2003 Seattle Annual Meeting (November 2–5, 2003)

Paper No. 5
Presentation Time: 2:30 PM


LIU, Gaisheng, Geological Sciences, Univ of Alabama, Tuscaloosa, AL 35487, ZHENG, Chunmiao, Geological Sciences, Univ of Alabama, AL and GORELICK, Steven M., Geological and Environmental Sciences, Stanford Univ, Bldg 320, Stanford, CA 94305, LIU022@BAMA.UA.EDU

The dual-domain mass transfer (DDMT) model, compared to the widely-used advection-dispersion model, has been suggested as more appropriate to reproduce average solute concentrations in extremely heterogeneous aquifers where small-scale preferential flow paths exert a critical control on flow. The primary goals of this study are 1) to evaluate, in a fundamental sense, the efficacy of the DDMT model to represent transport processes in aquifers affected by decimeter-scale preferential flow paths, and 2) to estimate DDMT model parameters without resorting to model calibration. Using an invasion percolation algorithm, hypothetical 3-D networks were generated to serve as surrogates for decimeter-scale preferential flow “channels” embedded in a lower-conductivity “matrix.” An advection-diffusion model employing a very fine high-resolution grid was used to obtain a series of “true” solute plumes under varied flow and transport conditions. Each “true” plume was then compared to concentrations resulting from the DDMT model. Based on the match between the reference plumes and DDMT-generated plumes, we demonstrate that under certain conditions the DDMT model can represent the asymmetric and non-Gaussian breakthrough curves that result from decimeter-scale preferential flow paths. The most critical factors that dictate the applicability of the DDMT model are the initial source configuration and the magnitude of the ratio between high and low hydraulic conductivities. We also demonstrate that the parameters required by the DDMT model can be estimated without calibration. Specifically, the mobile/immobile porosity ratio and the mass-transfer rate coefficient between the mobile and immobile domains can be estimated a priori. Estimates of these parameters can be based on the equivalent hydraulic conductivity of the channel/matrix system, the respective hydraulic conductivities of the channels and matrix, the average hydraulic gradient, and the molecular diffusion coefficient.