The transition toward renewable energy systems requires flexible energy conversion technologies capable of responding to intermittent power availability. A modular alkaline water electrolysis system offers a promising pathway for such flexibility, as it can dynamically adapt its production capacity to match the fluctuating output of sources like photovoltaics. While alkaline water electrolysis is technically mature, its deployment as a large-scale energy storage solution remains economically unattractive under current market conditions. To improve its viability in this role, targeted optimization of system operation and sizing is essential, particularly in direct coupling scenarios where real-time control is required to ensure both efficiency and safety. Existing literature has largely focused on component-level improvements, whereas system-level behavior under variable conditions remains underexplored.

This thesis presents a physics-based 0-D digital twin of a modular AWE system designed to enable optimal operation under varying power conditions. The model explicitly accounts for electrochemical behavior, heat generation and dissipation, and gas crossover dynamics. To the best of the author's knowledge, this is the first model that integrates these coupled physical phenomena within a modular alkaline electrolysis architecture. The model is validated against experimental data reported by Brauns et al., ensuring accurate representation of steady-state and transient system responses.

Based on the steady-state behavior across a range of voltages, flow rates, and temperatures, a novel voltage tracking strategy is developed that defines a safe and efficient operating voltage window. This approach enables the controller to adjust the number of active stacks and regulate operating conditions in real time, based solely on the measured busbar voltage. The method avoids direct intervention at the stack level, allowing for simple and robust implementation even in systems without power electronics. The digital twin thus serves as the foundation for a low-complexity yet effective control strategy and sizing methodology tailored to directly coupled photovoltaic systems. Its performance is evaluated through dynamic simulations using measured solar irradiance and ambient temperature data from multiple real-world locations. Compared to a conventional system equipped with maximum power point tracking (MPPT), the proposed modular configuration achieves approximately 45% higher power utilization and hydrogen yield over the course of a year, demonstrating a clear performance advantage under real-world solar conditions.