New Formulations for Prediction of Buckling Loads in Steel Plate Girders Through Linear and Nonlinear Stability Analysis

Abstract

Buckling analysis of steel structures can be performed using either linear (LBA) or nonlinear buckling analysis (NLBA). Each method is only effective in particular conditions and improper use might lead to inaccurate predictions. Despite the extensive use of these two methods, there is a lack of research on the accuracy of each method and their limitations regarding the range of application. In this paper, these two methods are evaluated and compared for steel plate girders with a wide range of geometry and the limitations of each method are presented. The numerical modeling of steel plate girders is developed using Abaqus software. The numerical models are verified with the existing experimental results in the literature. An extensive parametric study is conducted using 123 models to investigate the effect of web height, web thickness and beam length on the results. The buckling loads predicted from the numerical models using different analysis methods are compared with those determined based on the AISC 360-16 and EN 1993-1-5 formulations and recommendations are made. Results indicate that while the nonlinear buckling analysis provides accurate predictions of buckling load and mode shape of steel beams, the linear buckling analysis is accurate only for the case of elastic buckling mode without tension field action. A new formulation is proposed to modify the LBA results in order to obtain the advantages of both analysis methods. In addition, based on the NLBA results, a much simple yet accurate equation is proposed to predict the buckling load using plate geometry and material properties instead of the complicated code equations. The accuracy of the proposed equation in comparison with the code equations is also demonstrated for 500 plate girders selected with random geometry.