Studies of condition monitoring methods for system health assessment : health diagnostics and prognostics

Abstract: Increasing interest in productivity, safety and environmental issues have highlighted the area of maintenance and reliability. The increasing cost of maintenance covers both preservation and sustainable exploitation of resources and awareness in maintaining equipment in a way to ensure return on investment both in the short and long run. The information obtained from condition monitoring of existing turbine, plant, rails and pumps can provide an important basis for dimensioning of future systems and components. The main objective of this research work is to develop and apply methods for efficient condition monitoring, and hence reduce maintenance costs and provide a framework for development and implementation of computer based decision tools. Furthermore, methods enabling existing process data and cost effective transducers to be used together with modern data analysis and diagnostic tools for condition monitoring of complex mechanical systems have been examined and prototypes developed. The areas of investigation covered in this work are hydropower turbines, rails and the main cooling pumps in a nuclear power-plant. The interest in diagnostics for hydropower turbines was driven by the obvious risk of contamination of water by oil leaks and expensive refurbishments caused by wear of the Kaplan turbine vane bearings. The intrest in risk analysis was motivated by Vattenfall's intrest in gaining knowledge about the state of all turbines in the company. The aim of this project was to develop a generic model of hydropower turbine behavior using physics-based models based on material properties, load tolerances, etc.. An important question was whether it was possible to predict the wear rate and plan predictive replacement or maintenance. A systematic approach to find failure modes, their effects, their causes and consequences in combination with Fault Tree Analysis was needed. The objective of this project was to examine a systematic approach to map failure modes and their causes in an hydropower turbine. We have restricted the study to turbine units of the Kaplan, Francis and tube types. The objective of the study concerning rail track was to develop methods and equipment for detection of surface damage in rail track rail in addition to the present system of practice of visual examination. The equipment developed has to be used to obtain objective statistical data for evaluating maintenance methods and efforts. We have restricted the study to spalling and headchecks on the rail head surface and running edge. The method developed enables measurements of different types of surface damage such as spalling and shelling to be made with inductive transducers sensitive to the distance to the measured object. The assumption here is that the damage being detected is characterized by the absence of material from rail surface. The main object of condition monitoring of the cooling pumps was to be able to detect bearing wear in order to be able to plan and carry out restoration well ahead of breakdown or bearing seizure. The study was restricted to the main cooling pump motor and its main bearings. Condition monitoring of the pumps was done using a method based on current measurements. Analysis of the currents on the main cooling pump of the power plant proved that it is possible to monitor the condition of the pump in spite of the presence of electronic frequency converters which distorts the signal.