A material-based model of initial damage states for predicting fatigue life
The role of constituent particles in fatigue crack nucleation in 2024-T3 aluminum alloy has been well-documented. In polished 2024-T3 aluminum these particles are dominant nucleation sites for fatigue cracks. The cracks form early in the fatigue life and the initial crack size is related to the cross-sectional area of the nucleating defect. The particles are often cracked or broken into smaller pieces when the material is rolled to the desired sheet thickness. The size and aspect ratios of nucleating defects in this material have been quantified by other researchers. However, the damage states of the nucleation sites have not been characterized. If the particles that form cracks are themselves cracked, then they may be the critical distribution of initial cracks that could ultimately cause failure. If this is the case, then the proper model of the material would give good fatigue life predictions without the use of a crack growth threshold and without requiring large numbers of tests to determine a database of initial flaw sizes. The object of this study is to characterize the material properly and use this to predict fatigue lives at different stress levels and in different materials without the use of a crack growth threshold. ^ For this study, the damage states of nucleation sites in three aluminum alloys were observed and found to be cracked constituent particles. The population of cracked particles in the material was compared with the overall population of particles and found to be significantly different. The cracked particles were measured in two metallurgical planes and used as a database of existing cracks in the material. Fatigue life predictions were done using this material model and found to be in reasonable agreement with existing experimental data. The material-based model of initial damage states is a valuable tool in fatigue design, as it enables the designer essentially to measure a threshold ΔK value for the material without the need for numerous time-consuming fatigue tests. ^
Major Professor: Ben M. Hillberry, Purdue University.