ABSTRACT
The purpose of this chapter is to show on a theoretical basis and to illustrate by practical examples that reliability demonstration tests necessarily include an element of risk and to explain how they can be organized in order to minimize that risk. The underlying theory is that of hypothesis testing. Risks of type I and type II (or, in this context, producer’s risk and consumer’s risk) are complementary. Good test procedures aim at making those two risks acceptable for both parties while complying with the time and cost constraints. Reducing the risks usually entails a higher test cost. Reliability demonstration test procedures are derived, in the general case and in the constant-failure-rate case, and illustrated on examples. The notion of operating characteristic curve is introduced, i.e., the function which maps the true value of a dependability parameter to the probability of test acceptance. Erroneous methods (unfortunately sometimes found in calls for tenders), which entail excessive risks, are denounced. Time-terminated and sequential tests are compared.
