Development of an les methodology for complex geometries

Traditional steady-state CFD simulations fail to accurately predict the distributions of velocity and temperature in rod-bundles when the pitch (P) to diameter (D) ratio P/D is smaller than 1.1 for triangular lattices of cylindrical pins ([1]). This deficiency is considered due to the failure to predict quasi-periodical flow oscillations that are established between sub-channels as first observed by Hooper et al. ([2]). In fact, traditional turbulence modelling is not always reliable. Even in simple flows, the results can be not accurate when particular conditions occur. Examples are buoyancy, flow oscillations, and turbulent mixing. A complete unsteady simulation of turbulence might be necessary in order to achieve accurate results. Traditionally two main approaches are employed to study unsteady flows: the direct numerical simulation of the Navier-Stokes equations (DNS), and Large eddy simulation (LES). In the latter only the large scale of motion are solved, while the higher part of the Kolgomorov spectra has to be modeled. This approach has gained popularity in recent years due to its lower computational cost if compared to DNS, and to its maturity. Another possible solution is the adoption of an Unsteady Reynolds Averaged Navier Stokes (URANS) simulation. However, this approach has the drawback of being at an early stage of theoretical understanding ([3]). Moreover the use of LES and DNS allows for a deeper analysis of the flow field and the use of stochastical tools like POD ([4]). The present work presents the development of an LES methodology viable for complex geometries and suitable for the simulation of rod-bundles. Its main features include multi-block capability and the use of the fractional step algorithm. As an SGS model, a Dynamic Smagorinsky model has been used, because of its generality and its better results in the area of the narrow gap. The code has been tested on plane channel flow and the flow in concentric annular ducts. The results are in excellent agreement with experiments and previous calculations.