PROVIDING GEOLOGICALLY REALISTIC ESTIMATES OF FAULT ZONE PROPERTIES TO CONSTRAIN MODELS OF SUBSURFACE FLUID FLOW

Faults usually strongly influence subsurface fluid flow in deformed areas. For example, faults can strongly influence hydrocarbon accumulation and extraction, contaminant migration, carbon dioxide sequestration, and extraction of water from aquifers. The requirements for making a flow model in any scenario are based on first accurately positioning the faults in the model, and second assigning realistic properties to them.

Errors in fault interpretation are a longstanding issue. Fault characterization is further complicated by the observation that even accurately interpreted faults show significant spatial ambiguity. Both factors contribute to difficulties in accurately placing faults in flow models.

Assigning realistic properties to faults is also difficult. The magnitude of these properties and their heterogeneous spatial distribution in a fault zone, are key controls on fluid flow, but the scale of this spatial distribution is well below the limit of seismic resolution. This means it cannot be directly measured on the data that provide the means for making most subsurface models.

Given the inherent geophysical ambiguity and subseismic geological heterogeneity of fault zones, it is critical to calculate a realistic range of properties instead of making unique predictions. The first step in a method to do this is empirical compilation of measures of fault seal potential like buoyancy pressure against a proxy for fault rock composition. This proxy, often the shale gouge ratio, must combine parameters that can be measured at the seismic scale like displacement and subseismic-scale data such as logs of clay concentration. These compilations show considerable scatter since fault zones that appear identical on reflection seismic data can exhibit very different flow properties due to the subseismic-scale distribution of fault properties. In other words, the scatter is not noise, but provides useful information since it implicitly incorporates the effects of geological heterogeneity and geophysical ambiguity. The final step is to fit probabilistic curves to the empirical compilations. This provides a means to estimate a range of empirically validated fault seal potentials that can be estimated from routinely available subsurface data.