Abstract
Electric vehicle (EV) adoption in cold climates is hindered by reduced charging efficiency and the risk of battery degradation at low temperatures. Traditional smart charging strategies often treat battery heating as a separate pre-processing step, failing to exploit the thermal-electrical synergies of the system. This paper presents Temperature-Controlled Smart Charging (TCSC), a novel optimization framework that co-optimizes battery heating and charging power. By leveraging the thermal inertia of the battery and internal waste heat, TCSC reduces charging costs by up to 18.4% and improves solar energy utilization. To ensure scalability for large charging stations, a decentralized algorithm using Reduced-Order Dual Decomposition is implemented, providing significant computational speedups without sacrificing performance.
Introduction
In cold climates, EV batteries face significant operational hurdles. Low ambient temperatures increase internal resistance, which slows charging rates and reduces the effective range. More critically, charging a cold battery at high currents can lead to lithium plating, a form of permanent physical damage that compromises battery health and safety. While on-board heaters can mitigate these issues, they are energy-intensive and typically operated independently of the charging schedule.
Existing smart charging algorithms generally focus on grid price signals or renewable energy availability, treating battery temperature as a fixed constraint rather than a decision variable. This decoupled approach misses the opportunity to use the battery's thermal mass as a form of energy storage. This paper introduces the TCSC framework, which treats battery temperature as a flexible variable to be optimized alongside electrical demand.
Methodology
Thermal-Electrical Co-Optimization
The TCSC framework is built upon a unified optimization model that couples the electrical and thermal dynamics of the EV battery. The model incorporates three primary components:
- State-Space Thermal Modeling: The system accounts for ambient temperature, heat dissipation, and the internal "waste heat" generated by the charging process itself. This allows the optimizer to predict how much active heating is required versus how much can be provided by the charging activity.
- Temperature-Sensitive Charging Limits: Rather than assuming a constant maximum charging rate, the model adjusts the allowable current based on the instantaneous battery temperature. A "Practical Rule" is enforced to strictly limit high-current charging in sub-freezing conditions, ensuring long-term battery health.
- Mixed-Integer Linear Programming (MILP): The co-optimization problem is formulated as an MILP to determine the optimal timing and intensity for both heating and charging, minimizing total costs while satisfying user-defined State of Charge (SoC) requirements.
Scalability via Decentralization
To manage large-scale charging stations (30–50+ vehicles), the authors developed a decentralized approach. Using a Reduced-Order Dual Decomposition algorithm combined with a vehicle rescheduling heuristic, the station can solve complex optimization problems efficiently. In this setup, individual vehicles compute their own optimal charging and heating profiles, while a central coordinator manages system-wide constraints, such as total grid capacity.
Experimental Setup
The TCSC framework was validated using real-world data from Boston, Massachusetts, covering the period of November to December 2022. The dataset included:
- Ambient Temperature: Sourced from the Automated Surface Observing System (ASOS).
- Solar Generation: Real solar profiles from ISO New England, scaled to the capacity of a representative charging station.
- Electricity Pricing: Time-of-use (ToU) retail prices typical for the Boston metropolitan area.
- Fleet Parameters: A simulated fleet with base battery capacities of 37 kWh, including ±5% stochastic fluctuations to model real-world diversity.
Key Results and Insights
Cost and Energy Efficiency
The TCSC framework demonstrated superior performance compared to traditional "SmartChg&Heat" methods (where heating and charging are optimized separately). TCSC achieved a 12.5–18.4% reduction in total charging costs. Furthermore, the energy overhead specifically dedicated to heating was reduced by 0.4–6.8%, as the system maximized the use of internal waste heat and thermal inertia.
Solar Integration and Thermal Inertia
A key insight of the study is the role of thermal inertia. The battery's mass acts as a thermal buffer; TCSC learns to "pre-heat" the battery during periods of high solar abundance or low grid prices. This allows the vehicle to "coast" on that stored heat during expensive peak hours. Consequently, TCSC improved solar usage rates significantly—reaching approximately 79% compared to only 38% in a baseline scenario without active thermal management.
Computational Performance
The decentralized algorithm proved highly effective for large-scale applications. For a scenario involving 50 vehicles, the decentralized approach was 20–30% faster than the centralized MILP solver while maintaining nearly identical cost-saving performance, making it suitable for real-time station management.
Discussion and Future Work
The TCSC framework represents a significant step toward climate-resilient EV infrastructure. By integrating thermal management into the smart charging logic, operators can reduce costs and protect battery health simultaneously.
However, several areas for future research remain. The current model assumes a dedicated charger for every vehicle; future iterations should incorporate queuing dynamics and physical vehicle shuffling. Additionally, exploring the impact of Vehicle-to-Grid (V2G) services under cold-weather constraints and applying Reinforcement Learning (RL) to handle non-linear battery degradation models could further enhance the system's utility and accuracy.
Summary updated