基于四维变分的油藏数模自动历史拟合方法  

Automatically matching of reservoir development history based on 4D-variational data assimilation

作　　者：杨元明[1] 李治平[1] 赖枫鹏[1] 黄甫[2] 

机构地区：[1]中国地质大学(北京)能源学院非常规天然气能源地质评价与开发工程北京市重点实验室,北京100083 [2]长庆油田分公司第二采油厂,甘肃庆阳745100

出　　处：《西安石油大学学报（自然科学版）》2014年第2期70-73,2,共4页Journal of Xi’an Shiyou University（Natural Science Edition）

基　　金：国家科技重大专项课题"新一代数值模拟软件系统"(编号:2011ZX05009-006)

摘　　要：应用基于最优控制理论的四维变分资料同化方法,将油藏初始参数场和历史拟合时间段内的动态资料作为约束条件,将历史拟合过程转化为非线性规划中求解定义在多维空间的最优化问题,采用Levenberg-Marquardt优化算法求解目标函数极小值,从而建立基于四维变分资料同化的油藏自动历史拟合模型。将某小断块油藏渗透率和油水相渗曲线作为被估参数,自动拟合油藏产油量和含水率历史数据。结果表明:应用四维变分资料同化方法进行油藏数模自动历史拟合是可行的,可大幅提高历史拟合的效率。An objective function for the automatically matching of reservoir development history is established using four-dimensional variational data assimilation method,and the initial parameter field and the dynamic data in the matching time interval of the reservoir are as the constraints,which will make reservoir history matching process transfer an optimization problem in a multidimensional space. The minimum of the objective function can be solved using Levenberg-Marquardt optimization algorithm. Taking the permeability and the oil-water relative permeability curve of a small faulted oil reservoir as estimated parameters,the history data of the oil production and the water cut of the reservoir are automatically matched. The results show that the automatically matching method of reservoir development history based on 4D-variational data assimilation is feasible,and it can greatly improve the efficiency of reservoir history matching.

关 键 词：油藏数值模拟 自动历史拟合 四维变分 优化算法 拟合步骤 状态向量 

分 类 号：TE319[石油与天然气工程—油气田开发工程]