Inscrição na biblioteca: Guest

ISSN Online: 2379-1748

ISBN Flash Drive: 978-1-56700-431-1

ISBN Online: 978-1-56700-430-4

First Thermal and Fluids Engineering Summer Conference
August, 9-12, 2015 , New York City, USA


Get access (open in a dialog) pages 379-382
DOI: 10.1615/TFESC1.cmd.013177


Recent studies on fluid flow and heat transfer in a matrix of surface-mounted cubes demonstrated that the Reynolds-Averaged Navier-Stokes (RANS) models are inadequate for such cases. Measurements, which were explained in detail in various publications, were used as a reference point to the present calculations. At the measurements section, the flow is declared to be fully developed and only one cube in this region was heated. This means that the case can be set up with periodic boundaries for all equations except for the energy equation. This also means that the error is accumulated in each periodic cycle and therefore RANS models deliver too large errors in final results. Therefore, this paper will compare alternative methods. As the representative approach of variable-resolution methods, the Partially-Averaged Navier-Stokes (PANS) was chosen. The PANS method belongs to so called bridging or seamless methods. This approach adjusts seamlessly from the Reynolds-Averaged Navier-Stokes (RANS) to the Direct Numerical Solution (DNS) of the Navier-Stokes equation. The results are largely improved by using the PANS. On the other hand, the variant of Large Eddy Simulation (LES) model known as the Coherent Structure model which has some advantages in calculating the near wall region is used for the current comparisons as well. A special focus of this work is to analyse the influence of the unresolved or modelled part of the turbulent kinetic energy especially in the near wall region on the overall accuracy of the heat transfer calculations.