This conference paper was submitted for presentation at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

Fatigue estimation remains a complex and imprecise tool for predicting the behavior of machines and their components under real-world service conditions. The wide range of factors that influence fatigue crack initiation and growth'”many of which are often unknown or difficult to quantify'”means that fatigue estimation models must inherently account for uncertainty and variability. Over the past century, the challenge of incorporating these factors into reliable fatigue predictions has been extensively studied. However, the solutions developed so far often rely on relatively small datasets, which are now outdated. While these models have been broadly accepted, the verification processes intended to update them with newer data have stagnated. Collecting sufficient new datasets for verification can take years, and few institutions are willing to undertake such lengthy efforts. To address this, the FABER project (Fatigue Benchmark Repository) was launched as a collaborative initiative under the COST (European Cooperation in Science and Technology) framework. This large-scale effort aims to compile curated datasets from existing experimental fatigue studies. These datasets will serve multiple purposes: validating and verifying existing fatigue models, assessing new computational methods, and comparing different approaches to fatigue analysis. A network of researchers and engineers is essential to ensure consensus on the interpretation and application of these datasets. The goal is to aggregate enough data to support the application of advanced methods, such as artificial intelligence (AI) and machine learning, in fatigue analysis. To accelerate the transition from concept to practical application, FABER is also developing an open-source, Python-based library of fatigue analysis tools. This library will allow for rapid adoption, adaptation, and extension by the broader research community. The availability of such a solver will ensure that the fatigue models published in research papers produce the expected results when applied to real-world inputs'”something that is not always guaranteed with current methods. This presentation will explore these challenges and objectives, explaining the rationale behind these efforts and outlining the strategy to achieve these ambitious goals.