Reliability testing of power units

Abstract

The author discusses the trade-offs in formulating a plan for reliability testing of power products to gain more accurate knowledge of their mean time between failures (MTBF). Ways to estimate the sample size, test duration, and cost of such a program are presented in graphical form. The results are based on the constant-failure rate model. Results are presented in terms of an experimental plant to show that the MTBF of a power unit exceeds a certain lower bound with a 65% confidence level. Correction factors to adjust sample size to account for different confidence levels and ranges are shown. A factor to account for testing to show that an MTBF is bounded both above and below is also identified.<>