Creating the optimum Complex Fracture Network (CFN) during hydraulic fracturing by propagating Hydraulic Fractures (HF) stimulating and connecting the existing Natural Fractures (NF) is of great importance to fractured shale reservoirs' development. However, it yet haven't obtained enough research. Facing this urgent demand, a comprehensive CFN configuration, and optimization numerical investigation, considering geomechanics and economic effects, is proposed. An HF propagation model is constructed coupling the Displacement Discontinuity Method (DDM) and the fluid flow function, together with a power-law probability NF pattern, to simulate the CFN propagation considering the rock and fluid mechanics. Then the CFN productivity in shale reservoirs is simulated by the Embedded Discrete Fracture Model (EDFM) with shale gas flow model, in which the complexly distributed fractures and the adsorption and slippage effect of shale gas are described. Based on the above, a framework for CFN optimization is developed, in which the net present value (NPV) is adopted as the objective function to consider the economic viability. A state-of-the-art Neural Network Algorithm (NNA) is employed and further improved into a Modified Neural Network Algorithm (M-NNA) in this study, which presents 89% optimization searching accuracy tested by the complex multi-extreme benchmark function, far better than the 28% accuracy of the original PSO algorithm. The M-NNA is then adopted as the optimization algorithm in the CFN optimization framework. Finally, as an example, a case of CFN propagation and production is simulated under the coupling effects of geomechanical and fluid parameters, results show that the CFN provides nearly 90% of the total gas production; furthermore, the CFN for aligning and alternating fracturing placements are simulated and optimized considering the geological and economical influence, the results demonstrate that the alternating fracturing placement constructs better CFN configuration and leads to higher economic revenue.
Read full abstract