Initial Condition for Wind Directions #3226
Replies: 2 comments 6 replies
-
|
Dear @Parrotyeon, Can you clarify, when you change the wind direction, are you also changing the nacelle-yaw direction to align the rotor with the wind? Regarding your specific questions:
Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear , @jjonkman Yes Dr. Jonkman, I set NacYaw (ElastoDyn) and YawNeut (ServoDyn) to the negative of PropagationDir (InflowWind) based on discussions in the NREL forum. For the mean values, I extracted them by truncating the time-series data from a certain time to the end of the simulation. I suspect that this approach may have introduced some errors, as even small discrepancies in the initial conditions seem capable of triggering fatal errors. Recently, I have been running previously failed cases using a wind ramp-up approach and was able to identify more appropriate initial values for each wind speed and direction. I will further validate the interpolation method and provide an update in this thread. I also have another question. I noticed that cases with slightly inaccurate initial conditions (typically leading to a “0.3 Mach” fatal error) show significant instability in the control response. In particular, the blade pitch angle oscillates rapidly between 0° and 90°. My hypothesis is that an inappropriate initial condition may cause the blade nodes to experience an instantaneous large inflow velocity, which in turn destabilizes the pitch control system. Does this interpretation seem reasonable? Lastly, I am having difficulty determining the static equilibrium position of the 5MW semi-submersible model under still water and no wind conditions. I would expect the system to settle into a negatively pitched equilibrium configuration due to the offset between the center of gravity and center of buoyancy. However, even after a long simulation time, the platform continues to exhibit very small-amplitude oscillations. I would have expected the response to fully converge to a steady state due to system damping. Could this residual oscillation be related to the coupling methodology among submodules (e.g., ElastoDyn, HydroDyn, MoorDyn, etc.), or might there be another underlying reason? Those are results from 50,000 sec simulation. I disabled GenDOF to reduce the effect of rotating blade. 
I would greatly appreciate your advice. Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @Parrotyeon, Thanks for clarifying. The control is often sensitive to the initial conditions of rotor speed and blade-pitch angles and the use of poor initial conditions can result in the controller instability you are describing. You mention "slightly inaccurate", which sounds more sensitive than I would expect, but perhaps you can clarify exactly what you seeing. Regarding the static equilibrium, all of the plots you are showing are quite zoomed into a small range of values. So, it seems like you already know what the static equilibrium condition is (I would not think small fractions of meters or degrees in initial platform displacements would make any difference on the solution). I'm not sure I fully understand what is driving this in your case, but I suspect these would be reduce if you switch from single to double precision, reduced the time step, and/or added a correction step. Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @jjonkman, Thank you for your advice. Specifically, in the case of still water, 12 m/s wind speed, and a 50° wind direction, I found two different initial conditions: one case runs successfully, while the other fails. [Successful]
[Abort]
Error message:
Based on the error message, I suspect that an instantaneous increase in the local inflow wind velocity, caused by an inappropriate initial condition(especially blade pitch angle), triggered a control instability. In other words, the rotor may have experienced a sudden change in aerodynamic loading, which caused the pitch controller to behave abnormally. Does this interpretation make sense? (And I also found out that the OpenFAST runs well smaller blade pitch angle(0.9xxx deg) under 11.9 m/s 0~ 40 deg.) Regarding the static equilibrium, I observed a mean offset in the pitch and surge directions under regular wave conditions. I was curious about what might be causing this mean offset. Increasing the number of correction iterations did not improve the result, so I plan to revisit this issue later. I also plotted the initial-condition results. The trend seems reasonable: the surge displacement increases with wind speed, reaches a peak around the rated wind speed, and then decreases afterward. I assume the non-zero value at a 90° wind direction and more larger surge displacement when it comes to 180 deg wind may be related to the triangular configuration of the mooring system. Additionally, I noticed that the rated wind speed and blade pitch angle do not perfectly match the values reported in the paper “Definition of a 5-MW Reference Wind Turbine for Offshore System Development.” I suspect this discrepancy might be due to platform motion, which can modify the effective inflow wind velocity at the rotor (e.g., through vertical motion affecting the rotor inflow). Would this be a reasonable explanation, or is it possible that I have run OpenFAST under incorrect conditions? Thank you for your time and advice. Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @Parrotyeon, Thanks for clarifying. That is more sensitivity than I would expect for such small changes in initial conditions. But I'm also not sure why your initial blade pitch is above zero for rotor speeds below the rated speed of 12.1 rpm, which I would have expected with the baseline controller for the NREL-MW wind turbine, which is what I assume you are using. Does increasing I agree that the initial surge conditions you are plotting match my intuition. The turbine response may not match exactly with the values in NREL 5-MW specifications report because of small changes to FAST/OpenFAST over time and because on a floating platform, the heal of the platform can increase the effective shaft-tilt angle. Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @jjonkman , Thank you for the advice 😄 Yes, I'm using vanilla "Discon_OC3Hywind.dll" for the simulation. Based on my simulation results, I found that when I set the initial blade pitch angle to 0 deg, the simulation no longer produced the fatal error that occurred with the previous initial conditions. However, I noticed that even though the initial blade pitch angle is 0 deg, the blade pitch converges to approximately 1.5 deg in the steady state. Is there any reason why the steady-state blade pitch angle settles around 1.5 deg, even though it's under rated wind speed? Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @Parrotyeon, My guess is initializing the blade-pitch above rated with the rotor speed below rated is confusing the controller, resulting in an instability, while dropping the blade-pitch to zero (rated and below) is OK. Nevertheless, it looks like your solution is converging to 1.5 deg pitch at 12.1 rpm, which is probably also OK to start with. At 12 m/s, this wind speed is above rated. Best regards, |
Beta Was this translation helpful? Give feedback.
-
|
Dear @jjonkman , Thank you very much for your helpful explanation. I really appreciate it. Best regards, |
Beta Was this translation helpful? Give feedback.
Uh oh!
There was an error while loading. Please reload this page.
-
Dear all,
I am currently working on wind–wave misalignment cases using metocean data covering wind speeds from 0 to 25 m/s with various wind directions.
From my simulation experience, selecting appropriate initial conditions is very important — not only to minimize transient response time, but also to avoid potential fatal errors in OpenFAST.
To address this, I have been computing initial conditions under steady wind and still water conditions at every 10° increment of wind direction (from 0° to 180°) for each wind speed. However, my metocean dataset has a finer directional resolution, and I am unsure how to properly estimate initial conditions for wind directions that were not explicitly computed (for example, 53°).
Previously, I applied a simplified assumption:
I kept all states the same except for surge, sway, roll, and pitch, which I rotated as a vector according to the change in wind direction.
For example, if the original displacement was
[surgeo,swayo]=[8,0],
then for a 10° change in wind direction, I updated it as
[surgeo,swayo]=[8,0],[surgen,swayn]=[8cos(10∘),8sin(10∘)].
Then I ran OpenFAST until the mean values converged. However, this approach seems overly simplified, and in practice it required at least two iterations to reach convergence.
Therefore, I am considering constructing a more accurate estimation method by interpolating the initial conditions between two adjacent wind directions.
I am not fully confident that this is the correct approach. I also noticed that OpenFAST does not appear to provide a direct static equilibrium solver. I briefly considered using the linearization process, but I have not yet fully understood how it could be applied for this purpose.
Could anyone kindly provide advice on:
Any guidance would be greatly appreciated.
Best Regads,
Beta Was this translation helpful? Give feedback.
All reactions