UNCERTAINTY ANALYSIS OF RADIONUCLIDE TRANSPORT IN THE UNSATURATED ZONE AT YUCCA MOUNTAIN

This study is to assess parametric uncertainty of radionuclide transport in the unsaturated zone at Yucca Mountain using Monte Carlo method. The numerical code TOUGH2 is employed to simulate unsaturated flow and radionuclide transport. Matrix porosity, saturated hydraulic conductivity, and sorption coefficient are treated as statistically homogeneous random variables. Distributions of the three random parameters variables are determined based on site measurements, which are transformed by seven transformations including three transforms from the Johnson system (Carsel and Parrish, 1988). The transformed measurements are fitted to the normal distribution and the transformation giving that the best fit is selected using the Lilliefors test. For each of the three random parameters in each model layer, the Latin Hypercube Sampling (LHS) is used to generate 200 realizations based on the mean and variance of transformed measurements. The correlation between matrix porosity and permeability is incorporated in the random field generating by specifying the Spearman rank correlations calculated from the transformed measurements. 200 Monte Carlo simulations are conducted using the TOUGH2 code and the convergence of the Monte Carlo approach is thoroughly examined. Mean, variances, 5% and 95% percentiles of saturation, capillary pressure, and fluxes are estimated based on the 200 realizations. The 5% and 95% percentiles of saturation and capillary pressure bracket a large portion of site measurements, indicating the success of the uncertainty analysis. Uncertainty of the reactive and conservative tracer transport in the vadose zone is represented by the mean, 5%, and 95% percentile of the breakthrough curves. This study presents a comprehensive method of assessing radionuclide transport at the unsaturated zone at Yucca Mountain by rigorous random field generation and thorough Monte Carlo simulation.