SimScale CAE Forum

Improving Mesh for Project


#1

Project : https://www.simscale.com/projects/hmattoo/external_flow/
Mesh to investigate: Ln1-Rec

Hi
If you were to see the above mesh, its a car with vortex generators on top. For some reason after meshing the vortex generators do not get meshed properly and they loose their shape.

Please assist as this is one of my validation methods for my thesis.
Looking forward to hearing from you.


#2

Hi @hmattoo - nice project you are working on!

I would increase the refinement of the vortex generators or create a small domain (box) with a good refinement. For that I suggest using the parametric mesh option and not the automatic wind-tunnel/external flow mesh to have more flexibility in the pre-processing.

Try that out and let us know how it went!

Cheers!

Jousef


#3

Hi @jousefm

Thank you for your response but Im afraid I need more help.

When you say domain box - do you mean the cartesian box?

And yes I did use parametric meshing for more control as the automatic was worse than the current parametric mesh I made.


#4

Hi @hmattoo!

You can add a geometry primitive called “Cartesian Box” in the Mesh Creator tab. You could also add a refinement for your wake region - an example can be found here: Full Car Aerodynamics. Hope that helps!

Best,

Jousef


#5

Hi @jousefm

I have tried multiple combinations based on the sample project above however I still fail on meshing the Vortex generators on my car correctly. The only thing I did not use was feature refinement.

Please assist.


#6

Hi @hmattoo, A surface refinement on the car would benefit you here, try on in the region of level 6 to 8 in range. This should bring out more detail where required, it might also get rid of the weird internal meshing you are getting, Let me know how it goes.

Kind regards,
Darren


#7

Hi @1318980
Thankyou!

I know you just saw the project.

Sorry that I deleted the result. Re meshing it now with the new settings.
I would be grateful if you could have a look at it in 25 mins or later, as it should be done by then.

I won’t change the settings on Ln1-Rec


#8

Hi @hmattoo, so this is certainly going in the right direction. Firstly there is a double layer geometry used, it would be best if only the outer layer was present.

Secondly, the refinement for the vortex generators are close, maybe one more refinement, however, you might need region refinements on each of them to ensure that the vortex is actually resolved.

Good luck, let me know how it goes.
Darren


#9

Hi @1318980

I will look into the geometry issue.

I increased the region refinement around that region to a higher number.
Did you mean I have to make Cartesian boxes for each pair of vortex generator?

Once again I cannot thank you enough. This whole project helps me validate my experimentation.


#10

Yes I think that would be the most efficient way to get the results you need.

No problem, and good luck :slight_smile:

Darren


#11

Hi @1318980

My mesh has now improved substnially however after running a trial run with the mesh, the simulation is coming back with an error and referring to “mpirun noticed that process rank 2 with PID 152 on node exited on signal 15 (Terminated).”

Any assistance with this?
Thankyou!


#12

The error just says that the run has been given a signal that cancels the parallel run the origin of the issue could be many things, likely instability. with the new mesh there are some bad quality cells. I would firstly reduce the minimum cell size to -1e+30 and re-mesh with another refinement level on the vortex generators. After that run a new sim with non-orthogonal correctors. Failing that reduce relaxation factors.

Good luck,
Darren

Another likely cause would be to run out of memory, this happens if your mesh is too large, I would defiantly say it is plausible to get the mesh size under 20 million and maintain a g0od quality. I noticed you didn’t individually put boxes on the vortex wakes? and reduce the size of the car wake box would also reduce mesh size without affecting results. Lets try get the mesh to a more realistic size, The largest mesh I have ever run was around 30 million and the geometry was extremely complex.


#13

Hi @1318980

I did manage to fix the double car situation and bad mesh.

https://www.simscale.com/workbench/?pid=5138355731769837733#tab_0-0

If I could get your attention to mesh - Rec-STPbase
And then to simulation - SST HM

My mesh seems to be fine so far but the simulation gives out an error and I am not sure why?
Any assistance?


#14

Hi @hmattoo,

You need to set the maximum run time in “Simulation Control” to a higher number, around 20,000 should be fine. The simulation stops after a set time so you need to increase it. That should fix your issues.

Oh and your car isnt touching the floor, do ensure it does in order to have better accuracy.

Cheers!

Regards,
Barry


#15

Hi @Get_Barried

Thank you for the reply.

I did try your advise however I still have errors popping up with less than 1% completed but 50-85 min elapsed.
Could you please assist?

Simulations - SST HM and Trial SST HM , these are the ones I am referring to.


#16

Hi @hmattoo!

I saw in Trial SST HM that your car has no boundary condition (BC). Can you please add a no-slip BC and try rerunning the sim?

Best,

Jousef


#17

Hi @jousefm
I didn’t add the car BC as no slip because when I start the simulation, the error suggests any body that has not been defined will be assumed to be no slip. So it’s faster to do that than select all the 350 faces.

Do you think it is vital?


#18

Hi @hmattoo!

In general I would assign the faces to a boundary condition but that should not make a difference. We would have to see what else might be causing the error.

Best,

Jousef


#19

Hi @jousefm

What else can you suggest to help with having the simulation run through successfully?

Regards
H. Mattoo


#20

Hi @jousefm @Get_Barried

I have conducted several simulation trials run with minor changes to the settings.
Also dropped the mesh density and quality to enable completion of simulation firstly.

However, I have not progressed any further. Any advises that I could implement?
https://www.simscale.com/workbench/?pid=5138355731769837733#tab_0-0

In Simulation have a look at Trial SST-HM.

Looking forward to hearing from you.