FY22 Status Report on the ART-GCR CMVB and CNWG International Collaborations

Abstract

This work presents the numerical model for the high temperature test reactor (HTTR) loss of forced cooling (LOFC) experiment with the INL codes Griffin, BISON, and RELAP-7, based on the Multiphysics Object-Oriented Simulation Environment (MOOSE) framework. Promising results were obtained, with the overall behavior of the reactor successfully captured. Changes in the heat transfer coupling, as compared to the fiscal year (FY)-21 model, enabled drastic improvement of the steady-state solution, both in terms of computational time and global energy discrepancies. The former was reduced by a factor of roughly 60, whereas the latter decreased from 11% to less than 2%. Furthermore, the discrepancy in the steady-state multiplication factor was improved from +2,300 and +2,900 pcm (for the 30 and 9 MW cases) to -700 and +1,200 pcm, respectively, and now falls well within the large measurement uncertainties stemming from graphite impurities. Validation of the Monte Carlo model used to generate cross sections was also performed against available measurements. Though significant, the discrepancies remain acceptable overall in light of the large uncertainty.Numerous improvements are still needed to better compare with the experiment involving the 9 MW case and to instill greater confidence in the model's ability to accurately predict the 30 MW behavior. Specifically, the power levels predicted by the 9 MW transient simulation following re-criticality remain low, pointing to an underestimation of the passive cooling of the core. A key aspect of future work will be to better understand the flow pattern during the LOFC event, particularly to determine if natural or forced convection is occurring inside the reactor pressure vessel (RPV). More generally, additional validation data would be immensely useful for further enhancing the numerical model and better matching the experiments. In addition, a more sophisticated thermal-hydraulics model that simulates all the channels as a single system model should be considered to take into account the rest of the primary loop. Finally, even if the results are iv in better agreement with the experiments, sensitivity analysis and uncertainty quantification will be necessary to evaluate the model uncertainty. v

Improvements in High Temperature Gas Cooled Reactor Modeling Capabilities in the Pronghorn Code

“…• MOOSE-based thermal-hydraulics codes have the ability to couple one-dimensional (i.e., system level) fluid models with three-dimensional heat conduction domains. This capability has been used by a variety of researchers [23,[27][28][29]. The difference between these works is the level of resolution applied in the core-wide heat conduction model and the number of modeled coolant channels.…”

“…The difference between these works is the level of resolution applied in the core-wide heat conduction model and the number of modeled coolant channels. The heat conduction problem can use a heterogeneous geometry resolving all features [28] or use homogenized assemblies [27,29]. The former case incurs a large computational cost, while the latter requires an astute determination of effective thermal conductivities.…”

“…These can range from modeling all coolant channels [28] to modeling only a representative flow channel per fuel element [27,29]. However, regardless of the utilized model, conditions in the plena and vessel temperatures are difficult to compute.…”