The concept of safe life was first developed to deal with damage caused by repeated or cyclic loading of components. It recognized that in a large population of parts, failures would occur at widely differing operating times or number of load cycles applied. Lifetimes to failure would exhibit wide variability, some failures occurring early and some late. Taken as a whole population, the times or cycles to failure would exhibit a statistical distribution about some mean value. To avoid failure the practical and safe operating life would be set to the low end of the statistical scatter band, most usually at three standard deviations (-3𝜎) from the mean life. “Fatigue life” may be defined in any number of ways, but recognizing that the appearance of any crack would be deleterious to safety, fatigue was usually defined as the number of cycles to the detection of the first crack. By setting the safe life at –3𝜎 it will be recognized that most parts would be withdrawn from service containing no detectable cracks.
Only 1 in 1000 parts would be expected to contain a detectable crack. It is therefore a rather conservative approach to component lifing. The same concept is applied to many other forms of damage such as wear, stress corrosion, corrosion fatigue, creep and combined creep-fatigue environment interactions.