THE EFFECT OF SAMPLE SIZE ON THE CONFIDENCE OF ENDURANCE FATIGUE TESTS

Experimental determination of the fatigue endurance (S-N) curve of either a given material or a machine element imposes the choices of sample size and of stress levels to be tested. Since referenced recommendations do not apply when the data include run-outs or have variable scatter, this paper studies the confidence of fatigue tests, taking into account the effect of sample size and of a statistical model. In this research, a great number of rotary bending fatigue tests were carried out, using a heat treated carbon steel with relevant scatter of fatigue lives. The statistical analysis allowed one to obtain confidence limits of the S-N diagram estimates and to propose some criteria for a better formulation of test schedules.