ASTM E3498-26

5.1 This guide supports either of the following two analysis objectives:

5.1.1 Objective 1—ML estimation of the fatigue-life distribution quantiles at a given cyclic amplitude.

5.1.2 Objective 2—ML estimation of the fatigue-strength distribution quantiles at a given number of cycles.

5.2 The two analysis objectives are illustrated in Fig. 2 and Fig. 3. These examples are for illustrative purposes only; more complete examples are shown in 6.8.

5.2.1 Objective 1—Estimating fatigue-life distribution quantiles from the fatigue S-N data.

5.2.2 Objective 2—Estimating the fatigue-strength distribution quantiles at a given number of cycles.

5.3 This guide applies only to S-N fatigue data as defined in 6.1.

5.4 The use of this guide requires selection between analysis options. Fig. 4 provides an overview of the three main decisions. Left Panel (a): the modelling approach determines whether the model’s probability distribution specifies the fatigue-life or the fatigue-strength distribution. When a fatigue-life model is specified, the specified probability distribution lies horizontally and is located at each S by the S-N relationship. When a fatigue-strength model is specified, the specified probability distribution lies vertically and is located at each N by the S-N relationship. Center Panel (b): The probability distribution describes the data’s randomness. This is typically chosen to be lognormal or Weibull for fatigue data, although other options are compatible with this guide. Right Panel (c): The S-N relationship parameterizes the trend between cyclic amplitude and fatigue life. These three decisions are detailed in 6.2 – 6.4.

5.5 The use of this guide requires computation of model parameter estimates and confidence intervals based on ML estimation as described in 6.5 and 6.6. These techniques may be implemented using either commercially available software or programs written for specific applications.

5.6 This guide requires the user to make a critical assessment of the quality of the analysis using model diagnostics, such as the residual analysis and goodness-of-fit criteria described in 6.7.

5.7 Examples demonstrating the proper use of this guide are shown in 6.8.

5.8 Further considerations are described in 6.9.

1.1 This guide covers methodological options to analyze experimental fatigue data for the estimation of fatigue life, fatigue strength, or both.

1.2 These techniques enable the use of nonlinear fatigue-life (S-N) relationships.

1.3 These techniques are applicable to S-N fatigue data containing run-outs (also known as right-censored observations). This guide uses maximum likelihood (ML) statistical inference because it is the most well-accepted method for analyzing data containing run-outs. Least squares and Bayesian techniques are not covered.

1.4 In this guide, several analysis options are presented from which users may choose. It provides a procedure for the implementation of these method options, but it does not establish a standard practice to follow in all cases.

1.5 It is intended that users of this guide will implement the portions of the guide that are found relevant without any implication or necessity of adhering to the entirety of the content.

1.6 In this guide, a general framework for fatigue data analysis is provided. It contains commonly used S-N relationships and statistical distributions. The framework is applicable to other relationships and statistical distributions, which may be substituted as justified.

1.7 Experimental methods, sample size selection and choice of test levels are beyond the scope of this guide.

1.8 The analysis of data for variable amplitude cycling and variable mean stress/strain is not covered.

1.9 The methods in this guide assume that the data points being analyzed all belong to the same underlying statistical population. This guide does not cover the subject of pooling data from substantially different experiments.

1.10 Extrapolation—Caution should be used when estimating fatigue life and fatigue strength beyond the life and amplitude ranges of the collected data, as it may lead to non-conservative estimates.

1.11 Units—The values stated in SI units are to be regarded as the standard. No other units of measurement are included in this standard.

1.12 This standard does not purport to address all of the safety concerns, if any, associated with its use. It is the responsibility of the user of this standard to establish appropriate safety, health, and environmental practices and determine the applicability of regulatory limitations prior to use.

1.13 This international standard was developed in accordance with internationally recognized principles on standardization established in the Decision on Principles for the Development of International Standards, Guides and Recommendations issued by the World Trade Organization Technical Barriers to Trade (TBT) Committee.