Programmed fatigue testing of welded specimens

Abstract

1. When the loading process is schematized by the method of enlarged amplitudes the stress amplitude distributions can be taken as homologous. This greatly simplifies the methods of evolving fatigue testing programmes. 2. When programmed testing of welded specimens is conducted under symmetrical cyclic stresses higher than the endurance limit the accumulation of defects follows the hypothesis of linear summation of defects. It can be assumed, with sufficient accuracy, that the average limiting value of cumulative defectsa is equal to unity. (This assumption is valid for the stages of crack development and fracture.) The distribution fora follows the logarithmic-normal law quite closely. 3. The considerable concentration of stresses in welded specimens causes the early development of fatigue cracks. (Cracks 2–3 mm long have been noticed after an average period corresponding to 30% of the life before fracture.) This fact increases the destructive effect of stresses lower than the initial endurance limit. When fracture is the criterion of fatigue life the lower stresses should be taken into account during calculations. This is done by extending the summation of defects to a stress of 0.7 σ−1 after extrapolating the fatigue curve. When crack development is the criterion the effect of stresses lower than σ−1 should be neglected.