The optimization of the electrode manufacturing process is important for upscaling the application of Lithium-Ion Batteries (LIBs) to cater for growing energy demand. LIB manufacturing is important to be optimized because it determines the practical performance of the cells when the latter are being used in applications such as electric vehicles. I. ••Synthetic dataset generated by low-discrepancy sequences as inputs of the physics-based models.••Fast deterministic-assisted bi-objective optimization of the energy density and power density to determine the best set of manufacturing parameters.••Optimzation for different types of battery applications.Battery cell manufacturingBayesian optimizationMachine learningElectrodeIn our modern society, the demand for batteries has surged due to the widespread use of electric vehicles and portable electronic devices. Lithium-ion batteries (LIBs) have emerged as the most powerful technology for a fast energy transition,. Driven by the increasing demand for high-performance energy solutions with low-carbon emissions, the modern world is making efforts to establish gigafactories and recycling approaches to significantly reduce the production costs for LIBs and make them sustainable,. The manufacturing process is considered the most impactful part of battery design, and optimizing this process is crucial for improving overall battery performance. This complex fabrication process involves numerous interlinked steps and manufacturing parameters. The entire process includes electrode slurry preparation, coating and drying, calendering, and the cell assembly, electrolyte filling and formation. Certain parameters, such as the type of material, the amount of material, and the drying temperature applied to the slurry, have a significant impact on the final battery performance and must be optimized throughout the entire fabrication process. The electrode optimization in turn depends on the end application that can be categorized as: (1) energy-oriented and (2) power-oriented batteries,, which require different electrode design strategies. In general, energy-oriented batteries favor higher material loading, while power-oriented batteries favor lower material loading due to the nonlinea. 2.1. Data acquisitionDue to the high computational cost required to simulate the electrochemical performance for a continuum batch of various manufacturing conditions, we utilized the synthetic dataset generated from our previous work to obtain a highly representative dataset of the manufacturing parameter space. Specifically, we have generated quasi-random Sobol sequences with Saltelli extension based on three parameters: the amount of active material (AM %), the slurry solid content (SC %), and the electrode compression degree (CD %). These parameters are representative enough of the slurry preparation, drying, and calendering processes, as important parameters to assess when manufacturing electrodes,. Our focus was to properly probe the input manufacturing space and capture all of its sub-areas by varying these three parameters. This design of experiments (DOE) was used as input values for physics-based models to evaluate the properties that characterize the 3D electrode microstructures. More details on this can be found in section 2 of our previous work. It is worth mentioning that the DOE is large enough to generate data for further machine learning (ML) regression purposes while being efficient enough to avoid a too significant computational cost associated with generating all 3D microstructures for each manufacturing condition. In fac.
