Finite element analysis to simulate reinforced concrete corrosion in beams and bridge decks
University of Delaware
Current analysis techniques do not acknowledge the existence of load redistribution between girders, or system capacity, of bridges due to a lack of understanding on the redistribution mechanisms. This lack of understanding is the primary motivation for this research. Specifically, the deck as a load redistribution mechanism is analyzed. It is thought that including the system capacity of bridges would help to prioritize repairs and allocate the limited funding available for infrastructure. For this reason, this research aims to aid in quantifying the system capacity effects of bridges due to corrosion of reinforcement in the deck. Previously, a literature review was executed to determine the effects of corrosion in reinforced concrete. It was determined through this review of testing that the change in performance due to corrosion is best estimated as a strength decrease of 50% and an ultimate deflection increase of 82%, mimicking 25 years of corrosion. These expected performance metrics were used to create finite element models of uncorroded and corroded reinforced concrete beams. Different concrete material modeling techniques available within the commercial software ABAQUS were assessed; these include brittle cracking, smeared crack, and concrete damaged plasticity techniques. This was done for both 2-dimensional and 3-dimensional concrete elements, as well as for both 2-dimensional and 3-dimensional rebar elements within the 3-dimensional concrete elements. In the end, it was determined that using the concrete damaged plasticity approach with 2-dimensional beam and rebar elements produced the most accurate results and was also the easiest approach to implement in existing full-scale bridge models; this approach will also reduce computational effort in any future full-scale bridge models. After the modeling technique was determined, the input was calibrated to determine the optimal approach to model uncorroded and corroded reinforced concrete. It was determined which input values to use for the uncorroded concrete, and how to alter these values to simulate corrosion. Through this optimization process, it was determined that a 40% decrease in the modulus of elasticity of concrete, 40% decrease in tensile strength of concrete, 64% decrease in compressive strength of concrete, 20% decrease in area of compressive steel, and 61.5% decrease in area of tensile steel resulted in the optimum simulation of reinforced concrete corrosion. This uncorroded and corroded input was applied to 3 different full-scale bridge models which were previously created; these models were created and calibrated based on actual bridges located in Delaware that had been previously field tested, all having steel girders. It was found that this modeling approach created convergence difficulties in some of the bridge models when attempting to load the structures to their ultimate capacities and subsequently only the results of one of the bridges, for which convergence was obtained up through a peak loading, was analyzed in depth. This bridge is referred to as Bridge 7R and served as an exit ramp for Interstate 295 North, just south of the Delaware Memorial Bridge. Initially, convergence with corroded models was not reached. However, after changing the input parameters governing tension stiffening, convergence was achieved. The maximum loading and the distribution factors of the bridge models were analyzed. It can be seen in these results that the corrosion in the deck caused a more uniform stress distribution in the deck, and consequently the girders, than with an uncorroded deck. Contrary to the expected response, the corroded model resulted in DF values between those of the elastic and those of the uncorroded models; however, in all models the DF approached the theoretical inelastic values. The results of these models also indicated that the corroded models reached higher strengths than their uncorroded counterparts. It was thought that the cause of the differences in both the strength and DF values was due to greater load sharing in the corroded model when compared to the corresponding uncorroded model.