Analyse und Optimierung der Zuverlässigkeit und Effizienz von hybriden Batteriespeichern am Beispiel von M5BAT

The energy transition in Germany aims to enable a CO2-neutral future. In the field of electrical energy supply, the energy transition requires a shift away from fossil fuel power plants and an expansion of renewable energy such as wind and photovoltaic power plants. The volatile generation characteristics of renewable energies make short-term and long-term storage necessary in the energy system, with battery storage taking the place of short-term storage. This dissertation presents the analysis from the operation of the M5BAT large-scale battery storage system, which has been operated by RWTH Aachen University since 2016. With around 6 MW / 7.5 MWh and the hybrid design with ten battery units and five different battery technologies, M5BAT is unique in the research landscape. This allows the battery technologies used to be compared with each other and technology-specific advantages to be exploited. Research questions and theses are formulated on the topics of status analysis, reaction speed, efficiency and energy management and are answered by investigations on the M5BAT large-scale battery storage system. As part of the status analysis, the operational battery availability, the causes of failure and the degree of capacity utilization were determined. It became clear that the predominant failure times were caused by the respective battery management system or the violation of voltage limits. In case of lithium-ion batteries of the lithium-iron-phosphate battery unit of the M5BAT large-scale battery storage system, balancing is responsible for about 25 % of the unusable capacity, while in case of the lithium-manganese-oxide battery unit of the M5BAT large-scale battery storage system, only about 5 % of the capacity is unusable due to balancing. The reaction speed of large battery storage systems is mainly influenced by the inverters used and the control and measurement technology. With M5BAT, response times of 175 ms to 325 ms are possible in the best case. This means that faster applications than frequency containment reserve are possible, but grid-connected inverters and inverter-integrated control algorithms must be used for synthetic inertia. Efficiency ranges for inverters and transformers were determined as part of the efficiency studies. A benchmark test and long-term operational evaluations have shown that a round-trip efficiency of just under 88 % is possible with M5BAT; in frequency containment reserve and research operation, however, the round-trip efficiency is around 75 %. This is attributable to the fact that the efficiencies of the converters and transformers are loaddependent, with the lowest efficiencies occurring in low partial load operation. The efficiency evaluations were taken into account when developing the power distribution algorithm as part of the energy management system. A modular, rule-based power distribution algorithm is presented and tested in frequency containment reserve operation. The power distribution algorithm has shown that a round-trip-efficiency of around 80 % of the battery storage system can be reached while reducing the number of inverter switching operations and maintaining battery-specific state of charge ranges. The energy throughput per battery unit can also be controlled and used to reduce battery stress and consequently battery aging. This means that the power distribution algorithm is what makes it possible to exploit the technology-specific advantages. The algorithm presented can be adapted and used for hybrid or modular large-scale battery storage systems to ensure an efficient and aging-aware operation. The insights gained from the M5BAT large-scale battery storage system can also be transferred to other large-scale battery storage systems. However, the balancing problems identified must be solved by the respective manufacturers of the battery management systems or system integrators and can only be monitored in productive system operation. The development of a data-driven digital twin can improve the marketing and operation of M5BAT and other large-scale battery storage systems in the future and should be pursued. Continuous monitoring and datalogging on M5BAT made the analyses possible in the first place and are strongly recommended.