Fatigue Assessment Strategy Using Bayesian Techniques.

Area of Science:

• Materials Science
• Mechanical Engineering
• Statistical Modeling

Background:

• Existing fatigue strength models often use deterministic S-N fields, ignoring experimental data scatter.
• Probabilistic approaches are needed to account for inherent variability in fatigue testing.
• Accurate fatigue life prediction is crucial for reliable component design.

Purpose of the Study:

• To present Bayesian theory as a method for converting deterministic fatigue models to probabilistic ones.
• To enhance probabilistic fatigue models using statistical distributions for confidence bands.
• To provide a practical framework for analyzing fatigue S-N field data.

Main Methods:

• Application of Bayesian methodology for probabilistic modeling.
• Utilizing Open-Source software (OpenBUGS) for implementation.
• Employing the Weibull regression model for S-N field analysis.
• Statistical analysis of Maennig fatigue S-N field data.

Main Results:

• Demonstrated conversion of deterministic fatigue models to probabilistic ones using Bayesian inference.
• Successfully enhanced probabilistic models by incorporating confidence bands for S-N fields.
• Determined confidence bands for the S-N field using the Castillo and Canteli Weibull regression model.
• Provided a practical example using Maennig fatigue data.

Conclusions:

• Bayesian theory offers a robust framework for probabilistic fatigue analysis.
• The proposed method enhances the reliability of fatigue strength predictions.
• The study addresses the practical need for determining optimal experimental test numbers for reliable fatigue design data.