Time-dependent and time-independent CFD simulations of the flow through a staggered tube bundle were performed. This flow configuration partially simulates the anticipated flow in the lower plenum of a Very High Temperature Reactor (VHTR) design. To design a nuclear reactor with confidence, one needs strict benchmarking as part of a validation and verification exercise for any and all commercial CFD codes. Thus CFD simulations (FLUENT) of isothermal (at present), periodic flow through a tube bundle using both Steady Reynolds Averaged Navier-Stokes (SRANS) and Unsteady Reynolds Averaged Navier-Stokes (URANS) equations were investigated. Selected turbulence models for a single tube diameter and inlet velocity based Re-number, Re ∼ 1.8 × 104, were investigated. The first-order turbulence models were: a standard k-ε turbulence model, a Renormalized Group (RNG) k-ε model, and lastly, a Shear Stress Transport (SST) k-ε model; the second-order model was a Reynolds Stress Model (RSM). Comparison of CFD simulations against experimental results of Simonin and Barcouda was undertaken at five stations (x, y) locations. Under the SRANS, we found the ability of the models to predict the turbulence stresses (u′u′, v′v′, u′v′) generally marginal to poor. However, upon adapting a concept from Large Eddy Simulation (LES), our URANS simulation with RSM revealed a spatiotemporal, oscillating flow structures in the wake. In contrast, it appears that the URANS with (even a) RNG k-ε model is unable to simulate this flow phenomena. In fact, the data suggests that the RNG k-ε model is too spatiotemporally dissipative. Some aspects of the SRANS versus URANS and using the aforementioned turbulence models will be presented.

This content is only available via PDF.
You do not currently have access to this content.