Machine learning approach for predicting fire resistance of FRP-strengthened concrete beams

Abstract

This paper presents a data-driven machine learning (ML)-based approach for predicting the fire resistance of fiber-reinforced polymer (FRP)-strengthened reinforced concrete beams. To this end, a comprehensive database of fire tests on FRP-strengthened concrete beams reported in literature was compiled. The database comprised of varying geometric and material properties, sectional dimensions, steel reinforcement, FRP layers, and thermal insulation, as well as the load level. The compiled database was analyzed using three different ML algorithms including, support vector regression, random forest regressor, and deep neural network. The outcome of the analysis is a computational-based ML approach for predicting fire resistance of FRP-strengthened concrete beams. The results predicted from the ML approach were compared with those measured in fire tests to show reliability of the proposed approach. The results infer that the proposed ML approach can effectively predict the fire resistance of FRP-strengthened concrete beams with varying cross sections, properties of constituent materials, and applied loading.