Discussion Groups

Simulation CFD

Reply
Active Member
Posts: 6
Registered: ‎03-26-2013

Difference in solution when solved on different machine

70 Views, 2 Replies
10-29-2013 06:20 AM

Hello,

     I am facing one issue. I am using CFD-2013 for natural convection analysis. when i setup and run a model on our Server m/c (hiving higher configuration), i am getting some result (say 183 degC coil temp). But the same model when i setup and run it on my local m/c, the result i am getting is different (161 degC coil temp). There is a wide difference in the results. Boundry conditions are also same with Auto meshing. 

     Has anyone faced a similar problem? Could anyone plz explain the reason behind this and the solution for the same??

 

Thanks,

Mayur H

Please use plain text.
Product Support
apolo.vanderberg
Posts: 263
Registered: ‎08-31-2011

Re: Difference in solution when solved on different machine

10-29-2013 01:21 PM in reply to: mayur.harip

Are you setting the model up from scratch each time or are you using the support sharefile to move between each machine?

 

Are you running the model to convergence each time?

Is the mesh tight enough to capture the physics at hand, a default Automatic size may not be refined enough to capture the details and the two runs might be converging to slightly different solutions based on the mesh being too coarse to capture the gradients.

 

Please use plain text.
Active Member
Posts: 6
Registered: ‎03-26-2013

Re: Difference in solution when solved on different machine

10-29-2013 10:41 PM in reply to: apolo.vanderberg

 

Yes! on both machines i am starting from scratch. Right from importing the proe assembly--assigning material--BC--Automatic mesh generation--solving.

Automatic mesh is creating same no of elements on both the machines. For both the cases the model is converging (auto stop) at different iterations.

 

What is the criteria for a tight mesh? what convergence criteria is suitable?

Please use plain text.