openfast
openfast copied to clipboard
OpenFAST
FAST.Farm not outputting properly for Turbines 7 - 10
Bug description When running a FAST.Farm simulation with 10 turbines, the output files for turbines 7,8, and 9 are empty. The output file for turbine 10 has four entries for each timestep (assuming the entries for T 7,8, and 9). See screenshot
To Reproduce Steps to reproduce the behavior:
- Compile with FAST.Farm_x64_OMP.exe
- Run 10 turbine simulation
- Open output files for T7 - T10
- See the error
Expected behavior The output files for turbines 7,8, and 9 should have an output line for each time step. Turbine 10 should have one output line for each time step instead of 4.
Screenshots, if applicable
OpenFAST Version
FAST.Farm
Copyright (C) 2021 National Renewable Energy Laboratory Copyright (C) 2021 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.
FAST.Farm-v2.5.0-1214-gdc92d7c1-dirty Compile Info:
- Compiler: Intel(R) Fortran Compiler 1700
- Architecture: 64 bit
- Precision: single
- OpenMP: Yes, number of threads: 128/128
- Date: Jul 29 2021
- Time: 13:12:38 Execution Info:
- Date: 10/25/2021
- Time: 10:29:52-0600
System Information (please complete the following information):
- OS: Windows
@ebranlard were you commenting on this yesterday during the OpenFAST meeting?
@erickaloz -- Can you clarify what compiler you are using? It looks like the header is saying Intel, but in you post you mention GFortran. I also see that you are using an older version than what has been released (v2.5 versus v3.0); I'm curious if you see the same problem using FAST.Farm within OpenFAST v3.0. I'm also curious if you see the problem using the precompiled FAST.Farm executable for Windows provided with the OpenFAST v3.0 release? NREL has provided precompiled versions both with (FAST.Farm_x64_OMP.exe) and without (FAST.Farm_x64.exe) OMP parallelization from here: https://github.com/OpenFAST/openfast/releases. Do you experience the same problems with these versions of FAST.Farm?
@andrew-platt -- I think it was actually @mattEhall that made this comment during yesterday's meeting.
@mattEhall -- Is this the same problem you were seeing?
@jjonkman, it appears to me to be v2.5 + 1214 commits. So this is a version after v2.5 -- though I'm not sure which (can't find commit dc92d7c1 in the logs).
@erickaloz which branch did you complile?
@ebranlard were you commenting on this yesterday during the OpenFAST meeting?
No I'm not aware of this issue, but I haven't tried running more than 3 turbines in a while.
@jjonkman Yes sorry I think I am using the Intel Compiler. I just tried with OpenFAST v3.0.0 and am seeing the same issue. Here is the header:
FAST.Farm
Copyright (C) 2021 National Renewable Energy Laboratory Copyright (C) 2021 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.
FAST.Farm-v3.0.0 Compile Info:
- Compiler: Intel(R) Fortran Compiler 1900
- Architecture: 64 bit
- Precision: single
- OpenMP: No
- Date: Jun 23 2021
- Time: 21:41:28 Execution Info:
- Date: 10/26/2021
- Time: 11:22:55-0600
@jjonkman, it appears to me to be
v2.5 + 1214 commits. So this is a version after v2.5 -- though I'm not sure which (can't find commitdc92d7c1in the logs). @erickaloz which branch did you complile?
@andrew-platt I was using @mattEhall custom version but just tried again with v3.0.0 from https://github.com/OpenFAST/openfast/releases and am seeing the same issues
I am not able to confirm this bug on either Mac or Linux with either Intel or GCC compilers with a 12 turbine test case (development branch based on dev).
I'll try with the USFLOWT branch and see if I can reproduce it with that.
Hi all, this is the problem I mentioned yesterday in the meeting. We have seen this issue previously with the stock v3.0.0 FAST.Farm downloaded from the release as well as with the updates I've made in MDv2-farm, which I compiled with Intel Fortran. It occurs regardless of parallelization.
In the MDv2-farm version it occurs regardless of whether the shared mooring mode is on or not. When the shared-mooring mode is on, the time steps of each FAST instance are synchronized, so the outputs of Turbines 7-10 are combined in a consistent order in the one output file. In the normal mode, each FAST instances runs at its own pace for each FAST.Farm time step, so the order of the output lines from Turbines 7-10 in the one output file are unpredictable.
One note is that we have only looked at the .out files so far and have not checked if it occurs in when using binary output.
