Assessing Source Zone Mass Depletion in Heterogeneous Media: Application of a Multi-Rate-Mass-Transfer Approach Based on a Geostatistical Medium Description

Wednesday, 17 December 2014
Maria T Elenius, Eric L Miller and Linda M Abriola, Tufts University, Medford, MA, United States
Chlorinated solvents tend to persist for long periods in heterogeneous porous media, in part due to sustained sources of contaminant sequestered in lower permeability zones and sorbed to the soil matrix. Sharp contrasts in soil properties have been modeled successfully using Markov Chain / Transition Probability (MC/TP) methods. This statistical approach provides a means of generating permeability fields that are consistent with prior knowledge concerning the frequency and relative positioning of different strata.To assess source zone mass depletion in a suite of such geological realizations, the large computational burden may prohibit the use of direct numerical simulations. One alternative approach is the application of a multi-rate-mass-transfer (MRMT) method, an extension of the dual-domain concept that was first developed in the soil science literature for sorption modeling. In MRMT, rather than discretizing immobile regions, such as clay layers, the concentration in these regions is treated by explicit state variables, and the transport between mobile and immobile regions is modeled by first-order exchange terms. However, in the implementation of this approach, fine-scale simulations on subdomains are often necessary to develop appropriate parameters. Such simulations are tedious, especially when attempting to account for uncertainty in the geological description. In this work, the link between characteristics of MC/TP and transfer parameters in the MRMT is evaluated by regression based on fine-scale simulations, using the simulator MODFLOW/MT3DMS. Upscaled simulation results are obtained with the same simulator, linked to an MRMT module. The results facilitate efficient assessment of reactive transport in domains with sharp contrasts in soil properties and limit the need for fine-scale numerical simulations.