Energy Consumption Prediction Method of Oilfield Gathering System Based on Multi-Level Covariates

Abstract The oilfield gathering system converts electric energy and chemical energy from fuel into kinetic energy and thermal energy from fluids through rotating equipment and the heat exchanger, which is the main part of the energy consumption of the oilfield. To control the increase in energy consumption, it is necessary to predict the energy consumption in the future of the system. With the advancement of oilfield digitalization, data-driven intelligent prediction methods have received widespread attention due to their ability to effectively extract hidden features from energy consumption data. The data are analyzed to generate multiple covariates, including data from multiple pipeline network nodes and data from the union plant. The energy consumption prediction method based on multi-level covariates is proposed. Aiming at gas consumption prediction in an oilfield union plant, multivariable features are extracted from multi-level historical and future covariates by the recurrent neural network. The experimental results of actual production data show that the gas consumption in the gathering system can be predicted accurately. The proposed method can make full use of the collected multivariable data to achieve energy consumption prediction, which is helpful for oilfield staff to take pre-stage measures to achieve energy consumption reduction.