Development of Advanced Turbulent Flow Simulation Techniques for Use in Nuclear Reactor Safety Analysis, 20-R9680Printer Friendly Version
Inclusive Dates: 01/01/07 Current
Background - Safety analyses and accident simulations for nuclear reactors primarily use one-dimensional system-level codes. These system-level codes rely on correlations to resolve complex phenomena such as thermal mixing, cross flow, and turbulence, which are often encountered in complex fuel bundle geometries. Rapid advances in computational fluid dynamics make it feasible to analyze reactor components by solving the full Navier-Stokes equations and reducing dependence on empirical correlations. However, some specific technical issues such as turbulence related to reactor single-phase flow simulations remain to be resolved before they can be reliably used for safety analyses and regulatory purposes. A team of SwRI engineers is exploring and developing advanced multiscale turbulence models for application to reactor component flows to assess their robustness and applicability.
Approach - Existing commercial computational fluid dynamics (CFD) solvers are used to simulate the flow and thermal environment with modified turbulence models such as the traditional k-ε, k-ω, renormalization group k-ε, and the large eddy simulation (LES) models. Previous studies have shown that the Reynolds-averaged representations of the turbulence failed to capture the desired physics and the high-fidelity models such as LES are computationally expensive. Hybrid multiscale turbulence models such as detached eddy simulation (DES) are a promising new approach that strikes a balance between the traditional Reynolds-averaged-Navier Stokes equations (RANS)-type formulation and LES techniques. These multiscale models provide a better representation of turbulence compared with the available models in the commercial solvers and could be used to customize and extend the already implemented models in the solvers.
Accomplishments - The project team has tested a number of well-established turbulence models for a single fuel-rod configuration. The unsteady data obtained using the single-rod geometry have been analyzed to understand the frequency contents of the signal. Subsequently, flow around rod bundles was simulated to understand the effectiveness of the model in capturing the effect of vortex breakdown and interaction of vortical flow with downstream bluff bodies. Implementation of hybrid multiscale models that accounts for anisotropy is under way.