PSI - Issue 19

448 Kai Schnabel et al. / Procedia Structural Integrity 19 (2019) 442–451 Author name / Structural Integrity Procedia 00 (2019) 000 – 000 7 lines. For approximately > 5 ∙ 10 5 load cycles by using the ‘as - built’ geometry for the FKM estimation, the estimation of the fatigue strength is in this case conservative, Figure 6 left.

Figure 6 Fatigue strength of calculated and experimentally derived fatigue strength of notched flat specimen, left: comparison of WÖHLER -lines, right: comparision of numbers of cycles

This becomes apparent also in a comparison between the calculated and the experimental derived number of cycles until failure, Figure 6 right. In the range of the high cycle fatigue up to 10 6 cycles, all calculated results are not conservative and so unsafe. The use of real geometry improves the estimation significantly, even if the deviation can still be almost a decade. For the very high cycle fatigue range the use of the real geometry leads to a conservative estimation. It is important to mention, that the comparison was drawn to specimen without any removed support structures on the critical surfaces. Real components can have highly rougher surfaces because of support structures, which may lead to very different testing results. In summary, for a fatigue estimation auf AM components, the FKM guideline needs adjustments to considering the AM characteristics. With the results of the prior fatigue assessment it becomes clear, that an assessment concept has to be developed for AM components, which is able to include its specific characteristics. Ideally, fatigue-relevant characteristics like pores and rough surfaces should be modelled exactly in their size, shape and location in a finite-element model of the actual AM component. However, this is not advisable, because of high complexity of such a FE model and the required calculation time as a consequence of the needed ultra-fine mesh. Additionally, it will be difficult to get all the necessary information about the pores and surface conditions like exact positions, shapes and dimensions prior to the manufacturing process. Consequently, for this lack of information assumptions have to be made. It is very probable that regions with similar process conditions like same geometrical alignment to the building platform or same key process parameters in a fully controlled process result in a very similar structural behavior. The local stiffness, stress-strain behavior and fatigue strength than can be correlated to these regions. The structural behavior of each region can be pre-calculated and for each region so-called structural elements (SE) (Wagener, et al. 2019) can be derived. Each structural element contains information about the locally aligned stiffness, the local non isotropic stress-strain behavior and the local fatigue strength for the special type of modelled imperfections. With this information the structural element can be used in two different modes. The first mode is the element, where the defects respectively imperfections are directly modelled and the stress or strain increase can be calculated as a consequence of an external load as well as the total strain of the element. This first mode is intended for deriving the fatigue relevant values. The second mode represents the global translation and deformation behavior without modelling defects, using 5. Description of the structural behavior