Robustness of a class of bias update controllers for blending systems

The aim of this paper is to study the robustness of a class of control strategies with bias update for blending operation. The class of blender control systems consists of a linear programming endowed with a bias updating scheme to reduce (nonlinear) model/plant mismatch within a real-time operation procedure. It is proven that these control strategies are robust against fluctuations in the stream properties and blending nonlinearities as long as deviations from the ideal mixing rules are sufficiently small. Discussions are concluded with a demonstration of the results on simulations of a simple gasoline blending control problem.