Flow rate in and out does not match

[atm98]

I recently reviewed a ANSYS CFD report of a custom valve and noticed that the flow rate in and flow rate out did not match. More came in then out.

I am currently playing with the Comsol packages and did a quick model of a pipe with a reduced port in the middle. Again the report showed a difference. More flow in then out the back end.

(water used in both scenerios)

I am obviously new to this stuff so I am thinking there is something fundamental I am missing.

Any thoughts???

 

[MintJulep]

Most likely either your solutions haven't converged, or you have a gap in your mesh.

 

[btrueblood]

What Mint said. Or, you have a third port in the model, that you didn't know about?

Plot residual errors vs. iteration number, see if solution is converging.

Plot streamlines, look for funky behavior (streamlines diving into the wall and disappearing).

Possibly there is a leak in the CPU memory, and all the water is going thru there, look for a puddle under your PC.

 

[MintJulep]

gap in your mesh = third port in the model, that you didn't know about

Yes?

 

[btrueblood]

Kinda - but "gap in mesh" in my mind meant "I built the model, but it's f'd up". Third port = it's somebody else's model, of a valve type you are not familiar with (multi-port), and you didn't set the b.c.'s correctly. Same effect, different root cause.

Sorry to poach on your reply Mint, I'll shut up now.

 

[atm98]

Thanks for the input.

I am positive that in both cases there is no third port. I've looked at every possible chart or graph and they look fairly correct. I also believe the mesh to be correct. I have even refined it with minimal change to the results.

I'll see if I can find the residual error information.

btrueblood - I keep towels under my pc just in case of such a leak.

 

[mechengdude]

When you say that the flows don't match do you mean there is a signifigant difference? Or are we talking about a difference of say 1.50 GPM IN and 1.498895 OUT?

Typically with this sort of thing I see a very slight difference as described above when using COSMOS FloWorks. I don't consider the above example anything to fret about.

 

[atm98]

In the valve flow report the difference was about 10%
In my model its about 110% The traces look fine so I am a little stumped. I will provide updates once I return from my travels and have time to go over everything again.

 

[bagarbilla]

What is the size of your model ?

 

[FeX32]

Have you accounted for losses?

Fe

 

[hd007]

I've gone trough several problems that sounds the same as you have. And the troubles has always come from the boundary setting. Somehow, one or more nodes or edges on the mesh belongs to Neumann Bc's which should be Dirichlet Bc's.