CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   CFX (https://www.cfd-online.com/Forums/cfx/)
-   -   Transient axial rotor/stator convergence issue? (https://www.cfd-online.com/Forums/cfx/73768-transient-axial-rotor-stator-convergence-issue.html)

 Nicola Viscanti March 16, 2010 12:12

Transient axial rotor/stator convergence issue?

2 Attachment(s)
Hi all!

I made a transient analysis of an axial rotor/stator. First of all I made a steady analysis with frozen rotor interface and I reached a good convergence after 250 iterations (between 10^-5 and 10^-6 RMS).
Then I launched the transient analysis using the steady results like initial values.
I have the same rotor/stator blades number so I used 1:1 and I have a pitch ratio that's exactly 1.
I set the total time = (2pi/rotor blades)/omega where omega is rotational speed in rad/s and I used 50 time steps.
1) max coeff. loop =10 (default)
2) max coeff. loop =30

B.C. are:
Inlet -> Total pressure and Total temperature, turbulence intensity 5%
Outlet-> Average static pressure
k-epsilon

I enclose 2 plots:
1) massflow goes through the stage
2) axial force on rotor blade

I expected a constant massflow. What are oscillationts in the last time steps? Did my transient analysis reach convergence?
I expected a periodic behaviour for axial force (the same value or very similar at 250 and 300 timestep) because I have pitch ratio exactly 1. I'm right? Plots are the same using max coeff. loop 10 or 30

Thank you very much!

 longbow March 16, 2010 13:41

Your calculation probably hasn't been stabilized. It takes several cycles to develop a repeated pattern. By the way, specifying pressure at both inlet and outlet is not a robust setup.

 Bak_Flow March 16, 2010 18:14

Hi Nicola,

I am having a bit of trouble understading the set-up that you have discribed.

1. You say that you have the same number of blades on the rotor and the stator how many is that?

2. Are you modeling a full sector or a full-wheel?

4. What is the 50 steps for? 1 revolution or 1 blade pass?

5. Your figure show iteration...should this be time step?

i usually set one of these problems up with 10-20 time steps per blade pass. 3-5 iterations per time step. Things will begin to develop into a regular period after a few revolutions.

If you do not get good convergence (or accuracy of the transient features) then half the time step. The time step requirement depends on the macine and the strength of the coupling between the rotor and stator. Very sharp features such as shocks, separations, etc. from an upstream component need to be resolved with smaller time steps than "smoother interactions".

Why would you expect a constant mass flow rate with Ptotal in and Pstatic out on the stage?

Think of it this way: as you solve the transient interaction you are clocking and then solving the full transient equations...then clocking again right? What would you get if you just clocked and solved to steady state and then clocked and solved again......slightly different mass flows because it is a different configurations....right?

For boundary conditions it depends what operating point you are trying to solve but try to set boundary conditions for what is "most constant". On the flat part of the curve mass flow inlet and statc pressure outlet works well.....on the steep part and near choking (of it is compressible) then Ptotal inlet and Pstatic out works well. If you are having problems solving at a difficult point try solving an easier one for a few rotations and then "moving the solution over" to the point you want. All of the flow dynamics associated with rotation will be developed and you only need to solve for what is additionally challenging due to the rotor-stator interaction at the desired point.

I hope this helps,

let us know how it works out..........Bak_Flow

 Nicola Viscanti March 17, 2010 05:15

Hi bak_flow,

I have 73 stator blades and 73 rotor blades so I used a stationary domain made by 1 stator blade and periodic surfaces and 1 rotor blade with periodic surfaces. Pitch ratio is exactly 1!