Demonstrating Reliability for Actual Field Load Efficiently

Empirical life tests are used to demonstrate product reliability, typically defined by a fixed service life, reliability values, and a set load (e.g. for 95 % customer). However, actual field usage varies, requiring methods to estimate reliability based on real usage conditions and load variation. Physical tests are essential for reliability demonstration but are challenging to plan. The Probability of Test Success serves as a key evaluation metric to address those challenges. The established hypothesis test framework aids in developing easy-to-implement algorithms and procedures for identifying optimal tests in individual cases. This paper combines efficient reliability test planning with the estimation of actual field usage reliability, applicable to Success-Run tests, End-of-Life tests, and accelerated tests. The result is a method for identifying the best test to demonstrate and estimate a product's real-world reliability. This is achieved by establishing necessary equations and definitions and developing a Monte Carlo-based algorithm to evaluate these equations. Results show significant benefits in terms of test expenditure and highlight drawbacks of common test load profiles, which this approach can remedy.

Keywords: Reliability demonstration, Reliability testing, Probability of test success, Accelerated testing, Field load, Success run.