This paper was produced for the 2019 NAFEMS World Congress in Quebec Canada
Numerical fatigue analysis is an important part of modern product development. One of the basic inputs for fatigue-analyses is the detailed stress field over time-periods appropriately chosen for the particular load cases. Therefore, most computational cost is spent on performing dynamic analyses on large and complex finite-element (FE) models. Complex boundary conditions, nonlinearities and time-varying, multiaxial loads require transient dynamic calculations, associated with high computational effort. In addition to the cost intensive calculation of the component’s stress-history, estimating the fatigue life as a post-process connected to the finite element analysis is also associated with considerable computational effort. To reduce this expense, so-called hot-spot filters are implemented in commercially available software solutions (e.g., MSC-Fatigue). Based on a simplified analysis of the stress field, highly stressed component areas are identified and filtered. The subsequent complex fatigue analysis can then be limited to the identified hot-spots. Recent developments for more sophisticated hot-spot identification make use of the superposition of strain energy density (SED) fields on the base of a reduced modal space, as it is typically applied in Component-Mode-Synthesis (CMS) methods. The level of strain energy density within a component’s finite elements is observed to be directly correlated to the induced damage in that region. As the plain superposition of modal fields is leading to non-efficient results regarding falsely identified or missed hot-spots, a new approach has been investigated and presented in this paper. For the efficient calculation of the dynamic stress field, modal methods are widely used. Based on the solution of the decoupled system of equations, the modal coordinates and corresponding eigenvectors are superimposed to the system’s global solution. To further reduce the computational effort, the proposed approach addresses hot-spot identification before calculating the detailed stress field. Consequently, instead of explicitly calculating modal coordinates in a modal dynamic’s calculation step, a worst-case superposition of modal SED-fields is performed using qualitative assumptions based on modal load configurations and eigenvector properties. The resulting weighting factors can be interpreted as modal damage participation, taking aspects of loading position and direction into consideration. The validation of the approach is performed on a detailed FE-model from industrial application. Numerical fatigue analyses are performed for complex non-proportional multiaxial load cases; subsequently the identified hot-spots are evaluated.