Shear Buckling Analysis of Corrugated Web Steel Plate Girder with Random Material Properties

Abstract

The non-linear finite element method with initial geometric imperfection is compulsory to capture the shear buckling behavior of the Corrugated Web Steel Plate Girder (CWSPG). These initial geometry imperfections can come from the slender structure that cannot maintain its perfect shape or lousy quality during the assembly process. Most researchers generate the initial geometry imperfection from the elastic buckling modes that may not represent the randomness in the geometric imperfection. Therefore, there is a need to develop a method to generate random initial geometry imperfection without carrying out elastic buckling modes from the analysis. This paper investigated the shear buckling behavior of CWSPG using non-linear finite element analysis and proposed a method to generate the initial geometric imperfection using the random material imperfection. The random material properties for each meshed element follow a statistically random normal distribution of the material yield strength. The initial geometric imperfection is generated from the first second-order analysis with random material properties (using the in-house 3D-NLFEA package) to the point where the expected buckling shape is obtained. The final deformed geometry from the first second-order analysis is then scaled down to be used as the initial geometric imperfection. The proposed method requires the scaling value such that the first buckling load from the available experimental test result and the one from the numerical model are at the same level. The proposed method was able to capture the shear buckling behavior of the CWSPG and was sensitive to the element’s size. The minimum size of the element required normalized with the element thickness was found to be less than four to maintain the accuracy for both the peak and residual load of the CWSPG specimen. The proposed method shows excellent agreement in predicting the peak load and the post-buckling behavior of the available test results. Therefore, the proposed method can be used as an alternative method to capture the buckling load of the CWSPG specimen.