A New Methodology to Predict Cumulative Plastic Ductility Capacity of Steel Buckling-Restrained Braces

This paper presents a database of buckling-restrained brace (BRB) cyclic tests and predicts the cumulative plastic ductility (CPD) capacity of BRBs by using a regression-based machine learning (ML) model trained using the collected data. A summary of the past cyclic tests performed on BRBs is first presented. The hysteretic responses obtained from the test data along with influential, constitutional, and geometrical properties of tested BRBs are leveraged to develop the predictive model for the CPD capacity. The CPD capacity of prototype BRBs predicted using the predictive model proposed here agrees well with the test data, confirming the accuracy and efficiency of the ML-based technique employed here. Such a predictive model can be used in practice to size BRB cores in the preliminary design stage.