ASME NTB-4-2021 pdf – Background Information for Addressing Adequacy or Optimization of ASME BPVC Section II, Division 5 Rules for Nonmetallic Core Components.
The output from the tensile geometry FE model (VP-O0) was calibrated with the tensile experimental test results, summarized in reference [7]. The same methodology was applied to determine the predicted load factor for the other test geometries used for the VPs (see Table 4). The objective of the failure methodology is to predict the 50% POF load of each test case (VP). The results are shown in Figure 10 [34], where the 50% predicted failure load is normalized to the median 50% experimental failure load for each test case (as indicated by the predicted load factor data points in Figure 10). If a load factor is predicted to be greater than 1, it is considered not conservative; whereas if the load factor is less than 1, it is considered conservative. In the case of VP-00, the agreement is very good (as expected) because the model is optimized for tensile specimens. In the case of VP- 18, the agreement is poorer but is conservative because the predicted POF is lower than the mean reported failure strength.
A 50% POF was selected as a basis for comparison of the model output with the median experimental results. It was previously reported [50] that 50% of all tensile specimens failed, and that the highest number of failures occurred at the average experimental failure load, which corresponds to a 50% POF load predicted by the model. There can be several reasons for differences in the theoretical and experimental results, such as experimental error, error in failure prediction, and/or variability in the material itself. To illustrate materials’ billet-to-billet variation, experimental data were collected from 24 billets of NBG- 18. The assessment of the mean tensile strength from the 24 billets showed that only 50% of the billets fell within the reported ± 6% of the material mean, and 95% of the billets fell within the reported ± 18% of the material mean [20], [34]. These bands are plotted in Figure 10 and are compared with the calculated predicted load factor. On this basis, only the predicted load factors that fall within the ±6% band are considered accurate, and calculated results between ±6% and ±18% are considered acceptable. The predicted load factors that fall below O.82% are considered conservative. It is noticeable that all the VPs were conservative. The most accurate 50% POF failure load predictions were applicable to specimens subjected to a uniaxial (or approximate) tensile stress state.
Initially, Hindley et al. [50] made these comparisons of the actual VP failure load with the predicted 50% POF for the VP being modeled using a Weibull distribution (Figure 10). The selection of a Weibull distribution was based on prior work [511. However, the POF limits mandated by the code are 10-2 and iO [22], depending on the SRC as specified for the applicable service limit shown in Table 3.
Subsequently, the predictions of the model were extended to the code-prescribed limits of 10 2 and 10 [22]. Figure 11 shows the implementation of the method applying the code design—allowable POF loading as it relates to tests performed on NBG- 18. The results are conservative for all the VPs.
SUMMARY
The semi-probabilistic design approach for graphite materials defined for nonmetallic design, Subsection HH subpart A, is different from the delined deterministic approach relevant to metallic components defined in Section III, Division 5 of the ASME BPVC. This document provides a brief background and description that support the basis of the rules for graphite components. Details for typical graphite material properties and behavior are provided in the ASME STLLC technical report STP-NU-009 Graphite Jbr High Temperature Gas – Cooled Reactors [361.
The design rules define two analysis methods simplified assessment (two-parameter) and full assessment (three-parameter) based on a semi-probabilistic Weibull approach that describes the material reliability to accommodate the characteristic billet-to-billet and strength variability of graphite materials. The probabilistic failure criteria were verified, and it was shown that the failure methodology conservatively predicted real reactor components with larger volumes than tensile specimens.