Hello, I am currently making simulations for our FSAE team and i am having some problems with the mesh. I have already done about 8 successful meshes and simulations in the project links provided, one straight on and one cornering simulation. I noticed that some of the settings on simscale have changed and now the last 5 meshes i have done have failed. I am not sure what is wrong as the geometry hasnt changed much other then new aero elements and the meshes were ok before. Swttings are all the same. Can someone please take a quick look to see if they can find the problem.
Thank you,
Dan
In the picture below, the message in the event log has always been there with my previous meshes and they turned out fine.
[Deleted by author]
Wing3andDiffuser1 - mesh fine
Wing1withDiff2 - mesh failed
Wing3withDiff2 - mesh failed
similar results for the cornering mesh in this project
[Deleted by author]
For some extra info- I have tried my best to make a watertight Solidworks CAD model. Most parts were saved as Part files, not assembly files, and merged using the Combination feature. This saves the model as one solid body and no longer has features. I have merged parts such as the entire front suspension, a-arms, brackets, motors, etc… in this method to try and avoid intersections.
Then I proceeded to manually merge the STL files together (changing the names, no spaces, special characters etc…) such as Front and Rear suspension as well as the main roll hoop (because they should all receive the level 7-8 mesh refinement) as one STL file so everything can be selected in one click. I also made a master STL file, as many others have done, and simply added the Aero features to this.
this is the biggest intersection problem i have had. Its not really an intersection but the parts are so close to each other that the mesh thinks its one part and tries to close it. i am working on fixing this in the geometry but so far has not led to any failed meshes (until the last couple i have tried)
I would always make sure that the geometry is fixed in the first place to avoid such issues. The message in the mesh tree also indicates that there are problems with intersections in the CAD model. I have started a mesh operation myself now to see if I can be successful but to have no mesh related issues that might stop the simulation with an error I would meticulously fix the CAD model first before proceeding with the next step. Our PowerUser @Get_Barried might add his two cents here.
The previous meshes you ran, was there a significant time gap between you running those meshes vs the ones you just tried to run? Maybe a month?
What was the issue with the previous meshes? Not possible to re-use them?
Geometry wise, with the successful meshing, was the approach to generating them the same?
My suspicion on why you’re facing issues is 2 fold. One, as you’ve also noticed, SimScale has been significantly updated. It seems like the mesh algo has changed and it may be more sensitive to poor geometries than before. Two, you ran out of memory space. This is more unlikely, but could happen, even across different projects on the same account.
Moving forward from here, it is likely the issue you’re facing is issue One. To make it work again, you need to some serious work on the geometry. This you may need to go back to CAD and ensure all parts are separate with specific connecting joints that you can later assemble and constrain before exporting to STL.
However, as mentioned, lets probably try to see if problem Two is giving you issues.
Cheers.
Regards,
Barry
EDIT: Actually you can ignore some of what I just typed, its good for reference however. Your meshing actually completed, but was unable to save the file and hence the error. Refer to the meshing log of your latest meshing attempt.
It mentions “Finalising parallel run” which means your mesh, as mentioned, is actually successful. However, for some reason it was unable to save. So this does point to issue Two where you’ve actually run out of memory for whatever reason. Your mesh was only 16 million cells which is quite a bit but not too drastic. Try deleting all your unused meshes/projects/geometries/simulations and see if you’re able to mesh then. Of course if it is important, don’t delete it, or at the very least download it and save it.
Thanks so much @Jousef and @Get_Barried for all your help. Sorry for the long post but there is a lot to say
To start from the beginning, Jousef, i have tried my best to make the CAD geometry as watertight as possible. I know that this is the most important step for a good mesh and as i have said, the Solidworks part files create a non-intersecting solid body as far as i know. It is hard to pinpoint exactly where intersections occur if they are very small and apart from the picture i provided, i havent found any obvious ones.
For Get_Barried, addressing some of the things you said to ignore: (good for Reference i agree haha)
No the previous meshes were not done more than a week beforehand. Especially the mesh named Wing1andDiffuser1
As far as i know i have to re-mesh the entire car when i add aero features to the model. (because its a different STL file)
3, Yes i have made a set of instructions for making meshes just like the ones provided on the simscale platform. Unless i make a small mistake all of the mesh Parameters, boundary conditions, and surface refinements all have the same values.
I have also noticed the updates to the simscale platform and before i origonally created this forum post i ask the same question through the quick chat channel and got this response.
‘Hi Dan, Sorry for the late reply. This is a recent issue that some of our users are facing, we are looking into the problem and will let you update you soon. In the meantime, may I ask that you give your CAD model a second look? I noticed that the mesher does give an error as in the image below, please look for any open faces and self-intersections in the CAD.
Best,
Legin’
Regarding the memory space i was going to ask Jousef about this. I am currently signed up for the Student Plan with 10,000 Simulation hours through the sponsorship. However in the project selection page it states that i have 2235/3000 core hours used. First, i am not sure if this is a mistake because i would think this would say 2235/10,000 and second, even if this was true i should not have used up that amount of hours with these meshes.
I am a bit confused how i could have run out of memory unless it about the core hours issue. i would like to avoid if possible deleting meshes and simulatoins as it makes it much easier to reference everything in one place. Also, in the previous year simulating with simscale, my teammate had 29 meshes and 27 simulations and kept everything in one project.
let me take care of your core hour issue. I usually upgrade to 10k core hours when people reach out to me as some of the users never use that much and stay well below 3k. You should see an update in around 10 minutes.
So today i went back to try and figure out the problem with the geometry and did some tests. First i fixed some problems that i thought might be the cause (CFDtest1) - still mesh failure. Then i tried no contact patches because throughout all my meshes i am getting a low thickness%
This was normal and i know its not good, i already re-did the contact patch geometry once and it did not improve much, however the meshes still completed.
Then i tried without MRF zones as i read in another forum that it can help - (CFDtest3NoMRFzones) - Successful mesh
I am not sure why this helps or how i can fix it and i am still getting the illegal triangles message in my event log. For this i will start eliminating parts of the car until i find the problem
How do i fix my MRF zones? is it the geometry of them? because i havent changed them at all
Let me tag our PowerUsers and experts @pfernandez and @yosukegb4 here. They have very good knowledge about those kind of issues and they might give you nice tips on how to overcome this problem.
So i am still trying to figure out why i am getting this ‘Illegal triangles’ message. I also do not know how to fix the MRF zone issue.
I have started by trying to remove parts of the car until i can locate the source of the bad gometry.
First test - Pic 1
I first started with the contact patches because of the low thickness percentage. Also the nose and front splitter are reomoved Model is shown below:
Here is the original model geometry. i just used the same model as last year. I HAVE NOT CHANGED THE MODEL This is from before the Simscale update when the MRF zones worked and i could complete the mesh.
Pic 3
I also noticed that after i take out the MRF zones, the meshing of the MRF geometry has changed.
It seems that the small open surface area between the tire and MRF model in pic 3 has tried to close and the open hole once on the right side of the model is also closed. it also has bad mesh quality.
pic 10
Can anyone help with either the Illegal triangle problem or the MRF zone problem? @pfernandez and @yosukegb4 I do not know what problem solving tests to perform that will be most effective. Can either of you suggest some tests i can do to fix my problems? Maybe something i can change in the goemetry? I would really apprecaite any help as the deadline for CFD on our car is fast approaching and soon i will be forced to continue without the MRF zones to have some data for production.
From the initial thread, it seems that your inability to mesh the geometry has been fixed.
Moving on from here, you need to know that tuning a perfect mesh for a geometry that isn’t exactly simple and with an iterative mesher like SnappyHexMesh is very very time consuming.
What I would suggest instead, since you are tight on time is to negate the fact that illegal triangles are present and focus on overall mesh quality. The key parameters, which I have previously highlighted, is the no of illegal cells and your layering thickness.
For your case, the ideal standard would be to refer to the mesh in " Wing1andDiffuser1". That mesh has good layer generation and a low number of illegal cells which means you can directly simulate it straight away and get results.
Try simulating that and using that as your base results first. I believe that mesh I mentioned also has the MRF zone applied. We’ll discuss the results afterwards.
