Practical Methods for Analyzing Reliability Test Data in the Presence of an Unexpected Failure Mode 2016-01-0274

In order to accurately predict product reliability, it is best to design a test in which many specimens are tested for a long duration. However, this scenario is not often practical due to economic and time constraints. This paper describes a reliability test in which a limited number of specimens are tested with little time remaining before the scheduled start of production. During the test, an unexpected failure mode that can be mitigated through a product redesign occurs. Because the scheduled start of production is near, there is not enough time to perform a test with redesigned specimens, so the current test proceeds as planned. We discuss several methods and the associated assumptions that must be made to account for the presence of the unexpected failure mode in the test data in order to make predictions of reliability of the redesigned product.