This NAISS Medium Compute proposal covers all computing activities within Patrik Johansson’s research group -that can be summarized as being part of the PI's Distinguished Professor grant from Vetenskapsrådet (47.5 MSEK,2022-2031). The main goal of the project is to employ computational studies to rationalize the development and usage of rechargeable batteries. The specific goals for us at UU and Chalmers are: 1. To apply several modelling approaches based on both ab-initio, DFT, and COSMO-RS, and machine learning methods such as symbolic regression to rationalize the development of electrolytes. This will then be connected both to high-throughput screening and to large-scale facility experiments, for the best candidates. 2. To proceed with a detailed investigation of the proposed electrolytes by also understanding the underlying mechanisms of ion transport etc. 3. To understand how usage impacts battery ageing and how information about usage is encoded in a compact way. The initial stage of this project will investigate the structure and dynamics of small local models and then stretch to solubility calculations of salts in organic solvents. This will all be performed by employing the framework of DFT to predict Gibbs free energies together with the COSMO-RS approach to evaluate the solvation energy of the ions in a number of different organic solvents. This procedure will indicate, together with the computation of properties such as viscosity and flashpoint, attractive electrolyte compositions. Molecular dynamics (MD) simulations will be performed to gain deep understanding of the formation of decomposition products and the stability of the anion and solvent molecules on the interface surface. Moreover, important information regarding the kinetics of the electrolyte decomposition can be revealed from analysis of the MD trajectories. One particular goal for the DP grant is to in some way model electrolytes with a clear focus on the role of entropy. This is done by focusing on multi-component eutectic electrolytes, deep eutectic solvent-based electrolytes, liquid-solid gel electrolytes. In parallel with the studies above, and to approach goal 3, we are collaborating with several providers of field data relating to usage of lithium-ion and sodium-ion batteries. The data is provided in the form of time-series of voltage, temperature and current and will be used a starting point for both estimation of state-of-health (SOH), i.e. the remaining capacity of the battery, and to form aggregates of usage. The aggregates, also denoted “features”, will be tested for their power to predict SOH evolution of single battery cells by using several machine-learning and parametric model types and then successively permuting and adding the features used.