39 Moritz Diehl K.U. Leuven Department of Electrical Engineering ESAT/SCD, OPTEC Kasteelpark Arenberg 10 bus 2446 3001 Heverlee Belgium moritz.diehl@esat.kuleuven.be http://homes.esat.kuleuven.be/~mdiehl/
Automatic Control and Rotation Start of a Rigid-Wing
AWE System in Pumping Mode
Moritz Diehl*
KU Leuven
This talk presents the airborne wind energy (AWE) research at the University of Leuven that focuses on the automatic control and start-up of tethered rigid wing systems. Airborne wind energy systems with rigid wings promise high power output per wing area, good durability, reliable controllabil-ity in all weather conditions, and the possibilcontrollabil-ity to build on existing aircraft technology. However, automatically start-ing and landstart-ing them is a nontrivial task due to the fact that they typically need relative airspeeds higher than the wind speed to overcome gravity. In case of on-board turbines that can be operated in propeller mode, vertical take-off and land-ing (VTOL) is a possibility, but for a pumpland-ing AWE system without on-board turbines/propellers, the only possibility is to bring the airplane to sufficiently high speeds. Fortunately, this can be achieved by a rotation setup (see the left figure for the current setup at KU Leuven) that consists of a small rotating arm – a carousel - from which the main tether ex-tends to the airplane. Rotating the carousel will allow one to bring the airplane to sufficient speed in a small volume, and the tether length can then slowly be increased until tran-sition into power generating orbits is possible (see the right figure for an optimized power orbit in simulation). In addi-tion, a carousel setup is ideally suited to do indoors and out-doors testing of advanced control systems without the need to have large testing sites. The talk describes the experimen-tal setup and the control technology developed within the ERC Project HIGHWIND at the KU Leuven. In particular, we describe how nonlinear simulation models can be used not only for system and trajectory optimization, but also for on-line state and parameter estimation and for feedback control,
using the concepts of moving horizon estimation (MHE) and nonlinear model predictive control (NMPC). We show experi-mental flight results using MHE and NMPC and compare the flown trajectories with simulations.
*Joint work with Jan Swevers, Sebastien Gros, Andrew Wagner, Kurt Geebelen, Milan Vukov, Mario Zanon, Greg Horn, Joris Gillis