ABSTRACT
The compressive resistance of steel plates used in ship structures is studied using probabilistic modeling in an attempt to generate buckling curves that may be used within the framework of risk-based design. Uncertainty in the material properties (elasticity and plasticity) and production-related geometric distortions have been probabilistically modeled by considering realistic statistical structures. For time efficiency, a pre-defined number of detailed non-linear finite element models have been constructed and simulated based on design of experiments (DOE) sampling schemes in accordance with the response surface methodology. The generated surrogates were first validated and then used within a Monte Carlo Simulation (MCS) framework for uncertainty propagation. The probabilistic resistance of the plate populations of interest was obtained by identifying a parametric statistical model based on the sample statistics. The obtained statistical structures al-lowed for the analysis of the probabilistic buckling strength of steel plates, which further enables the comparison with empirical formulations found in rules and in the scientific literature.
