Process operational safety via model predictive control: Recent results and future research directions

Abstract The concept of maintaining or enhancing chemical process safety encompasses a broad set of considerations which stem from management/company culture, operator procedures, and engineering designs, and are meant to prevent incidents at chemical plants. The features of a plant design that take action to prevent incidents on a moment-by-moment basis are the control system and the safety system (i.e., the alarm system, safety instrumented system, and safety relief system). Though the control and safety systems have a common goal in this regard, coordination between them has been minimal. One impediment to such an integrated control-safety system design is that the traditional industrial approach to safety focuses on root causes of incidents and on keeping individual measured variables within recommended ranges, rather than seeking to understand incidents from a more fundamental perspective as the result of the dynamic process state evolving to a value at which consequences to humans and the environment occur. This work reviews the state of the art in control system designs that incorporate explicit safety considerations in the sense that they have constraints designed to prevent the process state from taking values at which incidents can occur and in the sense that they are coordinated with the safety system. The intent of this tutorial is to unify recent developments in this area and to encourage further research by showcasing that the topic, though critical for safe operation of chemical processes particularly as we move to more tightly integrated and economics-focused operating strategies, is in its infancy and that many open questions remain.