Economic Model Predictive Control: Handling Preventive Actuator and Sensor Maintenance and Application to Transport-Reaction Processes
- Author(s): Lao, Liangfeng
- Advisor(s): Christofides, Panagiotis D
- et al.
In chemical process industries, maintenance costs may comprise of up to 20−30%
of the operating budget, and therefore, improving maintenance practices can significantly impact plant economics and reduce production losses. Developing so-called “Smart” preventive maintenance policies/systems of key manufacturing components, especially those that may cause process upsets, losses and downtime, is therefore an important task. A high percentage of the day-to-day preventive maintenance tasks in the chemical process industry deals with control actuators and measurement sensors employed by process control systems. Motivated by these considerations, the first part of this dissertation focuses on the development of methods for integrating the on-line preventive maintenance of actuators and sensors with advanced process control system design. To accomplish these preventive maintenance tasks, economic model predictive control (EMPC), that optimizes economic process performance over an operating horizon by employing a dynamic process model to predict the evolution of the process, is employed to maintain stable operation of a process while dictating an economically optimal operating policy with respect to varying numbers of control actuators and measurement sensors. Novel EMPC schemes are developed that explicitly account for scheduled preventive control actuator/sensor maintenance programs, process economics and feedback control. In the second part of this dissertation, EMPC of transport-reaction processes is considered for the first time. Compared with lumped-parameter processes, no work has been done on the problem of designing EMPC for transport-reaction processes modeled by partial differential equations (PDEs). The thesis proposes EMPC schemes those are formulated on the basis of suitable reduced-order models to ensure input and state constraint satisfaction and economics optimization for both parabolic and hyperbolic PDEs. Finally, the thesis concludes with the presentation of multiscale, computational fluid dynamics modeling framework for an industrial-level steam methane reforming unit.