CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   SU2 (https://www.cfd-online.com/Forums/su2/)
-   -   Dimensionalization clarification (https://www.cfd-online.com/Forums/su2/113604-dimensionalization-clarification.html)

 osoucy February 22, 2013 18:37

Dimensionalization clarification

Suppose I want to solve for the turbulent flow past a three dimensional wing. The real physical length of the chord is 2 meters, but the grid has been scaled such that the chord is 1.0.

As such, I used the following setting:
CONVERT_TO_METER= 15.807

On the other hand, should I also set
REYNOLDS_LENGTH = 2.0

Or should the Reynolds Length be with respect to the original grid since it will be multiplied by CONVERT_TO_METER?

Regards,

 economon February 26, 2013 05:25

Hi Olivier,

Regardless of any non-dimensionalization that is applied using the supplied reference values in the configuration file (described here: http://adl.stanford.edu/docs/display...nsionalization), SU2 expects the mesh to be in meters for viscous flows. Please see the discussion on how to set up the conditions for a viscous simulation in SU2 within the following tutorial: http://adl.stanford.edu/docs/display...ulent+ONERA+M6.

In your case, if you have a Reynolds number based on a 2.0 meter chord length that you are trying to match but the mesh has been scaled to a chord of 1.0, you can simply scale up the mesh by using 'CONVERT_TO_METER= 2.0' while leaving the Re & Re Length as their original values. The value given for CONVERT_TO_METER is just a factor that multiplies the x, y, & z coordinates of all nodes in the mesh in order to scale the entire grid to a desired size. A new mesh file containing the scaled coordinates can even be written using the 'WRITE_CONVERTED_MESH= YES' option.

Hope this helps clear things up,
Tom

 dtucker April 23, 2013 00:44

Reference Length

2 Attachment(s)
Is it always possible to specify "lref"? ...I'm not clear on where that is coming from after looking at both references above.

I'm having trouble getting the Reynolds Number I'm expecting, I can see in the output file that density, velocity, and viscosity are all exactly what I predict and require. The length being used to calculate Re is 1, whereas my airfoil chord is 0.1524m, and that is what I'd like to use to calculate Re.

Is this doable?? ...I've tried:
CONVERT_TO_METER
REYNOLDS_LENGTH
REF_ELEM_LENGTH
...and even REF_LENGTH_MOMENT

Dave

PS: See attached for my config and output files.

 fpalacios April 26, 2013 00:12

Quote:
 Originally Posted by dtucker (Post 422475) Is it always possible to specify "lref"? ...I'm not clear on where that is coming from after looking at both references above. I'm having trouble getting the Reynolds Number I'm expecting, I can see in the output file that density, velocity, and viscosity are all exactly what I predict and require. The length being used to calculate Re is 1, whereas my airfoil chord is 0.1524m, and that is what I'd like to use to calculate Re. Is this doable?? ...I've tried: CONVERT_TO_METER REYNOLDS_LENGTH REF_ELEM_LENGTH ...and even REF_LENGTH_MOMENT Dave PS: See attached for my config and output files.
The incompressible SU2 non dimensionalization is coded between lines 5115 and 5130 in config_structure.cpp:

Pressure_FreeStream = 0.0;
Length_Ref = 1.0;
Density_Ref = Density_FreeStream;
ModVel_FreeStream = 0;
for (iDim = 0; iDim < val_nDim; iDim++)
ModVel_FreeStream += Velocity_FreeStream[iDim]*Velocity_FreeStream[iDim];
ModVel_FreeStream = sqrt(ModVel_FreeStream);
Velocity_Ref = ModVel_FreeStream;
Pressure_Ref = Density_Ref*(Velocity_Ref*Velocity_Ref);

if (Viscous) {
Reynolds = Density_Ref*Velocity_Ref*Length_Ref / Viscosity_FreeStream;
Viscosity_Ref = Viscosity_FreeStream * Reynolds;
}

as you can see by default the reference length is 1, in other words... if the grid and the config file parameters are dimensional, then the computed Reynolds number is based in a 1m reference length.

If you want to change the reference length, you should change Length_Ref in the config_structure.cpp file, but also non-dimensionalize the numerical grid (all the coordinates should be divided by the reference length). Another thing that you can do to obtain the desired Reynolds number with a dimensional numerical grid is to change the value of the viscosity.

Best,
Francisco

 All times are GMT -4. The time now is 02:03.