A large scale database MSDRD on mechanical properties of structural materials was constructed by the Research Group for Statistical Aspects of Materials Strength. In order to extract and analyze the material strength data from MSDRD, a software of STANAD was also developed by this research group. In the present study, fatigue data of aluminum alloys such as 2024-T4 and 5083BE-O were newly analyzed by the same software. Paying particular attention to the statistical feature of the crack initiation and propagation behaviors, statistical fatigue properties (P-S-N characteristics) were theoretically analyzed for smooth and notched specimens. Total fatigue life Nf is given by sum of the crack initiation life Ni and the crack propagation life Np. Thus we can derive the distribution function of Nf from the respective probability density functions of Ni and Np by applying the convolution integral. It was finally found that distribution characteristics of fatigue lives in a wide stress range were well explained by the present analytical model.
