Geostatistical and Stochastic Study of Flow and Transport in the Unsaturated Zone at Yucca Mountain

摘要: Yucca Mountain has been proposed by the U.S. Department of Energy as nation’s long-term, permanent geologic repository for spent nuclear fuel or high-level radioactive waste. The potential would be located in Mountain’s unsaturated zone (UZ), which acts a critical natural barrier delaying arrival radionuclides to water table. Since radionuclide transport groundwater can pose serious threats human health and environment, it is important understand how much fast travel through UZ groundwater. system consists multiple hydrogeologic units whose hydraulic geochemical properties exhibit systematic random spatial variation, heterogeneity, at scales. Predictions under such complicated conditions are uncertain, uncertainty complicates decision making risk analysis. This project aims using geostatistical stochastic methods assess flow Mountain. Focus this study parameter UZ. parametric arises since limited measurements unable deterministically describe variability parameters. In project, matrix porosity, permeability sorption coefficient reactive tracer (neptunium) treated variables. Corresponding propagation quantitatively measured mean, variance, 5th 95th percentiles simulated state variables (e.g., saturation, capillary pressure, percolation flux, time). These statistics evaluated Monte Carlo method, three-dimensional model implemented TOUGH2 code executed with realizations specifically studies caused multi-scale heterogeneity layer local Typically, Mountain, scale refers layers layer-wise average properties, variation within layer. While most have conducted scale, also important, affects path, velocity, time radionuclide. report first presents layer-scale permeability, coefficients tracers. Homogeneous fields parameters generated each simulations. referred homogeneous case. To local-scale sequential Gaussian simulator (SGSIM) GSLIB (Deutsch Journel, 1998) used generate heterogeneous layer, simulations conducted. cases. For cases, estimated assessment. addition, two cases compared investigate effect on transport. It found that increased predictive flux cumulative mass computational blocks below footprint repository, whereas mean predictions hardly affected. significantly times table both conservative early simulation period, delayed, influence diminishes during late period. Simulated more realistic than those one- dimensional models, due numerical characterize hydrological Therefore, we expect results directly facilitate DOE site performance analysis making.