Backflow problem in LES

[Muhsin]

Hi

I am working using LES and having an issue with my upward mixed convection with variable properties case. My calculations keep failing where I have backward flow from the outlet into my calculation domain. I've tried both isolib and ifrent as my outlet but both still giving me this problem. For the ifrent, i specified fix pressure with zero-flux for my velocities. Is there any suggestion on resolving this problem that I have?

[Yvan Fournier]

Hello,

What type of mesh are you using ?

Regards,

Yvan

[Muhsin]

Hi Yvan,

My mesh is fully structured, and it was running fine for my forced convection case.
The problem only started when I switched on my gravity.

I also checked the possibility of using convective outlet (icodcl = 2) for my boundary condition. Would it help of my unsteady mixed convection problem?
For this, how I'm going to specify the flux and CFL number to both rcodcl(ifac, ivar 1) & rcodcl(ifac, ivar 2). Because I really have no idea.

Regards,
Muhsin

[Luciano Garelli]

Hello Mushin,

Could you solve your problem? If not, could you upload a small problem where the case fails.

Regards,

Luciano

[Muhsin]

Hello Mushin,

Could you solve your problem? If not, could you upload a small problem where the case fails.

Regards,

Luciano

Hi Luciano,

Thank you for asking. No, I still haven't solved my problem. I am right now planning to impose my outlet with convective outlet. But still not sure if what I'm doing is right because i still having a backflow.

[Luciano Garelli]

Hello Mushin,

I was looking at your listing file. At start of your simulation you have an important density difference (481.81 MIN, 589.92 MAX) so, this could lead an important bouyancy (body) force. Did you try to reduce the value of the imposed gravity in order to reduce this forces? Because, as you mention the problem start when you activated
the gravity.

Also, in the cs_user_physical_properties you have a very "complicated" distribution of the physical properties of the fluid. Is this distribution of the physical properties spatially continuous?

Regards,

Luciano

[Muhsin]

Hello Luciano,

I tried to reduce my gravity to 2m.s-2. And still give me diverge calculations. I try to impose extremely low gravity at the moment. But I wondering if this will be the case when I increase the gravity manually later on.

Also, in the cs_user_physical_properties you have a very "complicated" distribution of the physical properties of the fluid. Is this distribution of the physical properties spatially continuous?

The physical properties are spatially continuous but there will have an extreme variation of the properties value about the critical point, which is why I have very "complicated" distribution.

I still thinking that my outlet BC maybe not implemented right. And try to get my convective outlet as it is spatially developing case.

Just an additional basic question. May I ask, if I use the standard outflow, how we could get negative mass flow. Please correct me if I'm wrong, is it from the pressure correction?

[Luciano Garelli]

Hello Mushin,

I don't understand your question..but in your listing file you have a negative mass flow at inlet and a positive mass flow at outlet. The sign of the mass flow at boundaries depends on the sign of .

When you activate the gravity, Are you also impousing a velocity at inlet?

Regards,

Luciano.

[Muhsin]

Hi Luciano,

My mass flow at the inlet is negative because of my streamwise domain is in a -z direction(sorry about this confusion).
And when I activate my gravity, the velocity at the inlet is still imposed as it is a mixed convection case.

Sorry for the bad question. I rephrase my basic question.
what causes the opposite flow in the calculation at the outlet? Is it from the pressure correction?

Regards,
Muhsin

[Luciano Garelli]

Muhsin,

In sections 5.2 and 5.4.4 of the theory guide of CS (http://code-saturne.org/cms/sites/defau ... theory.pdf) you can find a more detailed answer to this question.

Outlet BC " Warning: if the outgoing mass-flux is negative, i.e. if the outlet becomes an inlet, then the mass-flux is clipped to zero"

Free inlet/outlet BC : "it corresponds to the standard outlet when the flow is outgoing (see §5.4.4), but
to a free inlet when the flow is ingoing. The Bernoulli relationship is used to derive a boundary condition on the pressure increment to couple velocity and pressure at these free inlet faces. Note that no clipping on the velocity is imposed."

Regards,

Luciano