Modelling and real-time predictive control for hybrid cooling of lithium-ion batteries
Doctoral thesis, 2026
Lateral surface cooling can limit temperature rise in lithium-ion batteries through its large heat-transfer area, whereas tab cooling can reduce internal thermal gradients by exploiting the high thermal conductivity of the current collectors. This thesis investigates how these complementary characteristics can be integrated through modelling and predictive control for battery thermal management.
A control-oriented thermal modelling framework is developed using the Chebyshev spectral-Galerkin method and model component decomposition. The resulting reduced-order two-dimensional models enable independent manipulation of thermal boundary conditions at the battery tabs and lateral surface while reproducing the spatial temperature field at low computational cost. Building on this framework, tab and surface cooling are integrated via a constrained optimal coolant allocation problem in which coolant is dynamically distributed among the cooling interfaces. A first-principles coolant model and valve-actuation model are coupled with the electrothermal dynamics. The resulting nonlinear problem is solved using real-time iteration model predictive control (MPC), which closely reproduces nonlinear MPC while reducing mean computation time from several seconds to approximately 19 ms. The framework is further extended to degradation-aware control by reformulating a semi-empirical degradation model as dynamic states and incorporating spatial temperature non-uniformity through the volume-averaged temperature and temperature variance, reducing degradation model computation by more than 20×.
Across the control studies, integrated tab-surface cooling provides the best overall trade-off between temperature tracking, thermal uniformity, accumulated capacity loss, and coolant utilisation compared with conventional cooling configurations. Together, these contributions establish a basis for hybrid battery cooling combining spatially resolved modelling with real-time predictive control in battery systems.
A control-oriented thermal modelling framework is developed using the Chebyshev spectral-Galerkin method and model component decomposition. The resulting reduced-order two-dimensional models enable independent manipulation of thermal boundary conditions at the battery tabs and lateral surface while reproducing the spatial temperature field at low computational cost. Building on this framework, tab and surface cooling are integrated via a constrained optimal coolant allocation problem in which coolant is dynamically distributed among the cooling interfaces. A first-principles coolant model and valve-actuation model are coupled with the electrothermal dynamics. The resulting nonlinear problem is solved using real-time iteration model predictive control (MPC), which closely reproduces nonlinear MPC while reducing mean computation time from several seconds to approximately 19 ms. The framework is further extended to degradation-aware control by reformulating a semi-empirical degradation model as dynamic states and incorporating spatial temperature non-uniformity through the volume-averaged temperature and temperature variance, reducing degradation model computation by more than 20×.
Across the control studies, integrated tab-surface cooling provides the best overall trade-off between temperature tracking, thermal uniformity, accumulated capacity loss, and coolant utilisation compared with conventional cooling configurations. Together, these contributions establish a basis for hybrid battery cooling combining spatially resolved modelling with real-time predictive control in battery systems.
control-oriented thermal modelling
coolant modelling and allocation
integrated tab and surface cooling
model predictive control
battery degradation
real-time control.
battery thermal management
Lithium-ion batteries
EA (Room 4233), EDIT Building, Chalmers
Opponent: Professor Davide Martino Raimondo, Department of Engineering and Architecture, University of Trieste, Italy
Author
Godwin Peprah
Chalmers, Electrical Engineering, Systems and control
Keeping batteries '' just right ''
Do you remember the childhood story of Goldilocks and the Three Bears, where Goldilocks was always looking for something that was neither too hot nor too cold, but '' just right '' ? A lithium-ion battery is not so different. It also has a preferred temperature range in which it performs well and ages slowly. If it becomes too cold, less juice can be extracted, and charging can accelerate ageing. If it becomes too hot, it degrades faster and, in extreme cases, can become a safety risk.
Keeping the average temperature within a suitable range is not enough. Different regions inside a battery can reach different temperatures, creating hot and cold spots that can affect performance and ageing. The challenge is thus not only to keep the battery at the right temperature but also to maintain a reasonably uniform temperature. This thesis investigates how that can be achieved by combining cooling through the battery surface and its electrical terminals. These two cooling methods have different strengths, and instead of choosing only one, a controller decides how the available coolant should be shared between them to harness their individual strengths as the battery operating conditions change. Fast mathematical models are developed to estimate the temperature distribution inside the battery, while smart control is used to adjust the cooling in real time.
The work also considers battery ageing directly. The results show that making a battery as cold and thermally uniform as possible does not always minimise degradation. What matters is when, where, and how much cooling is applied. In short, the goal is to give the battery what Goldilocks wanted: the right conditions at the right time, helping it perform well, remain safe, and last longer.
Do you remember the childhood story of Goldilocks and the Three Bears, where Goldilocks was always looking for something that was neither too hot nor too cold, but '' just right '' ? A lithium-ion battery is not so different. It also has a preferred temperature range in which it performs well and ages slowly. If it becomes too cold, less juice can be extracted, and charging can accelerate ageing. If it becomes too hot, it degrades faster and, in extreme cases, can become a safety risk.
Keeping the average temperature within a suitable range is not enough. Different regions inside a battery can reach different temperatures, creating hot and cold spots that can affect performance and ageing. The challenge is thus not only to keep the battery at the right temperature but also to maintain a reasonably uniform temperature. This thesis investigates how that can be achieved by combining cooling through the battery surface and its electrical terminals. These two cooling methods have different strengths, and instead of choosing only one, a controller decides how the available coolant should be shared between them to harness their individual strengths as the battery operating conditions change. Fast mathematical models are developed to estimate the temperature distribution inside the battery, while smart control is used to adjust the cooling in real time.
The work also considers battery ageing directly. The results show that making a battery as cold and thermally uniform as possible does not always minimise degradation. What matters is when, where, and how much cooling is applied. In short, the goal is to give the battery what Goldilocks wanted: the right conditions at the right time, helping it perform well, remain safe, and last longer.
Subject Categories (SSIF 2025)
Fluid Mechanics
Energy Engineering
Control Engineering
DOI
10.63959/chalmers.dt/5967
ISBN
978-91-8103-510-0
Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie: 5967
Publisher
Chalmers
EA (Room 4233), EDIT Building, Chalmers
Opponent: Professor Davide Martino Raimondo, Department of Engineering and Architecture, University of Trieste, Italy