Implementation and application of the embedded discrete fracture model (EDFM) for reservoir simulation in fractured reservoirs
Fractured reservoirs have gained continuous attention from oil and gas industry. A huge amount of hydrocarbon are trapped in naturally fractured carbonate reservoirs. Besides, the advanced technology of multi-stage hydraulic fracturing have gained a great success in economic development of unconventional oil and gas reservoirs. Fractures add complexity into reservoir flow and significantly impact the ultimate recovery. Therefore, it is important yet challenging to accurately and effectively predict the recovery from fractured reservoirs. Conventional dual-continuum approaches, although effective in the simulation of naturally fractured reservoirs, may fail in some cases due to the highly idealized reservoir model. The unstructured-grid discrete fracture models, although flexible in representing complex fracture geometries, are restricted by the high complexity in gridding and high computational cost. An Embedded Discrete Fracture Model (EDFM) was recently developed to honor the accuracy of discrete fracture models while keeping the efficiency offered by structured gridding. By dividing the fractures into segments using matrix cell boundaries and creating non-neighboring connections (NNCs), the flow influence of fractures can be efficiently modeled through transport indices. In this work, the EDFM was implemented in UTCHEM, a chemical flooding in-house reservoir simulator developed at The University of Texas, to study complex recovery processes in fractured reservoirs. In addition, the model was applied in commercial simulators by making use of the non-intrusive property of the EDFM and the NNC functionality offered by the simulators. The accuracy of the EDFM in the modeling of orthogonal, non-orthogonal, and inclined fractures was verified against fine-grid explicit fracture simulations. Furthermore, case studies were performed to investigate the influence of hydraulic fracture orientations on primary depletion and the impact of large-scale natural fractures on water flooding processes. The influence of matrix grid size and fracture relative permeability was also studied. Finally, with modifications in NNC transmissibility calculation, the EDFM was applied to the modeling of a multi-lateral well stimulation technology. The accuracy of the modified formulations was verified through comparison with a multi-branch well method. The simulations carried out in this work confirmed the flexibility, applicability, and extensiveness of the EDFM.