In glass-bead experiments, Mayer and Miller [1993] reported discontinuous irregular NAPL blobs at residual saturation, with a few large ones containing more than half of the NAPL mass and many more small blobs with lengths spanning two orders of magnitude. The average dimensionless ratio of volume to surface area (1 for a sphere) was 0.6. This suggests the degree of departure from reality in the models of Geller and Hunt [1993] and Powers et al. [1994a], which both assume that blobs are spherical. The former study performed mechanistic analysis and glass-bead experiments to derive a model of nonequilibrium dissolution in homogeneous saturated media. It was pointed out that chemical heterogeneity can develop as more-soluble components dissolve, raising the equilibrium concentration of other components and creating nonequilibrium due to aqueous diffusion. The model developed a parameter called the mass-transfer zone, representing the minimum flow-direction length of a contaminated region in order for NAPL-water equilibrium to be achieved. For typical situations, the mass-transfer zone was predicted by the model to range from millimeters to meters, and it proved to be on the order of 10 cm in the experiments. Powers et al. [1994a] measured mass-transfer coefficients and specific surface areas independently, then assumed NAPL blobs with a range of sphere sizes and developed a model that could be calibrated to sufficiently rich experimental data. For surfactant-enhanced solubility of NAPLs, Abriola et al. [1993] compared calibrated simulations, assuming that trapped residual organics formed spherical globules, to laboratory experiments. Mass-transfer limitations were found to have a large effect on NAPL recovery in both simulations and experiments.
The assumption of spherical blobs was questioned in flow-visualization experiments with etched glass micromodels and short sand columns by Conrad et al. [1992]. Much of the NAPL in the saturated zone was in microscopic blobs, whose size, shape, and distribution affected dissolution rates. Spherical single-pore blobs were observed, but most of the NAPL mass in the residual-saturation zone (i.e., where NAPL was trapped by capillary forces) was contained in complex branching multipore blobs. After passing uncontaminated water through this zone, LEA would dictate that blobs at the upstream end should dissolve completely, leaving a sharp front at the point where the solubility limit was reached. This was not always the case, as mass-transfer rates were limited by diffusion through film surfaces of branched blobs, only the ``heads'' of which were exposed to flowing water.
Powers et al. [1994b] extended their earlier correlation by introducing a parameter corresponding to specific surface area. Previously this had been lumped into the mass-transfer coefficient, as in the correlation of Imhoff et al. [1994], where it could be separated out for a particular sand. Powers et al. [1994b] sought to generalize the separation with the new parameter, dependent on blob geometry and correlated to grain-size distribution. This was limited to homogeneous media and was verified by comparing experiments to numerical simulations incorporating it.