Key points are not available for this paper at this time.
Energy storage systems (ESSs) are a critical component of the electric grid, dispatching (charging and discharging) to performing grid applications such as frequency regulation, energy arbitrage, and peak shaving. Today, lithium-ion batteries are considered the best electrochemical ESSs in grid applications, and various cathode chemistries have been developed. On the electric grid, batteries can provide up to 13 ESS services, and different combinations of grid services and chemistries produce different battery aging and life performance under the given dispatch. Therefore, the characterization of each grid application dispatch can give an insight into optimal participation strategies for lithium-ion chemistries for each grid service. In this paper, an efficient algorithm is presented which uses a dispatch interval matrix to capture metrics in the ESS dispatch relevant to lithium-ion battery aging and performance, and implements unsupervised learning and dimensionality reduction on this matrix to produce characteristic duty cycles of the dispatch, from which synthetic duty cycles are produced that are suitable for laboratory testing and fast simulation. The algorithm is demonstrated for the dispatch under the grid application of peak shaving. Finally, an electrochemical-aging model is used to simulate a lithium-ion battery under both the original power dispatch and the synthetic duty cycle to validate the effectiveness of the method proposed in this paper to retain the operating stress factors.
Moy et al. (Thu,) studied this question.