Numerical Simulation of Involute-Plate Research Reactor Flow Behavior

Nuclear research reactors around the world are undergoing conversion from highly enriched to low-enriched uranium to mitigate proliferation concerns. Several of these research reactors use complex “involute” plate designs; it is critical these be understood from fluid flow and heat transfer perspectives. High-fidelity simulations can yield deep insights while saving tremendous capital resources needed for implementation of involute reactor designs but require leadership-class machines due to their steep computational costs. To this end, researchers led by Argonne National Laboratory leveraged ALCF computing resources to simulate the flow behavior in involute-plate nuclear research reactors with three levels of fidelity.

Challenge

Using the ALCF’s Theta system, the researchers simulated flow behavior in involute-plate nuclear research reactors with three levels of fidelity: Reynolds-Averaged Navier-Stokes simulation (RANS), large-eddy simulation (LES), and direct numerical simulation (DNS).

Approach

For RANS simulations, three widely used turbulence models were applied through use of the commercial computational fluid dynamics (CFD) code STAR-CCM+. LES and DNS calculations were carried out via the open-source CFD code Nek5000. A single coolant channel between involute fuel plates was chosen as the computational domain; velocity magnitude, lateral velocity, and turbulence kinetic energy from different models were compared and analyzed. ALCF staff support included help porting and optimizing the code for deployment on ALCF systems.

Results

The results from RANS were found to be in good agreement with LES and DNS data. Although some discrepancies were observed between LES and DNS results in some domain segments, the deviations were relatively negligible. Given that the computational expenses of DNS calculations are an order of magnitude higher than LES, using LES for benchmarking RANS models was determined to be a cost-effective approach for studying involute-plate research reactors.

Impact

This work directly supports several key thermal-hydraulic topics related to the conversion of the involute-plate reactors and demonstrates how high-fidelity simulations can add value where experimental data are sparse or nonexistent. Moreover, data collected from high-fidelity simulations can be used to justify or improve the coefficients or methods used in the low-fidelity methods. In this case, good agreement between reference high-fidelity solutions and more intermediate-fidelity approaches makes use of lower-computational cost modeling and simulation a reliable approach. This study supports the validation of modern methods for the analyses of involute-plate reactors, which may help simplify designs and therefore reduce the cost and time required to achieve conversion to low-enriched fuel.