Probabilistic Study of Corrosion Pit-Induced Fatigue

Abstract

Corrosion fatigue is the major damage mechanism responsible for the premature failure of the aircrafts and turbine parts, especially in the marine environment. It has been seen that even relatively mild corrosive atmospheres can reduce the fatigue strength of many structures considerably as compared to their fatigue strength in dry air. The process starts with the surface degradation due to the corrosion pits, that become fatigue initiation sites and the initial damage when the part is loaded. Microstructure and morphology of the surface pits is critical for the crack initiation. It is assumed that the cracks are initiated at the sharp corners and bottoms of the narrow “micropits”. The paper provides a statistical characterization of the crack initiation process based on these pits’ density and microstructure distribution. Based on the surface damage initiation analysis, new morphological characteristics combining the pit size and highest curvature are introduced and have shown to be the efficient metrics for pitting fatigue with the distribution of introduced morphological characteristics. The distribution of these characteristics is estimated from the measured data. Effects of pitting morphology are evaluated for various heat-treated Aluminum 2024 specimens, with varying distribution of pit shapes and curvatures. The statistical distribution of specimen life is estimated using the “weakest link” approach, that is, by computing the probability that at least one crack is initiated anywhere on the surface. The paper contains a detailed description of crack initiation’s statistical model, methodology of corrosion parameter estimation and representative numerical examples of statistical modeling. The surface pit characterization has been nondestructively measured on a Zygo ZeGage 3D profiler. The ZeGage uses a Coherence Scanning Interferometry (CSI) technique which uses the wavelength of light to define its high precision. A carefully calibrated CSI profiler can measure a larger range of surface qualities with repeatable results in sub-nanometer precision. High magnification (20x, 50x, 100x) lenses increase the versatility in measuring even smoother surfaces. Large area scanning was done using a segmentation approach. A large field of view of the selected lens and automated rapid scanning allowed for reasonably smaller data files obtained after stitching these individual segments.