I am trying to generate a turbulent wind file using Turbsim for some yaw control tests I am currently working on, and I need the wind file to cover a large range of wind directions. So is there a way of customizing statistical variables on the wind directions, e.g. standard deviation and upper/lower limits etc., before the wind file is generated?
Hope to hear from you soon, any help will be appriciated.
There is a wind direction input in TurbSim that changes the mean wind direction, but the direction must be constant for the wind file. I don’t recommend using this feature with grids to be used in AeroDyn because the turbulence grids march through the turbine, in the x-direction, at the hub-height mean wind speed, regardless of the mean wind direction. We recommend that people use an initial yaw error in their turbine simulations instead of the wind direction feature in TurbSim.
If you want to change the wind direction during a simulation, you can use hub-height (hh) files, but these files do not have spatial turbulence.
Thanks very much Bonnie for your help.
this subject is of interest also for us , exactly for the same reasons of Vicki (some yaw control algorithm simulations). I will not use the generated data for input to Aerodyn, but to a MathCAD application written by myself.
I read carefully the TurbSim input description, but I see that the input about the direction is just a constant value (if I understood correctly).
So I reformulate the question the following way: can you point me to some method (even algorithmic, that I could implement myself) that can reasonably simulate wind direction variations 360ï¿½ around, on a time scale of minutes to hours?
My idea would be to add to the hh files “WindDir” column, that represent small period direction variations, a longer period term, this I could do for example in MathCAD (my application reads the .hh file produced by TurbSim).
My proposal for the long term would be something like a stochastic walk, that is: adding to a running sum a small “step” that is taken randomly from a given gaussian probability distribution. I could not find any paper addresing this problem.
thanks a lot to all of NWTC for this forum and for the sharing of knowledge!