interFoam - where does the magic happen?
 

Hi all,

currently, I'm trying to understand the interFoam-solver. What I was looking for is, how the phase is transported near the wall. I found a very good explanation here in the forum (http://www.cfd-online.com/Forums/ope...interfoam.html) for the general workflow of the alphaEq.

Maybe I am missing the forest through the trees, but if my boundary Condition is a fixedValue (0 0 0) e.g. my fluxes should be zero parallel to wall. Therefore no phase-advection at the wall. This is nothing new - the "contact-line" movement is often modeled by a local slip condition or any other modification. How, and especailly where, is that handled in OF?

Thanks for any hints!

lindstroem,
Consider what a fixed value BC is doing. In the case of alpha it would be saying that it can't change at the boundary which unless you want to model a hydrophobic wall is probably not the right BC. If I recall correctedly a zeroGradient BC is probably what you would want (unless you have some special reason to want to modifiy the behavior of alpha at the wall). Hope this helps.
Regards,
Dave

Hi Dave,

thanks for your opinion. From my point of view it is not a question of "right" or "wrong" BC. Consider a withdrawing plate out of a box of water. Then one wall would have e.g. fixedValue (0 1 0) for one wall. A no slip BC would mean, that the moving wall carries the water to the top which it does not due to surface tension. So I am wondering where "this" happens..

Greetings

lindstroem,
You are correct that applying a fixed value of alpha to the plate would result in that nonphysical behavior. If I recall correctly interFoam is able to model surface tension effects though I am not familiar with the file setup required for it. A no slip condition is applied only to velocity and not to scalars transported in the flow. Consider if you had temperature as a scalar. There are certainly cases one can envision where a fixed value on a wall is appropriate , but there are many cases where the temperature of the wall must be allowed to change. This isn't in contradiction to the no slip condition of the velocity. Like temperature, alpha is a scalar that changes the density and viscosity of the fluid depending on its value (though obviously temperature rarely changes density by a factor of 800) and is transported with the flow but also affects the flow because of its effect on density and viscosity.
Regards,
Dave

Dave,

thanks again for your thoughts. Your right - alpha is just a scalar value for the phases but it is advected by the convective transport equation. Therein we have div(u*alpha) which would mean the velocity of the fluid "advects the phase".. The velocity at the wall is 0 (if we do not consider the withdrawing plate again, but lets say an advancing water front on a horizontal surface - damBreak e.g.). So if for the movement of the contact-line (between water, air and solid) the fluid velocity is used, the contact-line should not move at the wall (zero wall velocity). Usually one uses the velocity of the first cell-center to move the contact-line..

Am I thinking wrong?! :) Thanks again for the discussion so far!

The contact line issue is a bit of a conundrum since you are correct that if the no slip condition truely held then the contact line cannot move. Clearly we know this is not the case in reality. I do not know off hand know the physical justification for the no slip not properly accounting for the contact line movement. In a numerical case the issue becomes less pronounced since the value of alpha a cell takes on is that of the cell center which has a small but non-zero velocity. This tends to diffuse away the air at the wall. Unfortunately if you are attempting to resolve down to say y+=1 there is a tendenacy for the no slip condition to lead to what for lack of better terms is known as numerical ventilation. Effectively you are stuck with a single cell thick of low fraction mixture that plays havoc with skin friction results. I have not generally found it to be a problem with inerFoam but with some commercial codes when using large time steps. My interest is only in capturing large scale behavior of multiphase flows (ship scale), the small scale phenomna that this deals with aren't my forte I'm afraid.
Regards,
Dave

Dave,
thanks again for your time. I get your point. Actually there is no special treatment about the contactline. It is just "another value" of the cell-center, no matter if it is close to the wall or not.

As you said in large-scale computations this is not too important but im working on droplet and capillary problems, therefore i am interested :)

I already made some testcases where this phenomena can clearly be seen (with the withdrawing plate). I will spend some more work in this field to get into the details.

Thanks again!
Lindstroem

Hi guys I am wondering anyone can help me understand this line in the transport properties following line, where i dont understand what acap is and how can i find the boundaryField pactch?


fvsPatchVectorField& nHatp = nHatb[patchi];
const scalarField theta
(
convertToRad*acap.theta(U_.boundaryField()[patchi], nHatp)
);

Same Problem
 

Hi All,
I understand that this issue is old but I face same thing. for some unknown reason, the contact line moves even using (0 0 0) at the wall... have you solved this? any tips?

Were you able to find a way to pin the contact line?