Numerical Study on Retrofitting of Hot Rolled Steel Beams with Cold-formed Steel Encased Channels-Design Concept using Machine Learning Method

Abstract

The paper presents a numerical investigation on the retrofitting of Hot-Rolled Steel (HRS) beams using Cold-Formed Steel (CFS) encasing channels. The open cross-section HRS channel is transformed into a closed cross-section by encasing the CFS channel. This transformation increases the torsional rigidity of the structural member and helps in reducing the vulnerability of Lateral-Torsional Buckling (LTB). Parametric studies were carried out using numerical analysis. The existing experimental results were used for validation of the numerical model. A total of 600 numerical simulations including design parameters such as thickness of the CFS channel, intermediate spacing between the spot welds (connecting HRS and CFS channels), slenderness ratio, and the cross-sectional dimensions of the HRS beam were considered. The analyses indicated that the effectiveness of the retrofitting increases with an increase in the slenderness ratio of the HRS channel. Machine Learning (ML) method called Symbolic Regression (SR) was used to formulate an equation predicting the increment in the moment capacity as a function of parameters investigated. Finally, a simple design concept is suggested to determine the required CFS channel thickness and intermediate spot weld spacing to achieve the required increment in the moment capacity after retrofitting.