Use this URL to cite or link to this record in EThOS:
Title: Adaptive monitoring of health-state and performance of industrial centrifugal compressors
Author: Cicciotti, Matteo
ISNI:       0000 0004 6423 1406
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
Tens of thousands of centrifugal compressors are installed worldwide in chemical and petrochemical plants. The performance of these compressors degrade during the operation as a consequence of effects such as fouling, erosion, corrosion, and abrasion. A performance monitoring method that could detect and assess the magnitude of the degradation would be greatly beneficial to schedule production and maintenance leading to an economic profit for the operation. After searching the literature, it was concluded that such a method is yet to be developed for industrial centrifugal compressors. This thesis shows the development of an adaptive monitoring framework that simultaneously considers the degradation of the state of health and identifies the malfunctioning of sensors. Indeed, because of degradation, the state of the components changes over time and this change can be observed in the measurements, however, the measurements are also affected by random or persistent errors. The approach adopted in this thesis aims at reconciling two distinct features that are normally separated: (1) to account for the degraded state by recursively matching the model to the newly available measurements, and (2) to correct the measurements when these are biased by making use of the model. The leading hypothesis is that this aim can be achieved by employing models that establish causality relationships between the state of the components and the measured variables together with mathematical optimization methods. The thesis demonstrates: how performance can be systematically modelled even though a compressor is installed at the industrial site, how the degradation of performance can be detected and quantified in real-time, and finally, how the effects of degradation on performance can be modelled and monitored while simultaneously detecting and correcting sensor faults. The methods have been successfully applied to a 10 MW centrifugal compressor. When monitoring and modelling the degradation of its performance, it was observed that the difference between the performance of the compressor in undegraded and degraded state depends on the operation conditions. The implication of this observation is that the state-of-the-art practice of scaling the manufacturer maps to obtain degraded maps can lead to misleading conclusions about the performance of the degraded compressor.
Supervisor: Martinez-Botas, Ricardo ; Thornhill, Nina Sponsor: European Union
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral