This review addresses modeling of subsurface hydrological systems in which contaminants are transported in more than one fluid phase. The primary focus is on problems involving nonaqueous-phase liquids (NAPLs), a dynamic gas phase, or both, so that the unsaturated and saturated zones are both discussed. Basic research in simpler settings is included if it develops concepts that could be applied in modeling of such problems.
Recent developments in practical models are reviewed, along with experimental work and theoretical issues related to the formulation of mathematical models. It will be seen that the extent of validity of the local equilibrium assumption (LEA) for component partitioning among phases is a major question influencing the direction of these formulations. Also emphasized are upscaling to macroscopic and grid lengths, and the choice and coupling of primary variables in multicomponent systems. Relations to and possible use of developments in petroleum reservoir simulation are discussed.
The significance of the type of modeling reviewed here is growing, as the last several years have seen increased awareness of the complexity and difficulty of groundwater contamination problems that involve multiphase flow. For example, emplacement of NAPLs, such as gasoline or trichloroethylene, in the subsurface typically entails downward migration through the vadose zone, leaving some (or all) NAPL mass trapped behind by capillary forces in globules of widely varying shape and size. Depending on the NAPL density and other factors, the remaining mass will float at the water table or continue down into the saturated zone. Some contamination can occur via transport of the NAPL phase itself, but even when this phase reaches immobile residual saturation, its components can dissolve in groundwater in concentrations well above environmental thresholds. These components can then be transported in the water phase to contaminate water far from the NAPL source. Many such components, known as volatile organic compounds (VOCs), will also vaporize into the air phase in the unsaturated zone and can then be transported in that phase.
Conventional pump-and-treat techniques will not readily remove immobile NAPL from the subsurface, where it may persist for decades as a source of contamination. Other mechanisms, such as interphase mass transfer, must be exploited. One example of such a process is vapor extraction, pumping air through the contaminated zone and relying on volatility to partition some mass of the offending components into the air, which is then extracted from the subsurface. This technique may be enhanced by injection of steam, if the chemicals are such that a higher temperature will yield a more favorable partitioning. Partitioning can be quite sensitive to the multicontaminant compositions of the phases present. This discussion is merely illustrative, not exhaustive; for the purpose of this review, it suffices to realize that the whole range of mechanisms of multiphase multicomponent flow in porous media, including viscous, capillary, and gravity forces, diffusion, dispersion, and component partitioning, and reactive processes, are expected to be important in the remediation strategies of the future.
To keep within reasonable bounds, the scope of this review is limited, perhaps somewhat arbitrarily. Exhaustive literature surveys of this and other areas were compiled by Mayer et al. [1992, 1993, 1994]. Important topics not discussed here, except as they surface in papers within the scope delineated at the outset, include geostatistics, stochastic hydrology, flow and transport in fractured media, pore-scale models, unsaturated-zone modeling (based on Richards' approximation), and biorestoration modeling. Except for the last, these topics are addressed by other reviews in the present U.S. National Report. Microbial biorestoration is coupled to multiphase models, but is considered here to be more specialized than the subjects under review and so is not emphasized. ``Multiphase'' is taken to refer to multiple fluid phases, so that sorption by itself is not reviewed here, though it may appear in a wider context or in papers considered relevant to multiphase modeling. Unsaturated-zone modeling that allows for an air-pressure gradient is within the scope.
The review is organized as follows. The next section outlines some major issues in multiphase multicontaminant transport modeling to lay the groundwork for what follows. Further sections deal with physical processes, theoretical upscaling, and formulation of equation systems, primarily at the level of basic research, after which recent published models and their numerical methods are discussed. Finally, developments are summarized and directions for future research are suggested.