DNAPL Mobility in Heterogeneous Porous Media: Sensitivity of Migration Times to Source Characteristics and Release Location Parameters

This study examined the factors the influence the time required for a release of dense nonaqueous phase liquid (DNAPL) to cease migrating through heterogeneous porous media below the watertable. Using numerical simulation, the temporal and spatial sensitivity of DNAPL migration was evaluated for four DNAPL source characteristics - nonwetting fluid type (i.e., density and viscosity), interfacial tension (IFT), source strength, and volume released - and for three release location parameters - mean permeability, porosity, and hydraulic gradient. The study was conducted using the multiphase code DNAPL3D whose constitutive relationships were developed, and validated, for DNAPL migration in both space and time. All numerical simulations employed a single correlated random permeability field and identical boundary and source conditions to the base case, except for systematic variation of the parameter under investigation. It was found that all of the parameters examined had a significant spatial effect on the final DNAPL migration pattern, either on the overall volume of subsurface invaded (e.g., direct correlation to volume released) or on the amount of lateral spreading (e.g., direct correlation to IFT). However, only two of the parameters were found to have a significant effect on the time required to achieve the final, stable distribution of DNAPL pools and residual. Migration rates were very sensitive to DNAPL type, with predicted cessation times ranging from 30 days for the high mobility fluid tetrachloroethylene (PCE) to over 1000 years for the low mobility fluid coal tar. These simulations reveal that while density primarily influences the spatial extent of penetration and viscosity primarily influences penetration rate, the two effects are not independent due to interactions with site-specific heterogeneity. In addition, the mean permeability of the heterogeneous domain was found to be significant, with increases in mean k corresponding to decreases in both cessation time and degree of lateral spreading.