| 摘要: |
| 以大庆喇嘛甸油田葡萄花油组葡一段辫状河储层砂体为例,综合野外露头、测井、岩心等资料,在精细表征砂体结构基础上,总结辫状河储层砂体接触关系,建立辫状河储层砂体横向、纵向及内部3种连通模式,并应用支持向量机方法定量预测储层砂体连通性,在此基础上采用基于目标的地质建模方法,建立基于3～4级构型界面控制的单砂体模型,在夹层表征的基础上将砂体连通性定量评价结果作为确定性数据,分区分类赋值砂体间传导率,指导油藏精细化数值模拟。结果表明:分类优选砂体连通性评价参数,应用支持向量机算法,在统计学理论和结构风险最小化原则基础上,能较好地解决小样本、非线性、高维数和局部极小点等实际问题,实现了快速定量评价砂体连通性,通过动静态资料相互验证,预测准确率达到88%；实现了夹层及传导率耦合的砂体连通性定量化模拟,数值模拟效果更能反映实际生产情况,为后期油田开发优化及调整提供新的技术支撑。 |
| 关键词: 喇嘛甸油田 葡萄花油层 辫状河 砂体连通性 支持向量机 定量评价 |
| DOI：10.3969/j.issn.1673-5005.2021.02.001 |
| 分类号::TE 122.14 |
| 文献标识码:A |
| 基金项目: |
|
| Quantitative evaluation of sand body connectivity in braided river reservoirs based on support vector machine algorithm |
|
HE Hui1, LIU Chang1, LI Shunming1, XU Shouyu2, CHEN Xiu2, WANG Baining3
|
|
(1.Research Institute of Petroleum Exploration & Development, Beijing 100083, China;2.School of Geosciences in China University of Petroleum (East China), Qingdao 266580, China;3.School of Energy Resources in China University of Geosciences, Beijing 100083, China)
|
| Abstract: |
| Taking the braided river reservoir of Putaohua oil layer in Lamadian Oilfield as an example, the data of outcrops, well logging and core data were analyzed comprehensively. Based on the detailed characterization of reservoir architecture, this paper summarizes the sand body contact features of the braided river reservoir, and establishes the vertical, horizontal and internal sand body connecting models for the braided river reservoir. The support vector machine (SVM) algorithm can be applied to quantitatively evaluate and predict sand body connectivity. According to the object-based random modeling method, the reservoir architecture model based on the configuration interface was established. The evaluation results of sand body connectivity are used as deterministic data to guide and assign the sand body conductivity in the simulation model to obtain the quantitative characterization and simulation of the coupling between interlayer and conductivity. The results show that, based on the statistical theory and structural risk minimization principle, the optimized evaluation parameters of sand body connectivity can be used to quantitatively classify the connectivity by using the SVM method. The connectivity prediction accuracy can reach 88% through mutual verification of the dynamic and static data. The method can well solve the small sample, nonlinear, high dimension and local minimum data point and so on practical problems, which also can realize the rapid quantitative evaluation sand body connectivity with interlayer. The numerical simulation results based on the sand body connectivity evaluation model can better reflect the oilfield production status, and provide a new technical support for the later oilfield development adjustment. |
| Key words: Lamadian Oilfield Putaohua oil layer braided river sand body connectivity support vector machine quantitative evaluation |
