Reliability and sensitivity analysis of slender steel members utilizing nonlinear finite element analyses.

Several recent studies have revealed, that the structural resistance of stainless steel members under compressive loading determined in accordance with the current European standard does not satisfy sufficiently small level of failure probability. It has been observed, that currently utilized flexural buckling curves overestimate the ultimate structural resistance of stainless steel members of certain material grades and cross-sectional geometries. Not negligible differences have been achieved between the estimations based on standards and experimental data, mainly for specific range of member slenderness values. Compared to carbon steel, stainless steel has higher ductility, and in order to properly describe the stress-strain material behavior, higher numbers of material parameters are required. The subject of this article is to closely investigate the influence of these parameters (material, cross-sectional geometry and also parameters to describe global imperfection) on the ultimate resistance of the members exposed to flexural buckling. Previously validated numerical finite element models of the circular hollow cross-section (CHS) members are created in ANSYS Classic software utilizing parametric APDL macros. Statistical characteristics of all the input parameters along with their mutual correlations are obtained from recent comprehensive statistical research study. The sensitivity analysis is conducted for three different material grades of stainless steel: Austenitic, grade 1.4307, Ferritic 1.4003 and Duplex 1.4462, and for various values of the member slenderness. The purpose of the numerical models is to determine the ultimate limit state of the CHS members exposed to compressive loading utilizing geometrically and materially nonlinear analyses with initial geometrical imperfections. Influence of each input parameter on the model output (ultimate resistance) is investigated, and the results are well documented and discussed. The sensitivity analysis reveals, which of these parameters have the highest influence on the structural resistance, hence are the most significant for the monitored output, considering their statistical distribution and value range. Study also provides data for the subsequent optimization process of the flexural buckling curves, which can be feasibly conducted utilizing the first order reliability method. [ABSTRACT FROM AUTHOR]