15MW Simulink for IPC

Wind & Water Controls
1

Hello,

I was running the 15MW Simulink wind turbine for testing the IPC performance. After MBC, the bending moment was transformed to the stationary axis with M_tilt and M_yaw. Then they are controlled with PI and output theta_tilt and theta_yaw, seperately. The scope shows that the theta_yaw is very small and close to zero. I was wondering whether it is right for the whole simulation.

image
image1920×904 105 KB

I am using the ROSCO.mdl and with 15ms steady wind speed. The corresponding individual pithc angle is given as follows.

image
image1502×791 81.1 KB

image
image1914×994 146 KB

I would be so so grateful if you can help to check. Thanks in advance.

2

Hi Yongxiang,

This depends on your simulation and controller. For IPC, if no yaw moment is calculated from the blade loads, then the yaw pitch angle will be small.

Best, Dan

3

Hi Dan,

Thank you very much for your prompt response.

For the simulation setting, I just opened the IPC_ControlMode=2 and changed the IPC_KI parameters in DISCON.IN.

!------- INDIVIDUAL PITCH CONTROL -----------------------------------------
8.472000 10.590000 ! IPC_Vramp - Start and end wind speeds for cut-in ramp function. First entry: IPC inactive, second entry: IPC fully active. [m/s]
2 ! IPC_SatMode - IPC Saturation method (0 - no saturation (except by PC_MinPit), 1 - saturate by PS_BldPitchMin, 2 - saturate sotfly (full IPC cycle) by PC_MinPit, 3 - saturate softly by PS_BldPitchMin)
0.3 ! IPC_IntSat - Integrator saturation (maximum signal amplitude contribution to pitch from IPC), [rad]
0.000e+00 0.000e+00 ! IPC_KP - Proportional gain for the individual pitch controller: first parameter for 1P reductions, second for 2P reductions, [-]
0.200e+00 0.010e+00 ! IPC_KI - Integral gain for the individual pitch controller: first parameter for 1P reductions, second for 2P reductions, [-]
0.000000 0.000000 ! IPC_aziOffset - Phase offset added to the azimuth angle for the individual pitch controller, [rad].
0.0 ! IPC_CornerFreqAct - Corner frequency of the first-order actuators model, to induce a phase lag in the IPC signal {0: Disable}, [rad/s]

So, do you have some preference on how to set the optimal IPC_KP and IPC_KI? Or whether I ignore some parameter mode during the simulation?

Best,
Yongxiang

4

Hi Yongxiang,

I would tune those parameters to minimize the variation in blade root loading. I recommend starting with small values of IPC_KI and increasing until you see diminishing returns.

Best, Dan

5

Thanks Dan, it is really helpful.

If my understanding is correct, the ROSCO IPC for tuning the IPC_Kp IPC_KI is dynamically soft cutin with a sigma function (Controllers.f90).

    ! Soft cutin with sigma function 
    DO i = 1,2
        LocalVar%IPC_KP(i) = sigma(LocalVar%WE_Vw, CntrPar%IPC_Vramp(1), CntrPar%IPC_Vramp(2), 0.0_DbKi, CntrPar%IPC_KP(i), ErrVar)
        LocalVar%IPC_KI(i) = sigma(LocalVar%WE_Vw, CntrPar%IPC_Vramp(1), CntrPar%IPC_Vramp(2), 0.0_DbKi, CntrPar%IPC_KI(i), ErrVar)
    END DO

But I found that simulink model for intergral gain IPC_KI is a user-defined constant, which means in every sampling step, the IPC_KI is same. If it is true, then it is an optimal problem for tune the PI controller parameter. May I formulate an optimal problem with the decision variables as [IPC_Kp, IPC_KI], and then the objection function would be the sum of blade root moment DEL(M1+M2+M3)?

To implement this idea, what I need to do for the next procedure is that I change the fortran controller.f90 routine and then recompile it to work with cmake command?

Thanks for your attention.

Best,
Yongxiang