Quantitative Criteria for Correlating Simulation-Based Fatigue Life Predictions with Test Outcomes

Fatigue life of a component is influenced by multiple factors like material, manufacturing, load & geometric variations and due to this there is a huge scatter in both test & predicted life through simulations. There are different methods available to account for these variations while predicting fatigue life. However, whenever a fatigue simulation engineer tries to correlate predictions with test outcomes, he/she will face a challenge as to: How to account for scatter in test? How to compare predicted life through simulations with test data? How much difference between test outcome and predictions is acceptable? To address these challenges, authors have suggested two approaches in this paper – 1. Sample to Sample approach and 2. Statistical approach. This paper suggests a set of criteria under both these approaches to conclude confidently that the prediction model is able to match the test outcomes. Also, the paper highlights the kind of measurements that needs to be done and the statistical tools that needs to be used for such a correlation. This paper intends to help engineers to select appropriate design margin for fatigue of metallic components without being overly conservative and to validate their fatigue predictions with test outcomes.