[Ra0501]

Hello everyone,

For my master's thesis, I am conducting a study on the aero-structural optimization of an F1 wing. I'm not very experienced with CFD simulations and I’m struggling to complete the grid convergence. It’s a 2D simulation of a multi-element (mainplane + flap) with high curvature and low Mach (I need to validate the Cp from experimental data), and I’m using RANS with the SST turbulence model.

My main concerns are related to the numerical aspect, specifically about the correct convergence criteria to adopt and, at this point, the quality of the mesh. From what I understand, I should probably adjust the CFL number. However, I’ve run other simulations with CFL=10 and CFL=5; in the first case, I obtained roughly the same results, while in the second case, for finer meshes, the solution does not stabilize and keeps oscillating. I’ve set the convergence criterion based on the drag residual.

I’m attaching the config files and some of the results I’ve obtained. I hope someone can provide advice and tips to achieve the grid independence, especially regarding the numerical setup.

Thank you!

[bigfootedrockmidget]

First thing to do is to make sure that your simulations have converged sufficiently. All residuals should converge to at least R=-10, but preferably to machine precision.
Do you have some residual plots of the convergence? You can use paraview to load the history file with the residuals and make a plot.



You can have a look at the Verification and Validation results (and the setup) of our multi-element case:
https://su2code.github.io/vandv/30p30n/

[Ra0501]

Here there are the plot of the residuals, They don’t reach -10 but are flat. I also attached the y+ plot and the a screenshot of the Coarse and the Superfine mesh I used. Now I’ll try to use the setup you suggested.
Thank you for now.

[bigfootedrockmidget]

Can you share the entire coarse mesh somewhere, and is there a paper describing measurements or other simulation results for this configuration?

[Ra0501]

Thanks to the setup you suggested, I managed to achieve convergence, so I sincerely thank you. However, I have another very important question.

In the configuration of the boundary conditions, I distinguished, apart from the Farfield, two surfaces: mainplane and flap because I needed to evaluate Cl and Cd for each surface individually. The strange thing I can’t understand is that for one of these surfaces, specifically the mainplane, I get a negative Cd. I’ve double-checked to ensure the mesh is correctly oriented (wind in the positive x-direction), but I still get this result. The global Cd is positive, but I’m not sure how to interpret this.

I’m attaching a history.csv file with the last 1400 iterations and the config file.

The mesh in SU2 format:

https://drive.google.com/file/d/1FYU...usp=share_link

[EvertBunschoten]

Hi Ra,

The effect you observe has to do with the effect of the flap onto the main plane. The gap between the flap and main plane creates a low-pressure area which reduces the adverse pressure gradient on the suction side of the main plane, reducing drag and delaying separation in the process. As a result, the aerodynamic force vector is oriented more orthogonal to the main plane camber line, which may result in an overall negative CD for this surface, depending on the angle of attack.

If you have the opportunity, try to run a couple more simulations where you increase the spacing between the main plane and flap and see what happens to the CD of the main plane. It's an interesting effect to observe.

[Ra0501]

Understood, thank you so much.
At the beginning I thought I was doing something wrong with the reference frame but surprisingly this effect seams to be quite real.

Moreover I am conducting 2-dimensional simulation and I had data based on the 3D surface and 3D simulations in order to verify the quality of the solution.And even if the Cp distribution was satisfactory and consistent, the coefficients were not and I couldn't understand if it was due to 3D effects or something else.

My ideas are clearer now, thanks.