3. Test results and discussion
The test results focused on both the service and ultimate limit state behaviour. Given the low modulus of elasticity of GFRP compared to steel, the service behaviour of GFRP reinforced slabs such as deflection, crack width expansion, and stress on GFRP bars were of particular interest. Failure load and failure mode were monitored in each slab as 
3.1 Serviceability behaviour GFRP reinforced slabs
3.1.1 Load versus deflection behaviour
Load versus deflection of the test slabs is presented in Figure 4(a) and Figure 4(b). Test slabs displayed a linear response until the first crack formation. The post-cracking behaviour of the test slabs show that the nonlinear responses are influenced by the amount of reinforcement, position of the reinforcement, and concrete stiffness. Linear behaviour in the deflection response noticed up to the formation of the first crack. However, once cracked, single mid-depth reinforced slabs showed a higher deflection for the corresponding load than the conventional double reinforced slabs as presented in Figure 4(a). Higher deflection of single mid-depth reinforced slabs can be accredited to the loss of concrete stiffness resulting from the crack formation and the position of the reinforcement at the mid-depth.
Higher stiffness and lower deflection for a corresponding load in unreinforced slab compared to that of single mid-depth reinforced slabs as shown in Figure 4(a), can be attributed in part, to the higher concrete compressive strength compared to that of other slabs. Higher concrete strength increases the stiffness of concrete, hence results in increased arching action in unrestrained slab.
Once a crack had formed in the tension zone, the behaviour of the concrete slabs was influenced either by the position and the stiffness of the reinforcement, or stiffness of the membrane arch due to in-plane restraints. As shown in Figure 4(b), all of the slabs with top and bottom layers of reinforcement showed a similar load versus deflection response up to an applied load of 100 kN. Test slabs’ deflections are compared in Table 3 for BS EN 1991-241 maximum tandem system Load Model 1 (TSLM1) service wheel load of 150 kN and also at the failure load. This is an onerous comparison, since only a portion of the total wheel load would act on this slab strip and the stiffness of a continuous slab in a real bridge deck would be much higher. The two slabs with single mid-depth reinforcement showed higher deflection, exceeding span/250 that is recommended in BS EN 1992-1-136. Maximum deflection of span/250 for steel reinforced concrete slabs is allowed in Eurocode when the deflection cause no damage to the structure below. A similar limit can be found in ACI-318-0542 for steel reinforced slabs. All the two-layer GFRP reinforced slabs have satisfied the deflection criteria at the service load of 150 kN.
A comparison of the load versus deflection responses of the tested slabs indicates that conventional two-layer reinforcement method is required to control the deflection in GFRP reinforced restrained slabs.
3.1.2 Crack width and pattern
Tests on GFRP reinforced slabs disclosed two types of crack patterns. In total, three cracks were noted in unreinforced and 0.6% single mid-depth GFRP reinforced slabs [Figure 5(a) & (b)]. The first crack appeared directly below the loading point and the crack width increased with the load increment. Two more cracks were observed at the top surface, adjacent to the fixed end supports at higher loads. Lower stress on reinforcement and zero reinforcement can attribute to the single crack at the soffit of 0.6% single mid-depth GFRP reinforced slab and unreinforced slab respectively.
Two-layer reinforced concrete slabs and 0.15% single mid-depth GFRP reinforced slabs showed more cracking at higher load levels. In addition to the primary crack, some more cracks were visible at the soffit and parallel to the first crack, as the load increased [Figure 5(c) & (d)]. More distributed cracks appeared in two layer reinforced slabs and the 0.15% single mid-depth reinforced slab can be attributed to higher stress on GFRP bars and proximity of the bars to the soffit. Figure 6 (a) and 6 (b) show the crack width expansion according to load increment for the primary cracks on each slab.
Expansion of the primary crack was based on several parameters such as the stress on GFRP bars, size of the reinforcement, distribution of cracks, aggregate interlocking and influence of arching action (that is higher for higher concrete compressive strength and external in-plane restraint stiffness). The intention of measuring crack widths in this study was to understand the ability of GFRP reinforced slabs to control crack width expansion, as GFRP bars have lower modulus of elasticity than steel and the bond between GFRP and concrete can be different to that of steel and concrete.
Primary crack expansions for unreinforced slab and single mid-depth reinforced slabs are shown in Figure 6 (a). Observation of crack pattern and crack width expansion suggests that the crack widths were double in unreinforced slab and in single mid-depth reinforced slabs, compared to those of other slabs at an applied load of 65 kN (Table 3). Among the two layer reinforced slabs, the 0.6% GFRP reinforced slabs demonstrated a better ability to control crack width than 0.15% GFRP reinforced slab, as illustrated in Figure 6(b). All test slabs that have more than one crack at the soffit are compared in Figure 6(b) for crack width expansion of the primary crack. Table 3 presents the maximum crack widths measured at different load levels. Since some of the vibrating wire gauges failed to record crack width expansion at 150 kN, crack widths are compared at 65 kN for all of the test slabs.
Measured crack widths at EC1 TSLM1 load of 150 kN are compared against the allowable maximum crack width in Table 4. BS EN 1992-1-136 recommends the crack width to be less than 0.3 mm for steel. However, considering the corrosion resistant nature of GFRP bars, CAN/CSA (2006) recommends up to 0.5 mm crack width for GFRP reinforced structures. Since no guidelines are available for FRP reinforced structures within Eurocodes, CAN/CSA8 recommendations suggested in ACI 440-1R6 are considered to verify the crack width limits for the test slabs.
Crack width was within 0.5 mm level at 150 kN load for all the two layer reinforced slabs except G-0.15%-8 mm-300 T&B. Although the gauge failed before the 150 kN load for the test slabs G-1.2%-16 mm-125 T&B, it can be considered without any loss of generality that the crack width of the slab was within the acceptable limit as the crack width was 0.2 mm for 1.2% GFRP reinforced slab at 135 kN. The crack width expansion shows that crack width on 0.15% two-layer GFRP reinforced slab was greater than 0.5 mm and exceeding the recommended maximum crack width limit at the service load level. The slab tests revealed that two layers of 0.6% GFRP reinforcement is required to meet the recommended service load behaviour in concrete slabs. The maximum crack width for 0.6% two layer reinforced slabs were below 0.33 mm at 150 kN, which was just above the 0.3 mm crack width limit recommended for corrosive steel reinforced structures, and within the allowable 0.5 mm crack width for FRP reinforced structures.
3.1.3 Stress on GFRP bars
Stress values from GFRP bars were obtained using recorded strain values and measured modulus of elasticity for the test slabs. These are compared in Table 4 at a service load of 150 kN. It shows that the maximum service load stress on GFRP bars was less than half the rupture stress in all but 0.15% GFRP reinforced slab (G-0.15%-8 mm-300_T&B).
Due to creep rupture failure of FRP materials, CAN/CSA8 limits the maximum stress at the service load level to be 25% of the ultimate rupture stress of GFRP bars. Experimental investigation indicates that the maximum stress at the service load level was less than 25% of the ultimate stress in slabs reinforced with two layers of 0.6% GFRP bars. A higher stress than maximum allowable stress of 170.5 N/mm2 in 0.15% GFRP reinforced slabs, and 0.6% single mid-depth reinforced slab (G-0.6%-12mm-125-M) can be attributed to the lower amount of reinforcement in 0.15% GFRP reinforced slabs and reduced effective depth in single mid-depth reinforced slab G-0.6%-12 mm-125-M.
Higher stress values noted in the GFRP bars of slab G-1.2%-16 mm-125_T&B was unexpected, yet can be attributed to higher flexural strength contribution to the total strength of the slab as the slab was reinforced with more than the balanced amount of reinforcement of 1.1% and a slightly lower concrete compressive strength which would indicate lower arching action than that of other slabs. This phenomenon necessitates further studies to establish the behaviour of in-plane restrained slabs reinforced with more than the balanced amount of reinforcement.