USE OF MODELS IN DNAPL SOURCE ZONE REMEDIATION ASSESSMENT: EXPLORING THE POTENTIAL BENEFITS OF PARTIAL MASS REMOVAL IN NONUNIFORM FORMATIONS

Following the accidental subsurface release of a dense nonaqueous phase liquid (DNAPL), spatial variability of physical and chemical soil properties can exert a controlling influence on infiltration pathways and organic entrapment. DNAPL spreading, fingering, and pooling typically result in source zones characterized by irregular contaminated regions of variable saturation with complex boundaries. Spatial variability in aquifer properties also influences subsequent DNAPL dissolution and aqueous transport dynamics. Increasing interest in the evaluation of the benefits of partial mass removal from such DNAPL source zones has led to intensified efforts to model multiphase flow and transport behavior in these heterogeneous systems. This presentation provides an overview of recent advances in the application of multiphase models to source zone remediation assessment.

Ensembles of aquifer realizations were used in 2D simulations of DNAPL entrapment and mass recovery to explore the influence of physical and chemical property correlation on entrapment and remedial performance metrics. Numerical simulations for non-uniform, water-wet systems revealed notable differences among predicted metrics, linked to the correlation between physical aquifer properties. In particular, realizations that employed Leverett scaling behaved differently from realizations using capillary properties derived directly from grain size distributions. DNAPL behavior in mixed wettability systems was also evaluated, incorporating wettability-dependent constitutive relationships. The mass fraction of organic-wet aquifer material was correlated to aquifer permeability. Simulations revealed considerable differences in predicted source zone architecture that were shown to correspond to subsequent differences in remedial performance metrics. Metrics from individual realizations, however, did not exhibit a strong correlation with formation properties. Work is also underway to assess the influence of dimensionality on model predictions. Comparisons of DNAPL distribution and dissolution predictions in 2D and 3D suggests that 2D models can adequately represent ensemble DNAPL infiltration and entrapment behavior but may lead to conservative estimates of DNAPL recovery.