The FreeKiteSim project aims at the development of a free kite-power system simulator for educational purposes and research. It is inteded to be used to:
- Provide a learning environment for students and PhD researchers who are investigating airborne wind energy.
- Train kite pilots and winch operators. In this case kite and winch are operated with joysticks.
- Develop and test automatic control software for kite-power systems.
For more information, visit the project page on bitbucket at https://bitbucket.org/ufechner/freekitesim and subscribe to their mailing list to stay updated on the latest chagnes https://groups.google.com/forum/#!forum/free-kitesim
The most striking improvements which the latest version of FreeKiteSim brings are a 6-fold speed increase of the simulator which was achieved by updating the code to use numba 0.18 and reimplementing the core functions such that they do not have to allocate memory for temporary calculation results.
Furthermore a bug was fixed which makes the tether drag calculation significantly more accurate. The underlying calculations are explained in the paper “Dynamic Model of a Pumping Kite Power System” which was accepted for publication in Renewable Energy. The paper can be downloaded at http://arxiv.org/abs/1406.6218 .
Apart from the simulations itself we also updated the installation script and uploaded a meta-package to https://conda.binstar.org/ufechner, our conda package server. This way if you are already using conda the installation can easily be done by getting the freekitesim package. Obviously the installation script can still be used as a clean way without the whole conda environment system.
Here is the full list of Changes:
- Updated the code for the use of numba 0.18.2 leading to much faster execution and a large decrease of the startup time
- Reimplemented the core functions to not allocate memory for temporary arrays leading to an execution time decrease by a factor of 3, making the code 6 times faster in total
- Code enables the use of different wind profile laws
- Fixed a bug in the tether drag calculation which should be much more accurate now
- Next to the installation script the installation can be done by using a conda package
- Uses newest packages except for pandas and cgkit which need to be pinned with conda.