Database-Driven Modeling Error Assessment of Shear Design Methods for UHPC Flexural Members without Shear Reinforcement

Abstract In recent years, a significant amount of experimentation has been conducted to evaluate the shear behavior of ultra-high performance concrete (UHPC) flexural members. Several codal procedures, including Swiss society of engineers and architects (SIA) 2052:2016, French national standard (NF)-P18:2016, Korean ultra-high performance concrete (K-UHPC) 2014, China Association for Engineering Construction Standardization (CECS) 38:2004, e.construct prestressed concrete institute (eCPCI) 2021, and federal highway administration (FHWA) 2023, have also been devised to predict the shear capacity of UHPC flexural members. In this study, a qualified database of shear test results was extracted to estimate the statistics of the modeling error defined as the ratio of experimentally obtained shear capacity and shear capacity predicted using design procedures from different codal provisions available for UHPC. Initially, conversion factors for compressive strength among different geometries were established based on the available data to ensure the application of compressive strength of a specific specimen geometry for a particular design procedure. Subsequently, the test data were critically reviewed, and criteria were developed to achieve a reliable database. The refined database comprised 119 test specimens that exhibited shear-dominated failure. However, experimentally obtained tensile properties and shear crack inclination angles, essential for the evaluation of various shear design provisions, were available for only 85 tests. For each shear design method, the modeling error was quantified, and its statistical parameters—mean and coefficient of variation—were computed to assess accuracy and consistency. These results can be used to perform a reliability-based evaluation of structures designed according to the codal provisions, which can be a baseline for reliability-based calibration.