摘要
判断火驱稠油油藏注采井间连通关系进而计算燃烧前缘位置是火驱开发调整的首要工作,而现有注采关联模型难以表征复杂的火驱注采井间连通情况。以红浅1井区火驱先导试验区为例,建立了适用于火驱的多元热采关联模型。首先对原始动态数据进行了数据预处理;然后利用决策树、随机森林和AdaBoost 3种机器学习算法进行了特征变量的重要性比较,选择计算精度最高的随机森林算法确定出影响火驱产量的3个主特征为含水率、注气量和油压;在此基础上,建立了火驱多元注采关联模型,对典型井组注采井间连通性进行了计算;最后利用生产特征和示踪剂测试对计算结果进行了验证。结果表明,井间连通性计算值符合生产动态分析对生产井状态的认识,示踪剂监测结果也进一步验证了注采连通性的准确性。本文方法准确简便,有助于清晰认识火驱井间状态,并为后续火驱注采调整和产量预测提供决策依据。
Determining the connectivity between injection and production wells in fire flooding development of heavy oil reservoirs,and subsequently computing the combustion front location,is considered the primary task of fire flooding development adjustments.However,current injection-production correlation models struggle to represent the complex connectivity between fire flooding injection and production wells.A multi-dimensional thermal production correlation model suitable for fire flooding was established using Hongqian1 fire flooding pilot test area as a case study.Firstly,the original dynamic data was pre-processed.Then,the importance of feature variables was compared using three machine learning algorithms-decision tree,random forest,and AdaBoost,and the random forest algorithm was selected for its highest accuracy in determining the three main features affecting fire flooding production:water cut,injection volume,and oil pressure.On this basis,a multi-dimensional production-injection correlation model for fire flooding was established to calculate the connectivity between injection and production wells in typical well groups.Finally,the calculation results were validated using production characteristics and tracer tests.The results showed that the calculated values of inter-well connectivity were in line with the understanding of production well status through dynamic production analysis,and tracer monitoring results further verified the accuracy of injection-production connectivity.The proposed method in this study is accurate and simple,which helps to clearly understand the state of fire flooding wells and provides decision basis for subsequent adjustments and production forecasting for fire flooding injection and production.
作者
袁士宝
宋佳
任梓寒
杨凤祥
蒋海岩
YUAN Shibao;SONG Jia;REN Zihan;YANG Fengxiang;JIANG Haiyan(Petroleum Engineering Institute of Xi'an Shiyou University,Xi'an,Shannxi 710065,China;Research Institute of Exploration and Development,PetroChina Xinjiang Oilfield Company,Karamay,Xinjiang 834000,China)
出处
《中国海上油气》
CAS
CSCD
北大核心
2023年第2期93-100,共8页
China Offshore Oil and Gas
基金
国家自然科学基金项目“热-流-固-化多场耦合的火驱储层热次生孔道演化及盐沉析调控机理研究(编号:52274039)”
西安石油大学研究生创新与实践能力培养项目“基于机器学习的火驱注采井间连通性定量研究(编号:YCS22212014)”
陕西省自然科学基础研究重点项目“基于多点微波加热的油页岩原位热解开采方法研究(编号:2022JZ-28)”部分研究成果。
关键词
火驱
机器学习
主特征
注采关联模型
井间连通性
fire flooding
machine learning
main features
production-injection correlation model
inter-well connectivity
