Application of machine learning methods to forecast the rate of horizontal wells

Abstract

The paper summarizes and provides an overview of the analytical equations of fluid inflow to horizontal wells. Using the actual data, it was found that analytical equations do not allow reliably calculating and predicting the flow rate of horizontal wells and it is necessary to apply new approaches to solve this problem. The paper proposes a fundamentally new approach to forecasting the flow rate of horizontal wells, based on the application and training of machine learning methods. A fully connected neural network of direct propagation was used as a model. When comparing the actual and calculated using a fully connected neural network of direct propagation of horizontal well flow rates, their high convergence with a correlation coefficient of more than 0.8 was established. In further studies, it is planned to expand the sample and parameters included in the model to improve the calculation and forecasting of horizontal wells in various geological and physical conditions of their operation. Keywords: horizontal well; oil flow rate; linear regression; artificial neural network.