Large rotating machinery, such as centrifugal gas compressors and pumps, are widely applied as crucial components in the petrochemical industries. To enable in-time and effective maintenance of these machines, the concept of a health indicator is arousing great interest.
A suitable health indicator indicates the overall health of the machinery and it is closely related to maintenance strategies and decision-making. It can be obtained either from near misses and incident data, or from real-time measured data. However, the existing health indicators have some limitations. On the one hand, the near misses and incident data may have been obtained from similar systems, reflecting population characteristics but not fully accounting for the individual features of the target system. On the other hand, the existing health indicators that use condition monitoring data, mainly focused on detecting incipient faults, and usually do not include financial cost factors when calculating the indicators. Therefore, there is the requirement to develop a single system "Health Indicator", that can show the health condition of a system in real-time, as well as the likely financial loss incurred when a fault is detected in the system, to assist operators on maintenance decision making. This project has developed such an integrated health indicator for rotating machinery.
The integrated health indicator described in this thesis is extracted from a novel condition-based risk assessment strategy, which can be regarded as an integration of risk-based maintenance with improved conventional condition-based maintenance, with financial factors taken into account. The value of the health indicator is that it directly illustrates the risk to the system (or equipment), including likely financial loss, which makes it easier for operators to select the optimal time for maintenance or set alarm thresholds given the specific conditions in their companies or plants.
This thesis provides a guide to set up an integrated maintenance model for large rotating machinery. It provides a useful reference for researchers working on condition-based fault detection and dynamic risk-based maintenance.