Verification of a neuber-based notch analysis by the companion-specimen method

Measurements of actual notch-root stress-strain behavior by means of the companion-specimen method are compared with predictions of the notch analysis using both a constant (long-life) fatigue-notch factor and a variable fatigue-notch factor determined from fatigue-life data. The results, using cyclicly stabilized specimens, show thatK _f is dependent upon the stress level of the degree of local straining. The amount of error incurred in the use of either the variable or constantK _f was about 7 percent in the prediction of the local strains. In addition, test results using specimens that had not been cyclicly hardened, and specimens subjected to a random-load history, indicated the effects of material-deformation phenomena such as cyclic hardening, stress-strain hysteresis-loop shape, and material memory, on local stress-strain predictions.