Paper Number 26109-MS
Title The Use of Optimal Control Technique To Validate Reservoir Simulation Results
Authors Elemo, R.O., North Dakota State U.,; Lawal, A.S., BP Exploration (Alaska) Inc.
Source SPE Western Regional Meeting, 26-28 May 1993, Anchorage, Alaska
ABSTRACT:
The advent of fast digital computer has made numerical modeling of petroleum reservoirs very attractive, starting with a landmark paper published in 1953 by G.H. Bruce, et al. Reservoir simulation, a major numerical modeling technique, has now become an important tool used to study the performance of oil and gas reservoirs of various complexity where several variables are involved, and therefore, analytical solution may not be appropriate and/or closed form solution is not feasible. A number of sources of error in reservoir simulation application has been identified and documented in literature. One major source of error is frequent use of inaccurate reservoir description data. Therefore, model validation and calibration, meant to ascertain the correctness of the simulator and the applied description data, are essential parts of reservoir performance numerical modeling. Model validation and calibration are usually done by history matching of the field performance data and simulator calculated performance responses. In the last couple of decades, there has been considerable interest in the development of automatic history matching algorithms. Automatic history matching is preferred to the traditional trial-and-error approaches despite the fact that the automatic routines require some additional iterative optimization routines to the basic simulator because the investment of additional optimization code development time and computational CPU time is worth the effort.
This paper presents an efficient automatic history matching technique that has been found suitable for validating and calibrating reservoir simulation results for various types of field performances. Validation and calibration was done by use of a developed algorithm, based on optimal control theory, to perform automated matching of field observed performance data and reservoir simulator calculated data. Results of various studies involving black oil, multi-phase and compositional simulators for stress-sensitive and non-stress sensitive reservoirs that have been reported in literature were validated. In previously reported work by R.O. Elemo, history matching models have been developed for validating and calibrating numerical reservoir models, thereby estimating some pertinent description data of some types of reservoirs. Reliability of description and performance data estimates from reservoir simulations was also studied and reported. The technique discussed in this paper is an extension of these previous studies and is applied to validate reported field performance for reservoirs of various complexity. While previous studies have concentrated on individual types of field performance, the work reported herein provides an efficient method for verifying the accuracy of reservoir simulators for most types of reservoirs encountered in oil and gas fields.
题目:验证油贮仿真结果的优化控制技术的应用
快速数字计算机的出现为油贮的数值模型技术增色不少,自从1953年G.H. Bruce,etal 发表一篇有关数值仿真的重要论文开始,一项重要的数字模型技术-----油贮仿真,现在已成为研究包含多变量的复杂油贮、气贮运行情况的重要工具。而分析解法和闭合解法相对不太可行、不太适用。油贮仿真应用中的大量错误的来源已被证实并列入相关书籍。一项重要的错误来源如:常用错误油贮特性数据。因此模型验证与校准将验证模拟器和应用特性数据的正确性,并且做为油贮运算数字模拟的重要部分。模型验证与校准通常由野外使用特性的历史拟合与仿真运算反馈来完成.在最近几十年,自动历史拟合算法的发展带来了巨大的收益。这种算法更倾向于传统的经验算法,因为附加的优化编码运行时间和计算机处理器运行时间较短,因此尽管这种算法的自动例行程序需要一些附加的迭代优化过程,但还是被应用。
本文主要阐述了一项有效的自动历史拟合技术,该技术主要适用于验证和校准各种野外运行油贮的仿真结果,验证和校准功能由基于优化控制理论的高级算法完成,并且运行野外监视运行结果和油贮仿真计算数据的自动历史拟合。书中所列的包括重油、多相、和组分模拟的压力感应和零压感应油贮的各种研究结果已得到验证。在R.O. Elemo做的早期工作中,历史拟合模型被用来验证和校准数字油贮模型,因此可以估算部分类型油贮特性数据。油贮仿真的运算数据和可靠性也正在被研究和报导。本文中所讨论的技术是该领域早期研究工作的扩展和延伸,并且用来验证所报导的各种复杂情况的油贮野外使用特性.然而早期的研究主要集中于个别类型的野外使用特性,而其所作的工作为验证油气领域大部分类型贮罐仿真的正确性提供了有效的方法。