Different Behavior in Aerodyn v.14 and v.15

Dear all

This is my first post in the forum and I’m going to give a short summery of my work.

I am working on my PhD project on Maximum Power Point Tracking(MPPT) of wind turbines and I use FAST coupled with Simulink to simulate the dynamic behavior of the wind turbine and develop different control strategies to maximize the output power. The final prospect of the project is to develop an innovative predictive control strategy for small and medium size turbines.

So I used an almost linear model as my PMSG and I increase the generator torque to find the optimum TSR, to use for an simple TSR control strategy. However, I get different behavior when I use aerodyn v.14 and v.15.
The model used is TEST17 and the windSpeed is set to 12m/s steady and YawDOF to False.

1- By changing the pitch angle of the blade I get an optimum efficiency(Cp) in PitchAngle=2° in v.14 and PitchAngle=7° in v.15!! What is the optimum pitch angle for SWRT model? and why do I get different optimum pitch angles for different aerodyn versions?

2- The Cp curve reaches to almost Betz limit for V.14 but for V.15 is a value around 0.42!!

3- Another problem in v.14 is the really high fluctuation for the output RtrPwr (red line in pic below). The simulation time is around 10 mins and the generator torque slop is low enough. How may I find the reason behind this behavior ? The picture is attached below.

Thanks in advance for your help and sorry for the long post!!

While the solution algorithm in AeroDyn v15 is quite different than AeroDyn v14, I would not expect larger differences in gross aerodynamic loads between the two models. (I would expect some differences if the tangential induction is large, if the inflow-skew angle is sizeable, or if the structural deflections are large.) From your results, I would guess that there is a problem in the simulation set-up of one or both models. If you are seeing very high power coefficients (in a time-averaged sense) out of AeroDyn v14, have you perhaps disabled the induced velocity calculation by setting IndModel to NONE instead of WAKE or SWIRL?

Best regards,

Dear Jason

Thank you for your quick response. The IndModel is set to SWIRL. The behavior of Ayrodyn V.14 and V.15 are close for the pitch angles from 11° to 13°. However by changing it to values below 10° (e.g. 5° or 8°) the power coefficients goes unexpectedly high (in a time-averaged sense) and the fluctuations appear in the Rotorpower curve(red line in picture above), which makes me feel that Aerodyn V.14 fails to simulate some phenomenon in my simulation conditions.

I accidentally switched the InfModel from DYNIN to EQUIL and now, the simulation results are much closer to V.15 in my testing condition (for all pitch angles) especially for high generator torque and TSR values lower than optimum TSR. As DYNIN is a more recent algorithm comparing to EQUIL, would you recommend using EQUIL in my simulation?

I also receive these warnings using Aerodyn v.15.

Warning: block ‘MiladTest/FAST Nonlinear Wind Turbine/S-Function’:
FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption2:AD_CalcOutput:BEMT_CalcOutput(node 17, blade1): Compute_UA_AirfoilCoefs:UA_CalcOutput:Mach number exceeds 0.3. Theory is invalid. This warning will not be repeated though the condition may persist.

Would you please tell me the impact of these warnings on the validity of the results? Shall I be concerned and objective about them?

Kind Regards

The EQUIL option of AeroDyn v14 should match AeroDyn v15 better than DYNIN because dynamic wake is not yet included in AeroDyn v15. I would expect that the steady-state solution would be similar (but not identical) between EQUIL and DYNIN, with larger differences expected under transient dynamic conditions.

The warning regarding the disabling the unsteady airfoil aerodynamics routines at high angles of attack is not really a concern. See my post dated Mar 09, 2017 in the following forum topic for more information: http://forums.nrel.gov/t/windpack-1-5-mw/1571/2.

The warning regarding the Mach number exceeding 0.3 is potentially more serious. Are you expecting the Mach number to be so high?

Best regards,

Dear Jason

Thank you very much for all your help. In high wind speed and low generator torque Mach number exceeds 0.3. I modified the model and don’t get the warning anymore.

Kind Regards

Dear All,

To better investigate a 13.2MW two-bladed downwind turbine, I wanted to go from AeroDyn v14.05 to AeroDyn v15.03 (both with FASTv8) since there are more output sensors available in the latter. To do so, I changed the AeroDyn Input File of v14.05 (SUMR-13i_v6_AD14.ip) with the new one for v15.03 (SUMR-13i_v6_AD) and also created the new AeroDyn v15.03 Blade Definition Input File (FASTBlade_precomp_v6_s4.dat). All the values used in both .ipt files are based on a Blade Input File that was prepared for FASTv7 for the same turbine (FASTv7_Blade_Definition.dat). Due to some definition differences between the modules, I interpolated and extrapolated some points to have a refined blade shape and also to define the tip and root sections of the blade properly. Other than that, I used the following formulas that are taken from another discussion in the forum ([url]Definition of parameters] - Jun 14, 2017) to complete AeroDyn v15.03 Blade Definition Input File;

BlCrvAC = -(0.25-AeroCent)ChordSIN(AeroTwst)
BlSwpAC = -(0.25-AeroCent)ChordCOS(AeroTwst)
BlTwist = AeroTwst

Furthermore, PitchAxis values in ElastoDyn Blade Definition Input File (EDFASTBlade_precomp_v6_s4.dat) are calculated with the formula PitchAxis = 0.5 - AeroCent taken from README_FAST8.pdf (pg.41).

As I mentioned before, I ran FASTv8 with both AeroDyn v14.05 and v15.03. Other than the AeroDyn input files mentioned above, everything is kept constant for those test runs and All DOFs were disabled to have rigid conditions. For the comparison of the results, the blade section at 84.05m from the blade root is selected and sensors at that section are enabled for both versions.

I have a couple of questions regarding the results;

1. What could be the reason for the differences between the two runs, and especially for aerodynamic power, is it normal to have a 4% difference in it while there is not that much of a difference between other aerodynamic parameters shown in the figures?

2. The parameter used in Cp calculation ‘RtVAvgxh’ according to the formula specified in another discussion in the forum ([url]Pitch rate in Fast v7] - Feb 24, 2017) fluctuates while everything is supposed to be rigid. What could be the reason for those fluctuations? Furthermore, the wind is steady and its horizontal speed is set to 11.3m/s. Shouldn’t those fluctuations occur around 11.3m/s instead of 11.19m/s even if those fluctuations have to be there?

Sorry for the long post and thank you for your help in advance.

Best regards,

Note: Since the folder size is too big to upload here I shared a dropbox link for you to view the files.

Dear Emin,

In general, it sounds like your approach to develop and compare the two models sounds correct. To answer your questions:

1. I’m guessing the biggest cause of the differences between the AeroDyn v14 and AeroDyn v15 solutions (assuming you’ve set as similar as possible the aerodynamic properties distributed across the blade, which I have not checked) is related to the large 12.5deg precone angle of this turbine. There are many differences between AeroDyn v14 and AeroDyn v15 and it is difficult for me to remember all the details of the much older AeroDyn v14 code, but I believe the way the blade coning is handled between the two software differs. Generally speaking, we tend to believe AeroDyn v15 is more accurate. Do the results better align if you eliminate the precone in both solutions?

2. This is related to the high power-law shear exponent of 0.2 you are using. The rotor-disk-averaged speed is calculated not by analyzing the wind across the disk, but by averaging the wind speed along the aerodynamic analysis nodes distributed across the blades. As a result, the velocity in a 2-bladed rotor will be periodic in sheared flow. Furthermore, the average will be less than at hub height because of the nonlinear shape of the power law.

Best regards,

1 Like

Dear Dr. Jonkman,

Thank you very much for your quick reply. As you suggested, I ran both cases with 0º cone angle and, except for tangential induction (where the difference has increased very slightly), all aerodynamic parameters have matched almost perfectly.

Thank you very much for your help again.

Kind regards,