Control variates with splitting for aggregating results of Monte Carlo simulation and perturbation analysis

Estimation of second-order statistics allows characterizing the uncertainty associated with the response of stochastic finite element models. Two common approaches for estimating these statistics are Monte Carlo simulation and perturbation. The purpose of this paper is to present a framework to aggregate the results obtained by means of these two approaches under the umbrella of Control Variates with Splitting. This allows to produce estimates of the second-order statistics of the system's response with improved precision and accuracy. More specifically, Control Variates is implemented in such a way that the variance of the estimates of secondorder statistics is minimized. In addition, the application of intervening variables for enhancing perturbation is considered as well, showing substantial advantages by increasing the accuracy of the estimates of secondorder statistics. The application of the proposed framework is illustrated by means of an example involving the estimation of second-order statistics of a model involving confined seepage flow.