New Strategy Optimizes EV Charging Station Expansion
The global automotive industry is currently witnessing an unprecedented transformation driven by the rapid adoption of electric vehicles. As sales figures climb year over year, the infrastructure supporting this revolution faces intense scrutiny. While vehicle technology advances, the layout and capacity of charging facilities often lag behind, creating bottlenecks that frustrate users and strain investment budgets. A recent breakthrough study addresses this critical imbalance by proposing a sophisticated planning method that harmonizes operator economics with user convenience through active charging guidance.
This significant research emerges from a collaborative effort involving leading academic and industrial institutions in China. The study, titled Research on Capacity Expansion and Planning for Urban Electric Vehicle Charging Station Based on User-side Active Charging Guidance, was published in the prestigious Proceedings of the CSEE. The research team includes LI Hengjie and XIA Yuxuan from Lanzhou University of Technology, YU Sumin, ZHAO Xiaojin, FENG Donghan, and ZHOU Yun from Shanghai Jiao Tong University, along with FANG Chen from the State Grid Shanghai Electric Power Company. Their work offers a compelling solution to the complex problem of urban charging infrastructure planning, leveraging real-time data and advanced optimization algorithms to reshape how cities prepare for an electric future.
The Challenge of Urban Charging Infrastructure
The core issue addressed by the researchers is the unreasonable layout of charging facilities. In many urban environments, charging stations are not distributed according to actual demand patterns or traffic flow dynamics. This misalignment leads to a dual negative impact. For operators, it results in increased construction costs due to inefficient capital allocation. For users, it severely affects the charging experience and travel efficiency. Drivers often face excessive queue times at popular stations while nearby facilities remain underutilized. This phenomenon, often referred to as charging anxiety, is becoming as significant as range anxiety in hindering widespread electric vehicle adoption.
Traditional planning methods have often focused on single-objective optimization. Some studies prioritize the investment and operational costs of the charging station operators, while others focus solely on minimizing the charging time costs for users. However, the research team argues that looking at these factors in isolation is insufficient. A comprehensive approach must balance the investment economy of the operators with the travel convenience of the users. This requires a model that can simultaneously account for grid capacity constraints, construction budgets, and the dynamic behavior of drivers navigating a complex urban road network.
A Bi-Level Planning Framework
To solve this multi-faceted problem, the team proposed a bi-level planning method. This framework divides the planning process into two interconnected layers: an upper investment planning layer and a lower user charging decision production simulation layer. This structure allows the model to capture the interaction between infrastructure development and user behavior, which is often overlooked in static planning models.
The upper layer functions as the investment planner. Its primary objective is to minimize the economic costs associated with charging stations. This includes the costs of expanding existing stations and the costs of building new stations at candidate locations. However, this economic optimization is not performed in a vacuum. The upper layer operates under strict constraints, including the capacity limits of the power grid access points and the maximum allowable queuing time for users. The goal here is to determine whether to expand an existing station or select a location for a new one, ensuring that any investment made yields the maximum benefit in terms of capacity and service quality.
The lower layer simulates the real-world behavior of electric vehicle users. Based on the planning scheme generated by the upper layer, this model focuses on matching charging demand with charging stations. The objective here is to minimize the total time cost for the electric vehicles. This total time cost is a composite metric that includes the travel time to the station, the queuing time upon arrival, and the actual charging duration. Crucially, this layer utilizes real-time traffic information obtained through digital map interfaces. This allows the model to account for the spatial and temporal differences in road network traffic flow, ensuring that the calculated travel times reflect actual driving conditions rather than theoretical distances.
The Role of Active Charging Guidance
A standout feature of this research is the integration of user-side active charging guidance. In many existing planning models, users are assumed to simply choose the nearest charging station. This nearest-neighbor approach often leads to clustering, where popular stations become overwhelmed while others remain idle. The researchers identified that this behavior causes conflicts in station selection and fails to account for queuing times before the driver even arrives.
The active charging guidance strategy proposed in this study acts as an efficient matching mechanism between electric vehicle users and charging stations. By considering station congestion, selection conflicts, and total time costs, the model transforms the matching problem into a maximum matching problem with minimum weights. This is described using a bipartite graph where the weights include travel time, queuing time, and charging time. The system does not simply direct a car to the closest plug; it directs the car to the plug that offers the lowest total time cost, factoring in the current traffic conditions and the predicted queue at the destination.
To make this simulation even more realistic, the researchers did not perform a global match for all charging demands at once. Instead, they set up decision intervals based on time segments. Electric vehicles are grouped according to when their charging demand is generated, and guidance is provided in batches. After each interval, the queuing time at each station is updated based on the remaining charging time of vehicles already at the station. This dynamic updating process mimics the real-time nature of urban charging networks, where availability changes by the minute.
Solving the Complexity with Sensitivity Factors
One of the significant hurdles in bi-level planning models is computational complexity. If every possible combination of station expansions and new builds were enumerated, the calculation time would grow exponentially as the system scale increases. This would make the model impractical for large urban areas. To overcome this, the team designed an iterative algorithm based on a sensitivity factor.
The sensitivity factor is defined as the ratio of the reduction in overall vehicle queuing time to the cost of expanding or building a specific station. In simpler terms, it measures how much benefit is gained per unit of currency spent. A higher sensitivity coefficient indicates that a particular investment will significantly improve charging queues relative to its cost. The algorithm works by evaluating all possible expansion or new build options in a given round, calculating their sensitivity factors, and selecting the option with the highest value. This process repeats iteratively. In each round, the system expands one existing station or builds one new station, updates the network status, and then re-evaluates the next best step.
This successive planning method s the logic of greedy algorithms but is tailored to ensure that the queuing time is reduced below a specific threshold with the lowest investment. It effectively decomposes the complex planning goal into manageable steps. Additionally, if multiple options have the same sensitivity coefficient, the method prioritizes the station with a larger power distribution margin, ensuring grid stability is maintained alongside economic efficiency.
Case Study Analysis: Minhang District
To validate the feasibility and effectiveness of their method, the researchers conducted case studies using real areas in Shanghai. The first case study focused on a test area in Minhang District. This region contained eight existing charging stations and three candidate locations for new stations. The candidate locations were strategically chosen to represent different scenarios: a central location, a sparse area, and a remote location, each with varying construction costs.
The simulation was run based on typical daily data from 2020, involving 183 electric vehicles generating charging demands over a 24-hour period. Before any planning optimization, the simulation revealed significant inefficiencies. The maximum queuing time for users reached nearly 70 minutes, with an average queuing time of over 23 minutes. This highlighted the severity of the congestion caused by the existing layout.
After applying the proposed capacity expansion and planning method, the results were dramatic. The maximum queuing time dropped to under 20 minutes, and the average queuing time fell to less than 7 minutes. The planning result indicated that the optimal strategy involved expanding all eight existing stations to varying degrees. Interestingly, the algorithm did not choose to build any new stations in this specific scenario. The analysis showed that while new stations could provide more charging piles at once, their high construction costs resulted in lower sensitivity coefficients compared to expanding existing infrastructure. This demonstrates the model’s ability to prioritize cost-effective solutions over brute-force capacity addition.
The study also analyzed the impact of electric vehicle penetration rates. Looking ahead to a 2025 scenario with higher vehicle density, the model predicted that some stations would reach their expansion limits. In this high-demand scenario, the algorithm did select a new station for construction, indicating that the method adapts dynamically to changing demand levels. Furthermore, the research examined the impact of charging speeds. When fast charging piles were replaced with slow charging piles in the simulation, the sensitivity of expanding certain stations dropped significantly. This underscores the importance of considering charging technology types in planning, as fast charging offers a distinct advantage in reducing queue times that outweighs the higher equipment costs in high-traffic areas.
Case Study Analysis: Pudong New Area
To test the scalability of the method, a second case study was conducted in a larger test area in Pudong New Area. This region included 86 charging stations, comprising 65 existing expandable stations and 21 candidate sites. The scale of this test was significantly larger, involving over 6,000 electric vehicles in a typical day.
The results from Pudong mirrored the success seen in Minhang but on a larger scale. Before planning, the maximum queuing time exceeded 100 minutes. After the optimization process, which involved expanding 17 specific stations, the maximum queuing time was reduced to under 20 minutes, and the average queuing time dropped to just over one minute. Visualizations of the charging load distribution showed a marked improvement in congestion levels. Before planning, many stations exhibited severe congestion during peak hours. After planning and the application of active guidance, the congestion levels were reduced to mild or moderate across the board.
This large-scale validation is crucial for industry stakeholders. It proves that the sensitivity factor-based iterative algorithm does not break down when applied to complex, real-world urban networks. The ability to handle dozens of stations and thousands of vehicles suggests that this method is ready for practical deployment by utility companies and city planners.
Comparative Advantages and Industry Implications
The research included comparative scenarios to highlight the value of active charging guidance versus traditional nearest-neighbor decision-making. In the nearest-neighbor scenario, users simply choose the closest station. While this initially reduces queue times, the benefits plateau as demand increases. Popular stations become saturated, and because users are not redirected to less congested alternatives, the maximum queue time remains high. The iteration process fails to converge efficiently.
In contrast, the active charging guidance scenario showed a linear decrease in queue times as the planning iterations increased. The system successfully balanced the load across the network. Utilization analysis revealed that without guidance, some stations were heavily overloaded while others were underused. With active guidance, the total configured capacity and average utilized capacity were evenly distributed. This balance is vital for maximizing the return on investment for charging infrastructure. It means that fewer total charging piles might be needed to serve the same number of vehicles if they are managed intelligently.
From an economic perspective, the method offers a substantial improvement in computational efficiency. Enumerating all possible planning schemes for a large network is computationally prohibitive. The proposed successive planning method reduces the number of calculations significantly while maintaining near-optimal results. In a test comparing the method against full enumeration for a smaller subset, the proposed method achieved a result that was better than 96 percent of all possible schemes, proving that the sacrifice in absolute optimality is negligible compared to the gain in speed and practicality.
Future Directions and Conclusion
The study concludes by acknowledging areas for future research. While the current model focuses on social welfare, balancing user satisfaction and economic costs, future iterations could incorporate more complex factors. These include collaborative planning between charging stations and the distribution grid, coordinated optimization of planning and operations, game-theoretic interactions between stations and vehicles, and the uncertainties inherent in charging behavior. Integrating these factors would further enhance the robustness of the planning method.
The implications of this research extend beyond academia. For utility companies like State Grid, the method provides a tool to plan grid upgrades more accurately, avoiding over-investment in areas where demand can be managed through guidance. For charging operators, it offers a pathway to maximize revenue by reducing wait times and increasing station turnover. For city planners, it presents a data-driven approach to zoning and infrastructure development that aligns with actual traffic patterns. For the end users, the ultimate benefit is a seamless charging experience that mirrors the convenience of traditional refueling.
As the electric vehicle market continues to mature, the focus must shift from simply building more chargers to building smarter networks. The work by LI Hengjie, ZHOU Yun, and their colleagues represents a significant step forward in this direction. By integrating real-time traffic data, active user guidance, and economic sensitivity analysis, they have created a framework that is both theoretically sound and practically viable. The successful application in Shanghai’s dense urban environment serves as a compelling proof of concept for cities worldwide grappling with similar infrastructure challenges.
The transition to electric mobility is inevitable, but its success depends on the invisible infrastructure that supports it. This research underscores that the future of charging lies not just in hardware, but in the intelligent software and planning strategies that optimize how that hardware is used. As urban centers grow denser and vehicle numbers swell, methods like these will become essential tools in the toolkit of sustainable urban development.
Authors: LI Hengjie, XIA Yuxuan, YU Sumin, ZHAO Xiaojin, FANG Chen, FENG Donghan, ZHOU Yun Affiliations: Lanzhou University of Technology; Shanghai Jiao Tong University; State Grid Shanghai Electric Power Company Journal: Proceedings of the CSEE DOI: 10.13334/j.0258-8013.pcsee.220233