ABSTRACT
Reliability demonstration of very complex systems can be quite costly. Concept
or design prototypes are often built by hand and then tested until a catastrophic
failure event occurs. Can organizations justify the expenses associated with
design verification testing at the systems level? Can they afford not to? The
economics are such that companies are often unwilling to spare more than two or
three units for testing. Is this a correct strategy? Can reliability be demonstrated
with a specified confidence level when sample sizes are so small? Is it possible to
demonstrate reliability with a small sample, but by testing the small sample to two
or three times’ design life? These questions need to be answered.