[Isaac]

Hello,

I am trying to get a hang of OF and have been trying to set up some simple simulations. I have a simple case for a tank with a hole at the bottom draining water purely due to gravity.
It is a simple set up. Tank dia of 3 m, height 0.75 m, 0.1 m dia drain hole at the bottom center. The tank is initially filled with water upto height of 0.5 m and the rest above it is air. At time 0, the tank starts to drain purely under gravity.
What boundary conditions do I use for the atmosphere (top) and the drain, for U and p_rgh?
I would like to add that I have done a good bit of searching, including on this forum and while there are several "similar" questions and attempted responses, I feel that somehow the question hasn't been answered for someone at my level. I feel rather stupid not being able to figure this out but the truth is I could not. Each answer has confused me even more. The top boundary (atmosphere) is very similar to the dam break tutorial I think and is OK. But the bottom drain isn't that clear for me.
Can someone please help with a little more clear direction?

Thank you very much.

Isaac

[tomf]

Hi Isaac,

I think it may work if you put the same BCs for the drain as that you did for the atmosphere(totalPressure and pressureInletOutletVelocity), except maybe that you give the inletValue (and value) for alpha.water as 1 instead of 0. I suggest you test this first in 2D, maybe on a coarse mesh. You may get an error when the domain is (almost) completely drained since than there would be no pressure gradient anymore, but I am speculating here.

Best regards,
Tom

[floquation]

I reckon this tutorial (ch5) may help, at least for the atmosphere.

For the drain... From a theoretical point of view you know the velocity, hence you could set the velocity. This could be considered cheating if you want to see that OpenFoam correctly predicts what the theory predicts.
If you do not wish to do that, you know that to first-order approximation the pressure will be atmospheric pressure at the outlet (as the tank empties into an environment at atmospheric pressure). Then, also to first-order approximation, I think that Bernoulli can tell you the normal gradient of velocity. I don't know what you should then do for the tangential velocity though.

The proper way to do it, is to add a container underneath your tank, and also simulate the environment below your tank into which your tank is emptied. After all, you technically only know the correct BC "in infinity", and not right at your outlet.
This takes you to a situation that is almost identical to the mentioned tutorial.

[Isaac]

Hello,

Thanks Floquation and Tomf for your responses. I am not sure yet but I may have got it to work.

@Floquation: You are right in stating that the most realistic and perhaps foolproof way of simulating is to include a "receiving" reservoir or sorts underneath the tank. However, I am sure you know why I would prefer not to do that - extra computations. However, the example you quoted is a good one. I had indeed looked at it and learnt a few things from it. And, yes, in some ways I did want to check if I was getting the velocity I was expecting from Bernoulli.

@ Tomf: I think I got it to work with your suggestion. I can see vectors leaving the domain at the outlet. The exit velocity is about 1/2 of what Bernoulli predicts. However, that may be caused by something different and I should look into it further. Also, the level of the liquid does not seem to be coming down (I did about 5 s of flow time and the initial liquid height is about 0.5 m). Again, I need to look at my other BCs a little better.

Thanks for your help. If you have any other random suggestions, please feel free to throw it my way.

Isaac