Long life random fatigue behavior of notched specimens in service, in service duplication tests, and in program tests

The actual service fatigue life of notched steel and aluminum specimens was determined using a special testing device installed in a passenger car and operated by the vertical displacements of a car rear wheel; the specimens are subjected to reversed bending. The tests were carried out over a distance of more than 2 X 10 5 km. The testing device was equipped with counting instrumentation to ensure direct correlation between the individual test result and the relevant frequency distribution of level-crossings. The results of the service test series are statistically evaluated and presented as stress/life curves for SAE 5140 and for 2024 T 3 specimens. The corresponding stress/life curves, obtained in service duplication tests were not significantly different. However, the conventional eight-step program tests overestimate the fatigue life observed in service and service duplication tests by a factor which may be as high as five. The results of the test series were also compared to fatigue calculations. It was found that the application of a modified linear fatigue damage rule, which does not account for the sequence effect of load changes, yielded fatigue life data which are very much on the nonconservative side. This confirmed the now generally accepted fact that valid fatigue life prediction models must account for the sequence of individual peaks and troughs.