Experimental Evaluation of the Bond Dependent Coefficient and Parameters which Influence Crack Width in GFRP Reinforced Concrete

Abstract

Reinforcement of concrete flexural components has been traditionally provided by steel rebar; however, durability concerns and life maintenance costs of this product have powered the emergence of fibre reinforced polymers (FRP) as reinforcement in concrete. FRP products hold tremendous promise but their application can be constrained due to design challenges resulting from a reduced modulus of elasticity. The ability to meet serviceability behavior, such as crack width and deflection, is commonly the limiting factor for design. Therefore, the area of FRP reinforcement provided is often greater than the amount required for strength alone and this has significant impacts on the project economics. The bond dependent coefficient (kb) of FRP is required for serviceability design purposes in order to account for the bonding capability of FRP to concrete. The values of this coefficient reported in experimental studies are highly variable, resulting in unreliable crack response predictions. Therefore, a more consistent interpretation and calculation must be found for the bond dependent coefficient due to its critical importance in design. The bond dependent coefficient, as well as physical parameters which influence crack width in GFRP reinforced concrete, were investigated experimentally in this study using a total of 33 specimens. The test procedure was taken from a procedure being developed by the American Concrete Institute (ACI) Committee 440 and was evaluated and modified as required during testing. Phase I testing was used to investigate and determine the physical parameters which had the most significant influence on cracking behaviour and bonding capability. Using significant findings from Phase I, Phase II testing was structured to focus on the interpretation of the bond dependent coefficient and the statistical variation in a set of 5 identical test specimens. Current design equations, as recommended by ACI 440.1R-06 and CHBDC CAN/CSA-S6-06, were used for the calculation of the bond dependent coefficient for all specimens. Interpretation of the bond dependent coefficient was considered using the stress-level approach and newly developed slope approach. Results of the study indicated that the high variability of kb was likely due to its interpretation. Current design equations force a zero intercept, neglecting the fact that concrete does not crack immediately upon loading. In addition, clear definitions of service stress and maximum crack width are ambiguous, further complicating the calculation of the bond dependent coefficient. This resulted in a range of kb values for a given beam despite the fact that kb is inherently a material property of the bar. The behaviour of specimens following load cycling was also very different than the initial loading cycle and consequently, kb was also significantly different. As structures in the field will be subjected to continual loading and unloading, the effect of cyclic loading becomes a consideration in the calculation of kb.