CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Flow rate imbalance (https://www.cfd-online.com/Forums/cfx/83000-flow-rate-imbalance.html)

serezhkin December 10, 2010 17:18

Flow rate imbalance
 
Hi all,

I am running a steady state calculation where incompressible flow goes from one plenum to the other through several openings in the wall.

I specify the velocity on the inlet such that the flow rate is a certain number (465CFM). The outlet is a pressure boundary.

The problem is, when I go to CFX-Post and calculate massFlow()@Each Opening surface and add them up together, the flow rate through the openings does not match the inlet flow and is in fact much lower (392CFM).

Any ideas on why this is happening are much appreciated.

ghorrocks December 11, 2010 06:14

If your simulation is steady state then you have not adequately converged.

serezhkin December 12, 2010 00:56

It is steady state, and it has converged down to 5E-5 or so on all velocity residuals.
As far as I understand, the mass flow rate imbalance should be down to round-off error within the first 20-30 iterations (I ran 800) for steady-state.
I thought that maybe I am just not calculating the flow rate correctly.

Is there a situation where the result of call to massFlow() is open to interpretation?

ghorrocks December 12, 2010 05:54

Global balances are different things to residuals. You can have a simulation with very low residuals but the global balances are still miles off. Anyone who has done CHT simulations will have seen this. In this case you need to specify convergence on residuals AND imbalances. Residuals alone is not sufficient. This is discussed int he documentation.

How are you calculating the mass flows?

Quote:

the mass flow rate imbalance should be down to round-off error within the first 20-30 iterations
Rubbish. A well behaved simulation may have converged in 20-30 iterations but to converge to round off will take a lot longer than that. And if you ran 800 iterations then I suspect your simulation is not well optimised yet so even that may not be enough. And there are plenty of more complex cases when you need thousands of iterations for a steady state run to converge.

serezhkin December 13, 2010 14:16

I am calculating the flow rates using the massFlow() function on the opening surfaces where there is a 1:1 connection between mesh blocks. I am also monitoring the flow rates in solver, and there's less than 0.2% variation over the last 100 iterations or so.

By the way, the only reason I ran 800 iterations is to see if it will make a difference in the mass flow imbalance. The run is converged within 150 iterations.

ghorrocks December 13, 2010 17:33

What do the imbalances at the end of the output file report?

serezhkin December 15, 2010 19:34

All right, I figured it out, so I'll post here in case someone runs into a similar problem.

First of all, my thanks to ghorrocks for his help with the problem.

It looks like Ansys only checks the boundaries of the domain to make sure the flow rate balance is observed. Because all my mesh regions were in same domain, the flow balance between mesh regions was not enforced.

As soon as I split the domain into several domains (with GGI interfaces), CFX realized that there's a flow imbalance and fixed it.

Thus, on the boundary of interest (between mesh regions) the flow rate went up to what it should be (as specified on the inlet BC).


All times are GMT -4. The time now is 11:42.