Smart Charging Strategy Eases EV Range Anxiety with Multi-Factor Optimization
As electric vehicle (EV) ownership continues to rise across China, drivers are increasingly confronted with a familiar challenge: the fear of running out of battery power before reaching a charging station. Known as “range anxiety,” this psychological barrier remains a significant factor influencing consumer confidence and adoption rates. While advancements in battery technology have extended driving ranges, the supporting infrastructure—particularly the availability, efficiency, and cost of charging—has struggled to keep pace. In response, researchers are turning to intelligent routing and dynamic pricing models to optimize the EV charging experience.
A groundbreaking study led by Zhang Wei from the School of Information Engineering at Southwest University of Science and Technology offers a comprehensive solution to this multifaceted problem. Published in the Journal of Sichuan University (Engineering Science Edition), the research introduces a novel multi-objective optimization framework designed to guide EV drivers toward the most efficient and cost-effective charging decisions. By integrating travel distance, real-time charging costs, and station congestion levels, the model provides a personalized charging strategy that adapts to individual user preferences and dynamic network conditions.
The urgency of such innovation is underscored by China’s ambitious “dual carbon” goals—peaking carbon emissions by 2030 and achieving carbon neutrality by 2060. Transportation, a major contributor to greenhouse gas emissions, is a critical focus area. The government has heavily promoted EV adoption through subsidies, infrastructure investment, and regulatory measures. As of 2023, China accounts for over half of the world’s electric vehicles, with more than 14 million units on the road. However, the rapid growth in EV ownership has exposed systemic weaknesses in the charging ecosystem.
One of the most persistent issues is the mismatch between supply and demand at public charging stations. During peak hours, popular stations often experience long queues, leading to extended wait times that exacerbate driver frustration and undermine the convenience of EV ownership. Additionally, the pricing structure for charging services has traditionally been static, failing to reflect real-time demand fluctuations. This lack of dynamic pricing reduces the grid’s ability to balance load and increases the risk of localized overloads during high-usage periods.
Zhang Wei’s study directly addresses these challenges by developing a mixed integer linear programming (MILP) model that evaluates charging options based on three primary cost components: travel cost, charging cost, and time cost. Unlike previous approaches that focus solely on minimizing distance or total energy consumption, this model incorporates the operational dynamics of charging stations, including queue length, service rate, and capacity constraints.
The foundation of the model lies in its realistic representation of charging station behavior. Drawing from queuing theory, the research treats each charging station as a finite-capacity service system where EVs arrive according to a Poisson process—a statistical model that accurately reflects the randomness of human driving patterns. The model assumes a first-in, first-out (FIFO) service discipline, meaning vehicles are served in the order they arrive. Each station has a fixed number of charging points, and when all are occupied, incoming vehicles must wait in a queue. Crucially, the model imposes a maximum allowable waiting time; if the expected wait exceeds this threshold, the driver is assumed to abandon the station—a behavior known as “balking” in queuing theory.
This finite queue assumption is essential for modeling real-world conditions, where drivers are unwilling to wait indefinitely. It also introduces a natural limit on station capacity, which depends not only on the number of chargers but also on the ratio between charger count and maximum queue length. This relationship allows the model to simulate congestion effects and predict how station performance degrades under heavy load.
To account for variations in energy consumption across different road types, the study incorporates a detailed energy consumption model calibrated for urban environments. The model distinguishes between expressways, arterial roads, secondary roads, and local streets, each with distinct speed profiles and energy demands. By integrating average vehicle speed into the calculation, the framework can estimate the energy required to reach any given charging station, ensuring that route feasibility is evaluated under realistic driving conditions.
Perhaps the most innovative aspect of the research is its dynamic pricing mechanism. Rather than relying on fixed time-of-use tariffs, the model introduces a congestion-based pricing scheme that adjusts the charging rate in real time based on station occupancy. The final price consists of two components: a base rate determined by the time of day (reflecting peak, off-peak, and shoulder periods), and a floating premium proportional to the current congestion level. The congestion index is calculated as the ratio of the number of vehicles in the system (both charging and waiting) to the station’s total capacity.
This approach creates a feedback loop between user behavior and system performance. When a station becomes crowded, the price increases, discouraging additional drivers from selecting it. Conversely, less congested stations remain relatively cheaper, incentivizing load balancing across the network. This not only improves the overall efficiency of the charging infrastructure but also enhances the user experience by reducing wait times and preventing overloads.
To make the model adaptable to diverse user preferences, the researchers employ an entropy-based weighting method to determine the relative importance of each cost factor. Instead of assigning arbitrary weights, the entropy approach analyzes the variability in the underlying data to objectively assign greater significance to factors with higher discriminatory power. For example, if travel distances to available stations vary widely, the distance factor will receive a higher weight; if prices are relatively uniform, the price factor will be downweighted.
This data-driven weighting ensures that the model remains responsive to actual conditions rather than preconceived assumptions. Moreover, users can customize their preferences by adjusting the relative emphasis on distance, cost, and time. A commuter in a hurry might prioritize minimizing wait time, while a budget-conscious driver might focus on reducing charging expenses. The model accommodates these differing priorities through a configurable weight vector, allowing for personalized route recommendations.
The practical application of the model was demonstrated through a simulation study conducted in a 25.98-square-kilometer area of Kunming, a major city in southwestern China. The test network included 42 nodes, 77 road segments, and nine public charging stations, with topological data extracted from Baidu Maps using GIS tools. The simulation assumed a fleet of battery electric vehicles (BEVs) with a uniform 60 kWh battery capacity and initial states of charge ranging from 10% to 50%. Charging efficiency was set at 90%, with fast chargers operating at 30 kW and slow chargers at 10 kW.
Two comparative strategies were evaluated: a “nearest station” heuristic, where drivers simply choose the closest available charger, and the proposed multi-objective optimization approach. In one illustrative scenario, a driver departing from node v20 at 5:00 PM faced several charging options. Under the nearest-station rule, the driver would have selected CS6, located just 1.8 kilometers away. However, the optimization model recommended CS8, which was slightly farther at 2.4 kilometers but offered significantly better conditions.
The results revealed a compelling trade-off: although the optimized route increased travel distance by 33%, it reduced expected waiting time by 42.6% and lowered charging costs by 5.8%. This outcome highlights the limitations of simplistic routing strategies that ignore congestion and pricing dynamics. By accepting a modest increase in travel distance, drivers can achieve substantial gains in convenience and affordability.
Further analysis explored how the model performs under different user priorities. Three distinct scenarios were tested: one emphasizing minimal travel distance, another prioritizing lowest charging cost, and a third focused on shortest waiting time. In each case, the model successfully identified the optimal station according to the specified objective. For instance, when minimizing distance was the primary goal, the selected station was 60.5% closer than in the cost-minimization scenario. Conversely, when cost was prioritized, the recommended station offered a 7.0% reduction in charging fees compared to the distance-optimized choice. Similarly, the time-minimization scenario resulted in a 71.9% reduction in expected wait time relative to the distance-focused strategy.
These findings confirm that the model is not only effective but also highly flexible, capable of adapting to a wide range of user needs and operational contexts. This adaptability is crucial for real-world deployment, where drivers have diverse motivations and constraints.
Sensitivity analyses further validated the robustness of the approach. When the relative weight of charging cost (w2) was increased compared to waiting time cost (w3), the model became more responsive to price fluctuations, steering drivers toward cheaper but potentially busier stations. However, beyond a certain threshold—specifically, when the w2:w3 ratio reached 1:3—additional increases in the time cost weight yielded diminishing returns in terms of overall cost reduction. This suggests that there is an optimal balance between price and congestion signals, beyond which further emphasis on one factor provides little marginal benefit.
Similarly, the impact of charging infrastructure investment was examined by varying the number of charging points per station. As expected, increasing the number of chargers reduced average waiting times, but the improvements followed a law of diminishing returns. At low charger counts, each additional unit significantly improved service speed; however, beyond six or seven chargers, the marginal gain in performance became negligible. This insight has important implications for urban planning and infrastructure investment, suggesting that blindly expanding charger numbers may not be the most efficient use of resources. Instead, a balanced approach that considers both hardware deployment and intelligent demand management is likely to yield better outcomes.
The study also contributes to broader discussions about smart grid integration and demand-side management. As EV penetration grows, uncoordinated charging could place immense strain on local distribution networks, particularly during evening peak hours when many drivers return home and plug in their vehicles. By incorporating time-varying electricity prices and congestion-based pricing, the proposed model helps align individual charging behavior with grid stability objectives. It encourages off-peak charging and distributes load more evenly across stations, thereby reducing the need for costly grid upgrades.
Moreover, the integration of real-time data into the decision-making process paves the way for future developments in connected and autonomous mobility. As vehicles become more connected, they can continuously monitor nearby charging station status and dynamically reroute based on changing conditions. Autonomous driving systems could even initiate charging maneuvers without driver input, selecting the best station based on a combination of energy level, traffic conditions, and predicted wait times.
Despite its strengths, the study acknowledges certain limitations. The current model operates on a static snapshot of the network and does not fully capture the temporal evolution of congestion or the strategic interactions among multiple drivers. Future work aims to incorporate historical charging behavior data to better predict user preferences and station utilization patterns. Additionally, the researchers plan to extend the model to include station siting and capacity planning, helping cities optimize the placement and sizing of new charging infrastructure.
Another promising direction is the integration of vehicle-to-grid (V2G) technologies, which allow EVs to feed electricity back into the grid during periods of high demand. While the current study excludes V2G due to its limited commercial deployment, future versions of the model could incorporate bidirectional energy flows, transforming EVs from passive consumers into active participants in energy markets.
In conclusion, Zhang Wei’s research represents a significant step forward in the quest to make electric vehicle ownership more convenient, affordable, and sustainable. By moving beyond simple distance-based routing and embracing a holistic view of charging costs—including travel, monetary, and time components—the model offers a more realistic and effective solution to range anxiety. Its use of entropy-based weighting and congestion-sensitive pricing ensures that recommendations are both data-driven and user-centric.
The implications extend beyond individual drivers. For city planners, the model provides a powerful tool for evaluating the performance of charging networks and identifying bottlenecks. For utility companies, it offers a pathway to smoother load profiles and reduced infrastructure stress. And for policymakers, it demonstrates how intelligent systems can support broader environmental and energy goals.
As the world transitions to cleaner transportation, innovations like this will play a crucial role in ensuring that the shift is not only environmentally sound but also socially equitable and economically viable. By making EV charging smarter, faster, and fairer, this research helps pave the way for a future where range anxiety is a thing of the past.
Zhang Wei, School of Information Engineering, Southwest University of Science and Technology, Journal of Sichuan University (Engineering Science Edition), DOI: 10.19907/j.0490-6756.2024.017002