Model-based fault diagnosis via structural analysis of a reverse osmosis plant

• Autor: Göpfert, Johannes Georg
• Formato: tesis de maestría
• Fecha de Publicación: 2021
• Institución: Pontificia Universidad Católica del Perú
• Repositorio: PUCP-Tesis
• Lenguaje: inglés
• OAI Identifier: oai:tesis.pucp.edu.pe:20.500.12404/19043
• Enlace del recurso: http://hdl.handle.net/20.500.12404/19043
• Nivel de acceso: acceso abierto
• Materia: Agua--Tratamiento--Ósmosis inversa; Modelos matemáticos; Fallas estructurales; https://purl.org/pe-repo/ocde/ford#2.02.03

Water desalination is one approach to force water scarcity. One of the processes used for desalination is reverse osmosis. Like other systems, a reverse osmosis plant is susceptible to faults. A fault can lead to a loss of efficiency, or if the fault is severe to a total breakdown. Appropriate measures can minimize the impact of faults, but this requires in time fault detection. The following thesis shows a proposal for an online fault diagnosis system of a reverse osmosis plant. For the model-based approach, a mathematical model of a reverse osmosis plant has been developed. The model contains a new approach for modeling the interaction between the high-pressure pump, the brine valve, and the membrane module. Furthermore, six faults considered for fault diagnosis have been modeled. Two of the faults are plant faults: The leakage of the feed stream and membrane fouling. The other four faults are sensor or actuator malfunctions. The fault diagnosis system is developed via structural analysis, a graph-based approach to determine a mathematical model’s overdetermined systems of equations. With the structural analysis, 73 fault-driven minimal structurally overdetermined (FMSO) sets have been determined. The results show that all six faults are detectable. However, two faults are not isolable. Five of the FMSO sets have been chosen to deduce the residuals used for online fault detection and isolation. The simulations demonstrate that the calculated residuals are appropriate to detect and isolate the faults. If one assumes that only the considered faults occur, it is possible to determine some faults’ magnitude.