Multivariable closed - loop identification with constrained MPC control - a case study in an industrial depropanizer column

• Autores: Tisza Contreras, Juan F., Sotomayor, Oscar A. Z., Odloak, Darci, Moro, Lincoln F. L., Álvarez, Hernán D.
• Formato: artículo
• Fecha de Publicación: 2007
• Institución: Universidad Nacional de Ingeniería
• Repositorio: Revistas - Universidad Nacional de Ingeniería
• Lenguaje: español
• OAI Identifier: oai:oai:revistas.uni.edu.pe:article/379
• Enlace del recurso: https://revistas.uni.edu.pe/index.php/tecnia/article/view/379
• Nivel de acceso: acceso abierto
• Materia: identificación en lazo cerrado; control predictivo; mantenimiento del modelo del proceso; evaluación del desempeño del control; columna depropanizadora; closed-loop identification; model predictive control; process model maintenance; control performance assessment; depropanizer column

The objective of this work is the re-identification of the process model that is used in existing predictive controllers (MPC) using closed-loop operation data. The controller is assumed to have a two-layer structure, where in the upper layer a simple economic optimization algorithm determines a set of optimal steady-state values ​​("targets"), which are passed to the MPC for implementation. This is the case for several commercial MPC packages applied in the industry. This paper focuses on the case where the model represents significant benefits in the MPC re-commissioning procedure. A new methodology is proposed to excite the system in closed loop, introducing persistent excitation signals in the objective function of the upper layer of the MPC. This strategy allows the continuous operation of the system, respecting the constraints of the process and meeting the specifications of the product. The application of the method is illustrated through numerical simulations of a typical depropanizer column of the oil industry. The method is simple to implement in existing MPC controllers and the results show its great potential for practical applications.