CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   SIMPLE algorithm not converging. (http://www.cfd-online.com/Forums/main/79071-simple-algorithm-not-converging.html)

sgp August 11, 2010 12:36

SIMPLE algorithm not converging.
 
hi all,
I am new to CFD and i have started to code incompressible flow between two plates (given in anderson). The thing is, the code i have developed is not converging. the p' computed as given in anderson book, from the pressure poisson equation doesnot decrease. it keeps on increasing.

The boundary conditions i have given are

inlet: u=1, v=0, p'=0
exit: u(exit)=u(exit-1), p'=0// column before exit
top: u=0,v=0,p'=0
bottom: u=0,v=0,p'=0

Also, i dont know whether the way i am implementing staggering is correct or not.

i have
npx+2,npy+1 grid points for v-component
npx+1,npy grid points for u-component
npx,npy grid points for pressure.

i am computing u momentum at 1 to npx-1 and from 2 to npy-1
and v momentum at 1 to npx and from 2 to npy-1.

what could be the reason for p' not comverging.

thanks in advance,
sgp.

falopsy August 12, 2010 03:21

Quote:

Originally Posted by sgp (Post 271171)
hi all,
I am new to CFD and i have started to code incompressible flow between two plates (given in anderson). The thing is, the code i have developed is not converging. the p' computed as given in anderson book, from the pressure poisson equation doesnot decrease. it keeps on increasing.

The boundary conditions i have given are

inlet: u=1, v=0, p'=0
exit: u(exit)=u(exit-1), p'=0// column before exit
top: u=0,v=0,p'=0
bottom: u=0,v=0,p'=0

Also, i dont know whether the way i am implementing staggering is correct or not.

i have
npx+2,npy+1 grid points for v-component
npx+1,npy grid points for u-component
npx,npy grid points for pressure.

i am computing u momentum at 1 to npx-1 and from 2 to npy-1
and v momentum at 1 to npx and from 2 to npy-1.

what could be the reason for p' not comverging.

thanks in advance,
sgp.

I THINK (please note the word) your staggered grid is a bit wrong except you are using a different staggered grid than I know. Check the book by Versteeg and Malalasekera for more details. To my knowledge, using the most common and favoured partially staggered grid, you should have:
npx, npy+1 for v-components
npx+1, npy for u-components
npx, npy for the pressure.

You should then compute u-momentum from 1 to npx-1 ( I assume your starting index is 0) and v-momentum from 1 to npy-1 because the nodes 0 and npx/npy will coincide with points where the velocities are known. If you are using one of the other less common staggered grids, then maybe what you have now is correct. One thing that could cause convergence is your treatment of relaxation factor in the pressure correction equation. Are you incorporating it at all?

sgp August 12, 2010 12:44

under relaxation
 
@falopsy
thanks for the reply.
I am using under-relaxation factor for the pressure term as i read in Patankar's book that under-relaxation factor increases the stability of the code.
Is it right? I also want to know if there is any sample code for SIMPLE algorithm by patankar available in internet.

Thanks in advance,
regards,
sgp.

falopsy August 12, 2010 13:30

Quote:

Originally Posted by sgp (Post 271325)
@falopsy
thanks for the reply.
I am using under-relaxation factor for the pressure term as i read in Patankar's book that under-relaxation factor increases the stability of the code.
Is it right? I also want to know if there is any sample code for SIMPLE algorithm by patankar available in internet.

Thanks in advance,
regards,
sgp.

Yes, under-relaxation of the pressure equation improves stability but my question is how u are treating the velocity under-relaxation in the pressure equation or is it just neglected. You can download the code by Ferziger and peric from
"ftp://ftp.springer.de/pub/technik/peric/" - you can type the ftp directly in your browser. I think the code is based on collocated grid though but should still help somehow.

sgp August 12, 2010 21:06

I think the velocity correction in pressure correction is just neglected.


All times are GMT -4. The time now is 03:11.