Citation

Sayles RS, Moss TR, Daniels BK. Reliab. Eng. 1983; 6(2): 83-102.

Copyright

(Copyright © 1983, Elsevier Publishing)

DOI

10.1016/0143-8174(83)90037-9

PMID

unavailable

Abstract

Discriminant Function Analysis (DFA) techniques are particularly useful for analysis of data where the number of variables are large. Here a general-purpose DFA computer program has been applied to specific experience data from a large utility to investigate the effect of size on the availability performance of turbo-generator sets.

The analysis demonstrates that availability decreases with increased size of set. Control systems and a few other component groups pose particular problems for this utility. Although outage time/fault increases markedly on control systems, the important positive correlation appears to be between failure rate and unavailability.

The DFA techniques are presented and shown to have the ability to identify the relative importance of individual component groups to overall failure history, and therefore clearly emphasize areas where reduction in failure frequency will offer greatest return in improved overall reliability.