NON-ANDERSONIAN FAULT ANALYSIS IN RESERVOIRS

Andersonian faulting theory assumes that one of the principal stress (or strain) axes aligns vertically, and that faults align with the principal stresses (or strains). This simplification is used in common reservoir geomechanics applications such as determination of tectonic deformation episodes, uniaxial and biaxial in situ stress determination along the wellbore, and idealized hydraulic fracturing models. However, to understand three-dimensional fault evolution and fault activation under tectonic and present-day operational conditions, a non-Andersonian solution is required. We highlight three examples from real reservoir scenarios at different scales where the simplified Andersonian conditions cannot be used and fully three-dimensional solution techniques are required. The first case involves stress partitioning at Suban gas field, Sumatra, resulting from oblique activation of inclined regional faults at the converging Indo-Australian Plate boundary. The second case involves faulting in the Niobrara shale play, Denver Basin, where only optimally oriented overburden faults intersecting the reservoir are expected to locally enhance permeability. The third case is for focal mechanism analysis of microseismic events in the Barnett Shale, Texas, where small natural fractures and faults were seismically activated during hydraulic fracturing. In each case we use a similar analysis workflow that is scale-independent: (1) characterize the geometry of the faults and determine the fault normal vector at each surface point; (2) determine the acting stress state during a specific geologic period or for the present-day in situ state and rewrite the stress tensor in the global coordinate system; (3) solve for the traction, normal, and shear stress vectors on the faults; (4) solve for the instability parameters, such as critical pore pressure required to fail. The workflow can also be solved in the inverse to compute the acting stresses if they are unknown but the other parameters are known or inferred. In addition to using the tensor-based analytical solutions, we find that three-dimensional results are best-visualized using stereonet displays rather than polar plots. An example of the inverse solution and plotting techniques are highlighted for the Barnett Shale application.