Comparison of Pebble Bed Velocity Profiles Between High-Fidelity and Intermediate-Fidelity Codes

TL;DR

This study compares high-fidelity and intermediate-fidelity computational models of pebble bed flow to improve the accuracy of flow predictions in high-temperature gas reactors, highlighting the need for better near-wall drag modeling.

Contribution

It demonstrates the use of high-fidelity spectral-element simulations to evaluate and improve intermediate-fidelity porous media models for pebble bed flows.

Findings

High-fidelity simulations provide detailed flow data for pebble beds.

Intermediate models overestimate velocities near the outer regions.

Porosity alone does not explain velocity differences, indicating the need for better drag correlations.

Abstract

Recent interest for the development of high-temperature gas reactors has increased the need for more advanced understanding of flow characteristics in randomly packed pebble beds. A proper understanding of these flow characteristics can provide a better idea of the cooling capabilities of the system in both normal operation and accident scenarios. In order to enhance the accuracy of computationally efficient, intermediate fidelity modeling, high-fidelity simulation may be used to generate correlative data. For this research, NekRS, a GPU-enabled spectral-element computational fluid dynamics code, was used in order to produce the high-fidelity flow data for beds of 1,568 and 45,000 pebbles. Idaho National Lab's Pronghorn porous media code was used as the intermediate fidelity code. The results of the high-fidelity model were separated into multiple concentric regions in order to extract porosity and velocity averages in each region. The porosity values were input into the Pronghorn model and the resulting velocity profile was compared with that from NekRS. Both cases were run with a Reynolds number of 20,000 based on pebble diameter. The Pronghorn results were found to significantly overestimate the velocity in the outermost region indicating that changes in the porosity alone do not cause the difference in fluid velocity. We conclude that further work is necessary to develop a more effective drag coefficient correlation for the near-wall region and improve predictive capabilities of intermediate fidelity models.