Fault monitoring and fault tolerant control in distillation columns

Abstract

An investigation of fault tolerant control of distillation columns under faulty sensors and actuators is presented in this work. Real-time sensor and actuator fault detection, propagation and accommodation are all investigated. Dynamic principal components analysis is used to promptly and effectively detect and isolate actuator and sensor faults. Alternative control strategy is then implemented to accommodate the faults. Specifically, fault tolerant inferential control is employed to accommodate sensor fault using soft sensor developed through dynamic principal component regression. Dual composition control strategy used for normal column operation is switched to one-point control to accommodate actuator fault utilizing the remaining healthy actuator. The main contributions of this paper is the application of fault tolerant inferential control and one-point control strategy to accommodate sensor and actuator faults respectively in the distillation column operation. The effectiveness of the proposed approaches are demonstrated on a simulated methanol-water separation column.