Meta-parametrised meta-modelling approach for optimal design of power electronics conversion systems: Application to offshore wind energy

Abstract

In an offshore environment, the efficiency (η), the power density (ρ) and the powerto- mass ratio (γ) are of paramount importance in the design of wind energy conversion systems (WECS). Indeed, the optimisation of these performance indices can reduce investment costs, especially if most of the electrical conversion components are located in the nacelle or tower of the wind turbine. This thesis describes a simple procedure to calculate the η, the ρ and the γ of power converters via the calculation of power losses, volume and mass of the main components of the WECS: the power electronics valves, the magnetic components and the capacitors. In the proposed method, the system is first characterised, then a set of figures of merits are evaluated for a set of design parameters. Finally, a multi-objective optimisation is performed and the Pareto concept is used to present the set of solutions (for different sets of parameters) and the best trade-off for the performance indicators is identified. The approach does not identify a unique solution; instead, several solutions are obtained and other criteria can be used to choose the final solution, thus giving freedom in the design process and flexibility in the final decisions of conceptual design. The proposed method is applied to different offshore WECS topologies to illustrate the evaluation procedure. First, a well known topology used in AC-Grid connected wind turbines, the two level voltage source converter, is considered. Then, the meta-parametrised approach and the fundamental component models described in this thesis are used to compare six different WECS based on a medium frequency transformer for a 10MW WECS interfacing a permanent magnet synchronous generator suitable for offshore DC-grids. A modular approach in the power converter is considered and the impact of the number of modules and variation in nominal transformer frequency on performance indicators is studied. The proposed approach can be applied to identify key components/parameters to optimise a standard solution, to quantify the impact of a new component technology on state-of-the-art solutions for a given application, or to determine, at least in part, the best solutions for new applications regarding state-of-the-art converter topologies, modulation strategies, component technologies and system parameters.