all temperatures (Figure 8(b)). The variations in the compressive strength of the GFRP flange, GFRP web, FRP bolts and UFC with temperature are shown in Figure 9. The decrease in the compressive strength of the GFRP flange was minimal (approximately 7%) up to 60°C but became significant from 60°C to 90°C (48%). In contrast, the decrease in the compressive strength of the web coupons was almost linear from 20°C to 90°C. The different compression behaviours of the flange and the web coupons were caused by the amount of 90° oriented (to the longitudinal direction) glass fibres in the coupons, which contribute for the transverse mechanical properties of the coupons. As shown in Figure 3, there are high number of 90° oriented glass fibres in the web compared to the flanges, and hence, the compressive strength of the web coupons was not significantly decreased as the flange coupons at temperatures beyond the T_g of the vinylester resin. The GFRP compression tests conducted by Bai and Keller (2011) at room and high temperatures (20°C to 220°C) confirmed that the compressive strength is significantly affected by elevated temperatures as well as the high temperatures. Furthermore, the failure pattern of the compressive test coupons was changed from delamination failure to kink failure, which was also observed in this study as well. Because the length of test specimens in the study of Bai and Keller (2011) was much longer, buckling occurred in all specimens failed by delamination. The glass transition temperature of the GFRP material used in that study was 110°C.

Figure 8.

Failure patterns of compressive test specimens: (a) GFRP flange/GFRP web coupons and (b) FRP bolts.

Figure 9.

Compressive strength of GFRP flange, GFRP web, FRP bolts and UFC.

Regarding the FRP bolts, a large reduction in the compressive strength (71%) was observed at the temperatures beyond 50°C. The compression test results confirm that the compressive strength of the GFRP coupons and the FRP bolts is significantly influenced by the glass transition temperatures of their resin matrices.

The compression tests on the UFC slabs were carried out at 30°C, 50°C, 70°C and 90°C using standard cylinders (50 × 100 mm) cast during the manufacturing of the UFC segments. Prior to applying load, all the UFC compressive test specimens were gradually heated up to the test temperature and kept under the same temperature for 90 min. The UFC slab exhibited a high and nearly constant level of Young’s modulus and compressive strength up to 90°C (Figures 7 and 9). This result indicates that the mechanical properties of the UFC are virtually independent of the temperature over the examined temperature range in this work.

Shear strength of FRP bolts and epoxy adhesive

The double-lap shear tests for FRP bolts and epoxy adhesive were conducted at 20°C, 60°C and 90°C. The details of a FRP bolt shear test specimen are illustrated in Figure 10. The effective shear area of an epoxy adhesive test specimen was 10,000 mm² (100 mm × 50 mm × 2 sides) and the specimen details are shown in Figure 11. Before applying load, all the specimens were gradually heated inside a heat insulated steel box until the test temperature was reached and kept under that temperature for 90 min.

Figure 10.

Details of FRP bolt shear test specimen.

Figure 11.

Cross-sectional details of epoxy adhesive shear test (dimensions in millimetre).

The shear test results show that the shear capacity of both the FRP bolts and the epoxy adhesive significantly reduces at elevated temperatures (Figure 12). The governing factor for the sudden decrease in the FRP bolt shear capacity (approximately 29%) in between 60°C and 90°C is softening of the resin matrix of the FRP bolt. The effect of elevated temperature on the shear strength of the epoxy adhesive is very significant as its shear capacity dropped by approximately 82% between 20°C and 90°C.

Figure 12.

Shear strength of FRP bolts and epoxy adhesive.

Flexural testing of composite beams at room and elevated temperatures

Test variables and methodology

Flexural tests at room and elevated temperatures were carried out on two types of beams: (1) GFRP I-beams (control specimens) and (2) GFRP-UFC composite beams. Six large-scale beams (three GFRP I-beams and three GFRP-UFC composite beams) were tested under four-point static bending. The experimental variables are given in Table 5. The flexural and the shear spans of the beams were 700 and 1250 mm, respectively. Figure 2 illustrates the bending test set-up for the GFRP-UFC composite beams and the same loading configuration was used for the GFRP I-beams. Before applying the monotonic load, all the beams except G-20 and GC-20 were gradually heated up to the test temperature and held there for 60 min. The keeping time of the beams was selected as the time taken for the strain gauge readings to become constant once the test temperature was attained. The beams G-20 and GC-20 were tested under room temperature conditions. The other beams were heated inside a heat insulated steel box fixed with 10 electric heaters. The configuration of the electric heaters in the steel box is illustrated in Figure 13. The temperature of the flexural and the shear spans of the beams was measured using thermocouples attached to the UFC slab (only in the GC-60 and GC-90 beams), GFRP top flange, GFRP web and GFRP bottom flange (Figure 14). Heating of a beam prior to a bending test is shown in Figure 15. The changes in strain at the midspan of the GFRP-UFC composite beams were measured using seven strain gauges (G1-G7 in Figure 14), whereas in the GFRP I-beams there were only five strain gauges (G3-G7 in Figure 14). Maximum midspan deflection was recorded throughout the experiment using two wire displacement transducers, one fixed to each side of the bottom flange.

Table 5.

Test variables and summary of test results of beam flexural test.

Specimen name	Test variables			Test results
Specimen name	Beam type	Test temperature (°C)	Presence of UFC segments	Ultimate load (kN)	Deflection (mm)	Midspan compressive strain (µ)	Midspan tensile strain (µ)
G-20	GFRP I-shaped	20	No	108.9	67.4	6980	7492
G-60	GFRP I-shaped	60	No	96.9	64.4	6858	6386
G-90	GFRP I-shaped	90	No	41.1	29.7	2667	3020
GC-20	GFRP-UFC composite	20	Yes	199.8	64.1	3007	11400
GC-60	GFRP-UFC composite	60	Yes	169.9	72.8	2403	Gauge damaged
GC-90	GFRP-UFC composite	90	Yes	101.9	66.2	1794	6785

UFC: ultra-high-strength fibre-reinforced concrete; GFRP: glass fibre-reinforced polymer.

Figure 13.

Configuration of electric heaters in the steel box (elevation view).

Figure 14.

Configuration of the strain gauges and thermocouples in the GFRP-UFC beams.

Figure 15.

Heating of a beam prior to bending test.

Results and discussion

The temperature distribution across the composite beams’ depth was almost identical in the flexural and shear spans. This indicates that the temperature was efficiently maintained by the heat insulated steel box. Figure 16 shows the temperature distribution of the beams in the flexural span. There was a small temperature gradient across both the GFRP I-beams and the GFRP-UFC beams. The main reason for this is the upward movement of heat flows due to convection. However, the high temperature observed in the GC-90 beam’s web (Figure 16) may be caused by the overheating of the side heaters.

Figure 16.

Temperature distribution in the beams at flexural span.

Figure 17 shows the failure patterns of the GFRP I-beams. Delamination and crushing failure was observed in the GFRP I-beams, similar to that of the GFRP compressive coupons tested at 20°C and 60°C (Figure 8(a)). In both cases, the failure location was near the loading point in the shear span (Figure 17(a) and (b)). In the G-90 beam, failure was due to kinking of the compression flange and the web at the midspan (Figure 17(c)), caused by the glass transition of the vinylester resin. Kink failure similar to this was observed in the GFRP compressive coupons tested at 90°C (Figure 8(a)). The kink failure clearly indicates that when the vinylester resin softens at elevated temperatures, it reduces its ability to support the fibres in carrying the compressive load leading to micro-buckling of the fibres and resulting in the premature failure of the GFRP I-beam top flange.

Figure 17.

Failure patterns of GFRP I-beams: (a) G-20 beam, (b) G-60 beam and (c) G-90 beam.

Figure 18 shows the failure patterns of the GFRP-UFC composite beams at 20°C, 60°C and 90°C. As shown in Figure 18(a) and (b), both beams GC-20 and GC-60 failed due to crushing of the UFC segments in the flexural span. However, in the case of beam GC-60, a small slip of all the UFC segments in the shear span was also observed (Figure 18(b)). Failure of the GC-90 beam occurred through shearing of both the epoxy adhesive and the FRP bolts followed by kinking and crushing of the GFRP top flange and the web in the shear span (Figure 18(c)). There was no damage to the UFC segments, and four UFC segments in the shear span were completely detached from the GFRP top flange.

Figure 18.

Failure patterns of GFRP-UFC composite beams: (a) G-20 beam, (b) G-60 beam and (c) G-90 beam.

A summary of test results for all the GFRP I-beams and the GFRP-UFC composite beams is shown in Table 5. The ultimate load-carrying capacity of both the GFRP I-beams and the GFRP-UFC composite beams decreased as the temperature increased. Strain gauge data highlighted that the addition of the UFC at the GFRP I-beam top flange decreased the midspan compressive strain and increased the midspan tensile strain at all temperatures (Table 5). It was found that the use of the UFC slab eliminated the premature delamination or kink failure of the GFRP I-beams and improved the flexural capacity at all temperatures. The GFRP-UFC composite beams exhibited 83%, 75% and 147% flexural capacity higher than those of the GFRP I-beams at 20°C, 60°C and 90°C, respectively. Figure 19 shows the relationship between the normalized flexural capacity and the temperature of the GFRP I-beams and the GFRP-UFC composite beams. In both beams, more than 85% of their flexural capacity was retained up to 60°C. However, at 90°C, the retained flexural capacity of the GFRP I-beam and the GFRP-UFC beam dropped to 38% and 50%, respectively. This is attributed to the degradation of the material properties of the GFRP I-beams and the shear connectors (FRP bolts and epoxy adhesive) beyond their glass transition temperatures.

Figure 19.

Normalized flexural capacity of GFRP I-beams and GFRP-UFC beams.

All the GFRP I-beams and GC-20 beam exhibited almost linear load-deflection relationship until failure (Figures 20 and 21). On the other hand, beams GC-60 and GC-90 exhibited slippage of the UFC segments at the load of 144 and 89 kN, respectively (Figure 21). Similar to the flexural capacity, the stiffness of all the beams decreased as the temperature increased. The longitudinal strain distribution at the midspan section of the GFRP-UFC composite beams is shown in Figure 22. There was a very small slip between the UFC segments and the GFRP I-beam top flange in beam GC-20 and the strain distribution remained almost linear up to failure (Figure 22(b)). This confirms that there was almost full composite action between the UFC slab and the I-beam top flange in beam GC-20. The midspan strain variation was linear in beam GC-60 even after slipping of the UFC segments and this may be due to the residual shear capacity of the epoxy adhesive at 60°C (Figure 22(c)). However, as the slippage of the UFC slab happens in beam GC-60, the neutral axis exhibited downward movement, which can be interpreted as the loss of composite action between the UFC slab and the GFRP I-beam compression flange. In contrast, the strain distribution of beam GC-90 was non-linear and there was a large drop in the neutral axis after slipping of the UFC segments (Figure 22(d)). After the UFC segment slippage in beam GC-90, the composite action between the GFRP flange and the UFC slab was significantly reduced.

Figure 20.

Load-deflection relationship of GFRP I-beams.

Figure 21.

Load-deflection relationship of GFRP-UFC composite beams.

Figure 22.

Longitudinal strain distribution across midspan cross-section of GFRP-UFC composite beams: (a) strain gauge locations, (b) GC-20 beam, (c) GC-60 beam and (d) GC-90 beam.

The maximum midspan deflection of the GC-90 beam before applying load was very small (approximately 1 mm), which is only 1.5% of the deflection at failure. This result reveals that the deflection of the GFRP-UFC composite beams subjected to elevated temperatures caused by the bending effect due to the different coefficient of thermal expansion values of the GFRP and the UFC materials will be negligible.

Fibre model analysis

The flexural behaviour of the GFRP-UFC composite beams was predicted using a simple fibre model analysis (FMA) and verified by the experimental results. In this study, the FMA results were basically used to check the flexural behaviour of the GFRP-UFC composite beams within the serviceability limit state. The fibre model is a one-dimensional model and it has been used successfully by researchers to analyse the flexural behaviour of the reinforced concrete beams, FRP beams and FRP-concrete composite beams (Manalo and Aravinthan, 2012; Nguyen et al., 2013; Park and Paulay, 1975). In this method, the theoretical moment-curvature relationship of the GFRP-UFC composite beam sections is determined based on the strain compatibility and internal force equilibrium principles. The FMA is comparatively less time-consuming and cost-effective compared to the commercial finite element software packages.

In the FMA, the composite beam is divided into m longitudinal segments along the span (m = a + 2b) and the midspan cross-section is divided into n discrete horizontal elements (Figure 23(a)). The following assumptions were made in the analysis: (1) GFRP-UFC composite beam behaves under Euler-Bernoulli beam theory; (2) there is full interaction between the UFC segments and the GFRP I-beam top flange until beam failure, and hence, there is no slip at the UFC-GFRP interface; (3) the temperature within each part of the GFRP-UFC composite beam (UFC slab, GFRP top flange, GFRP web and GFRP bottom flange) remains constant and (4) the bending effect due to the mismatch of coefficient of the thermal expansion values is negligible. Two material models were used for the analysis: (1) UFC bi-linear model proposed by JSCE (2004) and (2) GFRP model (developed using material test results in section ‘Material properties at room and elevated temperatures’). Figure 24 shows the UFC and GFRP models used in the FMA at T°C temperature.

Figure 23.

Fibre model analysis: (a) GFRP-UFC beam cross-section, (b) strain diagram and (c) stress diagram.

Figure 24.

UFC and GFRP material models used in the fibre model: (a) UFC bi-linear model at T°C temperature and (b) GFRP model at T°C temperature.

Initially, the compressive and the tensile strains of top and bottom elements (ε_c and ε_t) at the midspan section are assumed in the fibre model. The average strain at the centre of i^th element (ε_i) is given by equation (1) (Figure 23(b)), where d_i is the distance to the centroid of the i^th element from the beam top (equation (2))

ε_{i} = ε_{c} - \frac{d_{i} (ε_{c} + ε_{t})}{h}

(1)

d_{i} = \frac{h (i - 0.5)}{n}

(2)

The stress in each element of the GFRP-UFC beam cross-section (Figure 23(c)) is calculated using the element strains and the material models (Figure 24) corresponding to the beam temperature being considered. The force equilibrium is checked using equation (3), where $\sum F_{c}$ and $\sum F_{t}$ are the resultant compressive and tensile forces acting on the GFRP-UFC cross-section, respectively

\sum F_{c} - \sum F_{t} = 0

(3)

If equation (3) is not satisfied, an iterative method is used in the fibre model to achieve the force equilibrium. Once force equilibrium is obtained, the resultant moment (M) and the curvature (ϕ) are calculated using equations (4) and (5), respectively

M = \sum F_{i} d_{i}

(4)

where F_i is the element force of the i^th element

ϕ = \frac{(ε_{c} + ε_{t})}{h}

(5)

The average curvature of each section of the longitudinal segments (ϕ_j) is calculated and then the maximum deflection at the midspan (δ_max) is calculated according to equation (6), where l_j is the length of the longitudinal segment

δ_{m a x} = \sum_{j = 1}^{0.5 a + b} ϕ_{j} l_{j}

(6)

The total shear force developed at the UFC-GFRP interface is calculated using the element strain values in the shear span (equation (7)), where D₁ is the number of horizontal divisions in the UFC slab cross-section, ΔA is the cross-sectional area of a UFC element and E_c is the Young’s modulus of concrete

F_{s h e a r} = \sum_{i = 0.5 a}^{0.5 a + b} \sum_{j = 1}^{D_{1}} E_{c} (ε_{i, j} - ε_{i + 1, j}) Δ A

(7)

In this method, loop calculations are carried out by increasing the ε_c and ε_t until the ultimate compressive strain of the UFC is reached or until the total shear force (at the UFC-GFRP interface) exceeds the total shear capacity of the FRP bolts. The total shear capacity of the FRP bolts and the epoxy adhesive at the beam temperature was determined from the shear test data.

FMA of the GFRP-UFC beams with a small temperature gradient

FMA was carried out for the GFRP-UFC composite beams with a small temperature gradient across the beams’ cross-section. The load-deflection relationships at the midspan section, obtained from the experiment and the analysis, are shown in Figure 25. Figure 25(a) shows a relatively good agreement between the experiment and the FMA results for beam GC-20. For beams GC-60 and GC-90, the load-deflection relationships were agreed up to the occurrence of slipping of the UFC segments (Figure 25(b) and (c)). This is because, in the FMA, full interaction between the UFC segments and the GFRP flange was assumed. The vertical load corresponding to the epoxy adhesive shear capacity in beams GC-60 and GC-90 was obtained by the FMA (see lines ② and ④ in Figure 25(b) and (c), respectively). In addition, the vertical load corresponding to the total FRP bolt shear capacity was analytically determined for beams GC-60 and GC-90 (see lines ① and ③ in Figure 25(b) and (c), respectively). According to the FMA, slipping occurred in both GC-60 and GC-90 beams when the vertical load is approximately equal to the load corresponding to the shear capacity of the epoxy adhesive. The difference in the vertical load corresponding to the slipping of the UFC segments between the experiment and the FMA was about 5.5% for beam GC-60 and 7.5% for beam GC-90.

Figure 25.

Load-deflection relationships of GFRP-UFC beams: (a) beam GC-20, (b) beam GC-60 and (c) beam GC-90.

In both GC-20 and GC-60 beams, the ultimate strain of the UFC slab was reached at midspan before the FRP bolt shear capacity was attained. Therefore, those beams failed by crushing of the UFC slab at the midspan. The FMA results of beam GC-90 showed that the total FRP bolt shear capacity at the UFC-GFRP interface was exceeded at the beam failure and that was confirmed by the experiment results (Figure 25(c)). The reason for the different failure modes observed in beam GC-90 was the significant degradation of shear capacity of the FRP bolts between 60°C and 90°C temperatures.

FMA of GFRP-UFC beams with a moderate temperature gradient

In contrast to the experiment described above, the GFRP-UFC composite beams may experience a moderate temperature gradient across the beam cross-section in actual circumstances. These GFRP-UFC composite beams were developed for outdoor constructions, where the main source of heat is solar radiation. Therefore, the beam top temperature will be the highest and the beam bottom temperature will be almost the same as the ambient temperature. The flexural performance of the GFRP-UFC composite beams under this condition was analysed by the FMA. Three temperature cases were analysed and Table 6 shows the temperatures of different parts of the GFRP-UFC composite beams for each case. Temperatures for Case 1 were determined based on the field investigation conducted by Sirimanna et al. (2011) and those for Case 2 and Case 3 were assumed by the authors.

Table 6.

Temperatures of the parts of the GFRP-UFC beams in the FMA.

Case	Temperature (°C)
Case	UFC slab	Epoxy adhesive	Top flange	Web	Bottom flange
Case 1	60	50	40	30	30
Case 2	70	60	50	40	30
Case 3	80	70	60	40	40

UFC: ultra-high-strength fibre-reinforced concrete.

The FMA results of the GFRP-UFC composite beams with a moderate temperature gradient are shown in Figure 26(a). For comparison purpose, the FMA results of beam GC-60 (small temperature gradient) are also included in Figure 26(a). Case 1 beam failed by crushing of the UFC segments and there was no slip of the UFC segments up to beam failure (Figure 26(b)). On the other hand, there was slipping of the UFC segments in the GC-60 beam, even though the maximum temperature at the GC-60 beam top was identical to Case 1 beam. As a result of the shear failure of the epoxy adhesive, Case 2 and Case 3 beams exhibited slipping of the UFC segments at the load of 151 and 127 kN, respectively (Figure 26(b)). This result confirms that the full composite behaviour of the GFRP-UFC beam is lost at relatively low vertical loads when the composite beam top is subjected to elevated temperatures.

Figure 26.

(a) Analytical results of GFRP-UFC beams with a moderate temperature gradient and (b) enlarged view.

As shown in Figure 26(a), the stiffness of Case 1, Case 2 and Case 3 beams was not significantly affected by the moderate temperature gradient. On the other hand, the stiffness of beams GC-60 and GC-90 with a small temperature gradient degraded significantly (Figure 21). Therefore, the GFRP-UFC composite beam’s stiffness may not be significantly affected by the actual hot temperature conditions, where there is a moderate temperature gradient across the beam cross-section. This is attributed to the fact that the temperature increment in the GFRP bottom flange and the GFRP web is comparatively lower than that of the top parts of the composite beam in the actual hot temperature environments.

Conclusions

In this article, the mechanical properties of the materials used in the GFRP-UFC composite beams and the flexural behaviour of the GFRP-UFC composite beams subjected to elevated temperature conditions were examined. The following conclusions were drawn:

The tensile and compressive strengths of the GFRP I-beams are significantly degraded by the temperatures beyond the glass transition temperature (T_g) of the vinylester resin. Similarly, the compressive and shear strengths of the FRP bolts are greatly reduced by the elevated temperatures. In contrast, there is no significant effect on the tensile strength of the FRP bolts as well as the compressive strength of the UFC slab for the temperatures ranging from 20°C to 90°C.

The flexural capacity and the stiffness of the GFRP I-beams and the GFRP-UFC composite beams are degraded by elevated temperatures. However, more than 85% of the flexural capacity (compared to the flexural capacity at 20°C) of both beams can be retained when the beam temperature is below 60°C. As the beam temperature increases beyond 60°C, the flexural capacity of these beams is severely degraded.

The use of the UFC segments significantly improves the ultimate flexural capacity and the stiffness of the GFRP I-beams at temperatures ranged from 20°C to 90°C. This is because the UFC segments can prevent premature delamination and kink failure of the GFRP I-beam compression flange. As a result, the tensile strength of the GFRP I-beam can be effectively utilized.

In the actual hot temperature environments, the GFRP-UFC composite beams may experience a moderate temperature gradient across beam depth. The beam top temperature can reach up to 60°C-80°C while the beam bottom temperature is almost the same as the ambient temperature. The flexural behaviour of the GFRP-UFC composite beams under this condition was analysed using the FMA. From the analysis, it was found that the stiffness of the GFRP-UFC composite beams, which is related to the main design criterion of the FRP bridges, is not significantly affected by the temperature gradient in the real situations. However, the flexural capacity of the GFRP-UFC composite beams at the slipping of the UFC segments is greatly influenced by the temperature of the beam top.

In short-span pedestrian bridges, the full flexural capacity of the GFRP-UFC composite beams cannot be utilized because of the deflection limitation. The flexural capacity of the GFRP-UFC composite beams at this deflection limitation was less than that of the same composite beams at the shear failure of epoxy adhesive. Since the FMA model can accurately predict the load–deflection behaviour of the GFRP-UFC composite beams up to the shear failure of the epoxy adhesive, this simplified analysis method is recommended for predicting the flexural behaviour of the GFRP-UFC composite beams subjected to elevated temperature within the usable range of the composite beam’s flexural capacity in actual applications.

Footnotes

Acknowledgements

The authors greatly acknowledge Dr. Atsushi Sumita, Mr. Yusuke Kanaya and all the other members of the FRP research group at Saitama University, who provided support for the successful implementation of the experiments. The authors thank the Fukui Fibertech Company for providing FRP materials.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: The authors gratefully acknowledge the financial support of the Japan Ministry of Economy, Trade, and Industry for this research project.

References

American Association of State Highway and Transportation Officials (AASHTO) (2008) Guide Specifications for Design of FRP Pedestrian Bridges. 1st ed. Washington, DC: AASHTO.

American Concrete Institute (ACI) (2008) Guide for the Design and Construction of Externally Bonded FRP Systems for Strengthening Concrete Structures (ACI 440.2R-08). Farmington Hills, MI: ACI, p. 14.

Bai

Keller

(2011) Delamination and kink-band failure of pultruded GFRP laminates under elevated temperatures and compression. Composite Structures 93(2): 843–849.

Bank

(2006) Composites for Construction: Structural Design with FRP Materials. Hoboken, NJ: John Wiley & Sons.

Bisby

Green

Kodur

VKR

(2005) Response to fire of concrete structures that incorporate FRP. Progress in Structural Engineering and Materials 7(3): 136–149.

Correia

Gomes

Pires

et al . (2013) Mechanical behaviour of pultruded glass fibre reinforced polymer composites at elevated temperature: experiments and model assessment. Composite Structures 98: 303–313.

Dai

Gao

Teng

(2012) Bond-slip model for FRP laminates externally bonded to concrete at elevated temperature. Journal of Composites for Construction 17(2): 217–228.

Farhey

(2005) Long-term performance monitoring of the Tech 21 all-composite bridge. Journal of Composites for Construction 9(3): 255–262.

Foster

Richards

Bogner

(2000) Design and installation of fiber-reinforced polymer composite bridge. Journal of Composites for Construction 4(1): 33–37.

10.

Foster

Bisby

(2008) Fire survivability of externally bonded FRP strengthening systems. Journal of Composites for Construction 12(5): 553–561.

11.

Gonilha

Correia

Branco

(2014) Structural behaviour of a GFRP-concrete hybrid footbridge prototype: experimental tests and numerical and analytical simulations. Engineering Structures 60: 11–22.

12.

Hai

Mutsuyoshi

Asamoto

et al . (2010) Structural behavior of hybrid FRP composite I-beam. Construction and Building Materials 24(6): 956–969.

13.

Hayes

Lesko

Haramis

et al . (2000) Laboratory and field testing of composite bridge superstructure. Journal of Composites for Construction 4(3): 120–128.

14.

Huh

Lee

Kim

et al . (2012) Effect of stress ratio on fatigue life of GFRP composites for WT blade. Journal of Mechanical Science and Technology 26(7): 2117–2120.

15.

Japan Society of Civil Engineers (JSCE) (2004) Recommendations for Design and Construction of Ultra High Strength Fiber Reinforced Concrete Structures, vol. 9 (JSCE guidelines for concrete). Tokyo, Japan: JSCE, pp. 1–106.

16.

Japan Society of Civil Engineers (JSCE) (2011) Guidelines for Design and Construction of FRP Footbridges. Tokyo, Japan: JSCE (in Japanese).

17.

Japanese Industrial Standard (JIS) (1987) Testing Methods for Transition Temperatures of Plastics (K7121). Tokyo, Japan: Japanese Standards Association.

18.

Karbhari

Wang

(2004) Multi-frequency dynamic mechanical thermal analysis of moisture uptake in E-glass/vinylester composites. Composites Part B: Engineering 35(4): 299–304.

19.

Keller

Bai

Vallée

(2007) Long-term performance of a glass fiber-reinforced polymer truss bridge. Journal of Composites for Construction 11(1): 99–108.

20.

Kwon

Dutta

Kim

et al . (2004) Fatigue studies of FRP composite decks at extreme environmental conditions. Key Engineering Materials 261: 1301–1306.

21.

Manalo

Aravinthan

(2012) Behaviour of glued fibre composite sandwich structure in flexure: experiment and fibre model analysis. Materials & Design 39: 458–468.

22.

Manalo

Aravinthan

Mutsuyoshi

et al . (2012) Composite behaviour of a hybrid FRP bridge girder and concrete deck. Advances in Structural Engineering 15(4): 589–600.

23.

Mendes

PJD

Barros

JAO

Sena-Cruz

et al . (2011) Development of a pedestrian bridge with GFRP profiles and fiber reinforced self-compacting concrete deck. Composite Structures 93(11): 2969–2982.

24.

Nguyen

Mutsuyoshi

Zatar

(2013) Flexural behavior of hybrid composite beams. Transportation Research Record: Journal of the Transportation Research Board 2332(1): 53–63.

25.

Nguyen

Mutsuyoshi

Zatar

(2014) Push-out tests for shear connections between UHPFRC slabs and FRP girder. Composite Structures 118: 528–547.

26.

Nguyen

Mutsuyoshi

Zatar

(2015) Hybrid FRP-UHPFRC composite girders: part 1 – experimental and numerical approach. Composite Structures 125: 631–652.

27.

Nishizaki

Takeda

Ishizuka

et al . (2006) A case study of life cycle cost based on a real FRP bridge. In: Proceedings of the 3rd international conference on FRP composites in civil engineering (CICE), Miami, FL, 13–15 December, pp. 99–102. Available at: https://www.iifc.org/proceedings/CICE_2006/CICE2006.pdf

28.

Park

Paulay

(1975) Reinforced Concrete Structures. New York: John Wiley & Sons.

29.

Sirimanna

Islam

Aravinthan

(2011) Temperature effects on full scale FRP bridge using innovative composite components. In: Ye

Feng

Yue

(eds) Advances in FRP Composites in Civil Engineering. Berlin: Springer, pp. 376–380.

30.

Strongwell (2002) Strongwell Design Manual (CD ROM). Bristol, VA: Strongwell, Inc. Available at: http://www.strongwell.com/

31.

Tanaka

Tazawa

Kurita

et al . (2006) A case study on life-cycle assessment of environmental aspect of FRP structures. In: Proceedings of the 3rd international conference on FRP composites in civil engineering (CICE), Miami, FL, 13–15 December, pp. 175–178. Available at: https://www.iifc.org/proceedings/CICE_2006/CICE2006.pdf

32.

Uno

Kitayama

(2003) Design, fabrication and erection of the pedestrian bridge in the road-park of Ikei-Tairagawa in Okinawa. IHI Engineering Review 36(2): 35–39.

33.

Van Erp

(2008) Mechanics and Technology of Fibre Composites. Toowoomba, QLD, Australia: University of Southern Queensland.

34.

Watanabe

Musha

Yoshinaga

(2007) Design and performance tests for bridge using ultra high strength fiber reinforced concrete. In: Proceedings of the 23rd US–Japan bridge engineering workshop, Tsukuba, Japan, 4–11 November.

35.

Wijayawardane

ISK

Mutsuyoshi

Perera

SVTJ

(2014) Innovative composite girders using FRP I-beams and precast UFC slab segments. In: Proceedings of the 7th international conference on FRP composites in civil engineering (CICE), Vancouver, BC, Canada, 20–22 August.

36.

Wijayawardane

ISK

Mutsuyoshi

Perera

SVTJ

et al . (2013) Development of GFRP and UHF composite girders. In: Proceedings of the 13th East Asia-Pacific conference on structural engineering and construction (EASEC-13), Sapporo, Japan, 11–13 September.