Generative Design and Optimization of Battery Packs with Active Immersion Cooling
Generative Design and Optimization of Battery Packs with Active Immersion Cooling
Embedded Files
Among different battery packaging technologies, cell-to-pack is a widely used method to reduce the cost and increase the volumetric density of battery packs. Unlike the traditional cell-to-module technology, it requires more robust management to keep the temperature uniformity of all cells within a desirable range to ensure good pack performances. Besides active cooling controls, the layout of cells within the battery pack plays an important role in cooling performances, and thus needs to be optimized for lower cooling costs considering the geometry limitations of the pack. This paper presents the layout optimization of the battery pack with active immersion cooling for the 21700 cylindrical battery pack under harsh loading conditions. Based on the experiment testing, the finite element model with electric and thermal couplings has been built in COMSOL Multiphysics. To reduce the high computational cost, a data-driven generative design method based on variational autoencoder has been developed, which could autonomously mine useful properties from the data set of existing battery layout designs and performance metrics. With the generative design method, candidate designs that optimize the layout decisions can be identified. Based on the computational studies, the cooling cost can be lowered by more than 90% with the identified optimal layout design.
Among different battery packaging technologies, cell-to-pack is a widely used method to reduce the cost and increase the volumetric density of battery packs. Unlike the traditional cell-to-module technology, it requires more robust management to keep the temperature uniformity of all cells within a desirable range to ensure good pack performances. Besides active cooling controls, the layout of cells within the battery pack plays an important role in cooling performances, and thus needs to be optimized for lower cooling costs considering the geometry limitations of the pack. This paper presents the layout optimization of the battery pack with active immersion cooling for the 21700 cylindrical battery pack under harsh loading conditions. Based on the experiment testing, the finite element model with electric and thermal couplings has been built in COMSOL Multiphysics. To reduce the high computational cost, a data-driven generative design method based on variational autoencoder has been developed, which could autonomously mine useful properties from the data set of existing battery layout designs and performance metrics. With the generative design method, candidate designs that optimize the layout decisions can be identified. Based on the computational studies, the cooling cost can be lowered by more than 90% with the identified optimal layout design.
Published and presented 1 paper as the 1st author on 2023 IEEE Transportation Electrification Conference & Expo (ITEC)
Battery tests performed on Samsung cylindrical cells. (a) 1-cell test without cooling, (b) 2C constant current discharge curve, and (c) 8-cell immersion cooling test setup.
The generative design method for battery pack layout optimization. (a) Overall flowchart of the generative design process, and (b) Architectures of the generator and estimator.
Case study results from experiments and simulations. (a) 1-cell without cooling, (b) 8-cell with immersion cooling, and (c) 8-cell with immersion cooling and optimized layout.
Page updated
Google Sites
Report abuse