I have a question regarding a 2 phase sloshing problem.

If we have a tank half filled with water and air, and the tank moves in a sinosoidal rolling movement, exciting the fluid. Is it valid to model the system by changing the gravity vector in a similar sinosoidal manner. Or should the whole tank (ie mesh), move in such a manner?

Could a similar assumption be made for a horizontal sinosoidal translation ?

Bob

Bob, I may miss a detail, but ...

If you lock your coordinate system to the tank, using a no-slip velocity on the tank walls and bottom, then the gravity vector shifts back and forth as the tank rocks. You'll want to check the equation form you use carefully, because this is a non-inertial coordinate frame and the momentum equations need to be modified. Actually you should start with momentum equations valid in the inertial frame and transform them into the oscillating frame. Your payoff is that you don't have to program the boundary condition motion of the walls.

The other way, solving in the inertial frame, you need to somehow program the velocity boundary conditions to follow the moving walls.

You didn't ask. But are you sure you need to include the air? I've been told that there's an 'impedence mismatch' between the two fluids at the interface, so that the influence of the air, which is mostly small, is poorly calculated in any case.

Hi Jim, yes we would need the air to be modelled I think. The reason I asked the question was a result from another conversation regarding this problem. Originally I suggested moving the gravity vector, however my boss told me that this was not the correct method as it did not take into accout the added mass of the water in the system. My relationship with my boss was no the best, and when I asked for an expalnation of this (as I didn't follow what he had said), he just waved his hand and didn't answer. Great ! So now I have left the company, I wanted to understand the issue hear and try and understand why my method would not have been applicable. We were using the CFX code for the simulation, and in the end we modelled the system by moving the mesh. So if there is any chance you could elaborate on your explanation I'd be most grateful, Cheers Bob

Bob, I'm not so great with the "added mass" myself. As I recall, it's a term that shows up in the equations of motion when you do the coordinate transformation. My recollection is that it's a result of algrebra, not physics.

If I remember correctly, it would have been useful to help interpret results from approximate solutions obtained before CFD became prevalent.

I'm sure some of the other folks on this forum will elaborate, correct, or both.

Cheers Jim, Can any one else help with the explanation please ? Bob