Abstract
Many failure accidents have indicated fatigue as the primary cause for the failure of a machine or structure. In general, the origin of failure is a structural discontinuous part such as a welded joint or a notched member that causes stress concentration. While designing such a component, a finite element analysis (FEA) has to be conducted, and the peak stress has to be compared with a design fatigue curve obtained from small-sized specimens to evaluate whether the component satisfies the design life. However, it is known that a fatigue life prediction at a stress concentration part based on a peak stress always provides an excessively conservative estimation. This is due to the stress gradient of the component. This paper discusses the stress-gradient approach to eliminate the conservatism and rationalize a fatigue design. Using literature test data, the relationship between the stress gradient calculated using FEA, and the fatigue strength reduction ratio was determined. Later, a fatigue test was conducted on a notched specimen of low-alloy steel to verify the stress-gradient approach, and the fatigue life of the notched specimen was predicted considering the stress gradient at the notch root. The predicted fatigue life agreed well with the experimental results.
