From Experimental Characterisation to Prediction Models of LIB Anode Drying Processes

In commercial lithium ion Battery (LIB) cell manufacturing, understanding and improving the drying process is key to the yield as well as energy-consumption, and thereby the CO2 emission footprint. As such, this study experimentally investigates convection drying kinetics of anodes alongside the use of (informed and non-informed) neural network models, in order to better understand and improve electrode manufacturing. In the experimental campaign we find that air velocity and temperature are the most important parameters in the initial phase (constant-rate period) before we reach the critical moisture content (critical point), whereas the subsequent phase (falling-rate period) is less sensitive to increases in air temperature and velocity. This supports the idea that drying during the constant-rate period is mainly driven by external heat and mass transfer while the drying during the falling-rate period is increasingly limited by internal mass-transport through the wet-coating thickness. In neural network modelling, we found it necessary to apply physics informed models in order to interpolate between the experimental observations. However, at the tested conditions, the non-informed neural network model gave a stronger fit to the data.