- The limiting value of the mean fatigue strength as the number of fatigue cycles (N) becomes very large (Graham, p. 23). Due the to the inherent scatter in fatigue data, individual fatigue failures may still occur at stresses below the fatigue limit. (See Maennig , particularly pages 638 - 642 - "The proof of the existence of a fatigue limit and the determination of Ng", where Ng is the number of fatigue cycles corresponding to the fatigue limit).
- The stress at which the median S-N curve attains zero slope -- i.e., the stress below which fatigue failure will not occur (infinite life). This stress depends on a number of factors.
Not all materials possess a fatigue limit in this sense (i.e., the possibility of infinite life). Ferrous materials and titanium are generally presumed to have a fatigue limit, although this may depend on the environment. Aluminum generally does not have a fatigue limit. As a result, aluminum will eventually fatigue if it is vibrated long enough, regardless of the stress level.
The minimum number of fatigue cycles that is needed to assure that a material does, indeed, have a fatigue limit depends on the material. Juvinall (p. 207) suggests the following:
- Steel - 107 cycles (p. 207).
- Titanium - 106 - 107 cycles (p. 218). However, other evidence suggests that this may be as high as 109 - 1010 cycles.
- Aluminum - 5x108 cycles (p. 213), even though aluminum may not have an actual fatigue limit.
Maennig (above) gives a statistical experimental method for determining the number of fatigue cycles corresponding to the fatigue limit, if indeed a fatigue limit exists for the particular material.