This project has been carried out on behalf of Nordtest in a co-operation between Det Norske Veritas(DNV) and Swedish National Testing and Research Institute (SP).
Fatigue S-N diagrams are usually determined from experimental data by linear regressions, e. g. according to ASTM E 739-91. This procedure does not allow run-outs. However, run-outs may represent valuable statistical information, and is often not to be avoided in practice.
Principles for including run-outs in the statistical assessment of S-N data based on the maximum likelihood method is outlined here, and the principles have been applied on some examples of experimental fatigue data.