Thank you very much for your reply,I’m trying to solve the problem of violating the boundary of wind array.
Best regards,
Thank you very much for your reply,I’m trying to solve the problem of violating the boundary of wind array.
Hi Binbin,
Regarding your error: Did you place a turbine (or any output probe) at Y = 1098.7 ? That would be outside your domain in Y direction because your domain seems like between Y = [-500, 500].
As a side note: I can see in your screenshot “OpenMP: No”
When I run FAST.Farm, I get there: OpenMP: Yes.
It looks like when you compiled the code, you did not use the OpenMP flag, meaning that your code cannot run in parallel. Maybe you can compile again with OpenMP, if you want to use multiple turbines, so you can run with multiple CPU’s.
Best regards,
1 Like
Hello,
I downloaded the OpenFAST v3.2.1 from github and compiled in Visual Studio 2017 (using release x64) and it works perfectly for my simulations.
Now, I want to utilize different cores of my PC by running the simulations in parallel using OpenMP. The steps I am following is as below:
First compile the openfast code using release_OMP, x64 in Visual Studio 2017. I have Intel oneAPI Base toolkit, HPC toolkit, DPC++/C++ compiler, Intel Fortran compiler runtime, Microsoft Visual C++ 2015-2022 redistributable (x64). I got the FAST.Farm_x64_OMP.exe in my bin folder. Then I try to use the same input file(.fstf) as I was using without the OMP and I could not get it running. I also tried running through Intel oneAPI command prompt. The program gets called but it doesn’t start the simulation.
I am not sure what is missing but I am not successful yet in making it work. Would greatly appreciate if any suggestions.
Thank you!
Regards,
Devesh
Davesh,
It does actually start with the OMP version you are showing: I see that it started two instances of turbines, but then it looks like it just ends. Typically this indicates some sort of memory access issue.
Have you tried using the pre-compiled OMP executable from the OpenFAST release: Release OpenFAST v3.2.1 · OpenFAST/openfast · GitHub?
If that doesn’t work, you could try building with static libraries. There may be some issue with the libraries in the version of the compiler you are running (or it could be something else, too).
Dear Bonnie,
Thank you for your reply. Yes, I started with the two turbines only for testing. I tried running with the precompiled OMP executable file from GitHub. This time I am getting below errors. I will try checking other things. In the meantime, please let me know if something else comes in your mind.
Thank you for your time.
Regards,
Devesh
Dear @Devesh.Kumar,
To run the precompiled FAST.Farm binary with OpenMP provided by NREL (FAST.Farm_x64_OMP.exe), you will need the redistributable libraries documented here: Releases · OpenFAST/openfast · GitHub.
Best regards,
Thank you for the reply Dr. Jonkman!
I understand I don’t have the version of compilers mentioned in the link to run the precompiled FAST.Farm binary.
But, when I try to build the FAST.Farm (OpenMP) in my PC, that also is not working. I do have those packages (with Visual Studio 2017) as below:
I am not sure what else is required.
Thanks and Regards,
Devesh
Dear @Devesh.Kumar,
My guess is that when you try to execute the OMP version of FAST.Farm that it cannot find the required Intel libraries, which are needed to run the OMP version and are not compiled into the executable statically. My guess is the proper environment paths are not set properly, e.g., you are not executing FAST.Farm from an Intel command prompt.
Best regards,
Dear Dr. Jonkman,
I am executing from the Intel oneAPI command prompt as mentioned in FAST.Farm - #386 by Devesh.Kumar.
Thank you!
Regards,
Devesh
Hi Devesh,
I took a look at the executable for Windows OMP and encountered the same issue. After some debugging, I traced it back to the way the OMP code handled some arrays. I don’t think that code was “wrong” per se (it does work fine on linux systems), but when I wrote it a little differently, it seemed to fix the issue on Windows+Intel.
The executable I built is here: https://github.com/bjonkman/openfast/releases/download/v3.2.1/FAST.Farm_x64_OMP_v2.exe. It uses the same version of the Intel redistributable as the the other binary assets on that release.
The code that I changed is on this branch: GitHub - bjonkman/openfast at test/OMP_FF, if you want to build it yourself.
3 Likes
Dear Bonnie,
Thank you so much for taking time out and looking into it.
I downloaded and compiled the code you gave. I am getting below error now. Do you have any suggestions related to this? Looks like may be some access error on my PC? I checked that location and didn’t find the fort.13 file. Is it related to some installation?
Regards,
Devesh
Dear @Devesh.Kumar,
Can you compile in debug mode, which should hopefully provide more clarity on what the actual issue is?
Best regards,
Davesh,
Do you get this same error every time you run the simulation?
When you open a file for reading or writing in Fortran, you have to provide a unit number. Typically that number is associated with a file name, but if it isn’t, Fortran assumes the file is called “fort.#” where # is the unit number.
I would guess that the OMP version is having some issues with file I/O and not actually supposed to be looking for fort.13. You could try to run the code again and see if this error persists, also try to change what kind of output file(s) you are writing and/or the frequency that it writes the files.
You could also add some print *, "text" statements at various parts of the code where it opens files so that you can pinpoint which files are causing the issue.
Dear Dr. Jonkman,
Thank you for your suggestion. I tried running in the debug mode. The simulation starts and runs as usual (taking more time as expected).
Regards,
Devesh
Dear Bonnie,
Yes, I am getting the same error every time.
Thank you so much for your attention. I will try doing as suggested and will let you know.
Regards,
Devesh
One other thing that may help you is to add debugging information to the Release OMP version you are running. I added it to the Release_OMP | x64 configuration of the FAST.Farm Visual Studio solution file on my test/OMP_FF branch of my OpenFAST fork. It may give you the name of the routine where it is getting the access error when it runs, and that may help figure out what file is causing the problem.
1 Like
Hi,
It is giving errors while creating vtp files. I turned the visualization off but it always creates a vtk folder at the location of fast farm input file (.fstf) and stores these vtp files. Getting error in different files now.
The OMP version ran correctly if I turned OFF the write VTK in the OpenFAST input files (.fst). So, it is working if I don’t save the animation.
Thank you so much for the help, this seems 4-5x faster.
Regards,
Devesh
Dear @Devesh.Kumar,
It looks like it is a known issue that enabling visualization of the wind turbines in OpenFAST can result in an error when OpenFAST is called within FAST.Farm. See: Bug report (FAST.Farm): Fortran runtime error: Cannot write to file opened for READ · Issue #822 · OpenFAST/openfast · GitHub. Hopefully we can get this fixed soon.
Best regards,
Thank you for the reply Dr. Jonkman!
Regards,
Devesh
Dear Dr. Jonkman,
I tried to finish the simulation regarding semisubmersible. However, I encountered a high thrust coefficient under a reference wind speed of 4.0 m and numerical stability with a reference wind velocity of 12.0m. Compared to the original example codes under the steady wind, I made a few changes in ElastoDyn.dat:
WaveDT 0.2→0.25; WaveHs 1.2646→3.75; WaveTp 10→7.50;
WvLowCoff 0.314159→0; WvHighCoff 1.570796→0;
PtfmSurge: 5→0; PtfmPitch: 1.9→0;
The error in the case of 4.0 ms
T1:FARM_InitialCO:FWrap_t0:FAST_Solution0:CalcOutputs_And_SolveForInputs:SolveOption2:SrvD_CalcOut
put:DLL_controller_call:Running with torque and pitch control of the NREL offshore 5MW baseline
wind turbine from DISCON.dll as written by J. Jonkman of NREL/NWTC for use in the IEA Annex XXIII
OC3 studies. Includes modifications for the Hywind spar.
T2:FARM_InitialCO:FWrap_t0:FAST_Solution0:CalcOutputs_And_SolveForInputs:SolveOption2:SrvD_CalcOut
put:DLL_controller_call:Running with torque and pitch control of the NREL offshore 5MW baseline
wind turbine from DISCON.dll as written by J. Jonkman of NREL/NWTC for use in the IEA Annex XXIII
OC3 studies. Includes modifications for the Hywind spar.
Time: 27 of 4000 seconds. Estimated final completion at 13:30:35.
Warning: SkewedWakeCorrection encountered a large value of chi (90.501 deg), so the yaw
correction will be limited. This warning will not be repeated though the condition may persist.
See the AD15 chi output channels, and consider turning off the Pitt/Peters skew model (set
SkewMod=1) if this condition persists.
Warning: SkewedWakeCorrection encountered a large value of chi (90.014 deg), so the yaw
correction will be limited. This warning will not be repeated though the condition may persist.
See the AD15 chi output channels, and consider turning off the Pitt/Peters skew model (set
SkewMod=1) if this condition persists.
Time: 28 of 4000 seconds. Estimated final completion at 13:58:00.
The BEM solution is being turned off due to low TSR. (TSR = -2620.2). This warning will not be
Time: 29 of 4000 seconds. Estimated final completion at 13:34:44.
T1:FARM_UpdateStates:NearWakeCorrection:Wake model is not valid in the propeller-brake region,
i.e., Ct > 2.0.
The selected outputs of 4ms
The error in the case of 12.0 ms
Warning: SkewedWakeCorrection encountered a large value of chi (90.352 deg), so the yaw
correction will be limited. This warning will not be repeated though the condition may persist.
See the AD15 chi output channels, and consider turning off the Pitt/Peters skew model (set
SkewMod=1) if this condition persists.
T1:FARM_UpdateStates:FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_
HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_I
Time: 18 of 4000 seconds. Estimated final completion at 01:18:44.
T1:FARM_UpdateStates:FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_
HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_InputOutputSolve:HydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4
radians.
FWrap_Increment:FAST_Solution:CalcOutputs_And_SolveForInputs:SolveOption1:ED_HD_InputOutputSolve:H
ydroDyn_CalcOutput: Angles in GetSmllRotAngs() are larger than 0.4 radians.
ED_HD_I
The results of 12ms
If possible, pls give me some tips when available. Many thanks
Best regards