Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.773792
Title: Fault tolerant control system design for distillation processes
Author: Lawal, Sulaiman Ayobami
ISNI:       0000 0004 7961 0337
Awarding Body: Newcastle University
Current Institution: University of Newcastle upon Tyne
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The complexity and sophistication of modern control systems deployed in the refinery operation, particularly the crude distillation unit as a result of increasing demand for higher performance and improved safety, are on the increase. This growing complexity comes with some level of vulnerabilities, part of which is the potential failure in some of the components that make up the control system, such as actuators and sensors. The interplay between these components and the control system needs to have some built-in robustness in the face of actuator and sensor faults, to guarantee higher reliability and improved safety of the control system and the plant respectively, which is fundamental to the economy and operation of the system. This thesis focuses on the application of frugally designed fault tolerant control systems (FTCS) with automatic actuator and sensor faults containment capabilities on distillation processes, particularly atmospheric crude distillation unit. A simple active actuator FTCS that used backup feedback signal, switchable references and restructurable PID controllers was designed and implemented on three distillation processes with varying complexities { methanol-water separation column, the benchmark Shell heavy oil fractionator, and an interactive dynamic crude distillation unit (CDU) to accommodate actuator faults. The fault detection and diagnosis (FDD) component of the actuator FTCS used dynamic principal component analysis (DPCA), a data-based fault diagnostic technique, because of its simplicity and ability to handle large amount of correlated process measurements. The recon gurable structure of the PID controllers was achieved using relative gain array (RGA) and dynamic RGA system interaction analysis tools for possible inputs { outputs pairing with and without the occurrence of actuator faults. The interactive dynamic simulation of CDU was developed in HYSYS and integrated with MATLAB application through which the FDD and the actuator FTCS were implemented. The proposed actuator FTCS is proved being very e ective in accommodating actuator faults in cases where there are suitable inputs { outputs pairing after occurrence of an actuator fault. Fault tolerant inferential controller (FTIC) was also designed and implemented on a binary distillation column and an interactive atmospheric CDU to accommodate sensor faults related to the controlled variables. The FTIC used dynamic partial least squared (DPLS) and dynamic principal component regression (DPCR) based soft sensor techniques to provide redundant controlled variable estimates, which are then used in place of faulty sensor outputs in the feedback loops to accommodate sensor faults and maintain the integrity of the entire control system. Implementation issues arising from the e ects of a sensor fault on the secondary variables used for soft sensor estimation were addressed and the approach was shown to be very e ective in accommodating all the sensor faults investigated in the distillation units. The actuator FTCS and the FTIC were then integrated with the DPCA FDD scheme to form a complete FTCS capable of accommodating successive actuator and sensor faults in the distillation processes investigated. The simulation results demonstrated the e ectiveness of the proposed approach. Lastly, fault tolerant model predictive control (FTMPC) with restructurable inputs { outputs pairing in the presence of actuator faults based on preassessed recon gurable control structures was proposed, and implemented on an interactive dynamic CDU. The FTMPC system used a rst order plus dead time (FOPDT) model of the plant for output prediction and RGA and DRGA tools to analyse possible control structure recon guration. The strategy helped improve the availability and performance of control systems in the presence of actuator faults, and can ultimately help prevent avoidable potential disasters in the re nery operation with improved bottom line { Pro t. Overall, the proposed approaches are shown to be e ective in handling actuator and sensor faults, when there are suitable manipulated variables and redundant analytical signals that could be used to contain the e ects of the faults on the system.
Supervisor: Not available Sponsor: University of Lagos, Nigeria ; Petroleum Technology Development Fund (PTDF)
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.773792  DOI: Not available
Share: