Rechargeable-battery systems are critical to two technologies that will help reduce the consumption of fossil fuels: electrically-powered transportation systems and energy storage systems for the grid. Despite great improvements in battery cell performance, battery integration into systems still faces significant challenges. Existing solutions are typically highly integrated with the target application, and cannot be repurposed. The systems are often not scalable, and the failure of a single battery cell can cause the entire system to fail. In addition, the power electronics of such systems is optimized for the nominal load, not for partial load, where the system typically operates. This work explores a radically new approach to designing energy storage and energy conversion systems by modularizing and integrating the battery with the power electronics to provide multiple functions using the same semiconductor chip area. The proposed battery technology will use a new multilevel inverter topology that allows dynamically reconfigurable series and parallel connectivity: the modular multilevel series-parallel converter (MMSPC). Modular design of identical sub-systems will enable the same modules to be used in multiple applications, making use of economies of scale to reduce system cost. In electric vehicles, the proposed system can replace hard-wired battery packs with a flexible, dynamically reconfigurable AC battery and replace multiple power electronics units, such as the drive inverter, battery charger, and battery balancing circuits, to provide the output directly from the AC battery. For grid energy storage, the proposed technology enables repurposing modules from various applications, such as electric vehicles, incorporation of cells of different capacity or age into one system, high output quality with substantially reduced or eliminated magnetic components, rapid dynamic response, and easy scaling of the storage and power converter systems by simple addition of AC battery modules.
To leverage the advantages of MMSPC, efficient control strategies have to be developed to optimize performance while minimizing system complexity and cost. The control of modular multilevel converters (MMCs) presents both challenges and opportunities associated with the large number of possible switch states. The MMSPC degrees of freedom are even more due to the additional parallel state, which allows widely flexible series-parallel configuration of the circuit, amplifying the need for a coherent control strategy. For instance, in both MMC and MMSPC the same output voltage can be achieved with a multitude of module configurations, providing the opportunity to optimize the switch states based on various additional constraints and objectives such as module balancing, efficiency, output quality, electromagnetic emissions, and switch and storage-element stress. Existing control approaches, however, do not fully exploit this opportunity as they typically reduce the number of objectives and treat the various constraints and objectives independently. Critically, established strategies are not designed to utilize parallel connectivity, precluding exploitation of the MMSPC advantages. Addressing these limitations, we propose to develop a real-time predictive multi-objective optimization framework that systematically unifies the treatment of multiple system constraints and objectives, and overcomes the exponential growth of degrees of freedom with system size and prediction horizon. This control framework will be applicable to both MMC and MMSPC, and will consider additional topology variations within each of these converter families. The advantages of the novel MMSPC topology and control strategy will be demonstrated with the development of a modular AC battery that incorporates multiple battery units, battery management, and inverter functionality for applications such as energy storage systems and electric vehicle drive trains. This innovative concept will improve lifetime, efficiency, and cost of battery systems, and is practical only when the capabilities of the MMSPC and the associated control are leveraged.
