We are pleased to announce that our partner TNO, working within the NEXTBMS project, has had a scientific paper accepted for presentation at the European Control Conference (ECC 2026), which will take place from 7 to 10 July in Iceland.
The paper presents important advances in the field of fast charging for electric vehicles, based on physics-based battery modelling developed within the NEXTBMS project.
Summary
Fast charging is essential for large-scale electric vehicle (EV) adoption but is fundamentally limited by battery degradation and safety risks, most notably lithium plating at the anode and excessive temperatures. Most existing fast charging strategies either rely on heuristic profiles (e.g. CCCV) or treat thermal effects purely as constraints rather than actively controlling them. This paper proposes and evaluates a degradation-aware fast charging framework that jointly controls charging current and thermal management to minimise charging time while respecting electrochemical and thermal safety limits.
Battery modelling framework
The study employs a physics-based battery model, the Single Particle Model with electrolyte dynamics (SPMe), augmented with a two-node lumped thermal model. This physics-based model enables prediction of critical internal variables that are not directly measurable, including:
- Anode potential (ϕ⁻), a key indicator for lithium plating risk (ϕ⁻ ≥ 0 V vs. Li/Li⁺)
- Cell core temperature (Tᶜ), which limits thermal degradation and safety
- Terminal voltage (Uᶜ), used for enforcing over-voltage constraints
Thermal management is represented through an actuated external thermal resistance, modelling different cooling regimes from natural to forced convection. This approach allows temperature to be treated as a control variable rather than a passive constraint.
Control architectures
Two multi-input multi-output (MIMO) control strategies are developed and compared using the same electrochemical-thermal model:
Classical PID control
- Parallel PID controllers independently enforce limits on anode potential, voltage, and temperature
- The applied current is selected as the minimum of the constraint-based currents
- Thermal resistance is regulated by a separate PID controller to maintain temperature near its upper bound
Non-linear Model Predictive Control (MPC)
- Simultaneously optimises charging current and thermal resistance over a prediction horizon
- Explicitly enforces electrochemical and thermal constraints
- Uses a dual-resolution horizon with fine time steps for electrochemical dynamics and coarser steps for thermal dynamics, reducing computational complexity
Methodology and case studies
Simulations are performed on a validated model of a Sony/Murata 18650 NCA–graphite cell, charging from 10% to 80% state of charge under different ambient conditions (0°C, 20°C, and 35°C). Performance is benchmarked against conventional CCCV charging at both manufacturer-recommended and aggressive current levels. Ageing effects are evaluated by increasing SEI resistance and reducing cell capacity.
Key results
- Coordinated electro-thermal control outperforms traditional CCCV charging, achieving a 42.2% reduction in charging time compared to the manufacturer’s recommendation without increasing degradation risk
- MPC consistently outperforms PID, achieving on average 5.2% shorter charging time by proactively exploiting thermal headroom and operating closer to safety limits
- Thermal actuation is critical: proactive cooling enables higher sustained charging currents, especially at elevated ambient temperatures
- MPC adapts effectively to battery ageing by automatically reducing current earlier as degradation progresses
- The main drawback of MPC is its high computational cost (approximately 77 minutes per simulated charge versus less than 1 minute for PID), which challenges real-time deployment
Conclusions and outlook
The paper demonstrates that joint electrochemical–thermal optimal control is a powerful enabler for safe and ultra-fast charging of lithium-ion batteries. While MPC delivers superior performance, its computational burden remains a barrier to practical implementation.
Future work will focus on reduced-order models, tailored solvers, experimental validation, and advanced thermal systems capable of both heating and cooling, to fully exploit predictive charging strategies.
We are proud to see this work from TNO accepted at ECC 2026. The paper will be published following the conference.
