A comprehensive and time efficient characterisation of redox flow batteries through Design of Experiments

Abstract

As the need for a sustainable economy rightly drives the share of renewable energy, electric grids and supporting infrastructure must flexibly adapt. As valuable building blocks in integrated systems, battery energy storage systems (BESSs) can provide the required flexibility for energy and power applications. Redox flow batteries (RFBs) are emerging as promising alternatives to lithium-ion batteries to meet this growing demand. As end-users, RFB operators must characterise the batteries to learn more about the battery's behaviour and performance and better integrate such RFB technology into energy systems. Characterisation experiments yield this information, which is essential to successfully operate and integrate redox flow battery systems. However, conducting classical characterisation protocols can take more than two weeks for large RFB modules (capacities >30 kWh), which is too long for an efficient RFB roll-out. Better characterisation methods are required to efficiently scale up, integrate and operate RFBs in an appropriate manner. Ideally, characterisation experiments would yield a more comprehensive understanding about the battery performance and behaviour in a shorter amount of time. In order to achieve this, statistical design of experiments (DoE) is explored as an RFB characterisation tool. DoE is a statistical method that makes optimal use of the available time and resources and increases the efficiency of experiments in a statistically sound manner. Designed experiments result in empirical models for the studied system, which can predict system outputs for a vast amount of operating points. This will enable optimal operation of the battery in terms of remaining capacity management and overall electrical efficiency. Through a number of such designed experiments, dominant RFB system variables could be identified, which allow reliable modelling of the RFB performance for different charge-discharge cycles. This facilitated the design of an optimised characterisation experiment. A 50% reduction of the required RFB characterisation time is achieved and the optimal experiment yields comprehensive information about the battery performance and behaviour. As such, a shorter and better RFB characterisation procedure is realised through DoE.