Initial situation, issue, motivation
“Green hydrogen” is still more expensive than “grey hydrogen”. In addition, the cost to transport H₂ to filling stations is a major price driver. Generating H₂ from PV electricity at the point of need or opportunity creates a solution for using excess electricity from PV systems for the efficient production of H₂. We view the current great potential for decentralising energy generation, increasing efficiency in terms of energy management, as well as the real-world, yet poorly researched field of using systems based on AI technologies specifically to optimise distribution and efficiency both as a problem, but also an opportunity. Another motivating factor is the global growth in the markets and application needs, especially in the area of fuel cells and hydrogen generation and particularly the use of green hydrogen. It really makes sense to install an H₂ filling station at PV locations to provide decentralized electricity for hydrogen applications. However, this can only be carried out efficiently in the long term with the use of intelligent management technology.
Goals: innovation vs state-of-the-art, results
The aim of this project is an intelligent, directly coupled, high-efficiency solar-powered hydrogen filling station consisting of
(1) the (regenerative) energy source (photovoltaic system) – already available
(2) a PEM electrolyser with high efficiency and service life,
(3) as well as an AI prediction & control module
The energy obtained from renewable energy sources is to be generated by means of electrolysis will be hydrogen, stored temporarily and then fed to a filling station. Intelligent control based on the AI prediction system will allow efficient, needs-based coupling with the energy supply infrastructure with the option of operating it independently. In the course of this project, a specific use case will, on the one hand, serve to verify the dynamic models and, on the other, underscore the (high) potential for further exploitation. Integration of a (self-)optimisation algorithm into the energy system’s control process taking into account load management, weather forecasts, and supply and demand (spot market). Due to the fluctuating input power when using renewable energy sources, i.e. the volatile load on the electrolyser, which also entails a high number of start/stop cycles, the electrolyser is additionally loaded and its service life restricted. The AI system is intended to expand this state-of-the-art technology and is therefore also a central USP in combination with this H₂ filling station. As a result, we foresee a high-efficiency H₂ filling station (new PEM stack), operated with PV electricity under the management of the AI prediction & control system, which we will operate in real conditions for at least 6 months to provide scientific and operational data in preparation for a market launch.
