Accurate prediction of the moment-rotation behaviour and deformation capacity of Cold-Formed Steel (CFS) members is essential for their design and performance assessment. However, reliable prediction of these characteristics remains challenging for slender sections due to complex interaction of the cross-sectional constituent plates and the resulting nonlinear post-buckling behaviour. Existing codified formulations, such as those in ASCE 41-23, rely primarily on cross-sectional local slenderness, which cannot fully capture the underlying mechanisms. This paper develops improved prediction methods for the local buckling flexural response of CFS members. To this end, detailed Finite Element (FE) models of CFS beams, incorporating material nonlinearity and geometric imperfections, were developed and validated against experimental results, and subsequently used to generate a comprehensive numerical dataset. This dataset was employed to assess the accuracy of the ASCE formulations, demonstrating that cross-sectional local slenderness alone cannot adequately characterise the flexural response of CFS members, leading to significant inaccuracies, particularly in post-buckling deformation predictions. Instead, the response is strongly influenced by the slenderness of individual plate elements, which governs both local buckling and post-buckling behaviour. Based on these findings, new ASCE-compatible predictive equations were proposed by incorporating web and flange slenderness parameters together with material strength characteristics. The proposed equations retain the practicality of existing design provisions while significantly improving prediction accuracy, reducing the average error in ultimate rotation estimates from approximately 62% to 13%. Finally, complementary Machine Learning (ML) models are developed to further enhance the predictive accuracy and are implemented within an open-source software platform for practical application.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)