Data assimilation with subchannel analysis code CTF on NUPEC BWR BFBT test matrix

In order to gain the reliability of subchannel analysis on three-fluid two-phase flow in nuclear fuel assemblies, implemented models are expected to describe the detailed three-dimensional two-phase flow in fuel assemblies. Especially, crossflow model limits the prediction performance of subchannel analysis codes, it is important to develop a model that can analyze the phenomenon appropriately. In this study, CTF results are validated against the NUPEC BWR Full-Size Fine-mesh Bundle Test (BFBT). In BFBT, void fraction distribution across 8×8 rod bundles was measured to confirm the effects of radial/axial power distribution and unheated rods. Moreover, uncertainties of void fraction were quantified. In order to evaluate the prediction performance of the CTF code for the BFBT, bundle-averaged void fraction difference and the residual void fraction difference for each subchannel were defined. Subchannels were classified into several groups considering the grid spacer pressure loss coefficients set for each channel and the characteristics of the subchannel considering location, such as corner subchannel, adjacent corner subchannel, etc. Sensitivity parameters affecting void fraction and/or cross flow were selected with the Kriging method, and the response surface model represented by these sensitivity parameters was created. the simulation-driven MCDA method using the alternative model was applied for optimizing sensitivity parameters by data assimilation, and a set of parameters to accurately calculate the bundle-averaged void fraction difference was identified with the Metropolis method. CTF analysis with the parameter set identified by the data assimilation was conducted, and it was confirmed that the average value of the bundle-averaged void fraction improved with the parameter set by the data assimilation so that the predicted value would match the experimental value.