Data analysis


#1

Hi,
In this simulation i wanted to analyse pressure on a windshield of a car due to air flow . The pressure values are not uniform. I want to learn average pressure applied on a specified surface. Is that possible?

https://www.simscale.com/projects/gokayguney/me492-15/

Ekran%20Al%C4%B1nt%C4%B1s%C4%B1asdasdasd


#2

Hi @gokayguney!

Make sure the increase the mesh fineness as well as the domain size - the bounding box is still too small to produce proper results.

Fix those things and let us know if you can see improvements in the post-processing!

Best,

Jousef


#3

Hi @jousefm,
Actually i am trying to optimize some parameters about this project and I have to simulate 25 different geometries. It takes too many times. Do you have any suggestion to do it in shorter time?

Sincerely,


#4

Hi @gokayguney!

If the assumption of a 2D/Quasi-2D simulation is valid for your model then you can use 1 cell in z-direction but still have to increase the size in x direction as well as in y direction. That might work. @Get_Barried and @1318980 , any input from your side here?

Cheers!

Jousef


#5

Hi @jousefm
Can you please take a look the link shared below? Thank you so much for your help.

https://www.simscale.com/projects/gokayguney/me492-1_1/


#6

Hi @gokayguney!

Can you explain me how this windshield is going to be used? Just to better grasp the concept of your simulation. On top of that Mesh1 is not good enough for the simultion. Mesh2 & Mesh3 could be used but I would rather use the parametric mesh option than the windtunnel for this type of geometry/project. Would be very thankful if you could give me some more input :slight_smile:

Cheers!

Jousef


#7

Hi @gokayguney,

Yup I agree with what Jousef has mentioned. A little more context would help.

From what you’ve mentioned, if its just simply wind values and the windscreen then you just have to have a large enough bounding box and the inflow perpendicular to the front of the windscreen. A fine enough surface refinement would allow you to see a nice gradient of pressure values. Do post-process offline though.

Perform a mesh convergence study for one windshield, obtain the ideal mesh that balances results vs node count, check and validate the results then upload all the different geometries, duplicate the setup parameters for the mesh you’ve chosen in the mesh convergence study, mesh then all in parallel, simulate in parallel by duplicating the setup in the simulation tab and you have all you relevant results for every other geometry with minimal workload.

If you’re truly talking about automated runs, there is no scripting function on SimScale so you will not be able to truly do automatic runs.

Cheers.

Regards,
Barry