Machine learning model for patch loading buckling coefficients for longitudinally stiffened curved plate girders

Slender plate girders are frequently employed in the construction of steel bridges due to their resistance to bending and light weight in comparison to reinforced concrete beams. In some cases, the webs of these girders are horizontally curved in order to overcome limitations presented during the installation. This paper aims at investigating the elastic buckling capacity of horizontally curved, longitudinally stiffened, steel plate girders subjected to patch loading. A linear buckling analysis is performed using the finite element method. Thereafter, a parametric analysis is conducted to investigate the effect of the girder curvature, the position and size of the stiffener, and the loading length. The results show that the buckling coefficients increase with both the girder curvature and size of the stiffener. In the end, an expression for the patch loading buckling coefficient is obtained through symbolic regression. Ultimately, two analytical expressions for the patch loading buckling coefficient (kF) are proposed: one for unstiffened girders, and another for stiffened girders, both derived through symbolic regression. The comprehensive results highlight the effectiveness of machine learning (ML) approaches in predicting buckling coefficients.