PSI - Issue 34

Sigfrid-Laurin Sindinger et al. / Procedia Structural Integrity 34 (2021) 78–86 S.-L. Sindinger et al. / Structural Integrity Procedia 00 (2019) 000–000

84

7

(b) FE HH : D ∗ = 0 . 9 mm

(c) FE MH : D ∗ = 1 . 14 mm

(a)

(d) FE MM : D ∗ = 1 . 11 mm

Fig. 5: ( a ) Experimentally determined and simulated load-displacement curves, each time step color-coded with maximum apparent failure index in the entire FE model and ( b - d ) quarter shell mesh with failure index contour plot at initial failure for the three considered model variants.

4. Discussion and Conclusion

In this research, a previously developed material modeling and property mapping approach for sti ff ness prediction in thin-walled additively manufactured structures displaying anisotropy and thickness dependency of mechanical mate rial behavior (Sindinger et al., 2021a,b) was extended for strength assessment. Tensile coupons were fabricated and tested. Thereby, for the first time the strong thickness dependency of ultimate tensile and shear strength in LS PA12 CF was disclosed, which generally implies its relevance for failure prediction. The obtained data was implemented fo structural FE simulations that were ultimately compared to physical experiments featuring a thin-walled part on sub component level. The comparisons of FE analyses with mapped inhomogeneous and conventional constant material properties re vealed considerably improved prediction of sti ff ness behavior as well as failure. For the latter, however, substantial deviations between simulation and physical experiment remained. The pre-mature failure was rooted in high stresses occurring in the simulations. This could have been attributed to di ff erences between the experimental setup and the FE models. However, the correct positioning of the parts on the supports was checked meticulously and highest e ff orts were undertaken to repli cate the setup as realistically as possible in the FE pre-processor. Furthermore, geometrical deviations, like a greater beam length or increased rib thickness of the physical compared to the virtual discretized part may have a ff ected the obtained results. Therefore, caliper measurements were performed and disclosed negligibly small deviations. Other factors, leading to an overestimation of stresses could have originated from the underlying shell element formulation. In this research, thin shell theory was considered, wherein transverse shear deformation is inhibited. The potential influence was examined by re-running the simulation based on thick shell formulation but only marginal di ff erences were found that cannot explain the substantial deviation between predicted and tested failure displacement.