Characterizing hydraulic and natural fractures properties in shale oil well in Permian basin using assisted history matching

Abstract

We performed assisted history matching (AHM) to an actual shale oil well in Permian basin using Embedded Discrete Fracture Model and neural network Markov chain Monte Carlo algorithm. Three main scenarios were investigated including hydraulic fractures only, hydraulic fractures with one realization of natural fractures, and hydraulic fractures with natural fractures. We integrated data from diagnostic fracture injection test analysis, PVT (Pressure Volume Temperature) report, nearby-well core data and petrophysical interpretation as initial input for AHM workflow. The uncertain parameters consist of fracture geometry, fracture conductivity, matrix permeability, matrix and fracture water saturation, and relative permeability curves. For the case with natural fractures, we included number, length and conductivity of natural fractures as the additional uncertain parameters. From history matching, we obtained the posterior distribution of each fracture parameter and reservoir property for all three scenarios. We found that the presence of natural fractures affects the posterior distribution of matrix permeability and fracture geometry of history match solutions, when comparing to case without natural fractures, by shifting to lower value distributions because natural fractures enhance the flow besides hydraulic fractures. Multiple realizations were then used for a probabilistic production forecast. We might underpredict oil production and overpredict gas production if the actual system was assumed with only hydraulic fractures. Lastly, the benefits from the study are that hydraulic fractures, natural fractures and reservoir properties are characterized in a probabilistic manner from production data. The results from the study can be used to improve future fracturing design and optimize well spacing.