Mixing ventilation flow in an enclosure driven by a transitional wall jet

Confined Flows

Underlying Flow Regime 4-20

Best Practice Advice

Key physics

Mixing ventilation flows are driven by high-momentum supply jets and generally consist of one or more recirculation cells, which can extend over the entire width of the enclosure (e.g. Awbi 2003). In this UFR case, one large recirculation cell is present which fills almost the entire cubical enclosure. Small secondary recirculation cells are present in the corners opposite to the supply jet and below the supply jet. The supply jet in this UFR is a transitional (low Reynolds number) wall jet. Transitional wall jets are highly transient, featuring the development of vortical structures in the shear layer due to Kelvin-Helmholtz type instabilities, as could be seen in Fig. 4. The detachment of the wall jet is Reynolds number dependent and occurred further downstream with increasing Reynolds number. A comparison with theoretical profiles for laminar and turbulent jets (Fig. 5) indicated that the measured flow was not fully turbulent, not even at Re ≈ 2,500.

Numerical modeling issues

A grid-sensitivity analysis is imperative for any CFD study and should be conducted to ensure the grid is providing (nearly) grid-independent results. In this case, the grid with 3,437,056 cells provided nearly grid-independent results. 3D simulations are needed for this specific case since 2D simulations resulted in different vertical profiles of streamwise velocity. Apparently, the side walls influence the overall velocity distribution in the cubical enclosure. Generally, discretization schemes of at least second-order upwind need to be applied to reduce the amount of numerical diffusion as recommended in numerous publications before. More information on the effect of discretization schemes on the results of CFD simulations of mixing ventilation driven by transitional wall jets can be found in van Hooff and Blocken (2017). Oscillatory convergence was present in the majority of the simulations. It is advised to carefully check the residuals and to monitor the parameter of interest in several points during the simulation. Subsequently, when oscillatory convergence is present it is needed to average the results over a sufficient number of iterations. It is also possible to resort to unsteady RANS simulations and average the results over a sufficiently large number of time steps. However, tests for another flow problem indicated that URANS provides similar results to steady RANS when both are averaged long enough. The effect of turbulence boundary conditions at the inlet on the detachment of the wall jet was observed and the boundary conditions were determined based on the obtained levels of turbulent kinetic energy. Care must be taken to specify/estimate correct turbulence levels at the inlet boundary since these can have a large effect on jet detachment and thus on the overall flow pattern.

Physical modeling

The results from this UFR case show that in this particular case, the low-Re k-ε RANS turbulence model by Chang et al. (1995) is capable of predicting mean velocities that are in a quite good agreement with the time-averaged velocities from the PIV experiments. However, the prediction accuracy of local turbulence levels in the boundary layer and shear layer by the tested turbulence models is considered to be less, most probably due to the fact that the tested RANS models were not specifically developed to model transitional flows (laminar to turbulent transitions); i.e. they are mainly developed for fully developed turbulent flows. The deviations between experiments and numerical results are larger for smaller Reynolds numbers, and the differences thus become less severe with increasing Reynolds numbers and increasing turbulence levels. The correct prediction of the detachment of the wall jet from the top surface is of primary importance for a proper prediction of the overall flow pattern. Therefore, the use of low-Reynolds number modeling for the near-wall flow is recommended and it is shown that it can lead to a fair to good prediction of jet detachment in this particular case. Further improvements can be expected when specific transition turbulence models are employed.

Application uncertainties

In addition to the possible systematic and random (repeatability) errors in the PIV measurements (see Section 2.2.2), the differences between the experimental results and CFD results can mainly be attributed to the assumptions needed to define the turbulence levels and exact velocities at the inlet. Due to practical constraints it was not possible to measure in the inlet opening and therefore assumptions are the only means to provide the inlet conditions.

Recommendations for future work

Future work can focus on a detailed comparison with LES simulations. Van Hooff et al. (2014) published results of such a comparison for Re ≈ 2,500, but additional research efforts on LES for other flow configurations (i.e. Reynolds numbers, inlet heights) and using other computational settings (e.g. subgrid-scale models, grid type and resolution) can be of added value. Furthermore, the assessment of other more advanced RANS turbulence models, such as the v²-f model by Durbin (1995), or the turbulence models specifically developed for transition modeling, such as the transition SST k-ω model by Menter et al. (2006, 2015) and Langtry and Menter (2009) is highly recommended. Especially the transition models should improve the prediction accuracy for this case in which the inlet flow is not fully developed and undergoes the transition from laminar to turbulent. The application of the transition turbulence models might thus lead to a better prediction of the turbulent kinetic energy levels and mean velocities in steady RANS CFD simulations. In addition, measurements of the same flow problem (mixing ventilation flows) for a longer geometry, i.e. extended in streamwise direction, would be of interest due to the reduced influence of the pressure gradient in that specific case which will lead to an even stronger influence of the vortex formation in the inner and outer layer of the wall jet on the jet separation.

Contributed by: T. van Hooff(*), B. Blocken(*), G.J.F. van Heijst(**) — (*)Dept. of Civil Engineering, KU Leuven, Belgium and Dept. of the Built Environment, Eindhoven University of Technology, the Netherlands.
(**)Dept. of Applied Physics, Eindhoven University of Technology, the Netherlands