On The Numerical Prediction Of Transitional Flows With Reynolds-Averaged Navier-Stokes And Scale-Resolving Simulation Models
Author Pereira, F.S. and Vaz, G.
Title On The Numerical Prediction Of Transitional Flows With Reynolds-Averaged Navier-Stokes And Scale-Resolving Simulation Models
Conference/Journal OMAE ASME 35th International Conference on Ocean, Offshore and Arctic Engineering, Busan, South Korea
Paper no. OMAE2016-54414
Month June
Year 2016

Abstract
The present work investigates the transitional flow around a smooth circular cylinder at Reynolds number Re = 140000. The flow is resolved using the viscous-flow solver ReFRESCO, and distinct mathematical models are applied to assess their ability to handle transitional flows. The selected mathematical models are the Reynolds-Averaged Navier-Stokes equations (RANS), Scale-Adaptive Simulation (SAS), Delayed Detached-Eddy Simulation (DDES), eXtra Large-Eddy Simulation (XLES) and Partially-Averaged Navier-Stokes (PANS) equations. The RANS equations are supplemented with the k-omega Shear-Stress Transport (SST) with and without the Local Correlation Transition Model (LCTM). The numerical simulations are carried out using structured grids ranging from 9.32e4 to 2.24e7 cells, and a dimensionless time-step of 1.50e-3. As expected, the outcome demonstrates that transition from laminar to turbulent regime is incorrectly predicted by the k-omega SST model. Transition occurs upstream of the flow separation, which is typical of the supercritical regime and so the flow physics is incorrectly modelled. Naturally, all Scale-Resolving Simulation (SRS) models that rely on RANS to solve the boundary-layer, called hybrid models, will exhibit a similar trend. On the other hand, mathematical models capable to resolve part of the turbulence field in the boundary layer (PANS) lead to a better agreement with the experimental data. Furthermore, the k-omega SST LCTM is also able to improve the modelling accuracy when compared to the k-omega SST. Therefore, it might be a valuable engineering tool if its computational demands are considered (in the RANS context). Therefore, the results confirm that the choice of the most appropriate mathematical model for the simulation of turbulent flows is not straightforward and it may depend on the details of the flow physics.


   Subject to copyright regulations
Disclaimer | print