Stochastic analysis of dense nonaqueous phase liquid dissolution in naturally heterogeneous subsurface systems

Abstract

Field-scale Dense Nonaqueous Phase Liquid (DNAPL) dissolution in three-dimensional heterogeneous subsurface systems is investigated using a stochastic approach that treats the variability of flow properties as three-dimensional random fields. A steady-state, quasi-static DNAPL saturation distribution in a source zone is derived, based on the previous research to describe the field-scale nonuniform residual DNAPL distribution. A local-scale dissolution model is generalized from the laboratory experimental results. Effective transport and dissolution properties are obtained by a stochastic analysis, which includes nonstationarity in the concentration field to address both boundary and downstream effects. An extrapolation of the effective properties is performed for the large spatial variability of hydraulic and dissolution parameters. The extrapolation is evaluated using a unique two-zone model that simplifies the continuous DNAPL distribution to two components: low permeability lens zone with high DNAPL saturation and a surrounding permeable zone with low DNAPL saturation. The agreement between the two-zone model results with stochastic solutions demonstrates the adequacy of the extrapolation of the latter. Four field sites with different geological settings and hydraulic characteristics are evaluated using effective properties: the Borden site, the Cape Cod aquifer, the Savannah River Site and the Hanford site.

(cont.) The theoretical prediction is compared with the DNAPL concentration data from the Hanford site, where millions of pounds carbon tetrachloride were dumped in disposal facilities. In spite of the error and uncertainty involved in the field data, there is reasonable agreement between the field observations and the predicted mean DNAPL concentration field. It is concluded that the dominant factor affecting the field-scale DNAPL dissolution is the variability of the dissolution rate coefficient, which is a function of spatial distribution of DNAPL and permeability. The bypassing effect, reflecting the diversion of water flow around zones of high DNAPL saturation with low aqueous relative permeability, is another important factor that can reduce the effective dissolution rate significantly. The limitations of the study are discussed regarding the data collection and further evaluation of the extrapolation.