时移地震约束油藏动态表征理论与方法研究

Time-lapse seismic can be used to monitor the dynamic change of reservoir parameters, which can provide high quality information for reservoir management. However, due to the noise and other factors, and limitation of seismic frequency, it is difficult to quantitatively describe the thin layer reservoir performance only by time-lapse seismic data, since the uncertainty of residual oil prediction is really strong. Using the past dynamic data of wells, reservoir numerical simulation technology simulates or recreation of the development oil, and it is also an effective mean to predict residual oil distribution and solve the practical problems of oilfield development. However, without lateral restraints in the numerical simulation processing and the heterogeneity of reservoir, the simulation results also have strong uncertainty, which increases the risks of drilling. In order to meet the needs of the residual oil prediction and development requirement of the oil field, this project will develop a systematic study on the theory and method for the time-lapse seismic constrained dynamic characterization of the reservoir. We will study the theory and method of time-lapse seismic nonlinear inversion based on 1-D wave equation solved by the reflectivity method. Moreover, the optimal selection of constraints as well as optimized regularization parameter adaptive acquisition can make the seismic difference nonlinear inversion more accurate and stable. Thus, a high precision time-lapse seismic difference nonlinear inversion technique may be developed. Coupled with the results of the time-lapse seismic difference inversion, reservoir numerical simulation can establish and calibrate the reservoir model more accurately. The study will improve the reliability and accuracy of the prediction of residual oil by making full use of the advantages of geophysical information and reservoir engineering data, and further promote the intersection and integration of geophysics and reservoir engineering.

时移地震能监测油藏生产引起的动态变化，从而为优化油藏管理提供高品质决策信息。但受地震资料频率限制及噪音等因素影响，时移地震差异资料难以实现油藏小层动态定量描述，预测不确定性强。油藏数值模拟是利用井开发动态数据实现或再现油田生产历史，是目前进行油藏剩余油预测、解决油田开发实际问题的有效手段，但由于实际油藏的非均质性以及开发动态模拟缺少横向约束，使得油藏动态模拟同样存在较强的不确定性。针对实际油田剩余油预测与开发生产需求，项目系统开展时移地震约束油藏动态表征理论与方法研究。探究基于波动方程反射率法求解的时移地震差异非线性反演理论与方法，并通过模型正则化约束条件优化与正则化参数自适应获取，实现时移地震差异非线性反演准确、稳定求解，实现开发油田的剩余油分布预测研究。研究将有效发挥地球物理信息与油藏工程数据各自优势，提高剩余油预测可靠性与精度同时，进一步促进地球物理学科与油藏工程学科的交叉与融合。

利用反演获取储层差异弹性参数是时移地震油藏监测的重要手段。针对实际油田剩余油预测与开发生产需求，本项目系统开展时移地震约束油藏动态表征理论与方法研究。项目基于波动方程反射率法求解的时移地震差异非线性反演理论与方法、时移地震约束油藏数值模拟理论与方法等方面开展了基于波动方程反射率法求解叠前波形模拟方法、基于衰减补偿波动方程叠前波形反演方法、基于波动方程反射率法的时移地震差异反演方法、时移地震差异反演正则化、参数自适应获取方法及影响因素分析、时移地震差异反演约束油藏动态表征方法实现与试验等研究工作。非线性差异反演实现过程中，提出了基于贝叶斯框架的超-拉普拉斯块约束的时移地震差异反演工作流程，使反演的差异弹性参数的垂向梯度上具有更强的稀疏性，能够形成更加清晰的差异参数块状边界。研究了GCV及改进的L曲线方法来自适应求取正则化参数，减少人为设定带来的误差，使得反演结果的精度、稳定性以及效率都得到提升，实现了相关算法，并对比了本项目研究方法与其他方法的优缺点，提出了算法的有效适用范围。项目按照研究计划，完成了项目研究内容，达到了预期目标。