Fault permeability from stochastic modeling of clay smears

Abstract In normally consolidated, shallow (depth <~3 km) siliciclastic sequences, faults develop clay smears. Existing models include the dependence of permeability on fraction, but improved predictions fault should account for uncertainty and anisotropy. We introduce PREDICT, a methodology that computes probability distributions directional components (dip-normal, strike-parallel, dip-parallel) tensor from statistical samples set geological variables. These variables, which geometrical, compositional, mechanical properties, allow multiple discretizations core to be populated with sand smears, can used upscale coarser scale (e.g., suitable reservoir modeling). validated our implementation experimental data applied PREDICT several stratigraphic sequences. show is controlled by smear configuration and, crucially, it typically exhibits multimodal due existence holes. The latter unique feature algorithm, build scenarios manage mitigate risk in subsurface applications.