Modelling Air Cooling in a Module of Prismatic Li-Ion Cells

Abstract

The control of temperature in lithium-ion stacks is of increasing industrial importance. In fact, an efficient thermal conditioning system ensures a homogeneous temperature distribution among the cells, preventing conditions which could hinder performance or led to temperature runaway and so a disruptive failure of the module. CFD models for thermal analysis can be useful in the design phase to optimize the module layout and to size the cooling system, as well as to support experimental campaigns in the definition of reduced-order thermal models to be implemented into the Battery Management System (BMS) for in-operation temperature prediction. The present work presents the development of a computational model of a 48V-20Ah 20-cell Li-ion module and its air-cooling system based on COMSOL Multiphysics software. Unlike most papers available in the literature, the individual cells are modelled by considering an internal ohmic heat generation term as a function of both state of charge, cell temperature and sign of the applied current (charge or discharge), exploiting experimentally validated data provided by the manufacturer in temperature-controlled conditions. In addition, the reversible thermal generation term due to entropic phenomena, which is often ignored in the literature, was computed alongside an irreversible term accounting for cell overpotentials. The entropic contribution, which was estimated experimentally and optimised computationally, resulted non-negligible when the module is cycled at ≤ 1C. To take into account the effect of the fan cooling system, the motion field was solved in COMSOL using a turbulent k-ε model. To ensure model accuracy, only minor geometrical simplifications were introduced to the fluid volume. The resulting geometrical complexity, made of manifold and narrow channels, called for special attention in the computational resolution. Assuming a decoupled behaviour, namely that the thermal evolution does not affect the flow field, allowed to use the stationary fluid-dynamic solution as the starting point for the time-dependent thermal simulation. The model was run under different experimental current profiles and the validation was performed on the entire module under effective operating conditions and not on a single cell under natural convection condition with the external environment, as is usually found in the literature. The computational data were found to be in good agreement with experimental measurements for all the cells even after charge-discharge cycles lasting more than two hours. The simulations provided insights to improve the positions of thermal sensors inside the module and to optimize the efficiency of the cooling system, that could not be done based on thermal measurements only. Figure 1