The Effects of Programmed Mean Stress on the Fatigue Strength of the Welded Joint of SM58 Steel

Abstract

Many informations on the effect of the constant mean stress have been reported by Munse and other investigators. But in the case that the mean stress is fluctuating in a long period, it is considered that we can not estimate simply from the result of fatigue test with constant mean stress.
In this report, the relation between the crack propagation rate and the mean stress was studied using the single notched specimens and that result was examined using the concept of effective stress in the vicinity of crack apex due to Erdogan and et. al.. After those studies, the effect of programmed mean stress on the fatigue strength and the relation between the crack propagation rate and the fatigue life about the both of the welded joint and the base metal were investigated by the analysis of fracture mechanics. The material used in this study was 14 mm thick plates of a steel for welded structure SM58. Fatigue tests were carried out from the high cycle region to the low cycle region in the axial loading condition with constant R-value or programmed R-value.
Firstly in the fatigue test with constant R-value, it was found that in the high cycle region, the fatigue test result of base metal was well agreed with the equation of aeff⋅Nαf=C used the effective stress, but that of the welded joint was not because of the scatter of the welded reinforcement. In the low cycle region, though the elasto-plastic DM model could not be adopted and so the effective stress was not used, formally we could seem to use the assumption with m=0 in the high mean stress condition.
The fatigue test with programmed R-value were carried out in the very small cycle ratio and it was found that the modified Miner's rule was coincided with the test result of the base metal, and we could estimate the fatigue life with programmed R-value from the result of that with R=0 using the effective stress. But in the case of the welded joint, the observed values had larger fatigue life than the estimated values from the modified Miner's rule. According to the crack propagation rule, the estimations of fatigue life were carried out on the fatigue test with programmed R-value, and it was found that considering the invalid cyclic portion in the vicinity of the threshold of crack propagation, we could have the more exact estimation of fatigue life with programmed R-value.