CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Info on local time stepping (http://www.cfd-online.com/Forums/main/242-info-local-time-stepping.html)

Frank Muldoon October 8, 1998 20:27

Info on local time stepping
 
Can anyone point me to some information on local time stepping for solving the steady state compressible N-S equations

Heinz Wilkening October 9, 1998 07:44

Re: Info on local time stepping
 
Ciao Frank,

from my experience when I made my diploma thesis, local time stepping means, that there is no common timestep in the hole computational domain, but each nodepoint has its own timestep, which is the maximum possible. eg. cfl = 1. depending on the solver used. But still you perform one advancestep for all nodes. You do not make for some nodes two or more timesteps when you do only one for others. Something like this might be usefull in a transient timeaccurate simulation.

As in the stady state du/dt gets zero, there isn't any error any more, because the solution gets independent on dt.

When I used local time steeping to the staty state solution I found it most simple (you save even to search for the minium timestep in the computational domain), and it was the most effective way to improve convergency to the stady state solution.

Ciao Heinz


All times are GMT -4. The time now is 15:33.