Evaluation of the role of heterogeneities on transverse mixing in bench-scale tank experiments by numerical modeling.

In this work, numerical modeling is used to evaluate and interpret a series of detailed and well-controlled two-dimensional bench-scale conservative tracer tank experiments performed to investigate transverse mixing in porous media. The porous medium used consists of a fine matrix and a more permeable lens vertically aligned with the tracer source and the flow direction. A sensitivity analysis shows that the tracer distribution after passing the lens is only slightly sensitive to variations in transverse dispersivity, but strongly sensitive to the contrast of hydraulic conductivities. A unique parameter set could be calibrated to closely fit the experimental observations. On the basis of calibrated and validated model, synthetic experiments with different contrasts in hydraulic conductivity and more complex setups were performed and the efficiency of mixing evaluated. Flux-related dilution indices derived from these simulations show that the contrasts in hydraulic conductivity between matrix and high-permeable lenses as well as the spatial configuration of tracer plumes and lenses dominate mixing, rather than the actual pore scale dispersivities. These results indicate that local material distributions, the magnitude of permeability contrasts, and their spatial and scale relation to solute plumes are more important for macro-scale transverse dispersion than the micro-scale dispersivities of individual materials. Local material characterization by thorough site investigation hence is of utmost importance for the evaluation of mixing-influenced or -governed problems in groundwater, such as tracer test evaluation or an assessment of contaminant natural attenuation.