Improvement of Localization Effect on Region Based Covariance Localization Ensemble Kalman Filter Method using Dynamic Parameters

摘要

Region based covariance localization ensemble Kalman filter is a method that incorporating the information of region to ensure that the updated parameters honor the region models such as facies, flow unit, rock type model, etc. Since, the model updated under specified regions, the adjacent parameters would not maintain its spatial correlation if it is under different regions. Therefore, the algorithm could freely update the parameters within the region without considering the values in another region. This approach would fit best in history matching that target reservoir-wide area. On the contrary, the significance of the fluid dynamics rarely follows such regions. The affected areas that influenced the production data is governed by the physics of fluid flow which incorporate the fluid types, relation of rock-fluid properties and so on. Since, history matching use production data as a measurement data, the parameters should only occur in the areas that affected by fluid flow in reservoir. These areas usually smaller than the area provided by regions model. Thus, it could be used to improve localization effect. In this study, we explore the formulation of localization based on the behavior of pressure and fluid flow combined with region based covariance localization ensemble kalman filter. The results show that, the combination of both methods could improve the localization effect while maintaining the defined regions. This method could be useful to improve the area within the wells that affects directly to the production forecast.