Translational periodicity with specified mass flow rate
 

Fellow CFX users,
I wanted to simulate a fully developed pipe flow using the translational periodicity bc with user specified mass flow rate. My geometry is a 90 degree section (taking advantage of the rotational periodic geometry) 4 diameters long. Fluid is water. As basic as a simulation can get.

When I apply translational periodicity with specified mass flow rate it appears impossible to obtain convergence as the solution starts to oscillate (pressure loss of the pipe section going up and down). This happens with different Reynolds numbers alike. Adjusting the pressure update multiplier even to small values like 0.001 cannot stabilize the solution. When I change interface model to specified pressure change the solution converges monotoniously and is completetely stable (resulting in same Re as targeted with specified mass flow rate).

Has anyone had similar experience with convergence using the specified flow rate? I am really not confident trying to use it for any real case if even this dummy pipe cannot be properly solved. Any ideas if I should check some particular settings?

have made the same experience. Don't know why and what is happening there. Did you solve the problem or have an solution? How do you specify the pressure change that you get the same Re number as with targeted mass flow rate? this would help me too

Have you investigated this a bit? Here are some ideas:

* save every iteration to a results file and run it. Try to see what the oscillation actually is - is it the axial velocity, a rotational velocity, or something else?
* Have a close look at the results file before it crashes. Do you have any unexpected flow features, like recirculations?
* If you artificially increase the viscosity by a large factor (say 1000) does that stabilise the results?
* Is your mesh quality good?

Hello,

When I tested this the simulation was very basic - just a straight pipe thus the mesh quality was very good. I tried different Reynold's numbers thus there shouldn't have been any fluid dynamic phenomena to cause instability in all cases.

For a pipe it was of course easy to set pressure difference for desired mass flow rate using the Moody diagram. When pressure difference was fixed intead of mass flow rate I was able to reproduce the Moody diagram throughout wide range of Reynoldäs numbers without any instability in the solution. This is why I smell a rat in this periodicidy using specified mass flow rate. Haven't tried this since cfx 14.5. Maybe this is fixed already in cfx 18

CFX 14.5 is many years old and many things have changed. Definitely give it a go in the current version.