I have noticed that, when reading turbulence files generated internally by HAWC2, the turbulence is incorrectly scaled when read by AeroDyn. I obtain better results when I modify HAWCWind.f90 to scale all values from the binary file by Uref * tint/4.0. (I added the turbulence intensity tint as an input in the .hwc file, and it is the value specified in the HAWC2 input file.)
Has anyone else encountered similar scaling issues?
Note: I have not examined files produced by the external Mann turbulence simulator, just those generated by the internal HAWC2 turbulence simulator.
The AeroDyn/InflowWind subroutines were developed using the IEC Turbulence Simulator and its documentation, which indicates the values in the binary file aren’t scaled at all. One of the reasons this feature is still undocumented is that I was never able to fully resolve a question I had with the order of the values on the x (time) axis, but it seemed to scale to the correct wind speeds. I do not have a copy of HAWC2 so I have not tested the routines with files generated by it.
You said you obtain “better” results with the scaling you developed. How are you comparing the values? Hopefully there is a way we can test and get the same values. Or perhaps we could compare the files generated by the IEC Turbulence Simulator with ones generated by HAWC2.
I’d recommend asking the HAWC2 developers if the files are being scaled (or if there is a new format they are using). If they aren’t scaling, could the issue be with the wind profile that is added to the turbulence in the file? (I don’t think they add the mean wind speed to the binary files, but perhaps they do.)
I have been comparing time series of the hub-height wind speed using my approximate correction, and there is still some error in the values (generally <5%). It appears to be a multiplicative error in the turbulent component, not in the mean, but it seems that my approximation isn’t perfect. I wouldn’t expect the wind shear effects to show up in the hub-height components (hub-height = reference height), but I haven’t checked that effect at all.
You’re comparing time series output of the hub-height wind vector from two aerolastic simulations? And you’re comparing the same components of undisturbed wind (i.e., as if the turbine wasn’t there) using the same coordinate systems?