A client developed a novel FMECA technique that I think has much to recommend it. A key difference lies in the way “Occurrence” is handled. Generally, when we develop a FMECA, we assign a somewhat-arbitrary number (1 – 10) to the occurrence factor. In this approach, the failure rate (in FIT’s) is used for each of the components. The total of the FIT’s for the components in the subsystem considered is then used to normalize the FIT for each component. In this way, the most critical components can be determined. The total can also be rolled up to the next higher level. Another feature of this approach is that the failure mode of the component (for example: open, shorted, parameter change) can be included in the analysis since, in some cases, a particular failure mode can have a more deleterious effect than another.
The main advantage of this approach is that it removes some of the arbitrariness of the standard approach. A challenge however, lies in finding the FIT values (and especially the ratios of the failure modes) for some of the components.