Stitch Together TurbSim Timeseries

[Sam.Kanner]

Jason/Bonnie/whomever,

We are interested in writing a wrapper for TurbSim that allows us to smoothly stitch together multiple wind timeseries that are generated from TurbSim. For instance, I want to generate a 1-hr timeseries with an avg wind speed of 10 m/s and then I want to add a second hour where the avg wind speed is 12 m/s. The question is making the transition between these timeseries physically representative. I'm wondering if anyone within NREL has done this previously or if there are some hidden features within TurbSim v2.0.0 that could enable this to happen.

I understand from another topic that you can generate periodic files by

Simply set TurbSim input UsableTime to the string, "ALL" and run TurbSim to generate periodic wind data files useable by FAST.

Source: https://wind.nrel.gov/forum/wind/viewtopic.php?f=13&t=1928&p=10012&hilit=extend+10+min#p10012

It seems like in making it periodic, you have constrained the timeseries such that U(t=0) == U(t=end). Perhaps this can be extended such that U(t=0), U(t=end) can be specified by the user, but I know there will be issues in ensuring that the statistical properties of the wind profile are kept.

Thanks for the help!

[Jason.Jonkman]

Hi Sam,

TurbSim essentially works by performing an inverse FFT to derive the turbulent fluctuations (about zero mean) and superimposing a mean wind speed profile (mean wind speed + shear/veer). The inverse FFT is naturally periodic (with zero) mean, and the UsableTime = "ALL" option simply ensures that all data is written to the file so that the TurbSim ouput is periodic. The wind speed profile naturally has no time variation, but you could change that so that the mean wind speed profile changes over time. This may be easiest approach to take, if you are OK with having the same turbulent fluctuations at different mean wind speeds. Would that solve the problem you are trying to solve?

Best regards,

[Sam.Kanner]

Hi Jason,

Thanks for the quick response. Yes, exactly-- we would want to be able to define how the mean wind speed (and also mean wind direction) profile changes over time. Ideally, we would also want to be able to define say, a lookup table that relates TI back to wind speed and wind direction. So in addition to specifying how the mean wind speed/direction evolves over a time, we could also specify how TI evolves.

But as a first pass we could just keep the TI constant for different mean wind speeds (and directions).

How do you recommend implementing such a solution?

Thanks

[Jason.Jonkman]

Hi Sam,

Can you clarify why you want to run a simulation with time varying mean wind speeds and directions? What is normally done is to simulate different mean wind speed and directions in separate FAST / OpenFAST simulations.

As I said in my prior post, TurbSim creates the turbulence (with zero mean) and superimposes that on a mean wind profile. If you want to change that, you'll have to change TurbSim. I'm not an expert on the TurbSim source code, but the superposition happens in SUBROUTINE AddMeanAndRotate(), which may be a good place to make your change. Please keep in mind that different TurbSim output file formats store the data differently, e.g. the TurbSim binary format (*.bts) stores the full 3D wind field (2 spatial + time), but it also stores the fixed mean wind speed at hub height as a parameter. And the Bladed format (*.wnd) stores the full 3D wind field, but with the mean hub-height wind speed removed (for compatibility with Bladed).

Even if TurbSim is changed so that the wind profile changes as a function of time, the use of this wind data in InflowWind will be approximate. This is because InflowWind creates a 4D wind field (3 spatial + time) from the 3D TurbSim-generated wind field (2 spatial + time) using Taylor's frozen turbulence assumption, i.e., the planes of 2D wind are propagated at a given fixed mean wind speed and direction, the wind speed of which is the parameter stored in the TurbSim output. It would be difficult to make this value time varying.

Best regards,

[Sam.Kanner]

Great-thanks for the tips! It looks like you are all in the midst of re-organizing where your software is stored on your website. Can you point me in the right direction of where to find the source code for the alpha version?

We're not looking to use this timeseries for FAST / OpenFAST simulations.

[Jason.Jonkman]

Hi Sam,

The version of TurbSIm in the OpenFAST repository (https://github.com/OpenFAST/openfast) is slightly updated from the alpha version of TurbSim on the old website. However, the documentation and tests from alpha archive of TurbSim have not yet been migrated to OpenFAST, as noted in OpenFAST issue #176: https://github.com/OpenFAST/openfast/issues/176. I can send you the alpha archive of TurbSim if you need it; just let me know.

By the way: even if you are not using FAST / OpenFAST simulations, I would expect similar software (Bladed, HAWC2, OrcaFlex, etc.) would have the same issue regarding use of Taylor's frozen turbulence assumption.

Best regards,