摘要
明确注采井间连通情况对油田稳产、控水、提高采收率具有重要意义,而传统的判断方法施工较繁琐,且费用高。应用机器学习方法,基于RNN-DTW进行井间连通性分析,通过训练RNN神经网络模型并预测油井未来产液量,避免DTW算法由于注水传播见效延迟导致的注水量与注水见效后油井产液量的差异,并使用DTW算法分析注水量与产液量的相似程度,确定井间连通状况。现场应用结果表明,该方法的计算结果符合井组物性特征,可以作为油田现场判断注采井间连通性的一种方法。
Making clear the connectivity between injection and production wells is important for oilfield stability,water control,and enhanced oil recovery,while traditional methods for judging the connectivity between wells are complicated in construction and high in cost.The machine learning method introduced in this paper is applied to analyze the connectivity based on RNN-DTW.By training the RNN neural network model and predicting the future liquid production of oil wells,the difference between water injection data volume and the oil well production data volume after effective water injection was avoided due to the delay of water injection propagation in the DTW algorithm.Therefore,DTW is used to analyze the similarity between water injection and liquid production.On-site application results indicate that the calculation results from this algorithm are in line with the physical characteristics of well groups,and can be used as a method for determining the connectivity between injection and production wells in oilfield sites.
作者
谭鑫龙
TAN Xinlong(Chongqing University of Science and Technology,Chongqing 401331,China)
出处
《石油地质与工程》
CAS
2023年第5期76-80,共5页
Petroleum Geology and Engineering
基金
重庆科技学院硕士研究生创新计划项目(YKJCX2120130)。
关键词
注采井组
连通性
RNN-DTW
机器学习
现场应用
injection-production well groups
connectivity
RNN-DTW
machine learning
field application