LAUSR

Abstract This paper proposes a reliable approach to evaluate the probabilistic residual ultimate strength of cracked unstiffened/stiffened plates by coupling nonlinear finite element analysis and Kriging-based reliability method. A numerical simulation scheme is established to evaluate the residual ultimate strength of cracked plates under uniaxial compression loading, considering the influence of uncertain initial imperfections, cracking damages and material parameters. To improve the efficiency of assessment, Kriging-based Monte Carlo simulation method combined with an active learning (AK-MCS) strategy is adopted to achieve the reliability analysis of ultimate strength. Firstly, a deterministic analysis is implemented to explore the influence of basic variables on the ultimate strength reduction characteristics of cracked plates. Then, probabilistic reliability analysis is performed to investigate the evolution of failure probability and reliability index of each uncertain parameter and to evaluate reliability index of design objective for ultimate strength. Sensitivity analysis of uncertain parameters on the reliability index and failure probability is studied, and the result reveals that its variation tendency is consistent with deterministic analysis. The reliability index is achieved by AK-MCS with a few of FEM calls, indicating the proposed probabilistic ultimate strength assessment method is accurate and efficient for structural reliability analysis of cracked plates.