Coupled Monte Carlo simulation and Copula theory for uncertainty analysis of multiphase flow simulation models

Abstract

Simulation-optimization techniques are effective in identifying an optimal remediation strategy. Simulation models with uncertainty, primarily in the form of parameter uncertainty with different degrees of correlation, influence the reliability of the optimal remediation strategy. In this study, a coupled Monte Carlo simulation and Copula theory is proposed for uncertainty analysis of a simulation model when parameters are correlated. Using the self-adaptive weight particle swarm optimization Kriging method, a surrogate model was constructed to replace the simulation model and reduce the computational burden and time consumption resulting from repeated and multiple Monte Carlo simulations. The Akaike information criterion (AIC) and the Bayesian information criterion (BIC) were employed to identify whether the t Copula function or the Gaussian Copula is the optimal Copula function to match the relevant structure of the parameters. The results show that both the AIC and BIC values of the t Copula function are less than those of the Gaussian Copula function. This indicates that the t Copula function is the optimal function for matching the relevant structure of the parameters. The outputs of the simulation model when parameter correlation was considered and when it was ignored were compared. The results show that the amplitude of the fluctuation interval when parameter correlation was considered is less than the corresponding amplitude when parameter estimation was ignored. Moreover, it was demonstrated that considering the correlation among parameters is essential for uncertainty analysis of a simulation model, and the results of uncertainty analysis should be incorporated into the remediation strategy optimization process.

http://ift.tt/2eTJKCc