Date of Award
Master of Science (MS)
Macrodispersivity estimates are crucial in predicting the subsurface transport of contaminants. These predictions have become increasingly important as the number of contaminated sites continues to increase. In addition, the longevity of many contaminants can affect water quality over extended periods of time.
Anionic tracers such as bromide and chloride are commonly used as groundwater tracers in dispersivity experiments because they are inexpensive, easily detected, and are commonly believed to behave conservatively. It has been found, however, that anionic tracers do not always behave conservatively; sediments with a net positive surface charge can adsorb some-of the tracer.
A weakly non-linear, experimentally derived Freundlich sorption isotherm was found to describe bromide adsorption in sediments from the Savannah River Site in South Carolina. The aim of this research was to quantify the effects of this sorption on apparent dispersivities relative to a conservative tracer. In addition, the effects on apparent dispersivities of injecting different amounts of tracer and omitting tracer concentrations below detection limits were evaluated.
Monte Carlo simulations were performed using SUTRA and W ATSUTRA. The models were developed to simulate natural gradient tracer tests. Two types of hydraulic conductivity fields were generated using the Fast Fourier Transform method. One represented a weakly heterogeneous flow field, with physical parameters similar to the Borden aquifer in Ontario, Canada; the other represented a strongly heterogeneous flow field with physical parameters similar to the Columbus aquifer in Mississippi. These hydraulic conductivity fields represent two extremes of aquifer conditions.
Forty simulations were performed in this study for eight cases. Five realizations were performed for each case; each realization was simulated for two years. Apparent dispersivities were calculated by the method of moments. Ensemble means were estimated by averaging the apparent longitudinal or horizontal transverse dispersivity values for the five realizations in each case. Null hypothesis testing was performed with a standard t-test to test the significance between the estimates for the ensemble means for various cases.
The following simplifying assumptions were used in this research:
1) Sorption according to the Freundlich isotherm was assumed to be a reversible equilibrium process. Thus, desorption followed the same isotherm as adsorption, only with decreasing concentrations, and both adsorption and desorption were rapid in comparison to advective flow.
2) The Freundlich isotherm was constant throughout the model domain.
3) The groundwater model employed was two-dimensional; vertical heterogeneity was not represented.
The results indicated the experimentally derived Freundlich isotherm, given the assumptions and limited number of realizations used in this study, did not significantly affect apparent plume dispersivity values. Thus, bromide was not found to be an unsuitable groundwater tracer. The quantity of tracer mass injected and omission of tracer concentrations below detection limits also did not have a significant effect on apparent dispersivity values. It was found that using tracer concentrations below detection level can adversely affect high-order spatial moments; this finding may, however, be unique to SUTRA and W ATSUTRA or similar finite-element programs.
Dodak, Eron J., "The effects of non-linear bromide adsorption on apparent macrodispersivity" (1996). Theses and Dissertations. 74.