Measuring and modeling the dissolution of nonideally shaped dense nonaqueous phase liquid pools in saturated porous media

[1] A three-dimensional physical aquifer model was used to study the dissolution of a dense nonaqueous phase liquid (DNAPL) pool. The model aquifer comprised a packing of homogeneous, medium-sized sand and conveyed steady, unidirectional flow. Tetrachloroethene (PCE) pools were introduced within model aquifers atop glass- and clay-lined aquifer bottoms. Transient breakthrough at an interstitial velocity of 7.2 cm/h, and three-dimensional steady state concentration distributions at velocities ranging from 0.4 to 7.2 cm/h were monitored over periods of 59 and 71 days for the glass- and clay-bottom experiments, respectively. Pool-averaged mass transfer coefficients were obtained from the observations via a single-parameter fit using an analytical model formulated with a second type boundary condition to describe pool dissolution [Chrysikopoulos, 1995]. Other model parameters (interstitial velocity, longitudinal and transverse dispersion coefficients, and pool geometry) were estimated independently. Simulated and observed dissolution behavior agreed well, except for locations relatively close to the pool or the glass-bottom plate. Estimated mass transfer coefficients ranged from 0.15 to 0.22 cm/h, increasing weakly with velocity toward a limiting value. Pool mass depletions of 31 and 43% for the glass- and clay-bottom experiments failed to produce observable changes in the plumes and suggested that changes in pool interfacial area over the period of the experiment were negligible. Dimensionless mass transfer behavior was quantified using a modified Sherwood number (Sh*). Observed Sh* values were found to be about 2–3 times greater than values predicted by an existing theoretical mass transfer correlation, and 3–4 times greater than those estimated previously for an ideally configured trichloroethene (TCE) pool (circular and smooth). It appeared that the analytical model's failure to account for pore-scale pool-water interfacial characteristics and larger scale pool shape irregularities biased the Sh* estimates toward greater values.