System identification of hydrocarbon reservoirs

Project J.F.M. van Doren (Jorn), prof. J.D. Jansen
Keywords:Identification and estimation, Process technology, Distributed and large-scale systems, Reservoir engineering
Sponsored by:Shell

Closed-loop reservoir management (see Figure 1) is an emerging topic in the oil industry. In this approach the system is the hydrocarbon reservoir including wells and surface facilities, and the goal is optimal performance of the system. Performance can mean a higher net present value of the reservoir or the reduction of uncertainty. Closed-loop reservoir management contains aspects such as model reduction, model-based optimization and system identification and updating.

The focus of this project is on system identification and parameter estimation within closed-loop reservoir management, with the aim to quantify and reduce the uncertainty of the reservoir model by assimilating information contained in measurements. In reservoir engineering this is called "history matching". The reservoir model is large-scale (possibly contains millions of states and parameters), nonlinear and has multiple inputs and outputs. One method that can deal with these kind of models is the Ensemble Kalman filter. This method is investigated together with possible improvements.

Figure 1: Closed-loop reservoir management

© Copyright Delft Center for Systems and Control, Delft University of Technology, 2017.