Ultra-High Performance concrete (UHPC) has gained popularity in recent years due to its strong tensile properties. Most notably, its tensile properties allow it to resist much higher loads in shear, with literature agreeing that it can greatly reduce beam sizes and steel reinforcement, potentially leading to savings in construction costs. However, how to design UHPC beams to resist shear is still not agreed upon. A new wave of prediction models for beam shear capacities using the Modified Compression Field Theory (MCFT) and empirical models have been produced with varying levels of accuracy depending on beam dimensions and loading, each with their strengths and weaknesses. However, none were developed using probabilistic data likely due to limited data on UHPC performances. Therefore, the proposed shear prediction models have an unclear level of safety, calling for an in-depth reliability analysis of the proposed models to assess their reliability, and economic feasibility. In this study, 11 UHPC models, 5 of which were developed using empirical methods, and the other 6 through the modified compression field theory (MCFT), are assessed using a database of 104 UHPC beams tested in shear. Out of 500+ beams considered for the database in this study, only 104 were considered to meet the criteria of all models. UHPC material statistics were also collected from literature and experimental work to describe UHPC materials into three classifications, labelled Class 1, 2, and 3, to better describe UHPC performance statistics. Each UHPC shear models are then calibrated. Among the shear models, the Federal Highway Administration (FHWA) model has the highest accuracy and consistency with an R2 value of 0.884, coefficient of variation (COV) of 0.396 and a bias factor of 1.52.A calibrated finite element model is then produced for 4 experimental beams and 5 beams from literature. Using the calibrated models, Rosenblueth 2n+1 point-estimation method was completed to produce a COV and bias factor representing the fabrication and material factor for reliability analysis. Using all calculated model factors, a reliability analysis is completed, producing a reliability index across span lengths for prestressed beams in bridge superstructures. A more accurate, calibrated model with an R2 value of 0.892, COV of 0.25, and bias factor of 1.07 and an increase in reliability up to 70% is proposed. An analysis is also conducted on the minimum steel reinforcement requirement based on a target reliability index defined by bridge girders currently in service, with new recommendations on span lengths for AASHTO girders. Results show that current typical girder span lengths do not require shear reinforcing steel, as the UHPC material can provide enough shear resistance alone. Moreover, UHPC can increase girder span lengths by as much as 73% using the same girder designed with high-performance concrete.
Reid Holland (Thu,) studied this question.