Troubleshooting FSAE Simulation


#1

Hey Simscale community,

First of all it is very nice to work with Simscale because it has a nice and friendly user interface and a very nice support.

I’m using Simscale to analyse a the aerodynamics of a FSAE racecar. The simulation is working, but there are some points where I your need help.

The simulation is not using Yaw studies, just a simple straight flow CFD simulation, with MRF zones and porous media.

First point:


You can see, there are some wrong airzones in the frontwheel. Is it based on an issue with the CAD-geometry? I use a STL master file where are all parts included.

Second point:


Usually in the marked area should be to see the radiator. In the Mesh Creator and Simulation Designer it is shown, but in the Post Processor it is missing. What could be the problem on this?

Third problem:


The simulation results are not convergent, what could be the reason for this? Are there more simulation time steps needed than 3000?

Fourth question:


Quiet Similar to #3, the force plot of the rearwheel is not convergent. How can the setup of the simulation be improved, to get nice and clear results?

Last point:


The force plot of the radiator shroud gives a negative force in x direction. This would mean, it gives a forward force instead of a breaking force. This can’t be right.

Well, a few questions and problems which need attention.

I’m looking forward to your ideas and tips for solving them.

Kind regards

H Koeckemann


#2

Hey @h_koeckemann and thanks for reaching out to us!

Tagging our PowerUsers (@PowerUsers_CFD ) here who might give you some detailed answers to your question. Will also check these issues out and see what might cause this behaviour.

All the best!

Jousef


#3

Hi @h_koeckemann,

I think I can answer 3 and 4 for you. :slight_smile:

three is the same as four except you are looking at two different ways of determining convergence. Increasing the number of iterations doesn’t look like it will help you. If you decrease the under-relaxation values you will hopefully see better convergence.

Let me know if this helps!
Darren


#4

Hi @jousefm and @1318980,

Than you for the very quick responses.
I tried to get the simulation run with changed paramters for the Relaxation factor equations p,U,k,Omega to 0.2. With this setup the sim chrashed.
Now I try it again with values of 0.2 for Rel. fact. eq. p,U and 0.5 for k and Omega.
I let you know as I have a result.

Do you have an idea why it is crashed?

To go back to my first question with the wrong airzone in the frontwheel:


This is the meshing log for the hole car, the mesh thicknes in percentage of the frontwheel is very low. Could this be the reason for the problem #1? How can it be improved?

Thank you again and best regards

H Koeckemann


#5

Hi @h_koeckemann can you share your project link please?

Cheers,
Darren


#6

Hi @h_koeckemann, I was unable to make a copy but looking at your mesh there were a lot of illegal cells. This is likely the cause of your divergence. Reduce the minimum cell size allowed and increase your quality iterations, try to have no illegal cells present.

Kind regards,
Darren


#7

Hey Darren,

at first thank you for your help again. I tried a few different setups for meshing and simulation.

For the mesh I changed the refinment level for the body, it seems that it brings better results and will solve the problem #1 of this topic.

The mesh has the following attributes:

In view to the illegal cellls it should be ok?

Furthermore I tried a few different setups for the simulation, but they all really need long time and the most common sim with the new created mesh and relaxation values of respectively 0.2 was killed, because the sim was running out of time.

Do you have more tips to get better results?

Do illegal cells mean illegal faces (in the meshing log)?

At the moment this is my meshing setup where I got the results above:

Maybe you can help me to set it up a little less computing-power-needing, but even more accurate in view to the non divergent results and meshing errors of #1.

Thank you for the nice support :slight_smile:

Kind regards
H Koeckemann


#8

Hi @h_koeckemann, is there any chance you can make the project public so I can change some settings and do a few mesh runs. The amount of non-orthogonality is concerning and likely the cause of numerical instability, and the number of illegal faces is a good measure as to how good your mesh is. At this stage I don’t know what is causing your poor quality mesh however the first things I would try would be to ensure the cells are the same size in all directions for the base mesh, ensure that minimum size check was off, increase quality iterations and check layer settings.

Kind regards,
Darren


#9

Hey Darren,

Because of the competitions we’re taking part on, I’m not allowed by the team to give the car’s details to public.

I changed some values of the meshing, as result I got no illegal faces, but there are stil improvements needed.


This is the meshing log of the last created mesh, it seems the mesh is almost ok but:

17 tesselation
There are problems with the wheelzones.

By checking the mesh, it appears to this behavior:


This is with hidden wheel, the mesh is insteady it seems to be.

Could it be of intersections between the wheels rimflange and the mrf zone?

We are startet this year to make the CFD analysis for our competition car with Simscale, so it is important for us to get familiar with simscale and all its settings to make best development for the future possible.

Thank you for helping us in this realy nice way.

H Koeckemann


#12

@h_koeckemann, just spoke to @jousefm. and if you want you could give me ‘can copy’ access (you have already shared ‘view’ access) I can then copy a private copy and give better feedback?

It does sound like geometry intersection, you should ensure that the MRF is not intersecting with the wheel.

Best,
Darren

edit: the way the MRF intersects the wheel mounting doesn’t look good.


#13

Good morning Darren,

I changed the setting for the support sharing, now you have can copy acces. For sure it would be nice to get feedback to get the simulation accurate and stable working :slight_smile:

To the MRF zone, should it not intersect with non turning zones?

I will try this soon with a changed geometry, because the simulation with the last created mesh is taking horrible much time and computing power…

Have a nice day

Hmann