Random forest-based optimization of UHPFRC under ductility requirements for seismic retrofitting applications

UHPFRC is a material that offers several openings within the building industry. One of those applications is the seismic retrofitting of existing non-ductile concrete structures. This paper aims to optimize a UHPFRC to meet ductility needs for seismic retrofitting applications at a lower cost. To achieve this objective, two Random Forest models were created to estimate the energy absorption capacity and the maximum post-cracking strain. Even though random forest models have been reported as one of the most efficient approaches for forecasting the properties of special concretes, with the purpose of demonstrating the efficiency of the proposed methodology, the random-forest-based optimization results were assessed by experimental works. The experimentally validated results showed that the appropriate combination of high-strength steel microfibers and normal-strength steel fibers with hooked ends meet the limit values (g >= 50 kJ/m(3) and epsilon(pc) >= 0.3%) at a lower cost. (C) 2021 Elsevier Ltd. All rights reserved.

Automated summary

This study aims to optimize UHPFRC for seismic retrofitting applications at a lower cost, with experimentally validated results showing that a combination of high-strength steel microfibers and normal-strength steel fibers with hooked ends can meet target values at a lower cost.