UFR 3-09 Best Practice Advice: Difference between revisions
No edit summary |
|||
| Line 18: | Line 18: | ||
Key physics | Key physics | ||
•The heat transfer in the impingement region is mainly driven by the turbulent heat flux in the wall-normal direction (<v theta>). | |||
Therefore, the prediction of wall-normal velocity fluctuations is of primary importance. | Therefore, the prediction of wall-normal velocity fluctuations is of primary importance. | ||
| Line 32: | Line 32: | ||
Grids of the order 100x100 cells appear adequate but a grid sensitivity analysis is highly advised. | Grids of the order 100x100 cells appear adequate but a grid sensitivity analysis is highly advised. | ||
<u>Discretisation method</u><u> | <u>Discretisation method</u><u> </u> | ||
Use a second order convection scheme (convection terms are crucial). | Use a second order convection scheme (convection terms are crucial). | ||
| Line 64: | Line 64: | ||
Use eddy-viscosity models reproducing the damping effect of the wall (V2F, k-eps-fmu and SST models) or linear RSM with modified wall echo terms and Yap correction. | Use eddy-viscosity models reproducing the damping effect of the wall (V2F, k-eps-fmu and SST models) or linear RSM with modified wall echo terms and Yap correction. | ||
''- Best Practice Advice for flow details (Reynolds stresses and turbulent heat fluxes)'''' | ''- Best Practice Advice for flow details (Reynolds stresses and turbulent heat fluxes)'''' '' | ||
Use RSM integrable down to the wall and accounting for the wall blocking effect (nonlinear Two-Component Limit model and linear Elliptic Blending Model) | Use RSM integrable down to the wall and accounting for the wall blocking effect (nonlinear Two-Component Limit model and linear Elliptic Blending Model) | ||
| Line 70: | Line 70: | ||
Turbulent heat fluxes modelling | Turbulent heat fluxes modelling | ||
<span><font color="black"><font face=""Times New Roman"">-<span style="font: 7.0pt "Times New Roman""> | <span><font color="black"><font face=""Times New Roman"">-<span style="font: 7.0pt "Times New Roman""> </span></font></font></span>''Best Practice Advice for overall quantities (Nusselt number distribution)'' | ||
Use a simple eddy-diffusivity model with constant Prt number (0.9) or with Kays & Crawford correlation. | Use a simple eddy-diffusivity model with constant Prt number (0.9) or with Kays & Crawford correlation. | ||
Revision as of 10:00, 18 January 2010
Impinging jet
Underlying Flow Regime 3-09 © copyright ERCOFTAC 2004
Best Practice Advice
Best Practice Advice for the UFR
Key physics
•The heat transfer in the impingement region is mainly driven by the turbulent heat flux in the wall-normal direction (<v theta>).
Therefore, the prediction of wall-normal velocity fluctuations is of primary importance.
These fluctuations results from a competition between generation terms (convection from the jet and production by normal straining) and destruction terms (wall-blocking).
Numerical Issues
Grid and grid resolution
Use grid clustering at the heated wall: y+=1 is necessary with low-Reynolds number models (tolerance depends on the model).
Grids of the order 100x100 cells appear adequate but a grid sensitivity analysis is highly advised.
Discretisation method
Use a second order convection scheme (convection terms are crucial).
Computational Domain & Boundary conditions for UFR Impinging Jet
Boundary conditions
Inlet conditions: use a separate fully turbulent pipe flow computation
Wall: avoid wall functions
Right outlet: pressure or convective outlet boundary conditions
Upper boundary: use constant pressure. Symmetry boundary condition is possible if sufficiently far to avoid confinement (x/D>5)
Computational domain
-Inlet at least 1D before the pipe exit
-Right outlet: r/D>10
Physical modelling for UFR Impinging Jet
Turbulence modelling
- Best Practice Advice for overall quantities (Nusselt number distribution and mean velocity)
Dont use eddy-viscosity models which do not reproduce the correct damping effect of the wall and RSM without wall echo terms or with standard wall echo terms (Nusselt number overestimated by a factor of about 2.8).
Use eddy-viscosity models reproducing the damping effect of the wall (V2F, k-eps-fmu and SST models) or linear RSM with modified wall echo terms and Yap correction.
- Best Practice Advice for flow details (Reynolds stresses and turbulent heat fluxes)'
Use RSM integrable down to the wall and accounting for the wall blocking effect (nonlinear Two-Component Limit model and linear Elliptic Blending Model)
Turbulent heat fluxes modelling
- Best Practice Advice for overall quantities (Nusselt number distribution)
Use a simple eddy-diffusivity model with constant Prt number (0.9) or with Kays & Crawford correlation.
- Best Practice Advice for flow details (turbulent heat fluxes)
Use heat fluxes transport equations
Application uncertainties
In this application, all the conditions are very well defined (boundary conditions, Reynolds number, geometry, ). The only uncertainties come from the measurements methods : the uncertainty analyses given by the authors lead to low values (typically less than 5%).
© copyright ERCOFTAC 2004
Contributors: Remi Manceau - Université de Poitiers