This repository contains the Python code for analysing vertical wind profile patterns in any dataset containing time series of wind speeds and directions for multiple altitudes. This analysis is the backbone of the Clustering wind profile shapes to estimate airborne wind energy production paper, which has been published in Wind Energy Science.
The code has been originally developed for analysing the Dutch offshore wind atlas (DOWA) dataset. The DOWA file reading functionalities are compatible with the time series files from 2008-2017 at 10-600 meter height at individual 2,5 km grid location. Additionally, file reading functionalities are provided for the raw output files of the Leosphere WindCube v.2.1.8 lidar. Measurements of this machine at a location near Köln in Germany for the first three months of 2020 are provided by GWU Umwelttechnik and are under analysis by an airborne wind energy (AWE) resource consortium, which aims for developing AWE system design load case standards. A request to publish the hour-averaged measurement in this repository is pending.
The code is tested in an Anaconda environment with Python 3.9.1, which can be created using the lower command:
conda create --name [env_name] --file requirements.txt python=3.9.1
replacing [env_name] by a name of your choice. Download the DOWA files of the desired location. Point with the data_dir variable in dowa.py to the download directory. In main of wind_profile_clustering.py change the grid point coordinates that are passed to the read_data function to those of the downloaded location. Activate the new environment:
conda activate [env_name]
Finally, run the script to perform the clustering:
python wind_profile_clustering.py