Turbulent Scalar Flux Models for the Mixing of Hot Gases in the Swirling Flow of High-Temperature Reactor Lower Plenums

With recent progress in high-temperature pebble-bed reactor programs research focus has started to include more ancillary engineering issues. One very important aspect for the realisability is the mixing of hot and colder helium in the reactor lower plenum. Under nominal operating conditions, depending on core design, the temperature of hot gas leaving the core can locally differ up to 210° C. Due to material limitations, these temperature differences have to be reduced to at least ±15° C. Several reduced-size air experiments have been performed on this problem, but their applicability to modern commercially sized reactors is not certain. With the rise in computing power CFD simulations can be performed in addition, but advanced turbulence modeling is necessary due to the highly swirling and turbulent nature of this flow. The presented work uses the geometry of the German HTR-Modul which consists of an annular mixing channel and radially arranged ribs. Using the commercial CFD code ANSYS CFX, we have made detailed analyses of the complex 3D vortical flow phenomena within this geometry. Several momentum transport turbulence models, e.g. the classical k-e model, advanced two-equation models and Reynolds-Stress Models were compared with respect to their accuracy for this particular flow. In addition, the full set of turbulent scalar flux transport equations was implemented for modeling the three components of turbulent transport of enthalpy seperately and were compared with the standard turbulent Prandtl number approach. As expected from previous work in related fields of turbulence modeling, the differences in predicting the mixing performance between models were significant. Only the full Reynolds-Stress model coupled with the scalar flux equations was able to reproduce the experimentally observed reduction of mixing efficiency with increasing Reynolds number. The correct scaling of mixing efficiencies demonstrates that the utilized turbulence models are able to reproduce the physics of the underlying flow. Hence they could be employed for the scaling and optimization of the lower plenum geometry. The results also showed that the original geometry used for the HTR-Modul is insufficient to provide adequate mixing, and that hence a not sufficiently mixed coolant for future reactor designs might be an issue. Based on this work, an optimization for future lower plenum geometries has become feasible.