PSI - Issue 57

Jan Papuga et al. / Procedia Structural Integrity 57 (2024) 79–86 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

85

7

DV and CROSS criteria, the first one showing the most extreme trend in this direction. The response to the phase shift of the QCP criterion is close to zero for the more ductile case (the right graph in Fig. 3) and milder than in the case of CROSS and DV criteria for the brittle material. The least change in the fatigue response to the phase shift effect is recorded for the integral LZ criterion, which is the only one among those investigated here, which assumes slight shortening of the fatigue life in the OP case compared to the IP variant for the brittle material. LZ criterion estimates the life extension for the OP case for ductile material, but this effect is much milder than for CROSS and DV criteria. If the observed response in  FI in Fig. 4 is analyzed and compared to Fig. 3 and to the last paragraph above, it correlates to the findings mentioned there. Note that to reach, e.g., 500,000 cycles for the IP and OP load regimes, the axial stress components of multiaxial loading are 52.2 MPa and 69.3 MPa, respectively. Due to the higher imposed stresses in the OP variant and milder phase shift effect, the initially non-conservative results for the IP case of LZ and QCP criteria were brought to better error values. The phase shift effect of the DV and CROSS criteria is too strong. The increase of the stress magnitude is not enough to bring the results close enough to the optimum response. Though the observation make sense in this aspect, it is obvious that none of the criteria respond well to both multiaxial cases. Even the QCP criterion tends to predict too non-conservative estimates for the IP situation. Various reasons for the observed deviations could be proposed and evaluated: 1. Other multiaxial fatigue strength criteria, see Papuga (2011) or Papuga et al. (2022), or shear stress path analyses (Papuga et al. (2021)) could be checked to see if there is no other solution providing better response universally. 2. The microstructure of AM specimens is quite specific, showing extensive occurrence of spherical pores at the skin layer, see Fig. 5. The same issues were observed in all specimens, including the uniaxial load cases. Therefore, it is questionable if some modification based, e.g., on the pore size, see Murakami et al. (2019), could improve the solution because it would also concern these uniaxial load inputs. 3. At least 8 specimens per each S-N curves were tested, and the scatter of the experimental data points is quite expectable and not excessive for AM material. More specimens would provide higher certitude about the correct position of the regression curves (projected to Fig. 2). 4. Being new to AM process, the team did not record the position of individual specimens on the platform. Further subsequent tests showed us that the specimens at the top and bottom position close to the wall (see Fig. 1, right), and at the atmosphere outlet (left) feature worse fatigue performance. If more specimens from these positions are chosen for a specific S-N curve, the lower than adequate fatigue response can result from such experiments, and the mutual positioning of individual curves in Fig. 2 can differ. 5. CROSS and DV criteria could benefit from using other enclosing entity to the shear stress path (or the load path in the Ilyushin deviatoric space in the case of the CROSS criterion). The use of the maximum prismatic hull or of the minimum circumscribed ellipse for defining the shear stress amplitude, see Papuga et al. (2021), would increase the shear stress amplitude in the DV criterion or the amplitude of the second invariant of the stress tensor deviator within the CROSS criterion. The same solution would however worsen the response in the case of LZ and QCP criteria, moving it to more conservative results.

Fig. 5. Photograph of a part of the typical cross-section at the fractured specimen with many pores in the skin layer and the approximate division of the cross-section to three thirds of skin-contour-skin layers.