Full text loading...
-
Ensemble Sampling with EnKF for Fast and Efficient Uncertainty Quantification
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 78th EAGE Conference and Exhibition 2016, May 2016, Volume 2016, p.1 - 5
Abstract
Reservoir characterization is a process to find unknown reservoir properties using data available. Successful reservoir characterization is important to make a reasonable decision in petroleum industry. Ensemble Kalman Filter (EnKF) was suggested for non-linear optimization by Evensen. It uses data sequentially to assimilate reservoir parameters. Multiple models are carried out for EnKF so that it is useful to quantify uncertainty ranges of reservoir parameters.
In EnKF, reservoir performances are well-predicted when many ensembles are used. Researchers have suggested that at least 100 ensembles are needed for reliable results. Not only several ensembles to be used, but also repetitive assimilation steps of EnKF increase total simulation time.
In this paper, we propose an ensemble sampling method to improve reservoir predictions as well as to reduce simulation time. We sample the ensembles using principal component analysis (PCA) and K-means clustering. Results are compared to cases with and without the sampling scheme. For the fast and reliable history matching, we demonstrate effectiveness of our sampling method.
Its purpose is to reduce total simulation time as well as to maintain prediction quality of reservoir properties. By choosing ensembles through the proposed method, we can carry out fast and reliable history matching with EnKF.