Production Simulation of Complex Fracture Networks for Shale Gas Reservoirs Using a Semi-analytical Model

Abstract

Abstract Interacting with the pre-existing natural fractures, multi-stage hydraulic fracturing along horizontal wells creates complex fracture networks. Gas flow for these fracture networks in shale gas reservoirs is significantly complex due to more flow mechanisms, including the interplay of flow between the interconnected fractures, seepage, diffusion, adsorption/desorption and stress-dependent permeability within the fractures. The objective of this paper is to present a semi-analytical model for production simulation of complex fracture networks coupled with above processes in shale gas reservoirs. The fractures within the networks are represented explicitly rather than idealized as a dual-porosity medium. The model dynamically couples a numerical fracture network model with an analytical reservoir model. This approach allows us to simulate the networks with complex geometry and varying fracture conductivity. Seepage, diffusion and gas adsorption/desorption in the matrix are taken into account. Fracture deformation is implicitly coupled with numerical network flow model to reflect stress-stress dependent of fracture permeability. A synthetic fracture network is simulated in the rectangular reservoir. The results indicate that for the same fracture network volume a very low conductivity will generate very large pressure drops in the fractures surrounding the wellbore resulting in less effective fracture network compared to high conductivity. The contribution of diffusivity flow of the matrix is not appreciable on the log-log plot for transient response for km=10−4md. The adsorption compressibility has a great influence on the middle and later production periods and gas recovery may be significantly enhanced by adsorption/desorption. When the fracture is closed with pressure drop, the permeability is still sufficient to transmit the volume of fluid available to flow due to the infinite-conductivity effect compared with the shale-matrix for km=10−4md. Finally, a field example from the Barnett Shale shows the capacity of the model to simulate more complex fracture networks in shale gas reservoirs.