Sensitivity Analysis of Geometrically Imperfect Single-storey Steel Frame Structure - A Comparison of Two Loading Modes

In this study, two methods to numerically analyse a single-storey vertically loaded steel frame structure with initial geometrical imperfections are compared. The first method is deterministic, where the initial imperfections, sway of the frame and local bow imperfections of the columns are based on the corresponding European standard to design steel structures, Eurocode 3 (EC3). The second, probabilistic method, where the imperfections are defined by the random stochastic parameters is using the first order reliability method (FORM) along with numerous numerical finite element analyses in order to estimate the ultimate resistance of the structure. In this FORM method, the statistical values of these input imperfections are derived from the European standard for allowed erection and manufacturer tolerances, and these data corresponds with the experimentally measured imperfections on real structures. Material parameters, as Young’s modulus and yield stress are also considered as stochastic variables. Design ultimate resistance based on EC3 is compared with the 0.1% quantile of the stochastic ultimate resistance of the FORM method. In general, assumptions of the deterministic EC3 approach are sometimes considered as overly conservative. The main objective of this study is to verify and evaluate these assumptions by comparison with more precise probabilistic method. Moreover, for both methods (EC3 and FORM), the resistance is determined under two loading modes, one by increasing of the force load, the other by prescribed displacement. The loading conditions of these two loading modes are applied analogically to each other, hence similar resistances for the corresponding method are expected. However, this assumption needs to be verified within the probabilistic analysis conditions, what is the secondary objective of this paper.