Modelling the current-dependent voltage behaviour of LFP batteries
This thesis investigates how a deliberately compact, TLM-inspired electrical model can reproduce the rate-dependent blurring of voltage plateaus in an LFP–graphite cell, while keeping complexity compatible with embedded applications. The focus is on moderate C-rates, where electrolyte and charge-tra...
| Autor: | |
|---|---|
| Tipo de recurso: | tesis de maestría |
| Fecha de publicación: | 2025 |
| País: | España |
| Institución: | Universitat Politècnica de Catalunya (UPC) |
| Repositorio: | UPCommons. Portal del coneixement obert de la UPC |
| Idioma: | inglés |
| OAI Identifier: | oai:upcommons.upc.edu:2117/449901 |
| Acceso en línea: | https://hdl.handle.net/2117/449901 |
| Access Level: | acceso abierto |
| Palabra clave: | Lithium ion batteries Electric batteries -- Mathematical models Energy storage Bateries d'ió liti Bateries elèctriques -- Models matemàtics Energia -- Emmagatzematge Àrees temàtiques de la UPC::Energies Àrees temàtiques de la UPC::Enginyeria elèctrica |
| Sumario: | This thesis investigates how a deliberately compact, TLM-inspired electrical model can reproduce the rate-dependent blurring of voltage plateaus in an LFP–graphite cell, while keeping complexity compatible with embedded applications. The focus is on moderate C-rates, where electrolyte and charge-transfer polarisation dominate over solid diffusion, and on preserving graphite staging features under load. First, an anode-side submodel is constructed using a multi-branch transmission-line architecture. Four parallel branches, each with its own coulomb counter, pseudo-OCP source and Butler–Volmer element, share an ionic rail whose resistance is initially estimated from geometry, porosity and electrolyte conductivity. With this physics-based configuration, the model preserves the staging sequence in both lithiation and delithiation. Lithiation RMSE remains within the range of 4−11 mV up to C/3, whereas delithiation exhibits larger asymmetries. A single scalar optimisation of the effective ionic resistance, anchored at C/3 lithiation with a combined voltage and DVA cost, reduces lithiation RMSE to below about 9 mV across C/25–C/3 and improves the representation of rate-induced plateau broadening. The optimised anode model is then combined with a positive-electrode half-cell LUT. A Dubarry-style alignment using LRini and OF Sini defines a full-cell pOCV at C/25. A lumped series resistance Rfull-cell is subsequently identified by multi-rate fitting. This reduces charge RMSE to 8–13 mV and discharge RMSE to about 20 mV over C/25–C/3, while preserving plateau shapes and SoC transitions. Finally, the static maps are embedded in a dynamic ECM with one ohmic resistor and two R–C branches. The model is validated under a realistic load current profile that exceeds the nominal C/3 range and includes frequent current reversals. Under these conditions, the ECM attains a global RMSE of 14.2 mV and captures the timing of voltage peaks and valleys with reasonable fidelity. The analysis identifies the main limitations—coarse parameter grids, operation outside the calibrated C-rate window and simplified hysteresis—and outlines a path towards battery models that remain physically informed yet deployable for diagnostics and control. |
|---|