Hyperbox Modelling for Externally Bonded Carbon Fibre Reinforced Polymers on Beams

Abstract

Carbon fibre reinforced polymers (CFRPs) are common retrofitting materials accounting for their high strength, light weight, durability, among others. Due to the lack of a worldwide consensus, much research about externally bonded (EB) FRPs on beams focus on determining the shear capacity contribution (𝑉𝑉𝑓𝑓), in which a parameter called the effective strain (𝜀𝜀𝑓𝑓𝑓𝑓) is often used. The 𝜀𝜀𝑓𝑓𝑓𝑓 is often limited by the governing failure mode (typically debonding). Factors like the complexity of shear phenomenon and composite systems hinder such consensus. Machine learning (ML) applications have been used to model complex behaviour using datasets. A hyperbox modelling ML approach with mixed-integer linear programming (MILP) is used, providing interpretability and versatility in results modelling. This study determines the 𝑉𝑉𝑓𝑓 sufficiency of EB CFRPs on beams while minimizing prediction errors through the 8 rule-based models produced for the EB CFRP configurations.