From precision to prediction in probabilistic FEA fatigue for structural reliability

Abstract Finite Element Analysis has become central to predictive structural assessment, yet its integration with fatigue-life modelling remains incomplete. This gap affects the reliability of predictions for advanced materials and complex component geometries, including additively manufactured structures. This study introduces a unified framework that links component-level finite element fields with microstructure-based crack-initiation models and incorporates Bayesian updating using in-service data.

The approach replaces single deterministic life estimates with probability-based distributions. A major barrier to consistent application is the absence of a standard process for identifying and propagating uncertainties across modelling scales. The framework addresses this through an uncertainty-aware algorithm, a worked example, and a reference table that connects key uncertainty sources to appropriate probabilistic models and calibration data.

Guidance on surrogate validation and computational considerations for safety-critical applications is also provided.