石油学报 ›› 2013, Vol. 34 ›› Issue (5): 947-953.DOI: 10.7623/syxb201305016

• 油田开发 • 上一篇    下一篇

油藏开发生产鲁棒优化方法

赵辉1, 李阳2, 康志江2   

  1. 1. 长江大学石油工程学院 湖北武汉 430100;
    2. 中国石油化工股份有限公司 北京 100728
  • 收稿日期:2013-01-05 修回日期:2013-04-08 出版日期:2013-09-25 发布日期:2013-08-01
  • 通讯作者: 赵辉
  • 作者简介:赵辉,男,1984年3月生,2006年毕业于中国石油大学(华东),2011年获中国石油大学(华东)油气田开发工程专业博士学位,现为长江大学石油工程学院副教授,主要从事油气田开发、油藏工程及优化控制工程方面的研究。Email:zhaohui-712@163.com
  • 基金资助:

    国家重大科技专项(2008ZX05024-004)及国家自然科学基金项目 (No.61004095)资助。

Robust optimization in oil reservoir production

ZHAO Hui1, LI Yang2, KANG Zhijiang2   

  1. 1. College of Petroleum Engineering, Yangtze University, Wuhan 430100, China;
    2. China Petroleum & Chemical Corporation, Beijing 100728, China
  • Received:2013-01-05 Revised:2013-04-08 Online:2013-09-25 Published:2013-08-01

摘要:

将油藏模型系统的不确定性考虑到优化设计中,建立了油藏生产鲁棒优化控制模型。针对伴随法求解模型梯度过于复杂的局限性,提出了两种基于随机扰动近似法(SPSA)的梯度的计算方法,然后利用对数变换和投影梯度法将所得梯度投影到可行方向上进行迭代求解,最终获得满足生产约束条件的最优控制方案。计算实例表明,其取得了与伴随法相一致的优化结果,且最终控制方案能够有效地改善油田开发效果,降低生产的风险性。该方法实现过程简单,能够和各种油藏模拟器相结合,为解决实际油藏的生产优化问题提供了新思路。

关键词: 生产优化, 鲁棒优化, 最优控制, 随机扰动近似, 投影梯度

Abstract:

Considering the uncertainty of reservoir model systems in the optimal scheme design, a robust optimal control model for reservoir production was established. Aiming at the limitation of excessive complication in solving the model gradient by the adjoint-based method, two algorithms were developed to estimate the gradient based on the simultaneous perturbation stochastic approximation (SPSA) method. Then the estimated gradient was iteratively projected to the feasible direction using the log-transformation and projection gradient methods, until the final optimal control content with production constraints was obtained. The experimental study showed that the final result obtained by the proposed approach is consistent with the adjoint-based algorithm and the corresponding optimal control plan can significantly improve the effect of waterflooding development with lower risk. It also showed that the proposed method is fairly simple and can be coupled with any numerical simulator to solve optimization problems in practical reservoir production.

Key words: production optimization, robust optimization, optimal control, simultaneous perturbation stochastic approximation (SPSA), projection gradient

中图分类号: