Design of a Real-Time Micro-Winch Controller for Kite-Power Applications

Airborne wind energy is a technology to extract energy from high altitude winds. This technology is under heavy development by several companies and universities. An actual problem with the commercialization of the technology is the reliability and safety of the system. In this thesis a real time environment suitable to perform research and further development of the prototype steering and depower control is proposed. Additionally, the overload prevention of the kite lines is researched. This thesis presents a method to estimate the tension on the kite control tapes using only one tension sensor. Thus, reducing the amount of hardware needed to protect the kite from overloads. The method relies on the characterization of the powertrain efficiency and can be used to estimate the tensions at high loads. An algorithm to limit the forces on the steering lines by depowering the kite is shown; it controls the depower state of the kite based on the desired depower state, the actual tension, and previous tensions on the KCU’s tapes. The tensions history is used to calculate a higher depower state to prevent future overloads, this reduces the amount of action needed by the motors and enable the system to use a brake to save energy. The limiter output is used as an input to a position controller, which allows the project to use off the shelf solutions to build the KCU prototype. The controller was implemented in a real time system and is able to run as fast as 20 Hz being the communication protocol the execution time bottleneck. The control algorithms were tested using a mathematical model of the kite, the environment, and trajectory control inputs from FreeKiteSim. Three scenarios were considered for the model test, normal operation, overload operation without tension limitation, and overload operation with tension limitation. The apparent wind speed during the reel out phase of the normal scenario is approximately 30 m/s and 35 m/s for the overload scenarios. During the overload scenario the limiter spent roughly 22% more energy than the normal operation scenario to counteract an increase of 5 m/s in the apparent wind during 3.5 hours of operation, but it spent 15% less energy than the overload scenario without tension limitation.