Why do Fluent 2015 and Fluent 2020 give the big differences of results?

- November 15, 2020 at 6:08 am
  
  SEANGHAI
  Subscriber
  
  Hello! I do not understand why the results provided by running Fluent 2015 and Fluent 2020 are too much different! The results are different around 5% to 20%, it is the big differences. If they are different under 1%, that is acceptable, but this is too much different, 20%. I need your advice!nUpdated! Now I found the differences, In Fluent 2020, it automatically chooses the Pseudo Transient Method to get the quicker results convergence faster that couple solver, but I am still curious why the values or results of Pseudo solver is bigger than couple solver around 20%? Thank You!n
- November 16, 2020 at 11:53 am
  
  Rob
  Forum Moderator
  
  Are both results fully converged? R15 is fairly old, and significant changes were made in the solvers, you'll also see R15 was (probably) run using SIMPLE as opposed to PBCS in 2020. I don't expect such large differences (and we run tests to confirm this) in well posed problems, but if one or other didn't converge well you would expect some change in the result. n
- November 16, 2020 at 12:33 pm
  
  SEANGHAI
  Subscriber
  
  Thank you for your reply! And Yes! Both are converged! 2020 is converged at 124th iteration and 2015 converged at 420th iteration. Bothe are being run by the couple solver. But for 2020, it is automatically chooses Pseudo Transient Method, but 2015 does not, so I think this is why the results are different. n
- November 16, 2020 at 2:33 pm
  
  Rob
  Forum Moderator
  
  What Courant Number & time factor did you use? The iteration count makes sense as the whole point of the change was to speed up the calculations. n
- November 17, 2020 at 1:09 am
  
  SEANGHAI
  Subscriber
  
  Oh! For Courant Number and time factor, I just use the default provided by Fluent, I saw the default is: Courant number=200 and Time scale factor is 1. I never change about these two things. How are they important for the results? I always appreciate all your advices! Thank you!n
- November 17, 2020 at 11:29 am
  
  SEANGHAI
  Subscriber
  
  And one more thing, In my problem, I think I do not need to choose Pseudo-Transient to define Under-Relaxation Factor because the solution without choosing Pseudo Transient is successfully converged. Is the result without choosing Pseudo Transient better than result of choosing Pseudo Transient if both are converged? I really want to know about this answer. Thank You! I am always waiting for your kind reply!n
- November 17, 2020 at 12:33 pm
  
  Amine Ben Hadj Ali
  Ansys Employee
  
  Create Monitor Points of key and relevant quantities and judge convergence behavior by those as well as Flux imbalances!n