Learning Forum

[Sardar]

Hi

Consider a mixing tank with its top being open for pressure-outlet BC.

The pressure-outlet window requires backflow turbulence specs.

There is neither inlet, nor downstream/upstream in my case, hence the only statement I found a bit related to my case in the User's guide was: "If the turbulence derives its characteristic length from an obstacle in the flow, such as a perforated plate, it is more appropriate to base the turbulence length scale on the characteristic length of the obstacle rather than on the duct size." which requires to enter the length od he baffle(?) as the length scale. but what about turbulent intensity. I do have an estimation of turbulence intensity in the entire tank but not for the tank top.

Or beyond all these, which "method" should I choose for my case and which values?

Thanks

[Rob]

If you're not including the head space it's more normal to use a symmetry plane at the free surface. If necessary you can put a small vent somewhere out of the way on that surface to let a small amount of material in/out to stabilise the model.
Otherwise you need to find the values at the boundary to avoid messing up the turbulence, species, temperature etc fields.

[Sardar]

Together with what sort of BC can I use a tiny vent? I am patching air on top area of the tank. So should/can I change the pressure outlet BC for tank top to wall BC and then put the vent on the tank top wall? Is that what you mean?
Thanks really

[Rob]

If you have the head space you just need to get the back flow conditions about right: if they're different enough to be effecting the free surface you need to rethink the values.
The vent approach is for when we don't use multiphase and assume the free surface is flat. Depending on the model (material properties etc) we sometimes need a pressure outlet somewhere out of the way to balance the volume in the domain.

[Sardar]

I am pretty confused, not making a final answer out of all this post. However, I kind of combined the head space (i.e. air on top part of tank) with symmetry BC and to my surprise, it worked. At least that is how I believe. This is contrary to Rob's first comment, but probably only apparently.
I do not know, maybe I am being a bit pedantic. Any ideas?

[Rob]

OK, let's look at the options for modelling.
Liquid (single phase).
Set the top as symmetry and nothing can enter or leave the domain. Assuming no volume is added to the domain it should run nicely. However, a small (1-2 cells) pressure boundary may be necessary to let the pressure stabilise.
Set the top as a pressure boundary. Flow will leave and enter the domain based on the flow field. If extreme vortexing is seen you may empty the tank.

Head space (VOF multiphase)
Set top as symmetry and nothing can enter or leave the domain. You will need good convergence and may need to monitor liquid volume to ensure you aren't gaining or losing mass.
Set the top as a pressure boundary. Flow will leave and enter the domain based on the flow field. If extreme vortexing is seen you may empty the tank, but typically you'll have air recirculating that'll mess with the residuals but won't do much to the liquid flow.

All approaches are valid, and I've used each for projects over the years. Single phase with a pressure boundary is the least stable, and the result is very dependent on the back flow settings.