Percolation-Based Techniques for Upscaling the Hydraulic Conductivity of Semi-Realistic Geological Media

by Idriss, Bilal

I tested three schemes for "upscaling" the hydraulic conductivity (K) on aquifers with bimodal K distributions. This bimodality (e.g., sand and mud deposits) was intended to capture typical geological conditions. Results were tested with a numerical model. Upscaling techniques used were inspired by schemes interpolating between arithmetic and harmonic means, but are based on percolation theory: 1) Critical path analysis (CPA), 2) Percolation path analysis (PPA, or standard scaling), and a novel scaling approach. Models chosen were both spatially correlated and uncorrelated, with important differences in critical percolation probabilities, Pc. Volume fractions Ps and 1-Ps (with Ps = sand volume fraction) were assigned to each mode of the distribution of K. I found that application of the requirement for CPA to interpolate smoothly between the parallel and series configurations makes it satisfactory only for Ps>Pc. PPA generally underestimates K. The novel scaling technique has the widest generality.