Predicting gas flow rate in fractured shale reservoirs using discrete fracture model and GA-BP neural network method

Abstract

In the process of shale gas exploitation, the hydraulic fracturing technology is generally used to generate the stimulate reservoir volume (SRV). The network connectivity between natural fractures (NFs) and hydraulic fractures (HFs) significantly affects gas production rate in the hydraulic fracturing stage. To study the influence of natural fracture characteristics on gas flow behavior, a discrete fracture model considering matrix and fracture behaviors is developed. It is found that the increase of natural fracture permeability and distribution density leads to the increase of gas production rate, whereas large orientation angle may cause its decrease. To determine the nonlinear relationship between the natural fracture parameters and gas production rate, the method of back propagation (BP) neural network with genetic algorithm (GA) was applied to make intelligent prediction for the shale gas production. The gas flow rate values for different natural fracture parameters (e.g., permeability, density, orientation) were calculated by discrete fracture model, and used as input data for training and prediction in the GA-BP neural network. Results demonstrate the applicability of forecasting model, which is suitable for shale gas prediction during the operation of horizontal wells if more data profiles could be collected.