Evaluating and applying contaminant transport models to groundwater systems

Abstract

This thesis examines the use of random walk techniques to model the transport of a contaminant in groundwater. These techniques involve the distribution of a plume of contaminant into a discrete number of particles. These particles are then individually subjected to advective and dispersive forces and their progress though the model domain tracked over time. In general, pollution of groundwater is characterised by: 1) being difficult to detect, 2) being complicated and expensive to investigate and monitor, and 3) being expensive to clean up. These factors make the modelling of groundwater contamination an important area of investigation. This thesis presents the principles of groundwater flow and contaminant transport, along with their governing equations (namely, the groundwater flow equation and the advection dispersion equation); methods for the solution of the advection dispersion equation are discussed. These methods include analytic, finite difference and random walk techniques. Three random walk techniques are presented and compared with the analytic solutions for the following cases: 1) one dimensional dispersion 2) one dimensional advection dispersion 3) two dimensional dispersion 4) two dimensional advection dispersion Results of the comparisons have showed that all three random walk schemes presented produce computed results which are consistent with the analytic solution in each of the cases considered. Two finite difference schemes are presented and applied to the case of two dimensional advection dispersion. Through doing so, the problem of numerical diffusion has been highlighted. Random walk techniques have been applied to two physical problems. In the first case, a model has been developed to simulate the movement of a plume of chloride in an aquifer in the province of Saskatchewan, Canada. Results are compared for each of the three random walk schemes, namely time histories and breakthrough curves which plot the concentration of particles at a location in space over the time period modelled. All three random walk techniques have produced results that are very similar, with each modelling the movement of the plume acceptably. The second model uses data from The South Australian Department of Mines and Energy to simulate the movement of salt in the groundwater in a vine growing region near Naracoorte, South Australia. This model produces results which are consistent with available measured data.