Design expressions for distortional lateral buckling of beams with T-sections

Abstract

T-sections possess unique characteristics that significantly influence their lateral-torsional buckling (LTB) capacity. Compared to typical I-sections, their lower warping constant and single-side stiffened stems make them particularly vulnerable to distortional LTB. Despite these vulnerabilities, favorable properties such as a high minor-to-major moment of inertia ratio can enhance their LTB capacity. Current North American steel design standards (ANSI/AISC360 2022 and CAN/CSA-S16 2019) simplify the critical moment equation by omitting several effects including (1) moment gradient, (2) load height, (3) Pre-Buckling Deformation (PBD), and (4) cross-sectional distortion. These simplifications are shown to lead to inaccurate predictions of the critical moments. A parametric investigation shows that lateral torsional buckling equations for T-sections in North American standards can lead to deviations of more than ±40 % from those predicted by shell-based finite element modeling. Towards developing improved solutions, the present study combines Rayleigh-Ritz approximate techniques with artificial neural network to develop a critical moment expression that incorporates moment gradient, load height, PBD, and distortional effects. An alternative analytical expression is proposed in the form of an interaction equation. The proposed solutions are shown to significantly improve the prediction of critical moments. The potential use of the expression in a design context is illustrated through an example. A comparison with shell finite element analysis reveals that, in a case where design standards overestimate the critical moment by 40 %, one of the proposed solutions provides a conservative estimate within 16 % while the other solution overpredicts the capacity by only 6 %.