【作者】 李虹；

【导师】 于海洋； 程时清；

【作者基本信息】 中国石油大学(北京) ， 油气田开发工程， 2022， 博士

【摘要】 裂缝性致密砂岩气藏非均质性强,裂缝发育,生产过程中因地层压力下降造成基质和裂缝渗透率下降,现有数值模拟方法没有考虑非均质条件下基质和裂缝两种介质内部应力敏感的不均匀性。嵌入式离散裂缝模型虽然提高了计算效率,但不完全适用于低导流能力的裂缝,且近井网格精度不够。本文针对上述问题,完善了裂缝性致密储层建模方法,建立了考虑非均匀应力敏感的数值模型,构建了基于多重混合网格的多尺度数值计算方法,为发展复杂气藏渗流理论、提高裂缝性气藏开发效果提供了理论基础和技术支撑。(1)建立了基于分形和机器学习的渗透率表征方法,在有限的资料条件下,可以快速建立反映实际储层特征的非均质储层模型。基于经典案例SPE10的孔隙度场,结合分形模型重构了渗透率场;根据须二裂缝性气藏的特征,建立了基于机器学习的渗透率预测模型,优选确定精细树算法构建的渗透率预测模型最接近实际。(2)针对结构化网格,改进了投影嵌入式离散裂缝模型(p EDFM)中裂缝的投影方式,拓展了基于颗粒变形的应力敏感模型,考虑基质、天然裂缝以及压裂裂缝应力敏感程度不一致,建立了非均匀应力敏感的数值模型。模型退化后与经典指数式模型模拟结果对比,验证了本模型的准确性。本模型可根据储层物性差异自适应获得应力敏感修正系数,揭示了基质、天然裂缝以及压裂裂缝在空间域和时间域的非均匀应力敏感规律。(3)建立了分层非结构网格、分段径向网格、嵌入式离散裂缝耦合的多重混合网格系统;提出了适用于非结构网格的p EDFM投影方法,修正了非相邻连接对的传导率。基于多重混合网格空间域,建立了多重混合网格多尺度数值计算方法。在单相流动条件下与常规有限体积方法的计算结果对比,验证了该方法的可靠性。在气水两相条件下,粗尺度划分不当会造成模拟结果存在较大误差,采用多重混合网格多尺度数值计算方法对近井区域网格细化能够提高模拟精度和计算效率。(4)将本文建立的数值模拟方法应用于川西须二气藏,建立了基质、天然裂缝、压裂裂缝的非均匀应力敏感图版;根据生产动态分析,建立了考虑不同水体能量、不同分层类别、不同裂缝模式的水侵模式,预测的气井产能与实际相吻合。本文发展了应力敏感型致密气藏数值模拟方法,研究成果对于确定应力敏感导致的空间域和时间域的渗透率不均匀分布、揭示裂缝性含水致密气藏复杂气水渗流规律、准确预测气井产能具有重要的理论意义和应用价值。更多还原

【Abstract】 During the production of fractured tight sandstone gas reservoirs,which have strong heterogeneity and developed fractures,the permeability of matrix and fracture decrease due to formation pressure drop.Existing numerical simulation methods for such reservoirs failed to consider the nonuniformity in stress sensitivity of matrix and fractures under heterogeneous conditions.Although the embedded discrete fracture model(EDFM)has improved the computational efficiency,it is not completely applicable to fractures with low conductivity,and the near-well mesh precision is insufficient.In view of the above problems,the modeling method for fractured tight reservoirs was improved,a numerical simulation method considering non-uniform stress sensitivity was established,and a multi-scale numerical calculation method based on multiple hybrid grids was formed,which have provided theoretical guidance and technical support for developing the seepage theory of complex gas reservoirs and improving the development effect of fractured gas reservoirs.(1)The permeability characterization method based on fractal and machine learning was established respectively.Under the condition of limited data,the heterogeneous reservoir model reflecting the actual reservoir characteristics can be quickly established.Based on the porosity of the classic case SPE10,the permeability field was reconstructed by combining the fractal model.According to the characteristics of Xu2 fractured gas reservoir,the permeability prediction model constructed by fine tree algorithm was selected to be the closest to reality.(2)For structured grids,by improving the projection mode of fracture plane in Projection-based embedded discrete fracture model,expanding the stress-sensitive model based on particle deformation,a numerical simulation method was established considering the non-uniform stress sensitivity of matrix and fracture.Co MPared with the calculation results of classical exponential model,the correctness of the proposed model was verified.With the model,the correction coefficient for stress sensitivity can be obtained adaptively according to the difference of reservoir physical property.The nonuniform stress sensitive law of matrix,natural fracture and fractured fracture was revealed.(3)Layered unstructured mesh,segmented radial mesh and embedded discrete fracture coupling multi-hybrid mesh were established.A projection method for p EDFM suitable for unstructured grids was proposed to correct the conductivity of unstructured grids.Based on the multiple hybrid grid spatial domain,a multi-hybrid grid multi-scale numerical method was established with the multiscale restriction smoothed basis method,which improved the computational efficiency.The reliability of the multi-scale numerical method was verified by co MParing single-phase simulation results of the conventional finite volume method.In the calculation process for gas water two phase flow,the improper coarse grid division will result in a large error in the simulation results,and the accuracy will be seriously distorted.Thus,grid refinement of near wellbore area could improve the simulation precision and computational efficiency by using the proposed method.(4)The models proposed in this paper were applied to the Xu2 gas reservoir in the western Sichuan Basin.The non-uniform stress-sensitive plates of matrix,natural fractures and fractured fractures were established.The water invasion models considering different water energy,different stratification types and different fracture patterns were established to predict the productivity of gas wells according to production dynamic analysis,and the predicted results were consistent with the reality.In this thesis,the simulation method for stress sensitive fractured tight sandstone gas reservoirs was developed.The research results have significant theoretical meaning and application value for the determining of uneven permeability distribution in the caused by stress sensitivity in spatial and time domain,revealing the complex gas and water seepage law in water-containing fractured tight gas reservoirs,and precisely predicting the productivity of gas wells.更多还原