Combined Dynamic Optimization, Optimal Control and Online Runaway Detection & Prevention under Uncertainty

Abstract

This contribution investigates the extent to which dynamic optimization and optimal control can provide reliable prediction and insure prevention of runaways under uncertainty. We limit our analysis to batch systems because both the issue of model uncertainty and that of hazardous loss of control are widespread within this type of processes. The paper shows that, surprisingly, a specific class of deterministic dynamic optimization and optimal control algorithms allows us to easily identify and prevent runaways, even when no accurate process model is available. The production process of 2-butyl propanoate via acid-catalyzed esterification of 2-butanol serves as case study to substantiate our claims.