Web crippling strength of cold-formed steel sigma beams with openings: numerical simulation and machine learning approach

Cold-formed steel (CFS) Sigma sections are widely used in modern industrial and residential construction. They are known for their enhanced load-carrying capacity, torsional rigidity, and lightweight profile compared to counterparts like Z and C sections. Despite these advantages, CFS elements are susceptible to web crippling, a critical failure mode influenced by the presence of openings. Modern design codes and specifications provide expressions for evaluation of web crippling strength of conventional Z and C CFS sections. However, they do not provide any expressions for evaluation of web crippling strength of CFS sigma beams especially with openings in their web. Motivated by the aforementioned advantages of CFS sections in general and CFS sigma sections in specific along with the lack of understanding of the buckling behavior of these sections especially with web openings, this study aims to numerically investigate the web crippling strength of CFS sigma beams with openings. Starting with the development and validation of a detailed finite element model (FEM) against experimental data, over four thousand parametric simulations are conducted to explore the effect of different geometric and loading parameters on web crippling strength of CFS sigma beams. The studied parameters include the beam height (h), cold formed plate thickness (t), flange width (b), lip length (S), the bearing length (i), number of holes (n), and radius of the hole to the beam height (r/h). Subsequently, machine learning (ML) techniques are applied to train five ML algorithms using the extensive dataset generated by the conducted parametric study. These ML models aim to accurately predict the web crippling strength based on geometric parameters and loading conditions, providing insights that can refine current design guidelines and enhance structural engineering practices in the use of CFS sigma sections with openings.