CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CD-adapco (http://www.cfd-online.com/Forums/cd-adapco/)
-   -   Discretization scheme: whih one is good in my case (http://www.cfd-online.com/Forums/cd-adapco/53333-discretization-scheme-whih-one-good-my-case.html)

Julie Polyakh June 3, 2003 14:53

Discretization scheme: whih one is good in my case
 
Hi,

The case I work on includes droplets injection and evaporation, gas combustion, conjugate heat transfer and so on. Results obtained with UD scheme do not completely agree with experimental data (calculated maximum velocity is much greater). I tried to set MARS scheme for all variables, but now the conjugate heat transfer residual doesn't converge.

Looking forward for any help.

Julie

Ossi June 4, 2003 02:37

Re: Discretization scheme: whih one is good in my
 
You should always use second order accurate schemes for at least velocity (momemtum). Otherwise your soution is basically rubbish. Of cource, second order accuracy for other variables too, if possible. Sometimes the convergence may be so poor that UD or some blended scheme is needed for e.g. turbulence. On the other hand, under-relaxation normally helps. In case you have steady state case, you may want to also try with the SMPISO pressure correction algorithm.

michael June 5, 2003 11:41

Re: Discretization scheme: whih one is good in my
 
MARS is a good choice. You may leave UD on k-epsilon. Try double precision to get converged solution for conjugate heat transfer residual.

skipio June 24, 2003 20:27

Re: Discretization scheme: whih one is good in my
 
Arguably if one has convergence problems one can start with ud on all variables and cd on density. Then after convergence is achieved restart with MARS on all variables. And of course probably use much lower under relazation than the default ones.


All times are GMT -4. The time now is 12:12.