Product Maturity, Volume 2

4Zero-Failure Reliability Demonstration

4.1. Purpose of zero-failure tests

The purpose of this type of test is to demonstrate a certain level of reliability with a given confidence level. When a reliability objective must be demonstrated, the common approach is to submit the components to testing and note the number of failures. As a result, the tests may prove time-consuming and expensive and thus, in some cases, a “zero-failure” test may be used.

While such a test has the advantage of being faster and less expensive, it also has disadvantages. These are generally accelerated tests and since only one test is conducted, assumptions must be made on the acceleration law. In addition, the demonstration may fail as soon as a failure is detected. Finally, when no failure is observed, reliability cannot be estimated; however, it is possible to estimate a lower bound (for a probability of correct functioning, also known as the survival function) or a lower bound (for an MTBF). If this bound is above the objective, this means that the reliability is demonstrated.

4.2. Theoretical principle

As already noted, there are two industrial application categories for which the reliability objective is different.

4.2.1. Non-maintained products

For this type of application, a Weibull distribution can be viewed as a reliability model. If R_obj is the probability that the mission will be completed successfully after a duration Tm, the duration of the required test is given by:

[4.1]

where: