Decisions for maintenance-intervals of equipment in thermal power stations, based on few data

This paper describes a maintenance scheduling for pump systems in thermal power stations in order to reduce the maintenance cost during the whole period of operation, while keeping the current reliability level of the pump system. The dimensional reduction method is used to solve the problem in which a few available data are used together with other factors relating to the failures of pumps. The problem, that the available data are limited to a short range of the operation period compared with the mean time to failure, is solved by extrapolating the whole distribution of period of failure using available actual data. Weibull distribution approximation is used for this purpose: the data analyzed for failure and population statistics are those of 569 pumps, during the past 17 years which were recorded at 19 thermal power stations owned by Kyushu Electric Power Company. The results of the analysis by the proposed method provided the most suitable maintenance interval for each pump, and hence are helpful to reduce the expense of unnecessary maintenance