ABSTRACT
Modeling is a tool which aids in assessing and predicting the likely impact of point sources of contamination on groundwater quality. In this chapter the results of recent modeling work on characterising the scales of hydrodynamic dispersion are reviewed. We also show in the chapter the usefulness of analytical models for initial assessment of the impact of a point source, and compare analytical models extensively to two and three dimensional numerical models for a test-case of a landfill-leachate plume in groundwater.
The review highlights the need to consider field-scale aquifer heterogeneity, since it complicates groundwater flow paths and leads to scale dependency of the hydrodynamic dispersion coefficient. Both deterministic and stochastic approaches to dispersion modeling suggest that the dispersion coefficient, although dependent on the scale of measurement, asymptotes at large measurement scales (in the far-field) to a constant value. For predictions close to the point source (in the near-field) fine detail of the variability of the hydraulic conductivity are required to adequately define the dispersion coefficient. It is clear that further research is needed to develop cost-effective means of identifying the appropriate scale of measurement of the hydraulic conductivity in heterogeneous aquifers.
Analytical modeling is shown to allow reliable testing of modeling assumptions. Analytical models are used to aid definition of the finite dimensions of a “point’’ source in relation to plume dimensions, and to show the effect on plume development of the occurrence of boundaries within the flow domain. Also the chapter highlights the advantages of three-dimensional modeling over two-dimensional modeling.
Finally three point-source modeling approaches (three-dimensional semi-analytical: SAM3D, and two- and three-dimensional numerical: MOC and HST3D) are compared for a test case of landfill-leachate contamination of groundwater. Uniform groundwater velocities are assumed but the irregular landfill shape is retained. Significant numerical error is observed for MOC for larger grid sizes (Peclet numbers ≥ 3.5). HST3D displays less numerical error than MOC and gives results comparable to those from the semi-analytical model SAM3D.
