ABSTRACT
This paper introduces an application of survival analysis to decision on the replacement interval for elevator components. We have developed a software tool to decide replacement intervals based on the maintenance records. We analyze more than one thousand elevator components to perform survival analysis. For each elevator component, we classified survival data into two groups by elevator operational parameters such as the number of starts, the rated speed and numbers of service floors. One group consists of the components that indicate higher tendency of failure, the other consists of the components that are not prone to failure with longer duration. We consider that the appropriate groups are chosen by the following method. (1) We compute a p-value of the two-sample logrank test of the two groups given by each classification. (2) We select the classification that gives the minimum p-value among all of the classifications. Finally, we obtain two survival curves corresponding to the two groups for each component. The software tool calculates failure rates and replacement costs by means of the survival curves of the two groups. From a trade-off between the failure rate and replacement cost, it is possible to determine intervals of components replacement. Evaluation of eleven elevator models indicates that application of replacement intervals decided by the tool reduces average replacement cost to less than 98%.
