Difference in the torque value computed from tangential forces and RotTorque output from Elastodyn_BDoutputs file

Wind & Water Rotor Aerodynamics
1

Hello everyone,
I am running FAST simulations for the NREL 5 MW turbine model at a steady wind speed of 12 m/s with aerodyn parameters shown in the below table. It rotates at a speed of 12.1 rpm with a pitch angle of 2.74 degrees when I used aeordynamic coefficients calculated for the aerofoils using CFD simulations.

AERODYN v15.03 input parameters

1 WakeMod - Type of wake/induction model (switch) {0=none, 1=BEMT}
2 AFAeroMod - Type of blade airfoil aerodynamics model (switch)
{1=steady model, 2=Beddoes-Leishman unsteady model}
True TipLoss - Use the Prandtl tip-loss model? (flag) [used only when WakeMod=1]
True HubLoss - Use the Prandtl hub-loss model? (flag) [used only when WakeMod=1]
True TanInd - Include tangential induction in BEMT calculations? (flag)
[used only when WakeMod=1]
True AIDrag - Include the drag term in the axial-induction calculation? (flag)
[used only when WakeMod=1]
True TIDrag - Include the drag term in the tangential-induction calculation? (flag)
[used only when WakeMod=1 and TanInd=TRUE]

When I run the NREL 5 MW onshore model (Test26 case) with above parameters, I get a distribution of Tangential force (to plane) per unit length along the Blade 1 (Using B1NjFy at 19 aero nodes as outputs in the Aerodyn file) as shown in the Figure 1 attached.

If I integrate this force distribution, I get a torque value of 1.3064e6 N-m.

Then its Rotor torque = 3*1.3064e6
= 3.9191e6 N-m

Rotor power = Rotor torque * omega
= 3.9191e6 * (2pi12.1/60)
= 4.9659 MW

Figure 1.png
Figure 1.png827×623 13.6 KB

I used ‘GenPower’ as an output from the Sevodyn file and ‘RotTorque’ as an output from the ElastoDyn_BDoutputs file, their values from the output file are given below.

GenPower = 5000 kW
RotTorque = 4200 kN-m

From the above output value of RotTorque, its rotor power can be calculated as
RotorPower = RotTorque * omega
= 420010^3 * (2pi*12.1/60)
= 5.321 MW

This turbine generates its rated power at 12 m/s and as per the generator efficiency of 94.4% these values are correct.

I have following questions on the outputs from FAST simulations.

  1. When I enable the tip loss factor, I would expect a zero value for the tangential force at the tip something like shown in the atatched Figure 2. Are there any mistakes in my inputs ?

Figure 2.png
Figure 2.png836×621 35.3 KB

  1. Why there is a difference in the torque value computed from the integration of tangential force Fy output values from Aerodyn and torque value output from the Elastodyn_BDoutputs file ?

I attached Aerodyn, Servodyn, Elastodyn_BDoutputs, InflowWind files used for the simulations to this post.

Can anyone help me where exactly the issue is.

Thanks in advance
Sudhakar Gantasala
Input_files.zip (10.8 KB)

2

Dear Sudhakar,

Just a few comments/questions:

  • How are you plotting the tangential force in N, when AeroDyn outputs this force in N/m? Have you someone integrated the distributed force to lump the forces at nodes?
  • It appears from your input files that blade/turbine structural flexibility, shaft tilt, the skewed-wake correction, and unsteady airfoil aerodynamics are still enabled for this simulation, which will result in a time-variation of aerodynamic loads. Have you time-averaged this time variation in order to generate your plot?
  • How are you considering the blade deflection in your integration of aerodynamic forces when hand-calculating the aerodynamic torque? How does this hand calculation compare to the aerodynamic torque computed and output by AeroDyn (RtAeroMxh)?
  • We’ve improved the blade-element/moment theory (BEMT) algorithm, as well as the hub- and tip-loss models in the release of AeroDyn v15.04–see e.g. FAST v8 "Mach number exceeds 1.0" for some wind cases - #15 by Jason.Jonkman. It would be interesting to see how your results change if you upgrade to the newest version of AeroDyn. Regardless, I would not expect the aerodynamic loads to be exactly zero at the tip, as the tip chordlength is finite–and while the axial induction factor tends toward 1 at the tip, the tangential induction factor tends toward zero (and thus there is still dynamic pressure at the tip).

Best regards,

3

Dear Jason,

Thanks for your reply and suggestions.

 I mentioned the y-axis units wrongly in those figures. You are correct, it should be N/m. The values used in those figures corresponds to one specific time step not the averaged values over time.

 I ran a new simulation with disabling all the degrees of freedom other than the generator in the Elastodyn file (chose CompElast as 1 in FAST input file) and also disabled skewed-wake and unsteady aerodynamics options in the AeroDyn15 input file. The tangential aerodynamic loads in this case are plotted in the below Figure.

Tangential_load_at_aeronodes_uncoupled_rigid_12mps.png
Tangential_load_at_aeronodes_uncoupled_rigid_12mps.png1920×1036 93.1 KB

They are time varying, and the load at the tip is zero (agrees with Prandtl tip loss factor model). In this case, I integrated aerodynamic loads along the blade to get the torque value and compared with the RotTorque output from ElastoDyn_BDoutputs file and RtAeroMxh from AeroDyn file in the below Figure.

Torque_comparison_uncoupled_rigid_12mps.png
Torque_comparison_uncoupled_rigid_12mps.png1715×791 17.9 KB

RotTorque and RtAeroMxh values are matching exactly and they are not time varying like the aerodynamic torque.

o What can be the reason for time varying aerodynamic loads in this case with rigid blades and tower, steady and uncoupled aerodynamics model ? I have removed the effect of gravity as well.
o How the RotTorque and RtAeroMxh values are calculated as both of them are not matching with the integrated torque values of the tangential aerodynamic loads output at aerodyne nodes ?

 I did not consider any blade deflection when integrating aerodynamic loads (from AeroDyn output) along the blade to calculate the aerodynamic torque in the case of flexible blades (i.e., previous post). Is there any way to output the local aerodynamic loads considering the blade vibrations ?

 I will try the same simulation with AeroDyn v15.04 if I see any difference.

I attached FAST files used in this case to this post.

Regards
Sudhakar
FAST_files.zip (14 KB)

4

Dear Sudhakar,

Output RotAeroMxh from AeroDyn v15 is the aerodynamic torque calculated by integrating the aerodynamic loads at the analysis nodes along the blade. My guess is there a small difference between the internal integration computed within AeroDyn and your hand calculation. The internal integration is computed using the mesh-mapping routines that allow Line2-to-Point mapping of loads (i.e. blade nodes with distributed aerodynamic loads mapped to a concentrated point load on the hub). The internal mesh-mapping routines permit blade deflection, but the integration would be difficult to do by hand.

I hope that helps.

Best regards,

5

Dear Jason,

Once again thank you for your quick reply.

I am convinced with the small difference in the hand calculation with respect to AeroDyn output can be due to the way it is integrated inside the code.
But, I have not found any reason for the periodic time-variation of the aerodynamic loads when blades and tower are considered as rigid structures, steady and uncoupled aerodynamic model is considered. How this peridocity disappeared in the aerodynamic torque output from the AeroDyn (RtAeroMxh).
Can you comment on the above observations of mine on the aerodynamic loads output.

Regards
Sudhakar

6

Dear Sudhakar,

I would assume that the small oscillations you are seeing in your hand calculation of aerodynamic torque are due to small inaccuracies in the hand calculation (I’m not exactly sure what those inaccuracies would be, but perhaps its as simple as numerical round off of the FAST output).

Best regards,

7

Dear Jason,

I am sorry that I didn’t present my question clearly in the last post.

I plotted tangential loads output (BiNjFy) at various nodal positions used in the AeroDyn file for one blade in the below figure. First of all, what can be the reason behind the time varying nature of these loads (when blades and tower are considered as rigid structures, steady and uncoupled aerodynamic model is considered). Even though these loads are time varying, the aerodynamic torque output from the AeroDyn (RtAeroMx) doesn’t change with time, it outputs a constant value at all time steps.

loads_variation.png
loads_variation.png919×613 53.2 KB

Regards
Sudhakar

8

Dear Sudhakar,

Have you considered a shaft tilt or wind shear (the FAST model you attached in your post above included shaft tilt)? These will induce a once-per-revolution (1P) oscillation in the aerodynamic loads in the rotating (blade) frame, but no oscillation in the fixed (tower) frame for a three-bladed rotor.

Best regards,

9

Dear Jason,

You are correct, i did not make shaft tilt to zero in my simulations.

Thanks for your help.

Regards
Sudhakar

10

Dear Jason,

I have written Matlab code for calculating aerodynamic forces; I am comparing with the FAST results ( AeroDynV15 used). I have NOT accounted for induction factors in my code. For a uniform wind speed case, I see good correlation between my results and FAST results, in rotor speed, alpha, inflow angle etc. at each section except the tangential force. Also, tangential force of FAST matches with my Normal force; But I believe I have used the correct expressions. Kindly confirm if the equations are correct? I couldn’t find the same in AeroDynV15 manual.

FAST outputs used:
B1N1Fx Normal force (to plane) per unit length at Blade 1, Node 1
B1N1Fy Tangential force (to plane) per unit length at Blade 1, Node 1
Phi: Inflow angle (Angle made by the relative velocity vector with the rotor plane). Sign conventions defined in FAST are assumed.

FX = (Lift Force) * cos(Phi) + (DragForce) * sin(Phi);
FY = (Lift Force) * sin (Phi) - (Drag Force) * cos(Phi);

Regards,
Kumara

11

Dear Kumara,

The correct equations are:

FX = (Lift Force) * cos(Phi) + (Drag Force) * sin(Phi)
FY = - (Lift Force) * sin(Phi) + (Drag Force) * cos(Phi)

See Figure 6.4 in the AeroDyn documentation on read-the-docs for an image: openfast.readthedocs.io/en/mast … t-channels.

Best regards,