Hi all,

I'm working on an incompressible solver. The pressure and velocity corrections that I use leads to a correct mass conservation.

But when I apply momentum equations with low viscosities, it seems that some unstable chaotic phenomena appears. There is first backflow, and then it make the process diverge catastrophically, even when using low under-relaxation factors for the momentum equations.

Did someone observe such phenomenas ? Do you have some tips to stabilize it ?

Kindest regards

Lionel

Are you using collocated grid?.What kind of numerical discretization do you employ for your computations?.Since you mention it occurs at low viscosities,I suspect that it might be an issue of stability of the scheme.Another possiblity might be the implementation of boundary conditions.You can add a fourth order dissipation term to stabilize your computations.But make sure that you do not have any other problems before doing it.Because you are introducing artificiality to the problem by adding the dissipation term and it should be used with care.The CFD book of hoffman and chiang has the details I think.

Hi Lionel,

Try reducing your element size in steps as far as your computer memory will allow & see what happens.

What velocity, geometry & viscosity (problem definition) are you using? Are you using FVM or FEM? Are you using a steady, or transient solver?

mw

Hi, thank you for answering.

I'm working by FVM and steady model.

But in fact, I think it's an error in the SIMPLE algorithm. It's said that I have to express velocities at cell faces in function of pressure field to inject in continuity equations.

The problem is that as I store all the values at cells centers, in steady flow I only can interpolate values at faces. I was using a modified algorithm to express these values at faces in function of pressures, but now I think it's wrong.

By the way, do you think that SIMPLER is better than SIMPLE ?

Lionel

hi Lionel, I suggest you to refer CFD by Mallalasekara & Versteeg.. In that text book, all the details: how to interpolate the values, implementing the boundary conditions and different schemes like SIMPLE,SIMPLER, SIMPLEC and PISO are explained very clearly..

Hi, when you say on low viscosity, how much it is low can you tell me what is your viscosity (Reynolds number). What is your test case and boundary conditions?

[MDPrasad]

Hai Mr. Lionel,
My name is Durga Prasad. I am also working on a incompressible solver based on SIMPLE algorithm implemented on a collocated scheme. I am using SIMPLE in conjunction with unstructured grid. I follow the interpolation scheme suggested by Rhie& chow, tailored for the unstructured grids (given by Dr. Jayathy Murthy).
I am facing the same problem you had, wherein unphysical oscillations sets in and divergence occur, when i am trying to solve problem for a low viscosity, high velocity combination.
How did you overcome the problem of this oscillations at low viscosity? I am using the FVM method. If you could share the corrective action, that will be a great help, as i am trying to solve this problem for almost a month.

Thanks in advance for the help.

[leflix]

Originally Posted by Lionel S.
;50610

Hi, thank you for answering.

I'm working by FVM and steady model.

But in fact, I think it's an error in the SIMPLE algorithm. It's said that I have to express velocities at cell faces in function of pressure field to inject in continuity equations.

The problem is that as I store all the values at cells centers, in steady flow I only can interpolate values at faces. I was using a modified algorithm to express these values at faces in function of pressures, but now I think it's wrong.

By the way, do you think that SIMPLER is better than SIMPLE ?

Lionel

I think it comes from a bad implementation of the Rhie and Chow interpolation as you seem to use colocated grid.