|
[Sponsors] |
February 28, 2006, 11:42 |
local time stepping
|
#1 |
Guest
Posts: n/a
|
Hello,
I would like to ask something about local time stepping and I hope somebody can help me. Would you say that this method is applied to a solver to increase stability - so a steady problem can be treated as "unsteady" with a time step which is dependent on the CFL number? Thanks. |
|
February 28, 2006, 15:03 |
Re: local time stepping
|
#2 |
Guest
Posts: n/a
|
No. Local time stepping is used as a solution acceleration method - it can occasionally lead to robustness problems as the solution is no longer entirely 'physical', except in the limit t -> infinity (i.e. the 'steady' solution).
|
|
March 1, 2006, 00:04 |
Re: local time stepping
|
#3 |
Guest
Posts: n/a
|
you actually lose the temporal accuracy if you use the local time stepping. It is mainly used when we have a grid clustered on one side. So for large cells the CFL does not demand the small time step that a small cell would need. Hence local time stepping is used to accelarate the steady state convergence. It actually means that the different cells move in different time regimes if at all it makes any sense !!!. No physics here. A considerable lot of people are not comfortable with the local time stepping due to this aspect. You can use Dual time stepping to regain the temporal accuracy and use local time stepping for unsteady problems.
|
|
March 1, 2006, 02:06 |
Re: local time stepping
|
#4 |
Guest
Posts: n/a
|
Sorry, but now I am completely mixed up.
Could you just try to give me another short explanation (or a free reference)? |
|
March 3, 2006, 04:26 |
Re: local time stepping
|
#5 |
Guest
Posts: n/a
|
Hello..
Whatever the informations which we are applying interms boundary conditions(Physics of the problem)... cant move more than one grid space per unit time. So we have to give optimum timestep sothat the information shouldnt jump more than one grid space per unit time... This is the funda of CFL number... Next Local and Global timestep.. Local timestep:: In each cell the available information moves with local velocity.. so Local timestep= (Size of the cell/ Local speed) Global timestep is calculated based on max speed of information otherwise min timestep required :: Global timestep= Min. all( size of cell/ corres.infor speed) I hope it will help Regards Mohammed |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
alphaEqn.H in twoPhaseEulerFoam | cheng1988sjtu | OpenFOAM Bugs | 15 | May 1, 2016 16:12 |
Dynamic moving mesh | Pei-Ying Hsieh (Hsieh) | OpenFOAM Running, Solving & CFD | 64 | June 7, 2012 10:04 |
Error while running rhoPisoFoam.. | nileshjrane | OpenFOAM Running, Solving & CFD | 8 | August 26, 2010 12:50 |
[blockMesh] BlockMeshmergePatchPairs | hjasak | OpenFOAM Meshing & Mesh Conversion | 11 | August 15, 2008 07:36 |
Info on local time stepping | Frank Muldoon | Main CFD Forum | 1 | October 9, 1998 07:44 |