The study on stochastic thermal-hydraulic-mechanical coupled model for porous media with dependent parameters

Abstract:

For the issue of safety assessment of final disposal sites for spent nuclear fuel, the study on coupled thermal-hydraulic-mechanical (THM) simulation seldom discusses the cross-interaction between random variables by using the stochastic concept with dependent parameters. Stochastic analysis is an important issue for the safety assessment whereas the THM system with dependent parameters are close to the reality. Therefore, this study used COMSOL Multiphysics to develop a fully-coupled THM model with dependent parameters and adopted Monte Carlo simulation with hydraulic conductivity being the random variable to assess the covariance functions between all the variables. Geostatistical code, GSLIB, is used to generate the random fields of hydraulic conductivity for a 1D fully-coupled THM model. Three variables in the THM model are displacement, pore water pressure, and temperature, which would interact with each other. The mean and covariance were calculated to evaluate the uncertainty and cross-interaction between these random variables. The results showed that the perturbation of hydraulic conductivity largely affects the variations of pore water pressure but less for displacement and temperature. The effect of dependent parameters is quantified by comparing the results with those of independent parameters. The results showed that the solutions of displacement, pore water pressure, and temperature generally had lower values comparing to those in the independent-parameter model. The independent-parameter model will overestimate the THM responses induced by driven temperature in the coupled THM system. Therefore, the dependent-parameter model should be considered to represent the real situation in the safety assessment of final disposal sites for spent nuclear fuel.

Keywords: Final disposal site for spent nuclear fuel, Fully-coupled thermal-hydraulic-mechanical model, Monte Carlo simulation, Dependent parameter, Covariance function