Uncertainty analysis in optimizing control of binary distillation columns. Case study

The paper discusses three different approaches to uncertainty problem in static optimization task performed within optimizing control of binary distillation column. Deterministic, probabilistic and set-bounded description of the model and input uncertainty is explored. One solves optimisation task with active constraints on output variable. Follow-up controller stabilises the output at the constraint level, thus compensating the effects of uncertain solutions of the optimization task. The optimal profit is compared with expected values of the performance index computed on a basis of probabilistic representation of the uncertain quantities with dynamic compensation losses taken into account. The results point the deterministic approach to be the most useful.