CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   subsonic airfoil simulation (https://www.cfd-online.com/Forums/fluent/50518-subsonic-airfoil-simulation.html)

 miguel soto February 9, 2009 14:12

subsonic airfoil simulation

Hi, my name is miguel, is there anyone who have some experience within subsonic airfoil simulations in fluent?

I donīt know what can i do to have an accurate polar for a 4 digit naca airfoil, at a reynolds number of 3*10^6. I try to do several simulations, with different types of turbulence models, such as spallart-allmaras, k-epsilon, k-omega and even reynolds stress, but the outcome was not accurate enough, taking into account real wind tunnel data. I realize that spallart-allmaras model is not the best one, due to it was designed for transonic speeds, but i think that k-epsilon model would be a nice way to have the first approach.

If someone of you have experience, and know how to help me. for instance, i donīt know wich are the reference values for the "model constants", or if it is better to use k-epsilon in "standard", ralizable or RNG.

best regards

 Alagesanj February 10, 2009 09:38

Re: subsonic airfoil simulation

Hi Miguel,

Try SST k - omega model with transitional flow. It is giving best results. I am using this for airfoils with Re = 0.1 x 10^6 to 1 x 10^06 flow.

Cheers, Alagesan.

 Alagesanj February 10, 2009 10:01

Re: subsonic airfoil simulation

Hi Miguel,

Sorry for the inadequate answer. No need to modify the model constants. But your mesh should be fine enough for this model (Y+ = 1). Give realistic value for the turbulence intensity (less than 0.1) and length scale in your velocity inlet boundary condition. You can refer this link http://support.esi-cfd.com/esi-users/turb_parameters/ for turbulence intensity and length scale or refer fluent user guide. Try Realizable k - epsilon model with enhanced wall treatment, this should also produce good results.

Cheers, Alagesan.

 miguel soto February 10, 2009 17:07

Re: subsonic airfoil simulation

At the previous simulations i get nice results for the lift coefficient (far from stall), but the drag coefficient was awfull. I think itīs a good idea to use "enhanced wall treatment", i didnīt realize about this option.

Thanks a lot Alagesan, youīre very kind

 miguel soto February 10, 2009 17:35

Re: subsonic airfoil simulation

Do you think it would be better use a discretization of "third order MUSCL" for the flow, turbulence kinetic energy and turbulence dissipation rate, or it runs better with the "second order upwind" discretization?

regards

 Alagesanj February 10, 2009 18:40

Re: subsonic airfoil simulation

Hi Miguel,

The difficult part in airfoil simulation is the drag prediction, since laminar turbulence separation involves. Start using 2nd order upwind. Once you are comfortable with your results then you can try QUICK. If still your drag results are not good then you can modify the source term (constant values)trial and error in the fluid boundary condition, which will improve your results.

Cheers, Alagesan.

thanks

I came across this forum by chance and discovered this is one of the best thread ever. Thanks u guys so much for the information you give. So cool.

[color=********]comparatif simulation assurance vie multisupport[/color][color=********] - simulation assurance vie ! Les sites de simulation assurance vie[/color][color=********]comparatif simulation assurance vie multisupport[/color]

 delfel August 19, 2009 23:15

Hi Miguel,

As someone mentioned, there's other things you should do before you start playing with the turbulence model tuning parameters -- the default values given by fluent should be the best values for a majority of cases.

The first step should be to make sure that you've achieved grid independence (i.e. that your mesh is fine enough that the solution doesn't change with changes to the mesh). Let me know if you need tips on how to do this. Also, like Alagesan mentioned, you're going to want to use the enhanced wall treatments or equivalent, depending on which model you use. Make sure your first grid point is at a y+ value of approx. 1.

Cheers,
-sean

 All times are GMT -4. The time now is 04:03.