Reliability assessment of composite column according to Monte Carlo Simulation and Latin Hypercube Sampling

When a structural element, like a column, is evaluated for resistance using the finite element method or analytical calculation, all parameters related to the resistance side and action effect are treated as deterministic. Nevertheless, in probabilistic analyses, these parameters are regarded as random variables, which are defined using various distributions like the normal distribution, log-normal distribution, beta, or gamma distribution. Monte Carlo Simulation (MCS) and Latin Hypercube Sampling (LHS) are two sampling methods widely used for generating random samples from statistical distribution. The number of samples generated influences the accuracy of the sampling methods especially the Monte Carlo Simulation; generally speaking, the more samples generated, the more accurate the results and this also implies a significant amount of work and computation time. Consequently, the purpose of this paper is to compare the two sampling methods for the reliability assessment of the resistance of the composite column to flexural buckling under compression force.