Model predictive fast charging control by means of a real-time discrete electrochemical model

Model-based fast charging control of Li-ion batteries requires real-time capable models with a spatial resolution to effectively prevent metallic lithium deposition. In a recent publication we showed that a discrete electrochemical modeling approach in form of a transmission line model is able to precisely describe the dynamic behavior of electrodes in time domain. The model is able to simulate and detect local lithium deposition and fulfills the requirement of real-time capability. On this basis, we study the feasibility of nonlinear model predictive control for fast charging of a graphite electrode. For this purpose, two different approaches for the cost function are introduced. The first approach controls the anode surface potential to a target value close to 0 V vs. Li/Li+, to charge at the maximum rate and thus preventing Li deposition. The cost function can be rewritten explicitly to achieve a minimum computation time. The second approach weighs the charge current, which should be maximized, against the deposition current, which should be minimized. A trade-off between charging time and lithium deposition can be achieved allowing for varied priorities while finding the optimum charging trajectory for the targeted purpose.

Automated summary

Model-based fast charging control of Li-ion batteries requires real-time capable models with spatial resolution to prevent metallic lithium deposition. A discrete electrochemical modeling approach using a transmission line model was able to simulate and detect local lithium deposition, fulfilling real-time capability. Two different approaches for cost function were introduced for nonlinear model predictive control, one controlling anode surface potential and the other balancing charge current with deposition current to achieve a trade-off between charging time and lithium deposition.