Bridging between 2 ( or more ) meshed regions with an analytical solution?
 

Just wondering if there is a way to bridge/join multiple meshed regions in Fluent with an analytical equation(s). For example, I would have a meshed flow controller fluid region and a meshed hydrostatic bearing fluid region with a long tube in between them. Instead of meshing the long tube ( which would have fully developed flow within it ) and having one continuous, meshed fluid domain, as usual, can I omit the meshed tube portion and replace it with an analytical solution in between the other 2 meshed regions? This would obviously save on computational time and memory. I suspect I might need a UDF, and I know next to nothing about those so far.



( Actually, I would have 3 flow controller regions, each with 2 long tubes, going to 6 inlets for 1 fluid bearing region. )


I'm using ANSYS 2022 R1, if that matters.

Instead of jumping straight to the most complicated solution, how about breaking it into simpler parts that can be digested?

How is your approach different than simply not meshing unneeded regions and providing boundary conditions? Actually that is what you would have to do anyway if even you go the UDF route. The point of this exercise is to get you to think small and not grandiose. The question is whether you actually need a UDF to specify these boundary conditions or is there some existing method you can just piggyback? A velocity inlet for example...

I was thinking of breaking it down into 2 separate problems initially.

The flow in the controller is radial, inward, turbulent flow between 2 parallel disks. That is, flow from the perimeter of these disks to a central hole. There are formulas for laminar flow rates as a function of pressure drop in this geometry, but not turbulent. ( And if it was outward flow, it would be even more complicated. )

As for the getting flow rates in the ( conical ) bearing by analytical means, forget it. Much too complicated.

So all I really have to work with are pressure BCs. The static pressure at the outlet ( of the bearing ) is always zero ( or some other low pressure that is known ), and the stagnation pressure "close" to the inlet ( of the controller ) is always a constant ( just have to make sure my reservoir is large enough so that the velocity is essentially zero )


Aside from the above, I'm still wondering if my original question is doable.

I'll think about this problem some more. But my original question was geared more towards decreasing the meshed fluid domain, if possible, by replacing an otherwise meshed tube(s) with a simple analytical solution of turbulent flow in a straight tube.

This problem is really a FSI problem, and that's how it will eventually be modeled. One of the disks in the controller portion can move in response to the downstream pressure in the recess of the hydrostatic bearing portion thereby changing the spacing between the disks and hence, the flow rate.

I've got 64 GB of RAM to work with. If I mesh the entire domain, I should have between 20 to 25 million elements. Just hope 64 GB is enough if it comes to that.

The frank answer to your actual question is a trivial yes, people do CFD all the time without meshing the entire universe. But that's not the issue...


Replacing things with known conditions from other sources of knowledge is just another way of saying you want to apply boundary conditions without actually saying you want to apply boundary conditions.