PSI - Issue 31

Vera Friederici et al. / Procedia Structural Integrity 31 (2021) 8–14

11

4

V. Friederici et al. / Structural Integrity Procedia 00 (2019) 000–000

Table 2. Heat treatment parameters to mimic different microstructures and hardness (condition) of the real bearing. Region Heat treatment Austeniti sation Temp. /°C Quenching medium Annealing Temperature / °C Hardness / HV1 Micro- structure Core as-received test specimen 1) - 2) - 1) - 2) - 1) - 2) - 1) 250-270 2) 260

Annealed Bainit (small amounts annealed martensite)

Hardened raceway Transition region

as- received test specimen as-received test specimen

1) - 2) 850°C

1) - 2) oil 1) - 2) oil

1) - 2) -

1) 618-719 2) 667±13 1) 440 2) in process

Martensite

1) - 2) in process

1) - 2) in process

Bainit, martensite

Crack propagation curves (Fig. 3) show the influence of hardness and applied stress ratio (R = 0.1, 0.3 and 0.5). Each R-curve shown in the diagram was obtained by testing at least samples. For both conditions, the curves move to the left with increasing R. At the same load level, crack propagation is faster in the hardened specimens than in the core specimens.

Fig. 3. Crack propagation curves for R=0.1, 0.3 and 0.5 with Paris lines for core and hardened material.

For small da / dN values below the linear region of the crack velocity curve, the rather ductile core material shows a large dependence on R, whereby the curves for the hardened material converge to one Δ K th value. The convergence to one threshold value for the hardened material corresponds well with the assumption by Schott et al (Schott, G., 1997), who stated that the threshold of the stress intensity factor can be estimated by: �� � � � �� � � � ∗ �� � � �� � � � � � ��� � ��� (1) and for high-strength martensitic steels γ → 0, so that the dependency of R is eliminated. For the linear region of the crack propagation curve, the parameters C and m for the Paris equations were calculated. � � � � � � ∗ � (2) The values are shown in Table 3 and displayed in Fig. 3. Displaying the results in different units, they can be correlated to literature data, see Fig. 4. Good agreement becomes clear for results by Göncz (Göncz, P., 2010) who