Methodology for assessment of statistical planning effects on the S‐N curve determination using Monte Carlo simulations

Abstract

One of the most important input data in fatigue analysis is the material fatigue properties. This research aims to present a methodology for assessment of statistical planning of fatigue experiments through a MatLab algorithm developed based on Monte Carlo simulations, which enables to simulate statistically the effects of main parameters used for defining the fatigue test setting and to verify their impact on the relative percentage difference (RPD) in fatigue properties estimation comparing to a reference material. The aspects treated here have not been clearly discussed in the standards. Therefore, the proposed recommendations combined with standards procedures is a tool for test engineers, permitting a fatigue test planning with more background and precision, which can help in decisions about which is the better setup, including the sample size, number of stress levels, stress value in each level, and replication. The methodology and good practices presented in this paper were demonstrated by means of actual data from the literature.

Small sample test approach for obtaining P-S-N curves based on a unified mathematical model

“…fatigue life). 1,2 However, due to the randomness of internal defects (such as bubbles, inclusions, microsegregation, and microcracks) and the uncertainty of fatigue damage mechanisms in mechanical components, the fatigue lives of identical mechanical components usually show a large dispersion even under the well-controlled constant amplitude cyclic loading conditions. 3,4 To characterize the significant dispersion of fatigue life data, the P-S-N curves (i.e.…”