Statistical significance of notch fatigue prognoses based on the strain-energy–density method: Application to conventionally and additively manufactured materials

[Display omitted] •The SED control radius R1 is deduced from the plain and notch fatigue strength.•Ad-hoc V-notched specimen to minimize the sensitivity to experimental uncertainties.•If the fatigue data are normally distributed, R1 has a skew-normal distribution.•Monte Carlo runs to determine mean, standard deviation and skewness of R1.•Material and inverse search related R1 variability affect notch fatigue assessments. The inverse search determination of the strain-energy–density (SED) control radius R1 devised in Benedetti et al. Int J Fatigue 2019;126:306–318 and based on the knowledge of the notch fatigue factor estimated using an optimal V-notch specimen geometry is here reformulated to take into account the statistical properties of the input fatigue properties. It was found that R1 exhibits a non-symmetric probability density function that is well represented by a skew-normal distribution. The uncertainty in R1 can be attributed to the uncertainty in the inverse search procedure and to the material variability in notch sensitivity. By applying the devised procedure to real experimental data, it was found that the former contribution is preponderant in the assessment of very sharp notches, while the latter dictates the fatigue strength of blunt notches, especially in the case of intrinsically flawed materials, such as those additively manufactured.