PSI - Issue 19

Marc J.W. Kanters et al. / Procedia Structural Integrity 19 (2019) 698–710 Marc Kanters et al./ Structural Integrity Procedia 00 (2019) 000–000

704

7

4.1.3. Load ratio dependence To study the influence of load ratio, defined as ��� / ��� , the fatigue lifetime was obtained for injection moulded tensile bars with high degree of orientation for a wide range of load ratios. The SN-curves are presented in Figure 7a. For this material, two failure mechanisms can be observed at this temperature and on these timescales: At , failure is dominated by plasticity-controlled failure (small slope), and at all the other load ratios by crack growth controlled failure (large slope). The R-value dependency of both failure mechanisms was modelled using analytical models, for more details see [Kanters2016, Kanters2018], resulting in the model descriptions in Figure 7a. These models allow to generate continuous Goodman/Haigh diagrams, as displayed in Figure 7b, displaying regions with little (plasticity-controlled) and strong stress dependencies (crack-growth controlled) of the lifetime. For illustration, in tension-tension fatigue ( 0 < ) the failure for short-cycles and large R-values is dominated by plasticity-controlled failure, while for higher number of cycles and smaller R-values failure is dominated by crack growth. In tension-compression fatigue ( − ∞ < ) failure is fully crack-growth dominated and in compression compression fatigue ( ) failure is completely dominated by plasticity-controlled failure.

Figure 7: SN-curves for injection moulded tensile bars for various load ratios (a) and the resulting Goodman diagram (b). Markers represent measurements, lines the analytical model.

4.1.4. Influence of anisotropy and load ratio The influence of anisotropy on SN-curves is calibrated using the Digimat software by fitting Tsai-Hill failure indicators, based on the analytical models for various samples with different microstructures, whilst considering the local stresses using the localization factor � . The regression lines from the analytical models are used for the injection moulded tensile bars (IM) and samples taken from the plaque at multiple angles. Figure 8a displays that both the analytical model (solid lines) and the Digimat material model (dashed lines) describe the experimental data accurately, even though the two steps come with a minor fitting compromise. For example, besides the good match, it shows that the Digimat model is slightly deviating from the analytical model used for calibration, caused by the fitting compromise required to fit the failure indicators on the analytical curves, while these curves already have a certain deviation. Besides the small compromise on offset of the curves, one might notice that the slope of the SN curve seems to deviate from the models at lower number of cycles, most likely caused by a change in mechanism due to crack closure or crack retardation due to plastic deformation in the matrix. However, as Figure 8b shows, for high cycle fatigue the anisotropy is captured well within a factor 3 of accuracy by the Digimat material model, although the material model is slightly conservative for all microstructures due to the combined compromises.