What is a fact? Since people generally come to failure analysts when they can’t figure things out for themselves, I have come to believe that it is important to not only have reliable facts but to be able to explain to people why our facts are indeed reliable!

If you are someone who has experienced a failure of your own product or a customer has brought such a failure to your attention, you have a lot of background information ready to help lay the foundation for a successful failure analysis. Presumably, you have engineering prints and knowledge of manufacturing, maintenance and use procedures.

When I get involved with a manufacturer’s quality department, especially if the failure involves a component that is failing on installation into a motor vehicle, the first activities of the manufacturer will often be to try to determine what lot number has failed, the percentage of bad parts and document review. For purchased components, the sub-supplier will often claim proprietary processes and refuse to provide some of the information. But even so, this type of investigation is much easier than trying to determine a service history of a machinery component that may have been in service for 15 years.

The basic facts in the first situation are usually not hotly disputed. For example, someone might remember that the total number of parts shown in a particular lot would be incorrectly shown because of a known glitch in the computer. But these facts are relatively straightforwardly recognizable as facts by most people.

As the investigation proceeds, it might be revealed that a metal stamping was made from 1008 steel instead of 1010. Or maybe, more concerning, it was made of carburized 1010 instead of through-hardened 1030. If this is discovered by the quality engineer who takes a piece to a chem lab for composition analysis, the first analytical result may be that the material looks like 1080. The chemist may keep grinding to deeper layers out of concern about contamination. Eventually, they may end up at a carbon value that is too low, rather than too high.

To be continued...