Supervisor:
Mohsen Soltani

Co-Supervisor:
Amin Hajizadeh

Assessment Committee:
Frede Blaabjerg (Chair)
Professor Mehdi Zadeh, Department of Technology, NTNU, Norway
Associate Professor Christoffer Sloth, SDU, Denmark

Moderator:
Mathias Mandø

Abstract:

To reduce global warming, at COP28 climate change summit, over 130 national governments, alongside the European Union, committed to collaborating towards tripling the global installed renewable energy capacity to a minimum of 11,000 GW by 2030. Reducing CO2 emissions necessitates this shift towards renewable energy, but also elevates electricity consumption. This poses challenges for transmission infrastructure due to variable renewable output and geographically dispersed resources. The intermittent nature of renewables strains grid stability, demanding robust transmission systems. Upgrading and expanding transmission networks become imperative to deliver renewable power efficiently. Balancing CO2 reductions, rising electricity use, and a growing renewable share demands strategic investments in smart grids, energy storage, and transmission technologies for a sustainable and resilient energy future. The DC microgrid is an energy system that operates on direct current (DC) instead of alternating current (AC). As the DC microgrid operates on DC, many renewable power sources such as wind turbines and PV can be interconnected to storage and loads, such as e.g. batteries, heat pumps, and electrolyzers through fewer power conversion stages. Thereby, the DC microgrids enhance efficiency, especially for devices that inherently use DC, fostering a more resilient and sustainable energy infrastructure. This thesis focuses on ensuring the robust operation of the power electronic converter system supplying challenging non-linear loads in the DC microgrid, through robust control design methods. For the control design, mathematical models are rarely known exactly. This phenomenon can be dealt with, by introducing an uncertainty in the system model to account for this. These modeling inaccuracies have the potential to negatively impact the stability and performance of the closed-loop control system. therefore, it becomes important to consider these factors when constructing the plant model. In this thesis, two different DC loads are considered, which are the constant power load (CPL) and the cyclic-operated Reversible Solid Oxide Electrolyzer Cell (RSOEC) stack. The CPL is a recognized challenging non-linear load, as it causes instability issues due to its negative incremental impedance. To achieve a constant output voltage of the DC/DC converter, a robust control approach, known as μ-synthesis was utilized to ensure robust performance and stability under uncertainty arising from unmodelled high-frequency dynamics, parameter tolerances, and changing operating conditions. The cyclic operation of an RSOEC stack is a novel challenging control design problem. Cyclic operation of an RSOEC stack implies that the stack switches between fuel-cell and electrolyzer operation cyclically, with a time interval in ms. To design the controller first, a mathematical model that describes the electrical dynamics of the stack was identified and validated based on experimental data from a commercial scale stack. The identified mathematical model was utilized in a comparative investigation to assess its impact on control design. This involved comparing its performance, specifically in bidirectional current reference tracking, with that of a simpler model commonly found in the literature for the RSOEC stack. The identified mathematical model was also used to develop an analog emulator, capable of representing the electrical dynamics of the RSOEC stack for rapid prototyping of power electronic converter systems. From extensive experimental data collected under various operational conditions of the commercial RSOEC stack, the uncertainties of the parameters in the established mathematical model were identified. These uncertainties, along with those related to the mathematical model of the bidirectional DC/DC converter, were combined to create a unified uncertain dynamic model for the system. μ-synthesis was employed once more to ensure robust performance and stability in the face of all identified uncertainties. The efficacy of the developed controller was subsequently validated through simulations and experimental tests utilizing the analog emulator