Structure-based resolution of turbulence for sodium fast reactor thermal striping applications

Thermal striping in sodium fast reactors (SFR) is characterized by oscillatory mixing of non-isothermal sodium coolant streams, and is a potential cause of thermal fatigue damage in upper-plenum materials. Accurate simulation of thermal striping is essential to support both reactor design and operation, but it is severely constrained by CPU requirements and uncertainty introduced by turbulence closures. Unsteady Reynolds-averaged Navier-Stokes (URANS) models are incapable of providing a reasonable level of description of flow fields, including variances and fluctuation spectra. Those parameters are needed to understand and mitigate the effects of thermal striping. Large-eddy simulation (LES) models are computationally restrictive for typical high-Reynolds-number reactor flow applications. Here, the performance of a recently developed structure-based (STRUCT) hybrid URANS/LES turbulence approach in predicting thermal striping is assessed on a triple-jet water case. The STRUCT approach is characterized by the identification of flow regions where the scale-separation assumption of URANS is not met. In those regions, partial resolution of turbulence is applied. Unresolved scales in the fully resolved and partially resolved regions are treated with a nonlinear eddy-viscosity model. The goal is to assess the robustness of the STRUCT approach for simulations of thermal striping. STRUCT demonstrated LES-like accuracy on computational grids typical of URANS simulations, reproducing experimental profiles and dominant temperature fluctuation frequencies. A reduction in computational cost of 70 times was achieved over LES, while maintaining the reliable grid-convergence behavior of URANS.