[hlf13]

Hi,

I'm trying to run a simulation on a propeller for an RC-helicopter, but facing some problems, mainly because of the solution will not convergence (even after 10 000 iterations)

The Reynolds Number is 350 000. Should the Y+ values be below 1, or above 30?

I have tried to vary the Y+ value, but still the residuals are oscillating, see the picture.



Mesh: 4 000 000 cells
I have used K-omega, segregated flow, implicit unsteady.
Time step: One time step per degree
Inner iterations: 8
RPM: 8000
I have used the Interface set up

Hope someone can help me out with this, it's very frustrating.

[fluid23]

First, your residuals will still oscillate in an implicit analysis. Every new time step will see your residuals spike then settle out.

Second, you may want more inner iterations, it's hard to tell from the residual plot you show. Try running it at say 50 inner iterations and see if your momentum and continuity converge better. You may not need to do this for the whole simulation, just for the first several time steps until things start to converge better.

Second, one degree per time step may not be your best bet.. What is your CFL number near the rotor tip. (I assume this is where you have the smallest cells and highest velocity.)

[fluid23]

Also... pardon my ignorance, but what do you mean by interface setup?

[hlf13]

Thank you for your reply.

I will try to run with more inner iterations

The CFL is around 10-20 on the tip of the blade. I found "one degree per time step" in the Star CCM manual

By interface setup I mean an interface between boundaries which produce a conformal mesh interface.
I created two regions, a rotating and a static domain. Subtracted the ‘’rotating cylinder’’ from the static domain and subtracted the propeller form the ‘’rotating cylinder’’. Then made an interface between the ‘’rotating cylinder’’ in each region. Chose the rotating domain to rotate with 8000rpm

[fluid23]

Got it... thanks.

I would also try reducing your time step to 1/2 degree. An implicit solver is usually more tolerant of high CFL, but it won't hurt to bring it down. Plus, it's usually a good bet that whatever the examples show in the manual is a fairly coarse choice. They aren't necessarily optimized for best results.