Development of Component-Level Damage Evolution Models for Mechanical Prognosis

This paper presents component-level empirical damage evolution regression models based on loads and damage information that do for mechanical damage prediction what the Paris law does for predicting crack growth under fatigue loading. Namely, these regression models combine information about the current damage state and internal system loads to predict the progress of damage to failure. One of the drawbacks of Paris-like crack evolution laws is that localized information about the loading (stress) and damage (crack length) is required. In structural health monitoring applications, it is not feasible to instrument every potential crack initiation region to collect this localized information. The component-level damage evolution regression models developed here only require global measurements that quantify the damage and loading at the level of the component rather than at the site of damage. This paper develops damage evolution regression models for an automotive sway bar link undergoing axial fatigue loading with two different damage mechanisms at a weldment and at an electrical discharge machining notch. Restoring force diagrams are used to calculate the load indicators as damage progresses and transmissibility functions are used to calculate the damage indicator during tests to failure. A component-level load intensity factor $(ΔK)$ is calculated during these tests so that the rate of damage accumulation can be used to predict the growth of damage and ultimate failure.