Duration of reliability growth and demonstration tests using mathematics and Physics of Failure

Abstract

This paper shows three different methods for planning reliability growth tests. One method is based on purely mathematical principles, and two additional methods are based on the Physics of Failure (PoF) principles. Our mathematical method determines test duration as a function of the initial failure rate of the design related failure modes that are being mitigated and the desired final, reduced failure rate. This method also includes the failure rate of random, but uncorrected design failure modes which were not accounted for previously. With this inclusion, the test may need to be accelerated to manage its potentially very long duration which introduces some limited PoF aspects into this test type. Fitting the required use exposure to the expected stresses may also present a challenge. The first PoF method, the Failure Free Test Approach, has a basic duration equal to the required product life, but it is then extended for validation of achieved reliability. This method establishes relationships among the stresses in use and the product life. The only deficiency of the method is the uncertainty about the duration of the failure free test extension. In the case of failures during the failure free period, sequential extensions may result in a very lengthy, even uncontrollable, test duration. The second PoF method, the Demonstrated Strength Approach, provides a fixed duration test where additional degradation of the product ensures a margin which is commensurate with the desired final product reliability. The test duration here is known in advance, so the test planning is easy and straightforward. The PoF reliability growth/demonstration tests are designed to represent a product as it is used and for the duration of its use. They therefore provide reasonable insight into its lifetime reliability. For that reason, they are preferred over the mathematically planned tests.