DEVELOPMENT OF REGION BASED COVARIANCE LOCALIZATION ON ENSEMBLE KALMAN FILTER FOR HISTORY MATCHING PROCESS IN RESERVOIR SIMULATION

• Main Author: AMBIA (NIM: 32212009), FAJRIL
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/22063
• Institution: Institut Teknologi Bandung

The Ensemble Kalman Filter (EnKF) method provides ability to fine tune static <br /> <br /> parameter distribution of reservoir, such as permeability. The distribution of <br /> <br /> permeability could be corrected along with the addition of new production data; <br /> <br /> therefore, the model could be updated continuously. This algorithm is also very <br /> <br /> efficient since it only needs the information from the previous timestep to update the <br /> <br /> model. Hence, reducing the computational time. However, the difficulty of using this <br /> <br /> EnKF algorithm is that the number unknown variables are significantly more than the <br /> <br /> available equations and turns the case into underdetermined problem. As a result, the <br /> <br /> obtained solution may not be good enough compared to the model yang sebenarnya. <br /> <br /> In order to reduce the effect of underdetermined system, the EnKF algorithm needs to <br /> <br /> be modified by imposing additional boundary conditions. One of the proposed <br /> <br /> boundary condition that could be used is facies information which guiding the Kalman <br /> <br /> Gain calculation, combined with implementation of spatial correlation model. The <br /> <br /> combination would give a correlation matrix that represents facies model. <br /> <br /> Additional effort to improve the localization effect is by introducing modifier of <br /> <br /> dynamic parameters to modify Kalman Gain calculation. It is implemented through <br /> <br /> selecting parameters which related to the production, such as pressure and fluid flux. <br /> <br /> The result shows that region based covariance localization could improve the updating <br /> <br /> process without altering geological concept which makes it suitable to be implemented <br /> <br /> near well area.