Experimental Characterization and Physics Based Modeling of Aluminum Alloy 7075

Extensive studies have been conducted to understand the fatigue behavior of aluminum alloys under unidirectional loading conditions over the years. An early description of fatigue damage can be found in the work of Ewing and Humfrey [1]. They found that repeated alterations of stress in metals showed up in the form of slip-lines on crystals that were very similar in appearance to those that occurred in simple tension tests. Further loading resulted in appearance of additional slip-lines. After many cycles, the sliplines changed into comparatively wide bands and continued to broaden as the number of cycles increased. Cracks occurred along broadened slip-bands from crystal and soon coalesced to form a long continuous crack across the surface of the specimen. Once the long crack developed, it took a few more cycles of loading to cause fatigue fracture. Therefore, it can be concluded that fatigue damage is a result of material structural change at the microscale. Thus, a comprehensive fatigue damage model should include important microscale features of the material, such as grain orientations, and reflect changes of those features at macroscale. In addition, engineering structural components are usually subjected to varying loads of different amplitudes and frequencies along different directions. This will give rise to biaxial or multiaxial stress state and a prediction of fatigue life should consider such loading condition for reliable assessment. Most published work on fatigue characterization of aluminum alloy was conducted under simple load conditions, such as uniaxial constant amplitude fatigue loading condition. However, the material behavior under complex loading conditions, such as in-plane multi-axial loading conditions, is still not fully understood. In general, fatigue process consists of three stages: initiation and early crack propagation, subsequent crack growth, and final fracture. The fatigue crack growth rate, da/dN, which determines the fatigue life of the components after crack initiation, has been extensively investigated experimentally. Considerable research efforts have also been devoted to developing methodology for modeling fatigue damage [2]. A comprehensive overview on fatigue criteria is found in Socies work [3]. In general, fatigue criteria can be categorized roughly according to the physical quantity upon which the criteria are based. Depending on different fatigue damage mechanisms, fatigue criteria are developed as based on stress, energy, and fracture mechanics. In earlier research, stress or plastic strain amplitudes were adopted for fatigue life prediction. For example, Gough et al. [4, 5] proposed empirical relationships that reduce to shear stress for ductile materials and principal stress for brittle materials. Since fatigue damage is found to be primarily driven by plastic strain energy, this parameter was believed to be a rational parameter for fatigue damage evaluation. However, most early attempts of fatigue model development based primarily on the energy concept without considering loading history related parameters seem unsatisfactory [3]. Modifications are also needed to apply those models for complex loading conditions such as multiaxial loading, and non-proportional loading. Since the fatigue phenomenon inherently involves multiple scales it is necessary to develop a scaledependent, physics-based model for accurate simulation in order to understand material performance/degradation in various operational environments and to ultimately assess the survivability of aerospace vehicles. This scale associated modeling approach, referred as multiscale modeling, must address important features at different scales, including multiple spatial and/or temporal scales. A significant amount of research has also been conducted on multiscale modeling [6-12]. For instance, the hierarchical approach based on the bottom-up description of the material structure has proved to be successful in a wide range of applications [7-10]. Unit representative cells are identified based on a multiscale decomposition of the material microstructure. A single macroscopic constitutive relation is built 2 hierarchically from one scale to another using cell-averaging technique. Microstructure parameters are included as variables in the resulting relation. An alternative approach to the homogenization is provided by the globallocal analysis [11, 12]. In this approach, the material response at a point is calculated simultaneously with the global simulation by performing a cell model. The key advantage of this method is that a homogenized constitutive relation is not needed, and therefore, no empirical determination of material constants is required. However, the microstructures used in such approaches are generated by Voronoi diagram which neglects the real grain size and shape effects. In addition, microvoids are arbitrarily introduced in the structure. Thus, damage initiation is not considered in those models.