INVESTIGATION OF DIFFERENT PLANE-AVERAGING TECHNIQUES IN NEK5000 FOR HIGH-TO-LOW INFORMATION OF CTF

When performing accurate thermal-hydraulic subchannel simulations, it is important to provide precise input values for pressure drop, friction velocity, and turbulent mixing models. These inputs can be calculated with high-fidelity Computational Fluid Dynamics (CFD) codes because of their highly resolved fluid flow domain. This is an example of the high-to-low (Hi2Lo) information modeling process. The CFD simulations in this work were conducted with the open-source spectral solver Nek5000 on the individual subchannels of a reactor core. Three types of subchannels were extracted from spacer grid geometries defined in the Pressurized Water Reactor Subchannel Bundle Test (PSBT) Benchmark, which includes mixing, non-mixing and simplified spacer grids. Plane-averaged pressure and velocity were extracted to ultimately inform the subchannel thermal-hydraulics code CTF. Several methods of planeaveraging are discussed and compared herein. The main method used in this work was a volume average weighted by a smoothed delta function to calculate an average over all the Gauss-Lobatto-Legendre (GLL) points within a thin volume. The results of this method are compared with averaged probe points in a single plane, the exact solution results, the linearized model predictions, and the post-processed visual data. The pressure average that was calculated using the main method closely matches the value obtained from visual post-processing. The density of the history points greatly affects the accuracy of the mean values; hence, an optimum number of points must be determined for any given plane.