Division of Scientific Computing
Department of Information Technology
Fully resolving the whole spectrum of high Reynolds number turbulent flows will not be possible in the foreseeable future due to limitations in computing power. Large Eddy Simulation is a turbulence modelling approach, where only the part of the spectrum corresponding to large structures is resolved. This allows to drastically reduce the amount of computational resources needed, but is, unfortunately, still too expensive to be used in case of high Reynolds number wall-bounded flows, typically occurring in applications.
An established approach for further reducing the computational demands is avoiding the need to resolve the near-wall region of the flow. A simple, and thus attractive, way to do this is to prescribe the values of the unknowns in this region based on some a priori known relationship. Such relationships are known as wall-models.
In our project we try to analyze the behavior of the existing wall-models and understand how their predictive capability is affected by the physics of the flow as well as other modelling choices such as grid resolution, choice of model-constants and used numerical methods. Additionally, we hope that the gained insights will allow us to develop new better models.