智能钻井多目标协同优化系统研究与应用
Research and Application of Intelligent Drilling Advisory System
雍锐1
作者信息
摘要
在深井超深井钻探过程中,井眼中高强度、高研磨性地层导致的钻头早期磨损、井下钻具振动剧烈、井眼清洁不足等,严重制约了深井超深井的安全快速钻进.针对上述问题,文章提出了一种智能钻井多目标协同优化系统,可以实时跟踪、优化钻井参数,提高钻井性能.同时基于机械钻速和机械比能,定义了钻井性能综合评价指标;结合地应力、钻具振动、摩阻扭矩、井眼清洁和破岩能效等地质、物理模型,提出了包含探索、学习和应用三种模式的钻井参数实时优化流程,训练了融合支持向量回归模型和随机森林回归模型的钻井参数实时优化算法.该系统在深地川科1井进行了现场应用,提速比达到了 41.5％,为深井超深井钻井优化提速提供了一种新的技术手段.
Abstract
In the process of deep well exploration and development,the high-strength and highly abrasive forma-tion in the wellbore will lead to early bit wear of the drill,violent vibration of downhole drilling tools,insufficient hole cleaning,and other adverse factors,it seriously restricts the safe and rapid drilling of deep and ultra-deep wells.Aiming at the above problems,an intelligent drilling advisory system for drilling is proposed to optimize drilling parameters in real time and improve drilling performance.Based on rate of penetration and mechanical specific energy,the comprehensive evaluation index of drilling performance is defined.Combining geological and physical models such as in-situ stress,drilling tool vibration,frictional torque,wellbore cleaning and rock breaking energy efficiency,a real-time optimization process of drilling parameters including three modes of exploration,learning and application is proposed.The real-time optimization algorithm of drilling parameters is trained by in-tegrating support vector regression model and random forest regression model.The system has been applied in Well SDCK-1,the ROP increase ratio has reached 41.5％,and provides a new technical means for the optimiza-tion of deep and ultra-deep well drilling.
关键词
智能钻井/多目标协同优化系统/机械钻速/机械比能/机器学习/深地川科1井Key words
intelligent drilling/multi-objective coordination optimization system/rate of penetration/mechan-ical specific energy/machine learning/Well SDCK-1引用本文复制引用
基金项目
中国石油集团关键核心技术攻关项目(2022ZG06)
出版年
2024
NETL
NSTL
万方数据