An Efficient Approach for Automatic Well-Testing Interpretation Based on Surrogate Model and Deep Reinforcement Learning

Abstract

The artificial well-testing interpretation is a good tool for parameter evaluations, performance predictions, and strategy designs. However, non-unique solutions and computational inefficiencies are obstacles to practical interpretation, especially when artificial fractures are considered. Under this situation, a new deep reinforcement learning (DRL) based approach is proposed for automatic curve matching on vertically fractured well-testing interpretation. Based on deep deterministic policy gradient (DDPG) algorithm, the proposed DRL approach is successfully applied to automatic matching of well test curves. In addition, to improve the training efficiency, a surrogate model of the vertically fractured well test model based on LSTM neural network was established. Through episodic training, the agent finally converged to an optimal curve matching policy on vertically fractured well-testing model through interaction with the surrogate model. The results show that the average relative error of the curve parameter interpretation is less than 6%. Additionally, the results from the case studies show that the proposed DRL approach has a high calculation speed, and the average computing time was 0.44 seconds. The proposed DRL approach also has high accuracy in field cases, and the average relative error was 7.15%, which show the reliability of the proposed DRL method.