This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 217311, “Energy-Based Fracture-Network Reconstruction of Shale Gas Reservoir,” by Ji Lu and Botao Lin, SPE, China University of Petroleum. The paper has not been peer reviewed.
Microseismic monitoring is a commonly used technique in characterizing hydraulic fractures. Fracture-network reconstruction remains challenging, however, because of the heterogeneity and complex stress fields in shale gas reservoirs. An energy-based 3D fracture-reconstruction method (EFR3D) is proposed to derive the complex fracture network from microseismic data in a shale gas reservoir. The proposed method has good adaptability and high accuracy in various fracture configurations.
Because the location of microseismic data has a closely related spatial and temporal relationship with fractures, reconstructing fractures directly from microseismic data has become an effective method of identifying fracture networks. Several 3D models have been established to model fracture networks. The authors write, however, that room for enhancement remains in the area of 3D reconstruction of hydraulic fractures using microseismic monitoring.
The authors propose the EFR3D method to reconstruct fractures with arbitrary orientations and shapes. The methodology of creation of the fracture geometry model, and of the EFR3D approach itself, is detailed in the complete paper. First, a radius filter was used to eliminate the noise of microseismic data. Subsequently, the Propose, Expand, and Re-Estimate Labels algorithm, the Density-Based Spatial Clustering of Applications With Noise algorithm, and the Alpha-Shape algorithm were used in combination to reconstruct fractures from the microseismic data. A simulation model based on the Monte Carlo method was adopted to verify the efficiency and accuracy of the proposed method. Finally, a field application was investigated in the southern Sichuan Basin where multistage hydraulic fracturing was performed to gain further insight into the complex configuration of the fracture network. The outcome of this study can provide fundamental and practical design guidelines for multistage hydraulic fracturing practices.
Fracture Reconstruction With Different Fracture Numbers.
To quantify the accuracy of the EFR3D method, the simulations were conducted in a 1-km3 study region. The fracture number gradually increased while other parameters were kept constant. The fracture scale ratio was set as 1.0 and the mean dip angle as 45°. The fracture number ranged from 1 to 50 at a step interval of 1.
Twenty-five hydraulic fractures were reconstructed. The EFR3D can reconstruct the fracture effectively with arbitrary orientations and shapes. When comparing the relative reconstruction errors of the EFR3D method and the random sample consensus (RANSAC) -based method at an increasing number of fractures, the errors of both methods experienced a significant fluctuation at the beginning and increased slightly as the fracture number increased. The EFR3D exhibited a lower mean error of 7.93% compared with the RANSAC-based method, which exhibited a mean error of 16.01%. Moreover, the former method displayed a lower error increase rate, implying that the EFR3D method has higher stability and accuracy in reconstructing fractures.
