Determining dispersion parameters to predict ground-water contamination

Abstract

Use of contaminant transport models to predict the potential for ground-water pollution at waste disposal sites has been hindered by lack of methods to determine dispersion parameters. As an aid to overcome this difficulty, methods have been developed in this study that permit the determination of dispersion parameters in a three-dimensional spreading field. Methods treat instantaneous and continuous source conditions, entail simplified analysis techniques and require a minimum number of observation points. Ananlyses are simplified by embedding two unknown parameters into an experimentally determined variable, the maximum observed concentration along the centerline of flow. Importantly, methods permit determining dispersion parameters independently of initial concentration and source volumetric flow rate, two parameters that are seldom known. To illustrate applications of methods, analyses were conducted using hypothetical tracer test and contamination cases. Literature data taken from a field tracer test and a contamination history were also analyzed. Methods were shown to simplify the determination of dispersion parameters and values obtained agreed closely with those determined in the literature studies. Also, methods were shown to permit estimation of contaminant velocity and retardation factor. Methods have been used to develop new techniques for determining transverse dispersion coefficients in laboratory column experiments. Experiments were performed in a column packed with glass beads and the transverse dispersivity was determined to be on the order of 10('-3) cm. This result agrees closely with flow tank test results for similar materials reported in the literature. Calculations and experiments suggested that the dispersion scale effect, generally attributed to heterogeneities, may be explained in part as an artifact of models used in field tracer experiments and contamination simulations. Analysis indicated that a scaling up of dispersivity will occur whenever an n-1 dimensional model is calibrated or otherwise employed to describe an n dimensional system. Scaling has also been shown to occur when point source models are used in the analysis of finite source conditions. The magnitude of induced scaling appears to be sufficient to encompass the many orders of magnitude range of differences between laboratory measurements of dispersivity and those reported for model calibrations.