Past Event: Oden Institute Seminar

The Ensemble Kalman Filter for Data Assimilation

Al Reynolds, University of Tulsa

Abstract

This talk discusses the use of the ensemble Kalman filter (EnKF) for integrating production and seismic dynamic data into reservoir models to estimate reservoir model parameters and quantify uncertainty in reservoir parameters and performance predictions. We provide an overview of EnKF and the underlying theoretical assumptions on Gaussianity and linearity necessary to show EnKF provides a correct assessment of uncertainty. Then we suggest methods to improve its performance for highly nonlinear problems and discuss covariance localization as a means to mitigate sampling errors and avoid filter divergence. Several examples based on reservoir simulation models are used to illustrate the theoretical ideas and the flexibility and efficiency of EnKF. Speaker Bio Dr. Albert C. Reynolds is the Director of TUPREP, McMan Chair Professor of Petroleum Engineering and Professor of Mathematics. A faculty member at the U. of Tulsa since 1970, Reynolds has served as associate graduate dean, associate director of research, and Chairman of the Dept. of Petroleum Engineering. He holds a BA degree from the U. of New Hampshire, an MS degree from Case Institute of Technology, and a PhD degree from Case Western Reserve University, all in Mathematics. Reynolds received the 1983 SPE Distinguished Achievement Award for Petroleum Engineering Faculty, 2003 SPE Reservoir Description and Dynamics Award, 2005 SPE Formation Evaluation Award, and Society of Petroleum Engineers Outstanding Technical Editor Award, 2008. He is a SPE Distinguished Member since 1999. His research areas include, reservoir characterization, well testing and reservoir simulation