| Summary: | Today, maintaining a mining equipment is an expensive affair. The two factors which influence significantly the occurrences of breakdowns are the operating reliability of equipment and the related human factors. The research work reported in the thesis is an attempt to control the occurrences of breakdowns and thereby the costs of maintenance by studying the operational reliability of the critical mining equipment. For the purpose of the reliability investigation, a fleet of load-haul-dump (LHD) machines deployed at Kiruna Iron Ore mine, Kiruna was selected. The times between successive failures of these machines and their main assemblies, namely the hydraulic, the engine, the brake and the transmission were analyzed for studying the reliability characteristics using the graphical methods in combination with analytical techniques. The results of the analysis show that it is very essential to test the field failure data for the presence of trends and serial correlations contrary to the approach adopted by most of the researchers today. It is shown that wrong conclusions can be drawn if such tests are not performed. When analyzed, the times between successive failures (TBFs) of LHD machines are found to be free from the presence of trends; and are exponentially distributed. However, most of the subsystems have shown increasing failure rate tendencies. The subsystems which have shown the existence of trends are analyzed using a time dependent model such as the power law process model. The growth in reliability over the years for the hydraulic systems is studied. contrary to the general expectation, no growth in reliability is noticed for such systems. Graphical methods for decision making in maintenance for both types of data, i.e. the TBFs with presence of a trend and the TBFs without a trend are presented. In this study simple and practical approaches are suggested for sorting and classification of the field failure data. The use of the homogeneous Poisson process model in place of the other best-fit models are ...
|
|---|