UFR 1-05 Best Practice Advice: Difference between revisions
Jump to navigation
Jump to search
David.Fowler (talk | contribs) No edit summary |
David.Fowler (talk | contribs) |
||
| Line 16: | Line 16: | ||
== Best Practice Advice for the UFR == | == Best Practice Advice for the UFR == | ||
* Distortions of the grid should be avoided in the immediate vicinity of the jet exit. | |||
* For large eddy simulations, dimensions should be around Δx=0.004D and Δz=0.006D. Scaled up by the shear velocity of the turbulent pipe flow, the dimensions should be around Δx<sup>+</sup><nowiki>=1.4 and </nowiki>Δz<sup>+</sup><nowiki>=2.2.</nowiki> | |||
* With LES, the following boundary conditions should be used: | |||
** Free slip on the lateral and top surfaces | |||
** No slip on the bottom | |||
** Laminar boundary layer velocity profile at inlet | |||
** Zero gradient for all variables at outlet | |||
** Instantaneous velocity profiles given by a temporal pipe flow simulation, at the pipe flow exit. | |||
* With high velocity ratios, turbulence modelling is not critical, particularly with respect to the region near the jet, where inviscid flow conditions dominate. | |||
* Particularly with low velocity ratios, where turbulence is important near the jet exit, eddy viscosity models should be avoided. However a clearly winner turbulence model has not been found yet for JICF and Baldwin-Lomax may be used with reservations. | |||
<u>Recommendations for future work</u> | <u>Recommendations for future work</u> | ||
* The Baldwin-Lomax turbulence model should be tested with a very well designed, undistorted grid at the immediate vicinity of the jet exit. | |||
* New experimental and computational work should be undertaken with velocity ratios R=1 or < 1, as these values are closer to those encountered in blade cooling. | |||
* In general terms, the thin layer approximation of the Navier Stokes equations seems to produce good quality results, as even the more complicated behaviour of the jet with swirl was reasonably predicted. However, more work is necessary to define the best choice of turbulence model. | |||
* The semi-elliptic scheme applied by Bergeles et al. worked well for velocity ratio up to 0.1 with injection angle of 90 degrees and for velocity ratio up to 0.5 with injection angle of 30 degrees. They have used a k-ε turbulence model, but recommend an algebraic stress treatment to be tried in future work, including all the secondary production terms associated with the three-dimensional flow field. | |||
<font size="-2" color="#888888">© copyright ERCOFTAC 2004</font><br /> | <font size="-2" color="#888888">© copyright ERCOFTAC 2004</font><br /> | ||
Revision as of 17:56, 10 March 2009
Jet in a cross flow
Underlying Flow Regime 1-05 © copyright ERCOFTAC 2004
Best Practice Advice
Best Practice Advice for the UFR
- Distortions of the grid should be avoided in the immediate vicinity of the jet exit.
- For large eddy simulations, dimensions should be around Δx=0.004D and Δz=0.006D. Scaled up by the shear velocity of the turbulent pipe flow, the dimensions should be around Δx+=1.4 and Δz+=2.2.
- With LES, the following boundary conditions should be used:
- Free slip on the lateral and top surfaces
- No slip on the bottom
- Laminar boundary layer velocity profile at inlet
- Zero gradient for all variables at outlet
- Instantaneous velocity profiles given by a temporal pipe flow simulation, at the pipe flow exit.
- With high velocity ratios, turbulence modelling is not critical, particularly with respect to the region near the jet, where inviscid flow conditions dominate.
- Particularly with low velocity ratios, where turbulence is important near the jet exit, eddy viscosity models should be avoided. However a clearly winner turbulence model has not been found yet for JICF and Baldwin-Lomax may be used with reservations.
Recommendations for future work
- The Baldwin-Lomax turbulence model should be tested with a very well designed, undistorted grid at the immediate vicinity of the jet exit.
- New experimental and computational work should be undertaken with velocity ratios R=1 or < 1, as these values are closer to those encountered in blade cooling.
- In general terms, the thin layer approximation of the Navier Stokes equations seems to produce good quality results, as even the more complicated behaviour of the jet with swirl was reasonably predicted. However, more work is necessary to define the best choice of turbulence model.
- The semi-elliptic scheme applied by Bergeles et al. worked well for velocity ratio up to 0.1 with injection angle of 90 degrees and for velocity ratio up to 0.5 with injection angle of 30 degrees. They have used a k-ε turbulence model, but recommend an algebraic stress treatment to be tried in future work, including all the secondary production terms associated with the three-dimensional flow field.
© copyright ERCOFTAC 2004
Contributors: Flavio Franco - ABB ALSTOM Power UK Ltd