Hello, I’m trying to use Turbsim with AnalysisTime set to 600 and UsableTime set to “ALL” to generate a wind field that I can use for a 4200s OpenFAST simulation (1h+600s transient to be discarded). I noticed that the resulting FFT amplitude of the wind velocity signal seems to differ significantly from the one generated with AnalysisTime set to 630, all other inputs the same, even though the time signals look alike:
The statistics of the two signals are also not identical:
I’m using OpenFAST to precompute the aerodynamic loading needed by my frequency-domain model of dynamics, and this discrete-like behaviour of the FFT amplitudes impacts my motion results down the path. Could you help me understand where the difference comes from, and whether this is something that should be avoided, please?
Dear @Kasia.Patryniak,
By changing AnalysisTime between separate TurbSim runs, you are effectively changing the frequency step (df) of the internal FFTs used by TurbSim, which will result in different time series.
The “discrete-like” behavior of the FFT is likely the result of resampling the time series, considering that it appears that you are extracting the wind time series not from the TurbSim output directly, but from the InflowWind module of OpenFAST, which has a a different time step (fmax) and simulation length (df).
I’m not sure I understand your point about a 3600-s simulation + 600-s start-up transient and why you are setting AnalysisTime = 600 s. This implies to me that you should set the AnalysisTime in TurbSim to 3600 s (with UsableTime = “ALL”).
Best regards,
Dear Jason and all,
According to IEC norms, when generating 10-minute wind fields using a TurbSim input file, the turbulence intensity is distributed over the full 600 seconds of the defined AnalysisTime. If this value is increased—for example, to 700 seconds—the turbulence intensity is proportionally distributed across the entire duration.
However, if my simulation runs for 700 seconds and includes a 100-second initialization period, I require a 700-second wind field. Generating only 600 seconds of wind data results in an FFT failure due to insufficient input length.
My question is: would it be possible to generate a 600-second wind field with accurate turbulence intensity and then repeat or reuse part of this wind field to cover the remaining 100 seconds?
I believe my query is relevant to this thread. Thank you in advance for any insights or suggestions.
Regards
Abhinay Goga
Dear @Abhinay.Goga,
I agree with your interpretation.
Yes, this is possible in TurbSim by setting AnalysisTime = 600 s and UsableTime = “ALL”, which causes TurbSim (whose turbulence is always periodic through the internal use of FFTs) to output the full dataset, ensuring its periodicity, which the InflowWind module of OpenFAST will treat periodically. In this way, you can generate 600-s of periodic data in TurbSim and then run simulations longer than 600 s in OpenFAST.
Best regards,
1 Like
Dear Jason and all,
Using the updated TurbSim version and input file, your suggestion worked quite well—thank you for the input.
I generated a 600-second wind array using TurbSim and ran my overall simulation for 700 seconds, accounting for an initial 100-second initialization period. As expected, after the first 600 seconds, the wind array starts from the beginning and covers the remaining 100 seconds. When examining the results, I do not observe any abrupt changes in wind speed at the 600-second mark. This seems to indicate that the initial and final points of the wind array match smoothly, allowing for a seamless loop without discontinuities.
However, this raises a question: If the initial and final points of the wind array are the same (or at least very close), enabling a continuous loop, how does TurbSim maintain the stochastic nature of the turbulence? In my understanding, a purely stochastic signal would not generally start and end with matching values.
I would appreciate your insights on this matter, and please feel free to correct me if I have misunderstood anything in this interpretation.
Best regards,
Abhinay Goga
Dear @Abhinay.Goga,
TurbSim makes use of FFTs, which are periodic in time to generate the synthetic turbulence. The randomness comes in through the use of random phases of each periodic component.
Best regards,
1 Like
