Study on the prediction method of ceasing-flowing for self-flowing wells Provisionally Accepted

Bo Kang¹ Zhongrong Mi¹ Yuhan Hu^2* Liang Zhang¹ Ruihan Zhang²

• ¹Chengdu North Petroleum Exploration and Development Technology Company Limited, China
• ²Southwest Petroleum University, China

At present, most of the wells in X-oilfield are self-flowing wells. In order to adjust the production system of oil wells in time according to the production requirements of oil fields, it is necessary to predict the ceasing-flowing time. Therefore, how to accurately predict the ceasing-flowing time is the main problem faced by the self-flowing well. Due to the conventional prediction methods only considering the influence of a single variable, the prediction results are not ideal. Combining the production prediction based on the long short-term memory (LSTM) neural network and the inflow and outflow dynamic curves, this study proposes a comprehensive method for predicting the ceasingflowing time of a flowing well by considering multiple factors. Using the minimum wellhead pressure prediction method, the changes in bottom hole flowing pressure and reservoir pressure are also considered. The practical application results in X-oilfield show that the calculated and predicted results are highly consistent with the actual production data, verifying the reliability of this method. This study can provide a reference for the prediction of oil well ceasing-flowing in other oilfields.