Turbulence Modeling Resource

Exp: CFDVAL2004 Lessons Learned

Return to: CFDVAL2004 - Intro Page

These "lessons learned" from the workshop have been compiled by the Page Curator. Many of these lessons can also be found in some of the documents and papers found on the SUMMARY OF RESULTS AND REFERENCES page (for example, see the summary paper AIAA Journal, Vol. 44, No. 2, Feb. 2006, pp. 194-207).
 
 

Case 1:

• For Case 1, including the entire internal as-built cavity in the CFD computations was very difficult and of questionable value
• Case 1 was a difficult experiment to simulate overall
  • Flow field may have been partially laminar or transitional
  • Piezo-electric driver and its effects (e.g., non-sinusoidal jet velocity at exit) were difficult to model in CFD
  • Ring vortices (3-D end effects) formed from slot ends probably influenced flow field away from the wall

• For Case 2, CFD missed some aspects of the flow field at the cavity exit
  • Experiment produced large cross-flow velocity component during part of the cycle
  • Need additional documentation of experimental orifice exit BCs over greater part of the face, as a function of phase

• For Case 3, CFD must account for side plate blockage to match CPs
• For Case 3, Reynolds-averaged Navier-Stokes (RANS) generally overpredicted separation length and underpredicted magnitude of turbulent shear stress in the separated region
  • This appears to be a turbulence modeling issue
  • Large-eddy-simulation and direct-numerical-simulation type methods hint toward improved trends, but these are difficult and time-consuming methods to run well

• In order to compare CFD methods with each other, they must employ the same BCs
• In order to compare CFD methods with each other, grid and temporal refinement studies must be performed (running on the same grid can be particularly helpful, as it removes one of the variables when trying to compare turbulence models, codes, etc.)
• Insuring that CFD has the same boundary conditions as experiment is necessary, but can be very difficult.

• CFD needs extremely well-documented BCs, especially at the flow-control hole/slot
  • Measured velocity profiles at hole/slot are particularly helpful
• CFD Needs thorough assessment of two-dimensionality if dataset is intended for 2-D computations
• CFD needs accurate assessment of as-built shape, including interior of hole/slot and plenum
• Use of multiple measurement techniques (e.g., LDV and PIV to measure the same thing) are very useful and can reveal that experimental uncertainty can be high for these types of synthetic jet flow fields
• Use of both high resolution and low resolution PIV to measure the same thing is useful for assessing experimental uncertainty
• The more details given, the better for CFD validation, because the biggest challenge is to get CFD to run the "same problem" as experiment
  • Thorough documentation of freestream conditions (both upstream and downstream), conditions in jet plenum, etc.
  • Details of jet driving mechanism (mechanical? electrical? frequencies? modes?)
  • Velocity and turbulence profiles at multiple stations, and especially at the flow-control hole/slot

Return to: CFDVAL2004 - Intro Page

Page Curators: Christopher Rumsey, Ethan Vogel, Clark Pederson
Last Updated: 05/15/2021