An efficient D-vine copula-based coupling uncertainty analysis for variable-stiffness composites

Abstract

This study suggests a coupling uncertainty analysis for investigating the uncertainty of stiffness characteristics for variable-stiffness (VS) composites. The uncertainty analysis is based on the Monte Carlo Simulation (MCS) and a novel one-step Bayesian copula model selection assisted D-vine sampling method (OBCS-D) is proposed to realize the coupling of random variables. Compared with other uncertainty analysis methods, the suggested method is capable of identifying suitable copula function and marginal cumulative distribution function (CDF) of random variables and obtain the correlated random variables. Because the coupling uncertainties are considered, the computational cost of the uncertainty analysis is significantly increased. Therefore, a fast solver reanalysis method for VS composite materials is developed to improve the efficiency of the evaluation. To further improve the efficiency of the analysis, a surrogate-based MCS is also developed. To select a suitable surrogate modeling method, multiple popular modeling techniques are compared. Finally, fiber angle deviation of the VS composite is investigated by the suggested strategies. Two numerical examples are presented to verify the feasibility of the suggested uncertainty analysis framework. Results show that the stiffness of VS composite is sensitive to fiber angle and the uncertain analysis is necessary and important for the analysis of VS composites.