Scholarlink

Abstract <p>An innovative approach is presented to optimizing the power consumption of oil wells equipped with submersible centrifugal pumps based on machine-learning methods. A comprehensive system combining simulation modeling and neural-network algorithms is proposed for predicting and managing energy consumption, taking into account the process and geological-well parameters. A mathematical model simulating well power consumption is developed in the SimInTech software environment. The model has an error of 2.8% and takes into account the interrelationships of parameters such as flow rate, dynamic liquid level, buffer pressure, and power-grid frequency, enabling analysis of energy consumption in various operating modes. A neural-network model for predicting and optimizing power consumption, trained using mathematical modeling data, was developed. A system for managing oil-well energy profiles with two neural-network controllers is proposed. A frequency controller for continuous operation ensures maintaining a specified flow rate with minimal energy consumption. A flow rate controller for periodic operation reduces equivalent power consumption by optimizing pumping-unit utilization. The proposed approaches are consistent with the digitalization strategy of the oil and gas industry and can be used to develop and improve automated intelligent oil-well control systems to reduce energy costs and energy-system maintenance at oil- and gas-production facilities.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine-Learning-Based Optimization of Oil-Well Power-Consumption Modes

  • 期刊论文

  • 发表日期：

    2025年11月1日

查看全文
  • A. B. Petrochenkov,
  • A. V. Romodin,
  • D. A. Dadenkov,
  • I. A. Schmidt,
  • A. S. Semenov,
  • D. Yu. Luzyanin,
  • V. I. Yuzhakov

摘要

Abstract

An innovative approach is presented to optimizing the power consumption of oil wells equipped with submersible centrifugal pumps based on machine-learning methods. A comprehensive system combining simulation modeling and neural-network algorithms is proposed for predicting and managing energy consumption, taking into account the process and geological-well parameters. A mathematical model simulating well power consumption is developed in the SimInTech software environment. The model has an error of 2.8% and takes into account the interrelationships of parameters such as flow rate, dynamic liquid level, buffer pressure, and power-grid frequency, enabling analysis of energy consumption in various operating modes. A neural-network model for predicting and optimizing power consumption, trained using mathematical modeling data, was developed. A system for managing oil-well energy profiles with two neural-network controllers is proposed. A frequency controller for continuous operation ensures maintaining a specified flow rate with minimal energy consumption. A flow rate controller for periodic operation reduces equivalent power consumption by optimizing pumping-unit utilization. The proposed approaches are consistent with the digitalization strategy of the oil and gas industry and can be used to develop and improve automated intelligent oil-well control systems to reduce energy costs and energy-system maintenance at oil- and gas-production facilities.