Automatic and adaptive localization for ensemble-based history matching

Abstract

Ensemble-based history matching methods are among the state-of-the-art approaches to reservoir characterization. In practice, however, they often suffer from ensemble collapse, a phenomenon that deteriorates history matching performance. To prevent ensemble collapse, it is customary to equip an ensemble history matching algorithm with a certain localization scheme.In a previous study, the authors propose an adaptive localization scheme that exploits the correlations between model variables and simulated observations. Correlation-based adaptive localization not only overcomes some longstanding issues (e.g., the challenges in handling non-local or time-lapse observations) arising in the conventional distance-based localization, but also is more convenient and flexible to use in real field case studies.In applications, however, correlation-based localization is also found to be subject to two problems. One is that, it requires to run a relatively large ensemble in order to achieve decent performance in an automatic manner, which becomes computationally expensive in large-scale problems. As a result, certain empirical tuning factors are introduced to reduce the computational costs. The other problem is that, the way used to compute the tapering coefficients in the previous study may induce discontinuities, and neglect the information of certain still-influential observations for model updates.The main objective of this work is to improve the efficiency and accuracy of correlation-based adaptive localization, making it run in an automatic manner but without incurring substantial extra computational costs. To this end, we introduce two enhancements that aim to avoid the aforementioned two problems, namely, empirical tuning and discontinuities. We apply the resulting automatic and adaptive correlation-based localization with these two enhancements to a 2D and a 3D cases investigated in the previous study, and show that it leads to better history matching performance (in terms of efficiency and/or estimation accuracy) than that is achieved in the previous work.