Simulations by Mayer and Miller [The influence of mass transfer characteristics and porous medium heterogeneity on nonaqueous phase dissolution, Water Resour. Res., submitted, 1994] considered the effects of heterogeneity along with mass-transfer models and residual NAPL distribution. Distinctions were drawn between the above one-dimensional column experiments and field results, which tended toward greater nonequilibrium [e.g., Mercer and Cohen, 1990] due to low NAPL saturation, heterogeneity, and aqueous-phase bypassing. Simulations of stratified aquifers differed from homogeneous cases significantly more than did runs using correlated low-variance data from the Borden site. Mass-transfer correlations were compared for homogeneous media, with the one of Powers et al. [1992] showing the only significant deviation from equilibrium. The simulations neglected NAPL dissolution while NAPL was mobile, relying on the brevity of this period, and found that it was important to reduce water relative permeability to account for residual NAPL.
Bypassing was also addressed in three-dimensional laboratory experiments by Anderson et al. [1992a], with water flowing partially around a zone of solvent at stable residual saturation. Because there was little reduction of dissolution due to low permeability in the solvent zone at a flow velocity of 1 meter per day, the authors proposed that below-solubility concentrations in field observations were caused by preferential formation of thin solvent pools, dispersion during transport, and dilution by uncontaminated well water. Another study [ Anderson et al., 1992b] noted that chlorinated hydrocarbon (CHC) solvents, which dissolve quickly in the laboratory, have concentrations far below solubility in field measurements. Using analytical transport models, the paper explained the difference through the accumulation of CHC below the water table in stagnant pools atop a low-permeability layer or aquitard, resulting in control of dissolution by vertical dispersion. Voudrias and Yeh [1994] flowed water horizontally under a light NAPL (toluene) pool and observed large vertical concentration gradients. Borden and Piwoni [1992] and Ptacek and Gillham [1992] compared laboratory-column data and column data from the Borden site, respectively, to simulations with equilibrium and nonequilibrium models; in both cases, the nonequilibrium model produced a better match.