[yzeyue]

Hi, everyone!

I recently use a CFD program to make simulation, but I found that my program is very unstable when I change some parameter, such as the grid. Mesh is the H-type, so there is not negative mesh, and the angle is not small--orthogonality is good.

When I use the CFD software, there is no such problem before, and the solving speed and the convergency is both fast and good.

So, I want to ask whether there is some method to make the CFD program stable. Is there paper to solve such problems?

Thanks!

Garry

[shahrooz.omd]

Are you using proper under relaxation factors?

[yzeyue]

Thanks for telling me this point. When I use the software I know the under relaxation factors, but I have not changed it in my CFD solver program. This solver is an old one, and I am the only person who use it in my team.

I want to ask u where the under relaxation is used in the CFD program; is it in the residual subroutine or the flux update subroutine or any other one?

thanks very much!

[yzeyue]

Originally Posted by shahrooz.omd

Are you using proper under relaxation factors?

Maybe there is no under relaxation factor in my explicit CFD program?????

[shahrooz.omd]

Originally Posted by yzeyue

Maybe there is no under relaxation factor in my explicit CFD program?????

if you have an explicit solver then other than relaxation factors you have to choose small time steps for stability.

under relaxation factors are in a part where coefficients are calculated. after calculating aP look for a line like this aP = aP / urfm or ap = aP / alpham then go change the amount of urf, alpha or whatever it is named in your solver to a lower amount.

[yzeyue]

Originally Posted by shahrooz.omd

if you have an explicit solver then other than relaxation factors you have to choose small time steps for stability.

under relaxation factors are in a part where coefficients are calculated. after calculating aP look for a line like this aP = aP / urfm or ap = aP / alpham then go change the amount of urf, alpha or whatever it is named in your solver to a lower amount.

I think I should try to change the time steps, or others.

Thank you for your reminding!

[shahrooz.omd]

I hope you get it to work this way, good luck and you are welcome.

[tanmayagrawal7]

In explicit schemes, CFL number is usually the criterion for stability. It should be lesser than one. If you can not find it really, then have a finer mesh and/or a smaller time step. That will do.