Good morning,
I have been running the 5 MW onshore model with ElastoDyn, and I have obtained 10 minute results for different load channels for 3 yaw conditions- 8 deg, 0 deg and -8 deg. I have used 6 turbulence seeds on TurbSim, NTM as the turbulence type and IECKAI as the turbulence model. The simulations I ran were from 4 m/s to 26 m/s in steps of 2.
While post processing the results, the way I have done it is for each yaw direction and wind speed, I have averaged the results for 600 seconds over the number of turbulence seeds, in my case 6. So essentially I will have 3*12 = 36 result files. My first question here is if my assumption of averaging over the number of turbulence seeds is correct (since the turbulence seeds are random, I did not see the point in assigning a particular weight).
After this, I have plotted some of the turbine loads for all 3 yaw conditions over the operating range. The load channels I have considered are- Tower Top Tilt, Tower Top Roll, Tower Bottom Fore-Aft, Tower Bottom Side-Side, Blade root flap and Blade root edge. The behavior of all the load channels except Tower Top Tilt and Tower Bottom Side-Side are looking as expected, but these two look very strange. I have attached the plots for these two load channels below.
My understanding is that the mean Tower Top tilt moment should follow the thrust curve (inverted V with the peak at rated speed) and mean Tower Bottom Side-Side moment should follow the power curve. However the Tower Top Tilt moment does not seem to follow the thrust curve (in fact resembling the power curve) and the Tower Bottom Side-Side moment does not resemble the power curve.
I have looked at the steady state behavior of these two load channels and they are as expected. My suspicion is that adding turbulence appears to have brought some instability, but it is not consistent with other tower loads (in that they behave as expected). Another suspicion is that I may have used the wrong channel name to get these loads. I have used “YawBrMyp_[kN-m]” for Tower Top Tilt and “TwrBsMxt_[kN-m]” for Tower Bottom Side-Side.
I hope to be able to get your guidance on this issue and what might be causing it.
Best regards,
Vishal S
Dear Vishal,
Are you plotting the mean of the time-series mean across all turbulence seeds or something else, like the mean of the absolute maximums across all turbulence seeds?
I agree with your general observations, but it is difficult for me to guess why your results are not as you expect. Can you plot other output quantities, such as the mean of means and mean of standard deviations of blade pitch, rotor speed, blade deflection, and tower deflection to ensure the turbine response is as expected?
Best regards,
Dear Jason,
I am loading the result files for all windspeeds, yaw angles and turbulence seeds, then removing the first 100 seconds due to transient events so I am left with 600 seconds of data (I simulated for 700 s). Then I average all the load channels over the turbulence seeds (add all the values and divide by number of turbulence seeds, in this case 6). So I am left with 600 seconds of data averaged over turbulence seeds, so 1 set of data per wind speed per yaw angle.
The plots that I had attached previously are done by calculating the mean of the 600 s data averaged over turbulence seeds per wind speed and per yaw angle. So in essence, these are the mean of means (unless my understanding of it is incorrect).
I have attached the plots of RPM, Pitch and Tip Defl in x direction means along with Std Deviation.
I have not added the tower deflection as one of the output channels so I don’t have them at the moment.
It seems that the turbine is behaving as expected. Just the tower top tilt and tower bottom side-side moments look strange.
Best regards,
Vishal
Dear Vishal,
Thanks for clarifying. These new plots are all as I would expect them to look.
Regarding your original plots, it is only the tower-base side-side moment that is a bit odd to me. The tower-top tilt moment has such a small moment arm from thrust, that much of the moment could by caused by asymmetry across the rotor, like shear. For the tower-base side-side moment, I question why the mean of the mean value does not plateau at wind speeds above rated. What does the generator torque look like?
Best regards,
Dear Jason,
The generator torque looks sensible, it plateaus out after rated wind speed. I have attached the plot below.
Best regards,
Vishal
Dear Vishal,
This generator torque input looks like what I would expect as well. Perhaps the additional tower-base side-side moment is coming from some side-side shear forces at the tower top, based on how the turbulent wind and associated aerodynamic loads have been applied. Do you have plots of that?
Best regards,
Dear Jason,
Unfortunately I did not have tower top side-side shear forces as one of the output channels during simulation. Do you have any other suggestion on which other output channel to look at? All the turbine operating parameters’ response are exactly as expected, this would give indication that the turbine is operating normally. The other load channels like BR bending and Tower bottom fore-aft are as expected. Wouldn’t side-side shear forces at the tower top also affect the BR bending as well?
Best regards,
Vishal
Dear Vishal,
It would be hard to see the side-side shear force from the blade-root bending moments, which are in a rotating frame of reference.
The tower-top side-side shear force would be easiest to look at. The tower-top side-side deflection may also be useful to review.
Best regards,
Dear Jason,
Apologies for the delay in replying. I have extracted the tower top side-side shear forces and side-side deflections.
Best regards,
Vishal
Dear Vishal,
It definitely looks like the slow increase in mean tower-base side-to-side moment above rated is correlated with a slow-increase in tower-top side-to-side shear force (and subsequent tower-top side-to-side deflection). This was not seen when Figure 9-1 from the NREL 5-MW baseline wind specifications report (nrel.gov/docs/fy09osti/38060.pdf) was first developed. I’m guessing this is related to the turbulent wind input you have used (the original work used steady uniform wind). If you want to examine this more, I would suggest comparing the mean response (motions/loads) between cases with steady uniform wind inflow and cases with turbulent wind inflow.
Best regards,
Dear Jason,
I have not focused much on this topic for the last few weeks, but I am trying to figure out what could be causing this abnormal load behavior. Like you suggested, I compared the mean loads for steady state condition like in the NREL 5 MW report with the mean loads for turbulent condition (both at 0 yaw). All the load channels are very close to each other except the Tower Top tilt and Tower bottom side-side. I tried changing the damping ratios for the Tower fore-aft and side-side modes in the tower ElastoDyn input but to no avail, the results still remained the same.
Best regards,
Vishal
Dear Vishal,
I would guess the key difference is between the mean wind speed profile between your simulations with steady wind and turbulent wind. How does the mean wind speed profile (shear, perhaps veer) compare across the rotor disk?
Best regards,
Dear Jason,
I have used the IEC recommended shear factor value for the TurbSim input when I generated .wnd files (I believe it is 0.2). I would have to look at the mean wind speed in X,Y and Z directions against azimuth to obtain a wind speed profile across the rotor disk, should I not?
Best regards,
Vishal
Dear Vishal,
I agree. You should also be able to see/compare how the shear is set through the input file(s).
Best regards,
Dear Jason,
I have plotted the time series of wind speed in X,Y and Z directions against azimuth for the 216 cases (4-26 m/s in steps of 2; 8,0,-8 yaw and 6 turbulence seeds needed for DLC13), but I am not able to very clearly observe the shear.
What did you mean when you said “You should also be able to see/compare how the shear is set through the input file(s).” in your previous post?
Best regards,
Vishal
Dear Vishal,
I meant that the desired shear is directly specified in the input file(s). It sounds like you’ve set a power-low law shear exponent of 0.2 in TurbSim, presumably based on a reference height of hub height. Have you set the same shear profile (exponent, reference height) for the steady wind simulations?
Best regards,
Dear Jason,
The power law exponent was actually 0 for the steady state simulations but the hub height was the same. I will resimulate steady state files with 0.2 power law exponent and make the same comparison again.
Best regards,
Vishal
Dear Jason,
I re-ran the steady state simulations with a power law exponent of 0.2 and plotted the mean values against the mean values for turbulent wind simulations at 0 yaw.
It is the shear that is causing uneven loading over the rotor.
Best regards,
Vishal
Dear Vishal,
Just as I expected; thanks for confirming!
Best regards,
