PSI - Issue 23

Hugo Wärner et al. / Procedia Structural Integrity 23 (2019) 354–359

357

Hugo Wärner / Structural Integrity Procedia 00 (2019) 000 – 000

4

Fig. 2. Sanicro 31HT specimen geometry after 1608 OP TMF- cycles. “1” indicates the area of the mid -LAGB calculation, “2” indicate s the area of the waist-LAGB calculation.

3.2. Modelling

To investigate if the test specimen geometry was contributing in the change of the diameter with cyclic load i.e. barrelling, a finite element analysis (FEA) was performed of the current specimen geometry subjected to TMF load. Material data was limited which necessitated some simplifications. The yield limits were taken at mid-life from LCF tests performed at room temperature and 700 ° C (since no data at 800 ° C was available). The yield limits at the two temperatures turned out to be similar (due to cyclic softening at room temperature and cyclic hardening at 700 ° C). The material was modelled as linear-elastic-ideal-plastic and creep was excluded from the analysis. The load and temperature were applied in accordance with an OP-TMF cycle. The temperature was cycled between 0 ° C and 700 ° C and the total strain averaged over a 12 mm gauge length was cycled between 0 % and 0.6 % (since no plastic deformation occurred for lower strain ranges at 700 ° C). The OP-TMF cycle was run for ten cycles after which the displacement constrains were removed, corresponding to the specimen being removed from the test rig. Fig. 3 shows results from the FEA; the figure shows a quarter of the cross section through the specimen after ten cycles with the displacement boundary conditions released (corresponding to the specimen having been removed from the test rig). As seen in Fig. 3 a), the permanent deformation of the specimen gauge region agrees reasonably well with experimental observations; i.e., the midspan diameter has increased and the diameter in the region above the midspan has decreased. The reason for the inhomogeneous deformation of the specimen gauge can be understood from Fig. 3 b) which shows the accumulated plastic strain in the specimen; as seen, the region immediately above the midspan carries most of the plastic strain. Fig. 4 shows the percent change in the midspan diameter during cycling; as seen, the midspan diameter increases continuously during cycling implying that the increased midspan diameter becomes more pronounced as the test proceeds. To confirm the approximation from the FEA, LAGB fraction calculation were carried out at areas specified in Fig. 2 for the Sanicro 31HT ( N = 1608) specimen and the other fixed OP-TMF tests were used as early stage references. In the present study, a misorientation angle of 1° - 10° was considered a LAGB. The density of LAGBs in an area was used to measure the plastic deformation and Lundberg et al. (2017) has verified this procedure to be accurate. The results from these calculations can be viewed in Table 2. Even though the values are not directly comparable to the results shown in Fig. 3 b), the fraction of the density of plastic deformation in the “middle” and the “waist” by the LAGB calculation are very similar to the plastic strain fraction of the FEA calculation at the same two areas. As can be expected, the fixed specimens that were stopped at N = 500 does not show the same severe plastic deformation and did not show any clear sign of inhomogeneous deformation as the continuous OP TMF-test. This means that for Sanicro 31HT, the barrelling effect does not initiate before at least one third of the OP TMF-life and this is also reasonable to assume in the case of Sanicro 25, which continuous OP TMF-tests failed by thermocouple detraction after approximately 3500 OP TMF-cycles, run with the same Δε mech as the fixed OP-tests. Given the results for aged IP TMF- testing from Wärner et al. (2018 b), the lower structural stability and oxidation resistance of Esshete 1250 compared to Sanicro 31HT should render it likely for Esshete 1250 to exhibit fewer cycles to failure for an OP TMF test condition. Although, no barrelling effect was visible before 500 cycles according to the LAGB calculations 3.3. Microstructural investigation