High-rate discharge behavior and augmented reduced-order modeling of 18650 sodium-ion battery

This study investigates the high-rate discharge behavior and reduced-order modeling of commercial 18650 sodium-ion batteries under combined temperature–humidity environments. Continuous discharge tests were performed at rates from 0.5C to 10C, with temperatures between 25 °C and 50 °C and relative humidity from 28% to 72%. At 0.5C, the cell delivered ∼ 1560 mAh (100% utilization), whereas at 10C the capacity dropped below 40%, with discharge time decreasing from 7254 s to 337 s. Additionally, differential capacity (dQ/dV) analysis revealed well-defined peaks at discharge rates up to 2C, whereas severe polarization and suppressed features were observed at ≥ 8C. Elevated temperature (50 °C, 50.3% RH) improved capacity and energy delivery by ∼ 25% compared to nominal conditions (32 °C, 38.2% RH), highlighting the coupled influence of thermal and moisture factors. Three reduced-order NTGK–PLR models were formulated and validated to predict terminal voltage V ( t ) : (i) a Simplified NTGK–PLR model, (ii) an Arrhenius–augmented NTGK–PLR model, and (iii) a Temperature–resistance augmented NTGK–PLR model. The Arrhenius model achieved the lowest RMSE, while the Temperature–Resistance model offered a simpler, computationally efficient alternative with sub-millivolt accuracy across 4C–10C discharge rates and thirteen environmental conditions. This work integrates high C-rate discharge characterization with temperature–humidity augmented NTGK modeling for sodium-ion batteries, bridging experimental insights and physics-based modeling with real-world applicability. The resulting framework enables accurate, real-time voltage prediction across diverse operating conditions, offering a physics-informed yet practically deployable tool for electric vehicle and grid-scale energy storage applications. • High C-rate discharge behavior (4 C–10 C rates) of 18650 sodium-ion batteries experimentally analyzed. • Temperature–humidity effects quantified on capacity, voltage, and dQ/dV profiles. • Three reduced-order NTGK–PLR models developed for real-time voltage prediction. • Arrhenius–augmented NTGK–PLR model captures thermal effects with sub-millivolt accuracy. • Modeling framework supports battery management systems for EVs and energy storage.