Rolling Load versus Pulsating Load Fatigue Behavior of a Full-Scale Bridge Deck Reinforced with GFRP Bars

Abstract

This paper presents a pioneering experimental investigation of the fatigue behavior of a full-scale (15.24 m × 3.89 m) concrete bridge deck reinforced with glass fiber–reinforced polymer (GFRP) rebar under rolling (R)-load cycles applied using the Rolling Load Simulator (ROLLS) at Queen's University, Canada. The study compares this type of fatigue with conventional fixed-point pulsating (P)-loads. The deck is supported by two Steel I-girders spaced at 3.05 m. A 3.81-m-long section on one end was subjected to R-loads and a similar section on the other end, to P-loads, both up to 3 million (M) cycles. Monotonic tests were performed periodically. The study showed that R-load results in far more fatigue damage than P-load, as reflected by 71% and 54% stiffness degradation (k/ko), respectively. This suggests that cumulative damage of one R-cycle is equivalent to 120 P-cycles. Live load deflection limit of span/800 was reached in the R-loading section after 0.78M cycles, and at 3M cycles was exceeded by 15%, but the P-loading section remained below the limit. GFRP reinforcement performed very well, with maximum strain remaining below 10% of the guaranteed tensile strain. The R-load section showed far more extensive and dense grid-pattern cracking with concrete pitting at the soffit, while the P-load section showed minor longitudinal and some radial cracks. A conversion factor (ξ) is established to enable researchers and designers convert (k/ko) from readily and easily available P-loading to an equivalent R-loading of GFRP-reinforced deck at any number of cycles, since P-loading is not conservative and R-loading capabilities are not easily available. In this study, ξ is 0.59 at 3M cycles and is projected to be 0.5 at 10M cycles. It has been well-established that fixed-point pulsating loads cannot accurately replicate the fatigue damage resulting from traffic loads on bridge elements. The degradation behavior of bridge decks’ stiffness should be studied using rolling loads. However, conducting a rolling load fatigue experiment often requires a specialized apparatus that is not readily available. This paper presents the establishment of a conversion factor (ξ) that can be used to convert the normalized stiffness data of bridge decks tested under traditional pulsating loads into real stiffness degradation data under rolling loads.