Author Topic: turbulentInflow mesh constrain  (Read 13247 times)

cdup

  • Newbie
  • *
  • Posts: 2
    • View Profile
turbulentInflow mesh constrain
« on: November 09, 2021, 01:36:23 PM »
First of all, thank you for this useful tool!!
I found your turbulentInflow tools very useful and I'm now trying to use them.

I’m using ANSA (BETA CAE Systems) mesh generator (instead of blockMesh). I’ve reproduced Jay's examples provided in the folder tutorials/inhomogeneousTurbulence/ DFDFSEM /channel395 with a different mesh size. My first goal is to make them run for 10 sec (single proc).

Therefore, I’ve created a fully structured hexa mesh and I ran it ten sec without any issue.
My second test is an unstructured hexa mesh which doesn’t run.
Code: [Select]
Turbulent DFDFSEM patch: inlet seeded 401 eddies with total volume 34.9364
mass flow correction coefficient: 1.00045
#0  Foam::error::printStack(Foam::Ostream&) at ??:?
#1  Foam::sigFpe::sigHandler(int) at ??:?

Then I’ve reduced the size of the meshes and they are now running fine.

Hence, here is my question, is there, according to your knowledge, any constrain for the mesh (size and type) to use your tools ?

By the way, I'm expecting to use it for wind load calculations.

If required, I’ll be pleased to send you my testcases. (fine mesh are quite heavy for the forum, 50Mo). I've attached the log files of the unstructured Hexa mesh.

Thank you in advance for your feed back.

Best regards,

Cedric

stan_lw

  • Newbie
  • *
  • Posts: 18
    • View Profile
Re: turbulentInflow mesh constrain
« Reply #1 on: November 11, 2021, 04:14:03 AM »
Seems like the case with fine mesh diverged due to large time step. The maximum Courant Number should be smaller than 1.0 for stability, but that case has a maximum Courant number of 2.46. You may try reducing your time step by 3 or 4 times.

cdup

  • Newbie
  • *
  • Posts: 2
    • View Profile
Re: turbulentInflow mesh constrain
« Reply #2 on: November 15, 2021, 08:35:12 AM »
Good morning Stan,
You're completely right.
I've reduced by 10 my time step and the calculation is now running.
I didn't expect the time step to be so influent for the calculation ....
Thank you again,
I’m trying to generate an atmospheric boundary layer and check if the turbulent structure survives along the domain.
Best regards,
Cedric