Reliability Data: Repeated Measured Degradation Data

In some reliability studies, the natural response is measured degradation (either performance degradation or physical/chemical degradation), as opposed to the traditional failure-time reliability data. Examples include light output from LEDs and lasers. crack length, and the amount of wear. Laboratory degradation tests are typically accelerated by increasing a variable like temperature to increase the degradation rate. Before statistical methods had been developed for repeated measured degradation data, it was common practice to wait for enough units to actually fail and then use statistical methods for failure-time data (Device-A in Chapter 18 of SMRD2 is such an example). There are, however, important advantages for direct modeling and analysis of repeated measures degradation data. In particular, it is possible to make reliability inferences with few or no observed failures and the degradation data contain useful information for checking important model assumptions.