Bug report in inflow wind openfast develop

[ptrbortolotti]

Bug description
Inflow wind on the develop branch suffers when .wnd files are loaded with a linear vertical shear (DLC 1.5)

To Reproduce

Steps to reproduce the behavior:

1. Compile develop with fortran
2. Run inflow wind WindType=2
3. Check outputs RtVavgxh
4. See the error

Expected behavior

Screenshots, if applicable

OpenFAST Version

**************************************************************************************************
 OpenFAST

 Copyright (C) 2020 National Renewable Energy Laboratory
 Copyright (C) 2020 Envision Energy USA LTD

 This program is licensed under Apache License Version 2.0 and comes with ABSOLUTELY NO WARRANTY.
 See the "LICENSE" file distributed with this software for details.
 **************************************************************************************************

 OpenFAST--128-NOTFOUND
 Compile Info:
  - Compiler: GCC version 9.2.0
  - Architecture: 64 bit
  - Precision: single
  - Date: Dec 15 2020
  - Time: 11:40:53
 Execution Info:
  - Date: 12/18/2020
  - Time: 18:45:01-0700
[outputs.zip](https://github.com/OpenFAST/openfast/files/5718760/outputs.zip)
fast d

**Additional context**
<Add any other context about the problem here.>

[ptrbortolotti]

This model (BAR design) with the wind file generating the error is attached

outputs.zip

[bjonkman]

@ptrbortolotti: When I tried to download the zip file you attached, my computer tells me it contains a virus or malware, so I can't actually check the model. (Perhaps a controller is included inside the zip file?)

I see only the AeroDyn output channel here, so I am wondering what checks have you done to say that this is caused by the linear vertical shear in InflowWind's uniform wind file and not something in AeroDyn or ServoDyn (or elsewhere).

[ptrbortolotti]

Thanks Bonnie for the support here. I'm running some more tests and I will upload new files. I had turned off servodyn, so that's out of the game, but now I do see changes in the outputs varying the aerodyn15 flags. I'll get back to you as soon as I have isolated the problem better. Thank you!

[ptrbortolotti]

outputs.zip

[ptrbortolotti]

Hello OpenFAST team, I've zipped the model again removing anything that is not a .dat file and a .wnd file.
I hope you can now download the folder
I ran some tests myself and everything works fine when WindType=1. When I switch to WindType=2, the blade deflections grow unrealistically and something is up. The problem could clearly be in the numbers of the .wnd file, but things look fine to me there
I'm happy to run any additional test you might suggest
Thank you!
Pietro

[andrew-platt]

I don't see a uniform wind file in that zip. There is a binary turbsim file (BAR01_3ETM_U4.000000_Seed1.0.bts), but nothing matching the format for the Uniform Wind file (WindType 2).

[ptrbortolotti]

that is surprising.. the folder should have two subfolders: "bar0" with the openfast model and "wind" with a .wnd file, which I attach here after changing the extension to .txt
bar0_EWS_V_P_U15.0.txt

[andrew-platt]

Thanks for the file. I'm not sure why it isn't in the zip (3 posts above).

Could you try adding another column of zeros at right side of the table? I wonder if we have a parsing error when only 8 columns are present instead of the expected 9 (the code in theory should handle this ok, but might have a bug).

[ptrbortolotti]

I don't think that has changed anything.

These are the plots when I use WindType=1 at 15 m/s and PLexp=0.2

And this what happens when I set WindType=2 and I load the file above

[andrew-platt]

I checked the bar0_EWS_V_P_U15.0.txt file with the IfW driver code and a grid of outputs. At 16.5 seconds, I see the following:

    # X               Y                 Z               Vx             Vy            Vz
     0.0000000      -20.0000000      120.0000000     -285.673        0.000        0.000
     0.0000000        0.0000000      120.0000000     -285.673        0.000        0.000
     0.0000000       20.0000000      120.0000000     -285.673        0.000        0.000
     0.0000000      -20.0000000      140.0000000       15.000       -0.000        0.000
     0.0000000        0.0000000      140.0000000       15.000       -0.000        0.000
     0.0000000       20.0000000      140.0000000       15.000       -0.000        0.000
     0.0000000      -20.0000000      160.0000000      315.623       -0.000        0.000
     0.0000000        0.0000000      160.0000000      315.623       -0.000        0.000
     0.0000000       20.0000000      160.0000000      315.623       -0.000        0.000

Are the values for linear vertical shear sensible? It is set to 1.000724 with an additional PL exponent of 0.2.

[bjonkman]

Was this a problem in the master branch (or whatever branch you were using)? i.e., is this a new issue?

Or could this be related to something in BEM that is triggered because of the shear you are adding?

I haven't looked in great detail, but one thing that sticks out to me is the RefLength used for the uniform wind files. It's set to 1 meter in the file you sent, while I would expect this to be closer to the rotor diameter. In your case, when the vertical linear shear is set to 1, you are probably adding hundreds of m/s to the velocity at the top of the rotor (and subtracting similar amounts at the bottom of the rotor). There is probably something weird happening in the aero with those kinds of velocities.

[ptrbortolotti]

Thank you Bonnie, it looks like the reference length is not setup correctly in WEIS! I'll keep this issue open for a little longer, but most likely the error is on our side then!

[ptrbortolotti]

hello openfast team, thank you for the great support. Bonnie was right, RefLength wasn't initially correctly in WEIS and it was left at a dummy value of 1 meter. This PR is fixing things on our side (together with some other changes)

WISDEM/WEIS#48 (comment)

thanks again, I'm happy to close this issue now