On selection of controlled variables for robust reservoir management

Reservoir management is subject to pronounced uncertainty, as the subsurface reservoir conditions are often poorly known, especially at the onset of production. Robust optimization is a method for handling uncertainty by optimizing control inputs over a series of model realizations. Well controls can be specified in several ways, most often by liquid rate or bottom hole pressure (BHP), and previous studies have used various types of well control selections. In this paper, we study how different choices of control variables affect robustness with respect to uncertainties. The analysis is supported by self-optimizing control, a well-known process control concept, which allows for systematic analysis of these effects. We observe that when the permeability is the main uncertain parameter, controlling both injectors and producers by rate is the most robust alternative. However, we also observe that this choice is less advantageous when other parameter uncertainties dominate.