TABLE OF CONTENTS When the times-to-failure distribution for specific units is not known, or only a few test units are available, so that the data are insufficient in quantity to determine the exact nature of the underlying times-to-failure distribution, nonparametric test procedures which do not presume a specific times-to-failure distribution may be used. Nonparametric analysis of the data utilizes more general assumptions about the nature of the underlying distribution; consequently, it does not have as high a power of making the right decisions as parametric testing analysis. The power of the analysis does increase as the sample size increases, however, such that nonparametric testing results may be made as powerful as parametric ones, and even more powerful if the data do not follow any known distribution.
Ten such nonparametric tests are presented in this chapter.
In this test the test duration is predetermined, and is usually taken to be equal to the mission duration for which the reliability of the units needs to be known. It is applicable when all units tested survive their mission; i.e., none of the units in the test fails or they all succeed in this test, hence the name "The Success-Run Test."
The lower, one-sided confidence limit on the reliability of equipment in a Success-Run Test, each one completing a mission of the same duration, t, without failure, at the confidence level of CL, is given by [1; 2] (See Appendix 5A)
| (5.1) |  |
where
