Uniformity Estimation of Hydraulic Fracturing by Neural Network

The quality of hydraulic fracturing directly affects the production of shale gas. An important indicator for measuring the quality of hydraulic fracturing is the uniformity of fluid inflow in the perforation clusters. That is, the more uniform the fluid inflow in each perforation cluster, the more gas will be produced in the future by the fracturing well. A conventional method is to deploy optical fibre temperature sensors in the fracturing well to measure the temperature of the liquid in the perforation cluster. The lower the temperature at which the perforation cluster is formed, the more liquid the cluster enters. The cost of this technology is very high, and the optical fiber sensors are easily damaged during the fracturing process. This paper adopts a data-driven approach to predict the uniformity of fluid inflow of the perforation cluster, thereby replacing optical fiber sensor technology. Specifically, we use historical data to establish a mapping relationship between fracturing operation parameters (i.e., casing pressure, displacement, etc.) and the uniformity of fluid inflow in the perforation cluster. For this mapping function, we use a neural network with a twotower structure that can learn the feature interaction between fracturing operation parameters. Experimental results show that the data-driven approach can effectively predict the uniformity of the liquid inflow of the perforation cluster, and the accuracy of our method is significantly better than that of the baseline method.