
[Sponsors] 
February 23, 2010, 10:42 
Is it possible to use PISO algorithm for steady case

#1 
Member
Jérémy Bulle
Join Date: Nov 2009
Posts: 93
Rep Power: 9 
Hello,
In fluent, when you are in a steady case you can choose SIMPLE or PISO for the solver. In openfoam, apparently, SIMPLE is for steady and PISO for transient. Is there a way to choose PISO even for steady state? Thanks for your answers 

February 24, 2010, 03:34 

#2 
Senior Member
matej forman
Join Date: Mar 2009
Location: Brno, Czech Republic
Posts: 117
Rep Power: 10 
well, steady state  strictly speaking is an average over time, or a transient with infinite time step. So this is also a way how to use PISO for steady state.
What is you motivation for using PISO for steadystate? matej 

February 24, 2010, 04:43 

#3 
Member
Jérémy Bulle
Join Date: Nov 2009
Posts: 93
Rep Power: 9 
Hello
Thanks for the answer My motivation was that with Fluent, PISO solver gave me better results than the SIMPLE solver so I wanted to try this in openFoam 

February 25, 2010, 05:39 

#4 
Member
Ulf Bunge
Join Date: Mar 2009
Location: Wolfsburg, Germany
Posts: 34
Rep Power: 10 
Hello,
in case the solution you find is steady, the temporal derivation should vanish and you could also deactivate this term in the code or use steadyState for ddtSchemes. However, alternatively and dependent on how good your initialization is, you might also want to compute the case as unsteady with a given ddtSchemes (other than steadyState). Dependent on the size of the time step you chose, you can interpret this as relaxation as you enhance diagonal dominance of your coefficient matrix. I.e., the smaller Dt, the larger the cellcentered coefficient becomes in comparison to the other coefficients. The procedure becomes stiffer but it might help in not so wellposed problems with a bad initialization. Best regards, Ulf. 

July 7, 2010, 04:03 

#5 
Member
David
Join Date: Dec 2009
Location: Spain
Posts: 61
Rep Power: 9 
Hi Ulf
Could you explain how to employ the PISO algorithm for a steadystate case so that it works as a relaxation? I am working with a SIMPLE algorithm without relaxation for the temperature and I am having problems due to that. So I have thought about use the PISO algorithm in the way you have suggested, but I don´t know exactly how to do it. My case problem is steady. Thank you very much Regards David 

July 7, 2010, 05:07 

#6 
Senior Member
matej forman
Join Date: Mar 2009
Location: Brno, Czech Republic
Posts: 117
Rep Power: 10 
David,
why don't you employ some underrelaxation for temperature? It seems to be much more simple solution than using PISO. The easiest way to use PISO for steady computation is to run transient solution, during the computation you may try lengthen the timestep. Once solution stops changing you have steady state solution. good luck 

July 7, 2010, 06:43 

#7 
Member
David
Join Date: Dec 2009
Location: Spain
Posts: 61
Rep Power: 9 
Thanks for your help
Yes I know that the best option is to apply a underelaxation for the temperature, but in the solver that I am employing (conjugateHeatFoam, in the 1.5dev version) the underelaxation is not defined and I am not able to do it (I don´t have programming skills). Thank you again, regards David 

July 14, 2010, 06:59 
Under relaxation factor

#8 
Senior Member
ata kamyabi
Join Date: Aug 2009
Location: Kerman
Posts: 323
Rep Power: 10 
Hi David By changing the delta t you are changing under relaxation factor for a steady state case from one time step to an other not in solving equations in a single time step. Do you understand me? Good luck


July 14, 2010, 09:40 

#9 
Member
David
Join Date: Dec 2009
Location: Spain
Posts: 61
Rep Power: 9 
Hi Ata
Thank you for your help. Well, I don't understand it very well. If I am solving a steady state case, how I can modify the time step? or, if the solver is transient, how I can specify that the case is steady? Do you mean that I can not employ the delta t of a transient solver as a underelaxation for a steady case? Thank you very much again. David 

July 15, 2010, 00:49 
Under relaxation

#10 
Senior Member
ata kamyabi
Join Date: Aug 2009
Location: Kerman
Posts: 323
Rep Power: 10 
Hi David
Consider you have a fully explicit scheme for a unsteady case. In this case you can use delta t as a under relaxation factor ti achieve steady state solution but if you have implicit terms in your scheme you need an iterative method for solving system of equations in each single time step in its relation is separate from delta t. So, you can use a full explicit scheme for solving a steady state problem but you must have enough time to steady state solution after transient solution achieve. I hope my explanation is clear for you. Best regards Good luck 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Doubt in unsteady case  vijay  FLUENT  1  April 24, 2006 11:11 
New case  Constantinos Zervides (Zervides)  OpenFOAM PreProcessing  5  October 24, 2005 03:34 
How to save a case running in background  us  FLUENT  0  July 6, 2005 10:43 
Whatbs best way to create new case  jvn  OpenFOAM PreProcessing  2  June 20, 2005 21:19 
SIMPLE algorithm  Jonathan Castro  Main CFD Forum  3  December 10, 1999 05:59 