基于随机配点法的油藏自动历史拟合与动态预测方法

项目名称： 基于随机配点法的油藏自动历史拟合与动态预测方法

项目编号： No.51204008

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 冶金与矿业学科

项目作者： 李恒

作者单位： 北京大学

项目金额： 23万元

中文摘要： 运用随机建模和数值模拟技术研究油藏数值模型的参数校正及不确定性量化。 一方面发展准确而高效的油藏随机模拟的方法，另一方面结合实际油气田生产实例，整合各 种有效的静态测量和动态生产数据，校正油藏地质模型，并且对生产预测中的不确定性进行 量化。运用随机配点方法进行正演问题的随机模拟，并将其结合集合卡尔曼滤波方法实现更 加高效的反演问题模拟。研究的核心思想是同时提高随机模拟方法的准确度和效率，并力求 保持与不同油藏模拟器的通用性和普适性，从而增加油藏模拟在大型油气田开发中的实用 性。摆脱传统的历史拟合方法对特定油藏模拟器源代码的依赖，解决众多随机模拟方法计算 成本高昂的问题，将随机模拟方法应用于大尺度油田的模拟。本研究课题旨在将油藏模拟的 自动历史拟合和不确定性量化有机结合起来，实现包含油藏模型校正和油藏动态预测的一套 完整的一体化模拟技术。

中文关键词： 随机建模；不确定性量化；随机配点；自动历史拟合；动态预测

英文摘要： In this project, techniques of stochastic modeling and numerical simulation are utilized to study the calibration of reservoir models and uncertainty quantification associated in reservoir simulation. Accurate and efficient approaches to stochastic reservoir simulation will be developed. On the other hand, field applications will be combined. The static and dynamic data from oil fields will be integrated as effective information to calibrate the reservoir models and quantify uncertainties in the reservoir performance forecasting. The probabilistic collocation method will be used for stochastic forward simulation, and the ensemble Kalman filter method will be combined for more efficient inverse-modeling. The main objective is to improve accuracy and efficiency of stochastic simulation at the same time. And the non-intrusiveness to reservoir simulators will be achieved in order to gain wide applications of reservoir simulation in real fields. The approach developed in this project is independent of specific reservoir simulators, and is computationally more efficient than traditional history matching methods. In this project, the automatic history matching and uncertainty quantification will be effectively integrated, and a comprehensive simulation approach to reservoir model calibration and dynamic prediction will

英文关键词： Stochastic modeling；uncertainty quantification；probabilistic collocation；automatic history matching；dynamic prediction