Abstract - The calculation of geologically meaningful fault transmissibility multipliers requires predictors of both fault rock thickness and permeability. Fault rock thickness, and its variability, can be predicted from the well established positive correlation with fault displacement. Definition of a fault rock permeability predictor is a much more difficult task due to the strongly heterogeneous and anisotropic nature of fault rocks. Permeability predictors are necessarily based on highly simplified and idealised conceptual models of fault rock which, it is hoped, capture the main controls. Here we describe a novel permeability predictor for faults offsetting a poorly lithified, sand/clay turbidite sequence based on outcrop studies and permeability measurements for faults exposed in Taranaki, New Zealand. The faults at Taranaki are characterised by granulation and cataclasis of sands and smearing of clay beds. Clay smears maintain continuity for high ratios of fault throw to clay source bed thickness. Clay smear thicknesses are highly irregular and gaps in clay smears can occur at any point between the clay source bed cutoffs. Although fault rock permeabilities may be significantly lower (~2 orders of magnitude) than those of the parent host rocks, the differences in permeability between different beds is several orders of magnitude greater (~7 orders of magnitude). The shearing of low permeability layers into fault zones is therefore a much more important control than fault related textural and mineralogical changes. The occurrence of continuous clay smears, combined with low clay permeabilities (10's to 100's nD) indicates that the primary control on fault rock permeability will be clay smear continuity. Therefore fault permeability predictors which implicitly assume that fault rocks represent a mixed aggregate of the faulted sequence, with equivalent average permeabilities, are not appropriate. Similarly, application of a method which assumes predictable clay smear thicknesses is not supported by outcrop observation. Instead we have devised a scheme whereby clay smears are continuous up to a particular ratio of fault throw to clay source layer thickness (i.e. Shale Smear Factor). Above this critical value the smear maintains a constant length and is positioned randomly between the cutoffs of the clay source bed. We compare this probabilistic approach with other predictors in terms of predicted across fault connectivity and permeability, both in cross-section and over fault surfaces. The approach combines the averaging effects expected at higher ratios of fault throw to bed thickness, with the bimodal permeability behaviour associated with clay smearing at lower ratios.
Abstract of talk given to:
Structurally Complex Reservoirs, Geological Society of London, February 2006.