Machine learning framework for predicting failure mode and shear capacity of ultra high performance concrete beams

Abstract

This paper presents a data-driven machine learning (ML) framework for predicting failure mode and shear capacity of Ultra High Performance Concrete (UHPC) beams. To this end, a comprehensive database on 360 reported tests on UHPC beams with different geometric, fiber properties, loading and material characteristics was collected. This database was then analyzed utilizing different ML algorithms including, support vector machine (SVM), artificial neural networks (ANN), k-nearest neighbor (k-NN), and genetic programing (GP), to identify key parameters governing failure pattern and shear capacity of UHPC beams. The outcome of this analysis is a computational-based ML framework that is capable of identifying failure mode of UHPC beams and simplified expressions for predicting shear capacity of UHPC beams. Predictions obtained from the proposed framework was compared against the values obtained from design equations in codes, and also results from full-scale tests to show the reliability of the proposed approach. The results clearly infer that the proposed data-driven ML framework can effectively predict failure mode and shear capacity of prestressed and non-prestressed UHPC beams with varying reinforcement detailing and configurations.