PSI - Issue 53

Benjamin Möller et al. / Procedia Structural Integrity 53 (2024) 190–202 Author name / Structural Integrity Procedia 00 (2023) 000–000

199

10

The outcome of these simulations showed some promising results:  Very good prognosis of the fracture location by the point of the maximum stress from FE simulation.  Both von Mises and maximum principal stress distributions show stress concentrations at the crack location of the fatigue test performed.  The maximum principal stress, as expected from an axial bending load, is the normal component ( σ zz ). In Fig. 12, the result of the FE simulation for specimen HB_2D as well as the cracked specimen from fatigue testing is shown as a typical example.

Fig. 12. Prognosis of the point of fracture from the FE simulation.

Additional FE analyses carried out were made to point out the effect of different shape defects on the stress concentration. By looking at the WAAM motorcycle handlebar structure, it becomes clear that their shape is characterized by two main types of geometrical defects (surface imperfections): the small waves, caused by the layer by-layer printing process, and the macro-geometric shape variations that are mostly related to non-axiality of the layers (positioning) of the WAAM welding head. To evaluate the effect of each one of these defects on the stress concentration, two different analyses for each hollow structure specimen were carried out (with load and constrains as defined before):  One analysis of realistic shape specimen in which the model used is the one obtained as showed above.  One analysis using a so-called smoothed specimen in which the model used has only the macro-geometric shape variations, while the small waves are eliminated by smoothening of the surface. Since all the analyses carried out are in the field of the elastic deformation is possible to obtain the effect of the large-scale defects from the second type of analyses and, together with the first type of analyses, according to the principle of superposition, is possible to find out the effect of the small-scale defects. To obtain the smoothened specimens the Blender functionality “smooth” was used. It reduces the angle in between adjacent triangles of a stl-file. The intensity of this smoothing process is defined by a value called “smoothing parameter” that must be chosen. To find the optimal value for the “smoothing parameter”, a simulation was repeated increasing this value until the peak stress value reached convergence as shown in Fig. 13b. The optimal smoothing parameter that was used for all the smoothed models was set to 40, since max. von Mises stresses do not show significant variations when increasing the value further. Lastly, to set up a reference for the simulations of both realistic and smoothed models, the simulation of the specimen with the nominal “ideal flat” geometry was carried out (simulation condition like load and constrains remained unchanged).