论文部分内容阅读
大多数油藏模拟研究都局限于静态范畴,因时间和预算限制,未知参数明显减少,这容易导致低估预测的不确定性或作出不明智的决策。马尔可夫链的蒙特卡洛(MCMC)方法已被用于静态研究以对预测参数空间的不确定性进行严格的量化考察。但这些方法在长期性和系列稳定性方面存在局限。文中将MCMC应用于实时油藏建模。较之传统方法,MCMC方法在某一特殊时间点上应用更少的模型来实现合理的概率预测,它也提供了一种随时校准不确定性预测的机制。
Most of the reservoir simulation studies are confined to the static category, due to time and budget constraints, a significant reduction in unknown parameters, which can easily lead to underestimation of forecast uncertainty or make unwise decisions. The Monte Carlo (MCMC) method of Markov chains has been used for static studies to rigorously quantify the uncertainty of the prediction parameter space. However, these methods have limitations in terms of long-term and series stability. In this paper, MCMC is applied to real-time reservoir modeling. Compared with the traditional method, the MCMC method uses fewer models to achieve reasonable probability prediction at a special time point, and it also provides a mechanism to calibrate the uncertainty prediction at any time.