Natural fracture mapping and discrete fracture network modeling of Wolfcamp formation in hydraulic fracturing test site phase 1 area, Midland Basin: Fractures from 3D seismic data, image log, and core

Abstract

Fractures control fluid flow in ultra-low permeability reservoirs due to their much higher permeability. Recognition, mapping, and characterization of both natural and hydraulic fractures at multiple scales are critical to predicting hydrocarbon production in tight rocks, including mudrock. However, this determination is very challenging due to data scarcity, varying data resolutions, and uncertainties in interpretations. In this research, we combined core-identified fractures, image log-identified fractures, and 3D seismic data to analyze natural fracture features of, and construct a discrete fracture network (DFN) model for, the Wolfcamp Formation in the Hydraulic Fracturing Test Site Phase 1 (HFTS-1) project area in the central Midland Basin of the United States. An azimuth-guided, ant-track seismic attribute called VarAntAntPas was developed to recognize natural fractures and extract their seismic discrete planes (SDP) at multiple dip azimuth ranges. The extracted SDPs were then employed to calculate the fracture length, height, and elongation, as well as the dip angle and dip azimuth. This process revealed a limitation of recognizing natural fractures shorter than 600 ft by 3D seismic data due to its resolution. To model fractures shorter than 600 ft, we calculated fracture intensity at different dip azimuths from the VarAntAntPas attributes and combined this variable with fracture dip angle derived from well data and the extracted SDPs. The extracted SDPs for larger fractures (length >600 ft) and the modeled smaller fractures (length <600 ft) were combined to build a DFN model of the Wolfcamp Formation in the HFTS-1 project area. Clearly, the DFN model was built with different ways for two different-size fracture groups. Meanwhile, the discrete fractures in different orientations were built independently and then combined together in the DFN model. Natural fractures predicted by the DFN model were then validated by comparison to fractures identified from core and image logs from two horizonal wells and one slant well in the HFTS-1 project area.