Model predictive control of a mixed battery array for electricity grid storage

My Discovery Grant program will develop a new model-predictive-control strategy to optimize performance of battery energy storage systems using repurposed electric vehicle batteries in a second-life application for electricity grid storage. Model-predictive-control of energy storage is in its infancy, and has not yet been applied to second-life applications.

Lithium ion batteries have become the choice technology for utility grade electricity grid energy storage due to long life and high energy efficiency. However, they are often subsidized due to high cost ($350/kWh). Most lithium ion batteries are presently used for electric vehicles (EV). At end of vehicle life the EV batteries may be repurposed into a second-life for grid storage. While the batteries may have degraded to 75% of their original capacity, this is acceptable for stationary applications, and the very low purchase price of used batteries is compelling (

My program proposes a \"mixed battery array\" that recognizes the continued evolution of EV batteries and their wide geographic dispersion as a function of population. It repurposes batteries from a wide variety of EVs, at different levels of degradation, into an array with individual power converters at a centralized facility. To date, we have tested energy performance of Nissan, Tesla, BMW, and Chevrolet batteries in electricity grid storage services of peak shaving and frequency regulation, achieving a range of results due to the diverse electrochemical materials, format (cylindrical, pouch), and thermal management (passive, air, liquid).

The next step is to create a model predictive control system that optimally apportions stacked-service calls (power, duration) across all the batteries, taking into account their specific characteristics. For example, using the most energy efficient batteries for peak shaving, the powerful batteries for frequency regulation, and accelerating usage of batteries that are near end of life so as to remove/recycle them and replace them with another used EV battery pack. The control system is a critical aspect to overall performance as it must pre-position the battery state-of-charge by discharging or charging in preparation for upcoming service call forecasts.

My Discovery Grant program has three phases: Phase 1 tests batteries to map trends in energy capacity, efficiency, power, and degradation to create discrete battery models. Phase 2 develops the model predictive control strategy to optimally operate each battery for technical and economic performance. Phase 3 demonstrates/validates the new control strategy on used EV battery packs with multi-channel power converters.

Research outputs are new energy storage architectures and control strategies for industry and utilities. This simultaneously brings value to owners of old EVs (to sell their battery packs), brings a new low-cost energy storage technology to utilities, and can be commercialized in all jurisdictions of Canada and beyond.