after simulating a 600 seconds simulation of a full field turbulent wind OpenFAST simulation and obtaining loads, is it possible to resimulate OpenFAST with lets say only 200 seconds of the exact same wind file?
Of course the simulation length can just be shortened, but if let say the turbulent wind of interest is in the range of 300 to 500 seconds of the wind timeseries, how can the OpenFAST start be shifted?
The reason why I am asking is, that for instance BeamDyn requires often still a very small timestep. Thus, it would mostly be more practically in load simulation to just resimulate for each sensor the related turbulent timeseries around the highest occurrances of load for this sensor, which has been obtained before with less fidelity simulation with ElastoDyn. The drawback is that this does not work for fatigue loads.
The only promising parameter to do this, which I have found, is the XOffset parameter in InfloWind input file. It describes the initial offset in +x direction (shift of wind box) using the HAWC-format binary files.
Please let me know if it is possible via the XOffset or any other parameter.
I agree that adding support for an input like
WindType = 5 (HAWC format) in InflowWind would be useful to add for other turbulent wind file formats including Bladed and TurbSim. This is not a feature currently available, but I would suggest adding it as a feature request on OpenFAST issues: Issues · OpenFAST/openfast · GitHub.
FYI: We are working on addressing the speed of OpenFAST when BeamDyn (or SubDyn) are enabled by introducing a tight coupling algorithm in OpenFAST around the structural modules (ElastoDyn, BeamDyn, SubDyn). I presented the status of this development last month at WESC and plan to present the final results at the NAWEA / WindTech 2023 conference in October.
Hi Jason Jonkman,
I was able to apply an offset of about 200 seconds for 10 m/s mean wind turbulent HAWC format wind file with the Xoffset parameter set to 2000 m in the InflowWind file.
However, the resulting time series in OpenFAST does show that there is a minimal variation in the maxima and minima, see attached plot and related statistics below:
Is there a specific source for this deviation? We would have expected it to be exactly the same if the turbulent wind time series are just shifted with the Xoffset.
Thank you for the update regarding the speed of OpenFAST if coupled with BeamDyn or SubDyn. This is indeed a very much appreciated development.
I would expect that you’d see exactly the time shift you’ve specified if you look at ambient wind-speed related outputs. But turbine-related outputs will also be impacted by the turbine operation, specifically different orientations of the rotor azimuth angle, which will impact how the rotor interacts with the wind.
Dear Jason Jonkman,
I am following the update of the coupling algorithm for the BeamDyn module with high interest. Unfortunately, I could not find any updates on the NAWEA conference website. Is there a presentation or paper or any other document available regarding this? Does NREL have a time plan when to incorporate these updates into the OpenFAST release, and is the code already available somewhere as development version? I could not find anything on OpenFAST Github yet.
We have made good progress on the implementation of a new tight coupling algorithm in OpenFAST and presented this update at NAWEA / WindTech 2023. The algorithm produces numerically stable results at larger time steps, resulting in dramatic speed ups of 10-100x faster execution time, depending on the model. The algorithm currently supports the InflowWind, AeroDyn, ElastoDyn, BeamDyn, ServoDyn, and SubDyn modules and we are working on supporting the other modules, including for offshore wind turbines. You can track the development (and make use of the capability) in the following pull request within the OpenFAST GitHub repository: Tight Coupling Solver by deslaughter · Pull Request #1850 · OpenFAST/openfast · GitHub.
Once the implementation is fully complete, we will complete our testing and submit a journal paper for publication. We are planning to release this capability in OpenFAST v5.0, currently planned for March 2024.