A Method of Fracture Prediction Across Multiple Stratigraphic Horizons in the Midland Basin, Texas, USA

摘要

Natural fractures influence the development of unconventional reservoirs in many ways. Production results indicate enhanced fluid rates associated with them. Microseismic and well interference during hydraulic stimulation suggest that they influence completions. Numerical models suggest that natural fractures may influence length and height dimensions of hydraulic fractures. Methods commonly used to quantify them can be costly, possibly risky or impractical. Here we present a method of fracture prediction using commonly available data in conjunction with commercial software. Two distinct fracture trends (NE-SW and NW-SE) are observed in the Wolfcamp and Spraberry unconventional reservoirs of the Midland Basin. A third, less prevalent E-W trending set, is observed in some intervals. These sets are consistent in orientation, spatially and stratigraphically, across the basin with mean trends varying by only a few degrees. Fracture intensity logs, calculated from a basin-wide set of horizontal image log interpretations, demonstrate an increasing intensity with proximity to faults in the middle Wolfcamp. In areas where multiple intervals are sampled, a similar relationship of fracture intensity to faults is observed in each. This suggests that the fractures are tectonic in origin and are coeval with faults or are the result of later fault reactivation. To predict fractures in under sampled areas and intervals of the basin, likely fracture formation mechanisms were evaluated in a study area with 3D seismic data (for horizon and fault surfaces) and image logs in multiple stratigraphic units. Mechanisms included: folding (curvature), geomechanical modeling and deformation related to fault reactivation under several potential stress regimes. Plausible paleo-stress regimes were determined by an inversion, varying differential stress orientations and magnitudes to maximize slip and dilation tendency of the observed fractures. Only fault reactivation models in a strike-slip regime (σ1 ~ 80°) predicted fracture orientations and mode consistent with observation. Proxy values for fracture intensity were also evaluated. Principal strains calculated for horizons in the fault reactivation model provide a means to predict fracture intensity. Comparisons of horizontal and vertical trends of maximum extensional strain (e1) to fracture intensity reveal similar trends. Strain intensity decreases exponentially away from faults and decreases with decreasing depth. This proxy relationship provides a promising means to estimate fracture orientations and intensity in areas where little image log data is available, whereas the more ubiquitous 3D seismic is available.