CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   implicit calculation (http://www.cfd-online.com/Forums/cd-adapco/54856-implicit-calculation.html)

PETE September 9, 2005 06:24

implicit calculation
 
I am wondering why Courant Number is a crriterion in an implicit transient calculation? Is Star really full implicit?

Tushar September 14, 2005 10:58

Re: implicit calculation
 
Hi

I think STAR-CD is not fully implicit since it uses SemiImplicit algorithm (SIMPLE,PISO) so at one stroke matrix is not inverted infact the difference equations are solved in transient or psuedo transin using iterative approach marhing in time, Hence cournat number to restrict the time step is required in place of underrelaxation parameters.

Regards

4xF September 15, 2005 13:27

Re: implicit calculation
 
STAR has an implicit time-integration scheme (at least with the standard procedure). Therefore you can allow yourself to have bigger time steps than with explicit time-marching procedures and have thus a Courant number > 1. Using a courant number is a good check on your time step size, since max. values over 100-150 at some cells can become critical and lead to divergence. You need also to have the Courant number < 1 when you are running free-surface flows with the VOF algorithm.

Hubert Janocha September 21, 2005 10:13

Re: implicit calculation
 
please imagine you have a flow which is so fast that during 1 timesteps a particle passes many (!!!) cells. So the code don't see this particle in some cells.

It is true that for a fully implicit scheme the solution progress itself is abolutly stable AND gives a correct solution for the velocities, but the code can crash because of some other reasons (numerical und physical), or the code is smearing the real solution.

So it is important to take a look to the coorant number even in a fully implicit scheme.


All times are GMT -4. The time now is 17:38.