Bath Engineering & Design Studentships: Enabling Physics-Based Battery Models for Next-Generation Battery Management

This project is one of a number in competition for a studentship from the Faculty of Engineering and Design at the University of Bath. If successful, this studentship is expected to commence 04 October 2027.

An alternative start date may be possible if agreed with your intended supervisors and the Doctoral College.

Project Description

The transition to battery electrification requires battery management systems (BMS) that are computationally efficient while retaining sufficient physical fidelity to predict cell behaviour, limitations and degradation. Physics-based electrochemical models, particularly the Doyle–Fuller–Newman (DFN), or pseudo-two-dimensional (P2D) model, provide a rigorous description of coupled lithium transport, electrochemical kinetics and electrical behaviour, but their computational complexity can limit their use in embedded real-time applications.

This PhD project will address a key challenge: how to retain the predictive capability of P2D-class models while making them suitable for microcontroller-based BMS, particularly under high-rate and dynamic operating conditions.

The project will build on recent advances in analytical and semi-analytical solutions of lithium transport. In particular, closed-form solutions for lithium diffusion within spherical active-material particles can substantially reduce the computational cost associated with resolving the solid-phase diffusion equations. Combined with efficient representations of electrolyte transport, these approaches offer a promising route towards analytical or semi-analytical P2D model suitable for real-time implementation.

The research will develop an embedded-ready P2D framework and an adaptive model hierarchy linking computationally efficient reduced-order models, such as the single-particle model with electrolyte dynamics (SPMe), to higher-fidelity P2D formulations. Model selection will be governed by physically motivated validity indicators, such as electrolyte concentration gradients, reaction-rate limitations, overpotential and dimensionless transport/kinetic criteria, rather than by fixed operating thresholds.

The project will further investigate descriptions of degradation, including SEI growth, lithium plating, mechanical cracking, and electrolyte depletion to enable physics-informed estimation of state of health (SOH), capacity and resistance fade. Parameter identification will combine physics-based inference with computationally efficient optimisation while explicitly considering identifiability, uncertainty and embedded resource constraints.
Models will be implemented in embedded C/C++ and assessed in terms of accuracy, stability, memory usage and execution time. Validation will combine high-fidelity P2D/DFN simulations with experimental data from commercial lithium-ion cells under high-rate and dynamic drive cycles, including UDDS and WLTP, with appropriate consideration of temperature and ageing.