Improving Economic Operation Performance of Self-Optimizing Control Structure by a Supervisory Control Layer.

BOTTARI, A.; BRACCIA, L.; ZUMOFFEN, D.

Buenos Aires

Congreso; WCCE11 - 11th WORLD CONGRESS OF CHEMICAL ENGINEERING; 2023

Asociación Argentina de Ingenieros Químicos

In the chemical and petrochemical industry, the involved processesare generally non-linear, multivariable, and subject to externaldisturbances. Finding the operating point that satisfies theproduction requirements (rate and quality) while the costs, energyconsumption and/or environmental pollution are minimized, withoutviolating any operational or safety restrictions, certainly requiresa deep analysis and implementing a proper strategy to this end. Thetwo-layer decision control system is the classical approach toaddress the optimal process operation. This system includes anoptimization top layer and a regulatory control bottom layer. Whilethe regulatory control layer rejects high-frequency disturbances andguarantees a safety and stable operation, the supervisory levelcommonly contains algorithms to pursue the changing optimal operatingpoint of the process (because of external disturbance, plant-modelmismatch or changes due planning and programming). Generally, theprocess control system is addressed by selecting the regulatorycontrol layer based on stability and controllability indices.However, the economic criteria or the existence of higher levels ofcontrol are ignored in these approaches. In this context, theeconomic-based regulatory control structure selection methods, likethe back-off (BO) approach and the self-optimizing control (SOC),have been appeared in the last decades. In the BO and SOC methods theoptimal control loops and set points values are selected by solving amathematical programing problem with an economical operation costobjective. Moreover, self-optimizing control literature shows thatthe economic performance can be improved using virtual controllablevariables that linearly combine output-input signals. However, afixed control structure and set points values strategy may be farfrom an optimal operation for processes under changing conditions. Recently in [1] theauthors formulate a mathematical programing problem whichincorporates the supervisory level in the search for economic-optimalregulatory control loops using a classical input-output pairing.Based on their findings, it is shown that the use of classic controlstructures with a supervisory layer might still be far from an idealoptimal economic operation for some processes. In this sense, the mainobjective of the present work is the economical optimal hierarchicalcontrol structure system design, by considering virtual controllablevariables.A novel mathematicalprograming problem for control structure selection is developed andtested for a (simulated) reactor separator recycle plant by usingavailable commercial software. Results show that the use of virtualcontrollable variables, based on lineal combination of measurablesvariables, together with a supervisory control layer could be theproper strategy to improve the economic performance of processesunder changing operation conditions.p { line-height: 115%; margin-bottom: 0.25cm; background: transparent }