Control system based loop and process monitoring

Chen, Jun (2000) Control system based loop and process monitoring. PhD thesis, University of Glasgow.

Abstract

For the sake of both economy and safety, the ability to diagnose a fault or disturbance is of great interest for an operator/engineer in process industries. To be practicable an on-line system with this capability must contain a suite of methods because no single method is likely to diagnose all possible faults. This thesis aims to contribute one novel component to this suite. This thesis envisages the situation where the detection and diagnosis of faults and disturbances would be distributed to separate modules, each associated with the individual control systems located throughout a plant. In particular the thesis addresses those plants whose control systems inherently eliminate steady state error. Thus it seeks to address the large proportion of process plants that have proportional plus integral action as standard. By reasoning about changes in steady state an approach is proposed that requires very little process specific information and therefore should be attractive to control systems implementers who seek economies of scale. Because the approach can be implemented as modules that are largely based on standard control systems, the implementation can be configured and commissioned using various generic programmes and hence has the potential to be commercialised. The approach is applicable to virtually all types of process plant, whether they are open loop stable or not, have a type number of zero or not and so on. It is founded on the application of both signed directed graph (SDG) and control systems theory to single and cascade control systems with integral action. This results in the derivation of cause-effect knowledge and fault isolation procedures that take into account factors like interactions between control systems, and the availability of non-control-loop- based measurements. Following on from a survey of the more relevant methods published in the literature, a theoretical analysis is carried out of what happens to control systems when they are subjected to various faults and disturbances. The main purpose is to derive equations to describe how control systems respond in the steady state to these occurrences. Although providing a foundation, these equations are unlikely to be suitable for direct use and a cause-effect analysis of the faults/disturbances involving signed-directed- graph (SDG) representation is then pursued. This leads to a search and test strategy for fault isolation involving interacting control systems, minimal knowledge acquisition and knowledge evolution. Since the approach is based on steady state deviations, a steady state change detection algorithm is proposed. The approach is tested by applying it to a continuous stirred tank reactor (CSTR) and to the Tennessee Eastman (T-E) process benchmark. Some recommendations are made for integrating the approach into a commercial software tool. In principle, the approach can form the basis for the diagnosis of faults/disturbances in both control systems and in the process itself. One of the key features is that the approach can work at different levels of detail. Diagnosis is based on knowledge of the signs of steady state interactions (gains) between individual control loops, non- control-system-related measurements and on the steady state effects of disturbances. Both faults and disturbances (e.g. a load change) can be diagnosed, although diagnostic detail, i.e. degree of isolation, is clearly dependent on the measurements and knowledge that is available. The concept of a distributed, control system based approach to the diagnosis of faults and disturbances, its development and application to various processes are all original, as are the integration aspects.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Keywords: Industrial engineering.
• Subjects: T Technology > T Technology (General)
• Colleges/Schools: College of Science and Engineering
• Supervisor's Name: Howell, Dr. John
• Date of Award: 2000
• Depositing User: Enlighten Team
• Unique ID: glathesis:2000-71275
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 10 May 2019 10:49
• Last Modified: 27 Oct 2022 11:14
• Thesis DOI: 10.5525/gla.thesis.71275
• URI: https://theses.gla.ac.uk/id/eprint/71275