TABLE OF CONTENTS Testing is a costly activity, but it is essential for the reliability assessment and acceptance into service of critical components and systems. A number of statistical techniques exist to demonstrate that the MTBF is not less than a specified number, and these are briefly described in this chapter's first section. The techniques require that a large number of articles be put on test at the same time (or that a long test time be scheduled), and they are therefore more suitable for parts and low-level components rather than at the system level.
An alternative approach is to use testing to verify or establish design margins and to transform these into a probability of failure, as discussed in the chapter's second section.
To select tests that are practical for assessing the reliability of a complex system, we survey the overall spectrum of tests (not restricted to those for reliability) that a system and major components can be expected to undergo. First we cover tests during development in Section 5.3 and then postdevelopment tests in Section 5.4. The importance of in-service tests (such as self-test and BIT) is steadily increasing, and the reliability implications of in-service testing are discussed in Section 5.5. Section 5.6 summarizes the chapter.
Before delving into the details of testing to prevent failures, the following apply to all types and levels of tests:
Early testing permits the results to be brought to bear on the design and development; problems found in late testing are much more difficult...
