# smoothSolver diverges - solution in using PBiCG solver?

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

 December 15, 2009, 14:49 smoothSolver diverges - solution in using PBiCG solver? #1 Senior Member   Dr. Alexander Vakhrushev Join Date: Mar 2009 Posts: 250 Blog Entries: 1 Rep Power: 19 Hi all! I have 2 points to discuss: 1st one: I wrote a heat transfer solver based on pimpleFoam, adding heat transfer equation. I use high Courant numbers (up to 5) to speed up calculations. Flow simulation runs without problems (GAMG solver for pressure, smoothSolvers for U and turbulent parameters). By analogy I used smoothSolver also for TEqn in my solver: Code: ``` T { solver smoothSolver; smoother GaussSeidel; tolerance 1e-05; relTol 0; nSweeps 1; }``` But it diverges just at the beggining of simulation with error in GaussSeidel smoother method. Then I just switched TEqn solver to PBiCG with DILU preconditioner Code: ``` T { solver PBiCG; preconditioner DILU; tolerance 1e-05; relTol 0; }``` and every thing runs Ok. I compared transient solution at different times with pisoFoam solution + heat transfer, and all fields correspond realy good to each other. So what is the reason for such behavior? 2nd point: Why am I using pimpleFoam? I solve more complicated task with solidification and turbulent flow. For some cases I search for steady state. One approache is to use steady state solvers, another - to apply transient or cvazi-transient solvers to reach steady state when solution will become stable. I tried to use simpleFoam just to test such posibility for my task. It reached solution which transient solvers achieve and do not change later, but instanteniously solution deverged! Residuals got a big jump and solution crashed... Am wondering why it happend? Stable solution for that simmetric geometry which I used was assimetric, because there are 3 jets and in 2D they cannot be stable because two zones of low pressure do not have a connection. Is assymethry of solution the case for steady state solver to deverge? So now I an forced to use PISO or PIMPLE technics to solve my problem... But it is time-consuming! So, waiting for your opinion! Have a good day! __________________ Best regards, Dr. Alexander VAKHRUSHEV Christian Doppler Laboratory for "Metallurgical Applications of Magnetohydrodynamics" Simulation and Modelling of Metallurgical Processes Department of Metallurgy University of Leoben http://smmp.unileoben.ac.at

 August 20, 2013, 16:22 #2 New Member   Jackie Chou Join Date: Mar 2013 Posts: 12 Rep Power: 13 Hello, Dr. Alexander VAKHRUSHEV, I have met the same situation as you had. Have you got an idea about it? Jackie

 August 21, 2013, 11:17 #3 Senior Member   Dr. Alexander Vakhrushev Join Date: Mar 2009 Posts: 250 Blog Entries: 1 Rep Power: 19 Hi Jackie! My post was from 2009, since that OpenFOAM improved a lot in numerics. So I do not observe that now in my simulations... Sometimes under-relaxation really helps to avoid simulation from explosion. Cheers, Alexander __________________ Best regards, Dr. Alexander VAKHRUSHEV Christian Doppler Laboratory for "Metallurgical Applications of Magnetohydrodynamics" Simulation and Modelling of Metallurgical Processes Department of Metallurgy University of Leoben http://smmp.unileoben.ac.at

 September 11, 2013, 12:44 Multigride #4 Member   Join Date: Oct 2012 Posts: 47 Rep Power: 13 Hi Dr.Alexander Can you tell me what the difference is GAMG algorithms algorithms AMG? THANKS