[Andreas2014]

Hi all,

I have a 2D channel geometry, whereby one zone is porous and another one is not. Normally I defined the interface between these two zones as interior to get the flow field and that worked fine. However, when I start adding in user defined scalars, I get values of these scalars to be very high in the porous region.

I think this maybe due to improperly defined conditions at the interface, since an interior boundary condition just assumes the scalar flux to be continuous right?

So my problem is that I need a boundary condition whereby I can maintain a continuous velocity and shear stress transition across the interface but also where I can define a particular flux for my user defined scalars.
An interior boundary condition does not let me specify the flux across the interface and shadow walls don't let me specify the velocity.

Could you please advise as to what type of boundary I might be able to use for my simulation?
Thanks,
Andreas

[pakk]

Originally Posted by Andreas2014

Hi all,

I have a 2D channel geometry, whereby one zone is porous and another one is not. Normally I defined the interface between these two zones as interior to get the flow field and that worked fine. However, when I start adding in user defined scalars, I get values of these scalars to be very high in the porous region.

Perhaps the scalars have this high value in reality? If at the inlet/outlet you define a flux for the UDS, and you are calculating steady-state, Fluent will try to find the steady-state solution. If your porous zone is 'blocking' the flow, the only way that the UDS can go through the porous zone with the correct flux is by having a really high value. (Assuming that diffusion in the porous zone is much lower than diffusion outside the porous zone.)

[Andreas2014]

Thank you for your reply,
You are absolutely correct,

It is true that diffusion is very small, so a buildup would be expected, it is just that the values are very, very high. Even in a transient simulation, after a very small time, the resulting scalar takes on very unphysical values, by an order of 9 higher.

I am just concerned with setting the correct boundary condition at this interface. If the scalar flux outside the porous region is set to be the same in the porous region, then there has to be a very high build up. The velocity in the porous region is much lower, so to maintain the same flux, the scalar has to be alot higher.

Therefore, I think that the flux has to be limited by mass transfer to prevent such a high buildup.
The only way I can set a particular flux in this region is by using shadow walls, but these do not allow the flow to pass through into the porous region.

So I am very stuck

[pakk]

I don't think that the problem is at your interface condition, but at your real boundary conditions.

[Andreas2014]

My boundary conditions are 0.1 m/s at the inlet, and 0 Pa at the outlet. The walls are no slip. The scalar inlet is set as 1 and the outlet specifies that the scalar flux at the outlet is zero. The diffusion is set to a very small value.

Could you please elaborate which one of the boundary conditions is wrong?

[Andreas2014]

I guess what I am looking for is something equivalent to the leaking wall BC in Comsol

[pakk]

Originally Posted by Andreas2014

My boundary conditions are 0.1 m/s at the inlet, and 0 Pa at the outlet. The walls are no slip. The scalar inlet is set as 1 and the outlet specifies that the scalar flux at the outlet is zero. The diffusion is set to a very small value.

Could you please elaborate which one of the boundary conditions is wrong?

So, at the inlet there is your scalar being transported into the domain (convective transport), and it cannot leave at the outlet (scalar flux is zero). What will happen with your scalar (imagine that it is a transient simulation)? It will keep on flowing into the domain.

You can only have steady state if there is just as much scalar going in the domain as out of the domain. The only way that a scalar can leave your domain is through diffusion at the inlet, and since your diffusion constant is so low, this will only be effective when the scalar value is really high.

[Andreas2014]

Originally Posted by pakk

You can only have steady state if there is just as much scalar going in the domain as out of the domain. The only way that a scalar can leave your domain is through diffusion at the inlet, and since your diffusion constant is so low, this will only be effective when the scalar value is really high.

Hi pakk,

Thank you very much for your help.

From my steady state simulations, the scalar flux in and the scalar flux out are the same so the scalar balance is maintained. So the scalar is leaving and entering the domain and I have a steady value for the scalar.
At higher and higher permeabilities, the scalar value increases.

I am just concerned about the interface, as surely there should be a mass resistance imposed across it?

[pakk]

Originally Posted by Andreas2014

From my steady state simulations, the scalar flux in and the scalar flux out are the same so the scalar balance is maintained. So the scalar is leaving and entering the domain and I have a steady value for the scalar.

Yes, because that is what the simulation tries to do. It will try to find a solution with a steady value. But you keep on pumping scalar into the domain, and the only way for the scalar to leave the domain is through diffusion, which will only be relevant at extremely high scalar a values.

I am just concerned about the interface, as surely there should be a mass resistance imposed across it?

Don't worry about the interface, really. (At least not yet.)

Think of it like you are modeling slowly putting feathers into a car (strange example, I know). Suppose that you put one feather in, every hour. What is the equilibrium? That is a situation in which the car is so stuffed with feathers, that something cracked, and one feather leaves the car every hour through that crack.
What you suggest now is similar to making the car stronger, bigger, and adding compartments inside. This will not change the fact that the equilibrium value will be that there is a crack in the car. Because the car is bigger and stronger, it will mean that more feathers are inside the car.
If you were not expecting to see that many feathers inside the car, reconsider your boundary equations. Are you really putting in one feather every hour? Is there really no naturaly way for the feathers to leave the car (maybe a window is open?)

In your case, you keep putting scalar into the domain, by convective flux, and you don't allow it to leave through the outlet. So, of course you will get an equilibrium value with high scalar values. It does not really matter what happens inside your domain. Reconsider your boundary conditions. Do you really want the scalar flux at the outlet to be zero?

[Andreas2014]

Hi,

Thanks for your help,
I am slowly trying to understand but I still have some queries.

If i was to solve this problem analytically, I would need to specify the boundary conditions at the interface. So my solute flux through this interface is proportional to the concentration difference and the hydraulic conductivity. A flux continuity I do not think would be right surely?

Also my flow is not completely impeded by the plug, I have attached the image. The high values are in the porous region itself and not in the free flow. So I don't see how the diffusion term is allowing the solute to leave via the inlet?

Thanks for your help

[pakk]

Why would the flux not be continuous over the interface? Is there a reason for your scalar to disappear or to appear at the interface?

This could happen if you have chemical reaction there, or something else special going on, but normally the flux is continuous.

Can you show a picture of your current scalar values, and a picture of your current diffusion constants, to see what is going on?

[Andreas2014]

Sure, I have attached an image of the scalars in the porous region, and an image of the scalars in the flow region. Note the different range of values.

Essentially, I mean that the flux on the fluid side of the interface should be

flux_fluid_side = -k(C_fluid - C_wall);

and flux on porous side of interface should be

flux_porous_side = k(C_fluid - C_wall);

So not all the scalar should pass through the interface, there should be some resistance via the hydraulic permeability no?

[pakk]

I don't see very high values in the porous region... Only at a very small spot, it is 75, is that troubling you? I thought you were talking about values that were like a billion times too high!

What makes you think that these values are unphysical? Look at the flow field close to that high spot, is there something special going on, like a recirculation zone?
(If for physical reasons you expect resistance at the interface, them by all means add it, but don't just do it because you want to get the results to agree to your expectations...)

[Andreas2014]

This is at a lower resistance that I normally use. The values go about 1000 times higher and sometimes the whole simulation does reach something a value that is a billion times higher, though i suspect that is when it goes unstable.

For the given conditions, I do think that they are physically correct, but I just do not think the conditions at the interface are right.
How would I be able to specify on Ansys Fluent a boundary whereby I can define both a velocity flux and a scalar flux? Shadow walls allow me to specify the scalar flux but not the velocity flux.

Thanks for your help Pakk,

[pakk]

Was the simulation you showed already converged, or was the scalar value in that extreme point still increasing? It might just be a mesh problem, causing an instability. Is it always at the same point, that these extreme values occur? The scalar diffusion is more sensitive to mesh issues than normal flow calculations.

I can not help you with choosing a suitable interface condition, sorry... I don't know that.

[Andreas2014]

Hi Pakk,

Yes the convergence is reached but the scalar values get very high in every cell. I suspect that it may indeed be a mesh problem but I've had the same problem for different types of meshes.

You don't happen to know of any BC similar to the leaking wall BC found in COMSOL?

thank you very much for your help Pakk,
You have given me plenty to think about.