[CFDleek]

Hi guys,

I don't have much experience with FLUENT yet, so this might be quite easy to most. I'm trying to simulate flow in a stirred tank. The fluid has a free surface at the top but I'm not sure what zero shear boundary condition type to use. Can I use "wall" type and set the shear to 0? Or can I use "symmetry" to get the zero shear effect as well? Are there any other options, and which would be prefered/ what is the exact difference? I tried to look for an explanation online but I can't find anything useful for FLUENT.

Thanks!

[jpina]

Why don't you let the wall be no-slip as usual?

[CeesH]

Because a free-surface between air and water is reasonably approximated as a zero-shear flow, due to the very low viscosity of air. Anyway, from the FLUENT 6 manual (but it still holds):

"Symmetry boundary conditions are used when the physical geometry of interest, and the expected pattern of the flow/thermal solution, have mirror symmetry. They can also be used to model zero-shear slip walls in viscous flows. This section describes the treatment of the flow at symmetry planes and provides examples of the use of symmetry. You do not define any boundary conditions at symmetry boundaries, but you must take care to correctly define your symmetry boundary locations."

And of course, a zero shear wall leads to a zero-shear slip surface as well. From that, I think the conclusion is not that hard for this particular situation. There may be other situations in which the two options lead to more distinct results, of course. (heat transfer or so)

(manual source: https://www.sharcnet.ca/Software/Flu...ug/node257.htm)

[jpina]

I have read some manuals related to VOF (dam break, etc.) and they never carry out such approach. Moreover, since the free-surface will move during the simulation, how could you fix a zero shear rate?

[CeesH]

Thing is, it won't move (in the average sense). For a single phase simulation of a stirred tank, it's a perfectly sensible and common thing to do. The surface is the only location where the fluid is in contact with air, and to set up a multiphase simulation just for the sake of getting some surface dynamics that you are not interested in in the first case is not a very desirable approach, multiphase simulations of stirred tanks take horrendously long to run (and are not very stable in convergence, typically). Knowing that the volume is conserved, the location of the surface is known and stationary. So why not make your life simple, and fix it in this way?

[jpina]

Oh, so you don't want to do a 2-phase VOF analysis... No idea then...

[CeesH]

Nobody ever said VOF, right? it's not even multiphase...
Edit: but even if it were VoF to predict the free surface shape, with no air flowing in and out of the domain setting a no-shear top would be a sensible idea - since the air inside the tank is most likely in direct contact with the surrounding air; it's unlikely there is a lid-top.

[jpina]

You're right, it's only that stirred tank calls for VOF. At least at my current knowledge

[CeesH]

depends very much on what you are interested in; if it's the surface dynamics, or air entertainment via the free surface, it certainly is important to use a multi-phase approach. But if you are interested in bulk mixing, exact assessment of the surface is not relevant (in fact, using a no-slip wall rather than a no shear wall will change the mixing time by less than 1%, so aside from the no-shear boundary condition being physically more realistic, result-wise it doesn't make a difference).

Of course, the multi-phase situation is a good example for a situation where a symmetry b.c. would not technically be a suitable boundary combination, as strictly, when mirroring, the vector of gravity is also mirrored in the symmetry plane. I have no idea what will happen if you would use a symmetry to mimic a no-shear surface in such a case; and in any way, it only applies when gravity is relevant for the problem (and has a component normal to the symmetry plane)

[LuckyTran]

If you want surface dynamics then you must treat it explicitly using a multiphase simulation or a model for the surface behavior. If you don't care about the surface waves and are only interested in bulk behavior then a single-phase simulation with zero shear walls is sufficient. For example, if you want to study how long it takes to empty a tank, you don't need to know much about surface waves.

The symmetry boundary condition is the most restrictive boundary condition because it applies the zero gradient condition to ALL variables. A wall with zero-shear imposes only a zero velocity gradient (but you can still have temperature gradients, species, etc). If you have only fluid flow and no energy equation (and no property changes) then the zero-shear wall and symmetry boundary condition are equivalent. But if you have other stuff in your simulation then you can end up with slightly different results because of the extra conditions imposed by the symmetry condition. However, in both cases, you should have the same zero-shear imposed. The differences are caused by velocity coupling with other properties.

[Kevin_Liu]

I tried zero shear condition with wall type. While I found some issue when turbulence model is used. Take k-omega as an example. The turbulence viscosity is zero at near wall region (laminar area). That also includes the free surface even zero shear condition is used. That means the turbulence simulation might be not wrong to some extent. If the flow you are simulated has strong turbulence at free surface, the setting might not be appropriate.