FLOW: Simulations of turbulent boundary layers with passive scalars

Caption: Contours of the streamwise velocity in the cross plane at the inlet together with the disturbances from the free stream.

The study of turbulent boundary layers with passive scalar transport has been an important research topic for the last few decades, since such a problem involves two fundamental aspects: the development of turbulence in a thin region adjacent to a solid wall, and the transport and diffusion of passive species by turbulent motion. The accurate prediction and eventually an improved understanding of the dynamic processes within this turbulent boundary layer are crucial factor when considering technical applications: Many appliances in industry contain such (or similar) geometries coupled to the transport of the passive scalars: See for example internal combustion engines, chemical mixing, but also large-scale geophysical environments. Furthermore, an understanding of the passive scalar mixing might also provide an impetus for the study of turbulence itself, since locally the passive scalar can be viewed as a marker for turbulent eddies and thus allow an in-depth study of that chaotic movement.

The traditional way to investigate turbulence problems is via laboratory experiments. However, the obtained results are only possible at moderately high Reynolds number, which might lead to inherent measurement difficulties and inaccuracies close to the wall. Also, there are certain quantities that are extremely difficult or even impossible to measure, such as budget terms or pressure correlations. With the help of parallel computers with fast inter-connection, fully resolved numerical simulations gradually become available at Reynolds number high enough to i) overlap with experimental results for a in-depth cross-validation, and ii) to provide insight into real turbulent motion away from low-Reynolds-number effects. The present project aims at simulating a turbulent boundary layers via both the large-eddy simulation (LES) and direct numerical simulation (DNS) approaches, extending previous DNS works to medium/high Reynolds numbers [2,3]. On-going are new simulations at even higher Reynolds number up to Re=2500 (based on momentum thickness and free-stream velocity) with three passive scalar with the molecular Prandtl numbers ranging from 0.2 to 2.0. With the present Reynolds number, the scale separation is more pronounced, and influences of the large-scale structures on the scalar transport will be studied. In addition, the scaling property of the previous findings based on the low Reynolds number data will be tested at high Reynolds number. The employed numerical code [1] is fully parallelised such that such simulations at relatively high-Re are feasible.

In addition, such simulations will also be used as baseline cases for studying the influence of ambient free-stream turbulence on the heat transfer on the wall (see [3] and Figure to the left). Previous experiments observed (unexplained) increased heat transfer under such circumstances. Therefore, it is also interesting and important to understand what the physical mechanism behind, since such situations correspond to the typical environment in gas turbine industry.

DNS Data


Qiang Li