石油科学通报2026，Vol.11Issue(3)：836-849,14.DOI:10.3969/j.issn.2096-1693.2026.01.020
时移地震多属性智能分析在CO2驱油波及范围识别与圈定中的应用—以胜利油田G89区块为例
Application of intelligent time-lapse seismic multi-attribute analysis to the identification and delineation of CO2 flooding sweep extent:Taking G89 area,Shengli oilfield as an example
摘要
Abstract
Under the"dual-carbon"goals,CO2 flooding,as a development method that combines enhanced oil recovery with emission reduction and efficiency improvement,is an important technological approach to promote efficient oilfield development and low-carbon transformation.Accurate identification of CO2 migration pathways and effective sweep extent are crucial for evaluating displacement effects and adjusting injection-production schemes.However,conventional production performance analysis and single monitoring methods have limited capability in characterizing the subsurface migration process and spatial distribution of CO2,making it difficult to meet the demand for fine-scale characterization.To address the problems associated with existing identification methods based on time-lapse seismic difference attributes,which are susceptible to noise,non-re-peatability errors,and reservoir heterogeneity and thus often lead to scattered anomalous responses and blurred boundaries,this study proposes a method for identifying the sweep extent of CO2 flooding based on intelligent integration of time-lapse seismic multi-attributes.Difference volumes were constructed from time-lapse seismic data,and sensitive difference attributes,including amplitude,phase,and attenuation,were selected.Then,a fuzzy neural network(FNN)was introduced to perform nonlinear fusion of multiple attributes based on fuzzy rules,thereby constructing a continuous response indicator characterizing the intensity of CO2 sweep.The results show that the proposed method can effectively suppress scattered false anomalies and improve the boundary clarity and spatial connectivity of the predicted results.Time-series comparison indicates that the predicted sweep extent expanded outward from the vicinity of injection wells as injection proceeded,and migrated upward along the up-dip direction toward structurally higher positions.The delineated sweep area increased from approximately 1.7 km² in 2010 to approximately 2.6 km² in 2022.Validation against production performance data further shows that strong predicted responses generally correspond to the vicinity of high gas-injection wells and high gas-production wells,indicating that the proposed method can more accurately reflect reservoir response differences during the CO2 flooding process and provide an effective geophysical approach for identifying CO2 migration pathways,quantitatively characterizing sweep extent,and evaluating displacement performance.关键词
时移地震/差异体/时移差异属性分析/多属性融合/模糊神经网络/CO2 驱油Key words
time-lapse seismic/difference volume/time-lapse difference attribute analysis/multi-attribute fusion/fuzzy neural network/CO2 flooding分类
天文与地球科学引用本文复制引用
刘浩辰,刘钰铭,曲志鹏,张伟忠,张冰冰,陈冠宇..时移地震多属性智能分析在CO2驱油波及范围识别与圈定中的应用—以胜利油田G89区块为例[J].石油科学通报,2026,11(3):836-849,14.基金项目
国家自然科学基金项目"盆缘过渡带坡度-流量双重控制下的辫状河成因机制与砂体构型模式"(42472205)资助 （42472205）
