Simulation of a field scale tritium tracer experiment in a fractured, weathered shale using discrete-fracture/matrix-diffusion and equivalent porous medium models

Abstract

Simulations of a tritium tracer experiment in fractured shale saprolite, conducted at the Oak Ridge National Laboratory, were performed using 1D and 2D equivalent porous medium (EPM) and discrete-fracture/matrix-diffusion (DFMD) models. The models successfully reproduced the general shape of the breakthrough curves in down-gradient monitoring wells which are characterized by rapid first arrival, a slow-moving center of mass, and a persistent ``tail`` of low concentration. In plan view, the plume shows a large degree of transverse spreading with the width almost as great as the length. EPM models were sensitive to dispersivity coefficient values which had to be large (relative to the 3.7m distance between the injection and monitoring wells) to fit the tail and transverse spreading. For example, to fit the tail a longitudinal dispersivity coefficient, {alpha}{sub L}, of 0.8 meters for the 2D simulations was used. To fit the transverse spreading, a transverse dispersivity coefficient, {alpha}{sub T}, of 0.8 to 0.08 meters was used indicating an {alpha}{sub L}/{alpha}{sub T} ratio between 10 and 1. Transverse spreading trends were also simulated using a 2D DFMD model using a few larger aperture fractures superimposed onto an EPM. Of the fracture networks studied, only those with truncated fractures caused transverse spreading. Simulated tritium levels in all of the cases were larger than observed values by a factor of approximately 100. Although this is partly due to input of too much tritium mass by the models it appears that dilution in the wells, which were not purged prior to sampling, is also a significant factor. The 1D and 2D EPM models were fitted to monitoring data from the first five years of the experiment and then used to predict future tritium concentrations.