Battery cells within battery energy storage systems (BESS) do not have homogeneous attributes, and the lowest capacity ones limit the performance and lifetime of the whole pack. Modern battery management systems (BMS) solve this problem with balancing, while providing the required service, and safe operation to the user. Reconfigurable battery systems (RBS) are BESSs that involve a BMS with reconfiguration. Reconfiguration uses feedback to determine the circuit switching logic. This paper presents a structured review of the control algorithms for RBSs. The RBSs are divided into groups according to their control strategies and control implementations. Finding the adequate control strategy requires well-defined objectives and control design. The control implementation focuses on physical and architectural aspects, like the reconfiguration frequency, the balancing operation and the control topology. The considerations and categories are discussed with the advantages, disadvantages and academic examples, and then an innovative industrial BMS is introduced.

This paper presents the experimental validation of a lithium-iron phosphate cell model. The modelling of dynamic cell behaviour is crucial to improve the performance of reconfigurable battery systems, in which monitoring of the dynamics allows more resoluted leveraging of the battery cells. However, the models of lithium-ion cells are generally inaccurate due to nonlinearities, measurement noise and because the most descriptive state, the state of charge is hidden. Furthermore, the parameter identification of the model requires time and precise measurements, while they differ among the cells, and change as the cells age. The burden of the detailed modelling of a battery system can be alleviated by modelling a single cell, and using the model for each cell in the system. In this work, this possibility is explored by validating a single-cell reference model for a reconfigurable battery system. The terminal voltage residual (error between model and measurement) is presented, and its correlations with internal and external variables are investigated. These correlations can also be used to alleviate the modelling errors. It is concluded that the reference model can qualitatively describe the cell behaviour. By applying small modifications, the model could be used for online estimation.

The paper proposes replicable testing procedures to verify active power functions for Grid Code compliance of battery systems. The investigated control functions are frequency response, absolute power constraint, and ramp rate constraint. The considered requirements are specified in the Danish Grid Code for battery systems with rated grid connection below 125 kW. With the designed procedures, battery systems can be characterized thoroughly, fully examining the three required success criteria: accuracy, sensitivity, and response time. The testing procedures are experimentally applied to a 33.5 kW / 95 kWh reconfigurable battery storage system. The testbed was built at PowerLabDK at DTU. Results from the testing campaign showed that the battery storage operates under the specified control functions, meeting accuracy, sensitivity, and response time requirements.

No abstract