摘要
深层煤层(岩)气井产量受地质和工程双重因素控制,其中可调控的工程控制因素对产量预测和提升尤为重要。水平井大规模体积压裂是提高产量的重要手段。实践表明,不同生产阶段的产气量差异显著,亟需对工程控制因素进行优化,以提高产量预测的精度和可靠性。以大吉区块煤岩气水平井为例,将连续生产300 d划分为短期、中期、长期三个阶段,结合随机森林、相关系数和灰色关联分析进行综合评价,通过交叉验证优选的不同生产阶段产气量的工程控制因素为输入,结合地质参数建立了基于XGBoost的分阶段产气量预测模型。结果表明:总液量、总砂量、平均施工排量、总簇数和段数这5个参数对不同生产阶段的产气量均具有重要影响;工程控制因素的优选使各阶段产气量拟合优度R^(2)均有大幅提升;建立的分阶段产气量多输出预测模型相较于LSTM和SVM准确率最高,将该模型迁移应用到多个不同区块,预测短期产量的R^(2)均超过0.81、预测中期和长期产量的R^(2)均超过0.71,能够为煤岩气井的压裂设计提供有效指导。
Production from deep coal bed(rock)gas wells is controlled by both geological and engineering factors,with adjustable engineering controls being particularly important for production prediction and enhancement.Large-scale volume fracturing of horizontal wells is an important means to improve production.Practice has shown that gas production varies significantly at different production stages,and there is an urgent need to optimize engineering control factors to improve the accuracy and reliability of production prediction.Taking the coal-rock gas horizontal wells in Daji block as an example,300 days of continuous production were divided into three stages:short-term,medium-term,and long-term,and combined with Random Forest,correlation coefficient,and grey correlation analysis to conduct a comprehensive evaluation,and the engineering controlling factors for the gas production in different production stages preferred through cross-validation were used as inputs,and a phased gas production prediction model based on XGBoost was established by combining the geologic parameters.The results show that the five parameters of total liquid volume,total sand volume,average flow rate,total cluster number and section number have important influence on the gas production at different production stages;the optimization of engineering controlling factors greatly improves the goodness-of-fit R2 of the gas production at each stage;and the multi-output prediction model of the phased gas production has the highest accuracy compared with the LSTM and the SVM,and the R^(2)of short-term production exceeds the R^(2) of short-term production in all the cases when the model is migrated to different blocks.The R^(2) of short-term production exceeds 0.81,and the R^(2)of medium-and long-term production exceeds 0.71,which can provide effective guidance for the fracturing design of coal rock gas wells.
作者
周立宏
李曙光
甄怀宾
郭子熙
白坤森
袁朴
王成旺
ZHOU Lihong;LI Shuguang;ZHEN Huaibin;GUO Zixi;BAI Kunsen;YUAN Pu;WANG Chengwang(PetroChina Coalbed Methane Company Limited,Beijing 100028,China;China United Coalbed Methane National Engineering Research Center Co.Ltd.,Beijing 100095,China;Southwest Petroleum University,Chengdu,Sichuan 610500,China)
出处
《钻采工艺》
北大核心
2025年第3期101-109,共9页
Drilling & Production Technology
基金
中国石油天然气股份有限公司攻关性应用性技术攻关项目“深地煤岩气成藏理论与效益开发技术研究”(编号:2023ZZ18)
“非常规储层改造关键技术研究”(编号:2023ZZ28)。
关键词
煤层(岩)气
综合评价
工程控制因素
产量预测
coalbed methane(rock)gas
comprehensive evaluation
engineering controlling factor
production prediction
