Upscaled Mass Transfer Correlations for Estimating Mass Discharge From DNAPL Source Zones: Comparisons to Field-Scale Numerical Simulations

In recent years, multiphase, multicomponent numerical simulators have been used in the literature to enhance our understanding of the potential distribution and persistence of dense non-aqueous phase liquids (DNAPLs), such as tetrachloroethene (PCE), in plume source regions. Incorporation of hysteretic capillary pressure - saturation relationships, nonuniform flow fields, and rate-limited mass transfer between phases in these simulators facilitates prediction of realistic, spatially variable DNAPL saturation distributions, and enables quantification of down-gradient aqueous-phase contaminant concentrations over the life of a source zone. Application of such numerical models, however, often requires extensive user training and a large amount of site-specific information. In contrast, upscaled mass transfer correlations have been used with simplified one-dimensional analytical solutions to approximate the temporal evolution of a source zone and resulting contaminant mass discharge. This work compares predictions of simplified methodologies for estimating the mass discharge from a PCE-DNAPL source zone with those generated using a three dimensional multiphase, multicomponent simulator. In the numerical simulator, local-scale mass transfer and mixing due to a nonuniform flow field are modeled directly. In the simplified approach these effects are consolidated into an upscaled mass transfer coefficient. Preliminary results for saturation distributions dominated by low saturation ganglia indicate that, for moderate levels of DNAPL mass removal (less than 80 percent), predictions of mass discharge using an upscaled mass transfer correlation are generally within 5 percent of those computed using the numerical simulator. However, as ganglia dissolve and the source-zone architecture becomes dominated by DNAPL pools (typical of higher mass removal conditions), the simplified methodology over-predicts mass discharge by a factor of 2 or more, resulting in the under-prediction of source longevity. These effects become more exaggerated as the fraction of pools in the initial saturation distribution increases. Thus, understanding the limitations of the upscaled modeling approach at high levels of DNAPL mass removal will be critical to the accurate estimation of long-term mass discharge. Simulations suggest that late-time predictions of contaminant mass discharge may be improved by incorporating additional information about the source zone into the upscaled mass transfer correlation; however, more work will be needed to confirm this relationship.