# pimpleFoam large time step solver?!!

 Register Blogs Members List Search Today's Posts Mark Forums Read

 February 25, 2019, 16:11 pimpleFoam large time step solver?!! #1 Member   Os Join Date: Jun 2017 Posts: 80 Rep Power: 8 Hello guys, i ve been looking around and i found that pimpleFoam is a Large time-step solver . Can anyone please explain to me What do we mean by large time step ? and what is the minimum time step that we can use with pimpleFoam ? and does this mean that we cannot use it for a simulation with a CFL <1 ? Clément_G and kac24 like this.

 February 26, 2019, 03:08 #2 Senior Member   Robert Join Date: May 2015 Location: Bremen, GER Posts: 292 Rep Power: 11 pimpleFoam can work even if CFL > 1. I found that it works well until CFL = 4. Tobias Holzmann describes the usage of pimpleFoam in his Book Mathematics, Numerics, Derivations and OpenFOAM®. According to his record pimpleFoam can work until CFL ~ 20 before it starts producing unrealistic results. Clément_G and kac24 like this. __________________ If you liked my answer to your question, please consider leaving a "Like" in return

February 26, 2019, 16:08
#3
Member

Os
Join Date: Jun 2017
Posts: 80
Rep Power: 8
Quote:
 Originally Posted by RobertHB pimpleFoam can work even if CFL > 1. I found that it works well until CFL = 4. Tobias Holzmann describes the usage of pimpleFoam in his Book Mathematics, Numerics, Derivations and OpenFOAM®. According to his record pimpleFoam can work until CFL ~ 20 before it starts producing unrealistic results.

Thank you for the Answer ,
does this mean even if my courant number is > 1 the results am getting are accurate?

2/-What happens is i use vert small timesteps (adaptive timesteps) that may even get to 1e-5 .Will this affect the final results or no?

3/-Also the last thing is , I ve been trying to make this 2D airfoil simulation for the last month and i keep getting values of Cd very far from the experimental Data , I ve tried to change mesh with multiple wall distance and refinement ,the turbulence model, the solver and i even used fluent to see if there is a problem with my OpenFoam settings But i still get the same result:

Cd_simu =0.01342 While Cd_experimental=0.00883
Cl_simu=0.3424 while Cl_experimental=0.2954

At some point while i was desperately trying to find the problem and find Where the issue is . I used a case and solved that case with pimpleFoam and i made maxCo =0.9 in the system/controDict file

Than i used the same case to run the simulation once again with maxCo=0.2 surprisingly i found that the final results for Cd were very different because for the last case i Got Cd=0.050 and for the first one i got Cd=0.0135

Do you have any idea About what is the problem for this situation ?

February 27, 2019, 02:58
#4
Senior Member

Robert
Join Date: May 2015
Location: Bremen, GER
Posts: 292
Rep Power: 11
Quote:
 Originally Posted by losiola Thank you for the Answer , does this mean even if my courant number is > 1 the results am getting are accurate?
As far as i understand, yes. And it worked for my for laminar and turbelent flows in the past.

Quote:
 What happens is i use vert small timesteps (adaptive timesteps) that may even get to 1e-5 .Will this affect the final results or no?
The results should be the same. Using pimpleFoam should circumvent the need for a very small timestep.

Quote:
 Also the last thing is , I ve been trying to make this 2D airfoil simulation [...]
I have no experience with airfoil simulations.
__________________

March 19, 2019, 09:10
#5
Member

mahdi
Join Date: Nov 2015
Location: Paris, France
Posts: 32
Rep Power: 10
Hello

Quote:
 Originally Posted by losiola does this mean even if my courant number is > 1 the results am getting are accurate?
It depends on the variable you are looking for. In your case when looking for Cd or Cl , it is better to use CFL<1.

Quote:
 Originally Posted by losiola What happens is i use vert small timesteps (adaptive timesteps) that may even get to 1e-5 .Will this affect the final results or no?
Well, It is not just about time step, your boundary layer mesh could be fine enough (Y+<1) to have better results

Quote:
 Originally Posted by losiola 3/-Also the last thing is , I ve been trying to make this 2D airfoil simulation for the last month and i keep getting values of Cd very far from the experimental Data , I ve tried to change mesh with multiple wall distance and refinement ,the turbulence model, the solver and i even used fluent to see if there is a problem with my OpenFoam settings But i still get the same result: Cd_simu =0.01342 While Cd_experimental=0.00883 Cl_simu=0.3424 while Cl_experimental=0.2954 At some point while i was desperately trying to find the problem and find Where the issue is . I used a case and solved that case with pimpleFoam and i made maxCo =0.9 in the system/controDict file Than i used the same case to run the simulation once again with maxCo=0.2 surprisingly i found that the final results for Cd were very different because for the last case i Got Cd=0.050 and for the first one i got Cd=0.0135
For this kind of simulations, it is better to use K-w SST as the turbulence model and very fine mesh close to the airfoil wall without using any wall functions. keep your CFL around 0.5 and let pimple solver select the relative time step. Also it is better to use nOuterCorrectors = 1 in fvSolution.