Learning Forum

[Dino]

Hi,

I get very good results and highly match with experimental data when my time-step is like 3600 sec or more. The data from the real experiments are collected every 3600 sec as well. But if I apply a smaller time step in Fluent, I get good results but do not match as good as the big time step. Does this mean I have a problem with my simulation since smaller time-step should give better results?

 

Thank you,

[Karthik Remella]

Hello,

Few questions:

• What kind of analysis are you running - multiphase, turbulent flow, etc.?


• What is your minimum grid size?


• What is your Courant number condition?


• Are you ensuring deep convergence in your analysis?


• Are you running a fixed time step or variable time step?


• Are you making sure that your solution is converging every time step and reaching your prescribed residual values?


• What is your residual criteria? Perhaps, you might want to lower your residual criteria further and re-run both your analyses.


• Have you checked your conservation - mass, energy?


• Do you use the same mesh when running a smaller time step?


• Did you use any monitors to make sure you are reaching steady state (irrespective of your residuals)?

I hope these questions get you thinking about your analysis and help you move in the right direction.

Thank you.

Best Regards,

Karthik

[konangsh]

Ideally your results should be time-step independent, just like grid independence.  But as Karthik pointed out, you need to provide some details on the type of problem that you are solving, discretization schemes used, etc., for us to provide any useful insights.