Fracture networks present a critical influence on to better assess oil recovery and to optimize production of hydraulically fractured reservoirs. We established a novel workflow for fracture uncertainty analysis and prediction, in which multiple fracture pattern realizations are created from the geostatistical analysis. The original fracture networks were created using geomechanics tools. Then the fracture network is reconstrued for uncertainty analysis and history matching. For different realizations of the fracture distributions created in this workflow, we successfully maintain the continuity of the fractures, as well as the connections from the matrix to fractures. The generated grid can be further used for treatment simulation to determine fractures geometry, height growth and respected proppant transport in the induced fracture network.
In this paper, we also apply Embedded Discrete Fracture Model (EDFM) to capture the realistic geometry of fractures. Within the EDFM, each fracture plane is embedded inside the matrix grid and is discretized by the cell boundaries. We study a series of reservoir simulation realizations, in which the complex hydraulic fracture networks are created in the workflow. We investigate different fracture realizations in both planar and complex fracture configurations while maintaining the continuity within the fracture networks. This study also includes the influence of the network geometry and fractures properties on the overall performance of the reservoir.
From the uncertainty analysis results, we find that the overall reservoir performance is controlled by the fracture connectivity and the distribution of conductivity within the network. A good match of production history can be achieved by adjusting the fracture connectivity. Modeling with multiple realizations of fracture networks acknowledge that a reliable production forecast is achievable using geostatistical analysis for fracture connection reconstructions. Applying Embedded Discrete Fracture Model (EDFM) on the fracture-related model simulation provides a robust and effective means of investigating multiple fracture realizations.