Fett, G., "The Limitations of Fatigue Testing," SAE Technical Paper 2010-01-1908, 2010, doi:10.4271/2010-01-1908.
Fatigue testing of components is used to validate new product designs as well as changes made to existing designs. On new designs it is common to initially test parts at the design stage (design verification or DV) and then again at the production stage (production verification or PV) to make sure the performance has not changed. On changes to existing designs typically the life of the new part (B) is compared to that of the old part (A). When comparing the fatigue life Weibull analysis is normally used to evaluate the data. The expectation is that the B10 or B50 life of the new part or PV parts should be equal to or better than that of the old parts or the DV parts. However, fatigue testing has a great deal of inherent variability in the resulting life. In this paper the variability of numerous carburized and induction hardened components is examined. The data indicates that the value of traditional DV to PV or A to B type testing is limited due to the high level of background variation. Instead the value of fatigue testing components is in comparing the data to a well established SN curve or database made from many lots of material tested over a long period of time.