Application of LPSS-EIS method for calculating failure probability of steel trusses using direct analysis

Abstract

Steel truss structures are very popular nowadays thanks to their efficient use of materials, good span, and beautiful and diverse designs. Along with the strong application of steel truss structure, direct analysis techniques are interested in researching and applying more and more to the design of this type of structure to more accurately evaluate structural nonlinear behaviors. Because the design parameters and materials are random in the reality, the structural failure probability of the structure is always considered in the design to ensure the reliability of the structure during use. In this study, the hybrid algorithm combined two techniques of "Latinized" Partially Stratified Sampling (LPSS) and effective importance sampling (EIS) LPSS-EIS is applied to determine the failure probability of the steel truss system using direct analysis. The effectiveness of the method is compared with the commonly used techniques of Monte Carlo Simulation (MCS) and LPSS. The analysis results of the 72-bar space steel truss show that LPSSEIS is very effective when accurately calculating the failure probability of the structure with a much smaller COV than the MCS and LPSS algorithms.

Abstract: Truss; Advanced analysis; Probability; Monte Carlo; Latin Hypercube.