Abstract

Reliability improvement should be a major consideration when conducting product life data analysis. One method of determining the failure mode responsible for failure is to perform separate analyses for each mode and combine the results, as opposed to doing a single omnibus analysis that treats all failures together. Analysis of individual failure modes allows you to quantify the impact of each failure mode and evaluate the impact on product reliability by removing each failure mode. This method is appropriate for designers or manufacturers looking for overall reliability representations, whereas an omnibus analysis may suffice for a customer interested only in total reliability. To do these analyses sound data on the failure mode of each failed unit is needed. Additionally, it must be assumed that the failure modes are independent of one another. So that reliable information may be correctly and consistently gathered, it may be necessary to conduct an autopsy on a sample of failed units. While there are software packages with special capabilities to separate failure modes and do this kind of analysis, some general-purpose statistical software packages can be modified to work as well. When appropriate, analysis by individual failure modes can result in better total product reliability estimates. A sidebar article lists important things to consider when conducting failure mode analyses.