Three-Layer Charging Strategy Balances Grid, Traffic, and Driver Needs
As electric vehicle (EV) adoption accelerates worldwide, the strain on power grids and urban transportation networks is becoming increasingly evident. With millions of EVs expected to be on the road in the coming decade, the challenge of managing their charging behavior has shifted from a technical curiosity to a critical infrastructure issue. Uncoordinated charging, particularly during peak hours, can lead to significant load fluctuations, increased power losses, voltage instability, and traffic congestion around charging stations. In response to these growing concerns, researchers at China Jiliang University have developed an innovative three-layer optimization framework designed to harmonize the competing interests of the power grid, urban road networks, and individual EV owners.
The study, conducted by Wang Haodong, Yu Jiangtao, and Zheng Di from the College of Mechanical and Electrical Engineering at China Jiliang University, introduces a holistic “vehicle-road-network” charging optimization method. Published in the Modern Electronics Technique, the research presents a comprehensive model that simultaneously addresses the stability of the distribution grid, the fluidity of the transportation system, and the economic burden on drivers. By integrating these three dimensions, the proposed strategy aims to transform EVs from potential disruptors into intelligent, grid-supportive assets that contribute to a more resilient and efficient urban ecosystem.
The core of the research lies in its multi-tiered approach, which recognizes that EV charging is not merely an electrical event but a complex socio-technical process involving energy, mobility, and user behavior. Traditional optimization methods have often treated these aspects in isolation. Some studies focus on grid-side management, using time-of-use pricing or direct load control to shift charging to off-peak hours. Others concentrate on navigation and routing, guiding drivers to the nearest or least congested charging station. However, these siloed approaches fail to capture the full picture. A route that minimizes driving time might lead to a charging station that is already overloaded, causing long queues and user dissatisfaction. Conversely, a charging strategy that reduces grid losses might inadvertently direct a large number of vehicles to a single location, creating traffic bottlenecks and negating the benefits of grid optimization.
The three-layer model proposed by the China Jiliang University team directly addresses this fragmentation. The upper layer of the model focuses on the macro-level operation of the power grid. Its primary objective is to minimize the variance of the system’s equivalent load over a 24-hour period. This “valley-filling and peak-shaving” goal is crucial for grid stability, as large fluctuations in demand can stress transformers, increase transmission losses, and raise the risk of blackouts. By optimizing the aggregate charging and discharging power of EVs across different time slots, the upper layer ensures that EV charging complements, rather than conflicts with, the natural load profile of the grid. This layer uses a Cuckoo Search (CS) algorithm, a nature-inspired metaheuristic known for its ability to escape local optima and find high-quality solutions in complex search spaces. The output of this layer is a schedule of total EV power demand for each hour, which serves as a guiding signal for the subsequent layers.
The middle layer operates at the level of the distribution network. It takes the total power demand from the upper layer as a constraint and then allocates this power among individual charging stations, optimizing both active and reactive power flows. This is a significant advancement, as most previous models have focused solely on active power. By incorporating reactive power, the model can more accurately represent the physics of the distribution system, leading to better management of voltage profiles and a more substantial reduction in active power losses. The objective of the middle layer is to minimize the total active power loss across all network branches, which directly translates to improved grid efficiency and lower operational costs for utilities. To solve this complex, non-linear optimization problem, the researchers employ Second-Order Cone Relaxation (SOCR) techniques, which convert the non-convex power flow equations into a convex form that can be efficiently solved using commercial solvers like CPLEX. The result is an optimal dispatch of power for each charging station, ensuring that the grid operates within safe voltage and current limits while minimizing energy waste.
The lower layer is where the strategy connects with the individual driver. It focuses on the user’s total cost of charging, which is a composite metric that includes not just the price of electricity but also the value of the driver’s time. This is a critical insight, as it acknowledges that for many EV owners, time is a precious commodity. The total cost is defined as the sum of the time spent driving to the charging station, the time spent waiting in line, and the direct cost of the electricity consumed. The model uses a Dijkstra algorithm to find the optimal path through the road network, represented as an adjacency matrix where each link has a weight based on travel time and congestion. The charging station’s queue length is modeled as a function of the arrival rate of vehicles, which is influenced by the decisions of all other drivers in the system. This creates a feedback loop: the choice of a charging station affects its queue, which in turn affects the total cost for future drivers. The lower layer uses the power allocation information from the middle layer to determine which stations are available and how much power they can deliver, ensuring that the driver’s chosen station can actually meet their charging needs. The output is a personalized charging plan that minimizes the driver’s total cost, balancing the trade-off between a longer drive to a less crowded station and a shorter drive to a potentially busier one.
The integration of these three layers is what gives the model its power. The layers are not independent; they communicate and influence each other. The upper layer sets the overall power schedule, which constrains the middle layer. The middle layer’s power allocation determines the availability and capacity of charging stations, which is a key input for the lower layer’s path planning. The lower layer’s decisions on which stations drivers will use feed back into the model, allowing for a more accurate prediction of load distribution. This closed-loop system enables a coordinated optimization that would be impossible with a single-layer approach.
To validate their model, the researchers conducted a detailed simulation study based on a real-world scenario in Hangzhou, China. They modeled a specific urban area with a complex road network divided into residential, commercial, educational, and recreational zones. A fleet of 1,000 EVs, all modeled as Zeekr 001 vehicles with a 100 kWh battery, was introduced into this network. The electrical infrastructure was based on the standard IEEE-33 node distribution system, augmented with five EV charging stations, two photovoltaic (PV) solar farms, and two wind turbines. This setup allowed the researchers to study the interaction between EVs and renewable energy sources, a key aspect of a sustainable future grid.
The simulation compared three different scenarios to demonstrate the effectiveness of their three-layer method. The first scenario was a baseline with no EVs connected to the grid. The second scenario represented a “vehicle-grid” approach, where EV charging was optimized for the grid’s benefit but without considering the road network or driver costs. The third scenario was the full “vehicle-road-network” model proposed in the paper.
The results were striking. Compared to uncoordinated charging, the three-layer optimization reduced the system’s equivalent load variance by 72.82%. This dramatic smoothing of the load curve means the grid experiences far less stress, reducing the need for expensive peaking power plants and improving overall reliability. More impressively, the system’s active power losses were reduced by 83.41%. This is a direct measure of efficiency; less energy is wasted as heat in the power lines, which translates to lower carbon emissions and lower electricity costs. The model also significantly improved voltage stability, reducing voltage deviation by over 72% compared to the uncoordinated case, ensuring that all users receive power within acceptable quality standards.
From the driver’s perspective, the benefits were equally compelling. The researchers compared three different charging behaviors within the optimized framework. In the first, drivers chose the station with the shortest driving time, ignoring queue lengths. In the second, they chose the station with the shortest waiting time, ignoring driving distance. In the third, they used the full model to minimize their total cost. Drivers who only minimized driving time ended up with long waits, making their total trip time and cost higher. Those who only minimized waiting time had to drive much farther, increasing their energy consumption and travel time. The drivers who used the three-layer optimization, however, achieved the lowest total cost, reducing it by 8.89% compared to the shortest-driving-time strategy and by 4.51% compared to the shortest-waiting-time strategy. This demonstrates that the model’s holistic approach leads to better outcomes for individual users by helping them make smarter, more informed decisions.
The success of this research lies in its practical and balanced approach. It does not seek to maximize one objective at the expense of others. Instead, it finds a Pareto-optimal solution where the grid, the transportation network, and the drivers all benefit. This is a crucial step toward building a truly sustainable and user-friendly EV ecosystem. The model’s ability to reduce grid losses and load variance makes it attractive to utilities and system operators. Its ability to prevent traffic congestion around charging stations makes it valuable for city planners and transportation authorities. And its ability to lower the total cost of ownership for drivers makes it appealing to consumers, thereby encouraging further EV adoption.
The implications of this work extend beyond the immediate technical results. It provides a blueprint for how future smart cities can manage the integration of new technologies. As more and more devices become connected and intelligent, from EVs to smart homes to autonomous vehicles, the need for integrated, multi-layered optimization will only grow. This research shows that by building models that reflect the interconnected nature of modern infrastructure, we can achieve outcomes that are greater than the sum of their parts.
While the current study is a significant advancement, the authors acknowledge its limitations. The simulation was conducted on a relatively small-scale network, and the model assumed a homogeneous fleet of EVs. In reality, urban areas have diverse vehicle types with different charging needs, battery sizes, and driving patterns. Future work will need to scale the model to larger, multi-regional networks and incorporate a more diverse set of vehicle and user behaviors. Additionally, the model could be enhanced by integrating real-time data from traffic sensors, weather forecasts, and grid monitoring systems to create a truly dynamic and adaptive optimization platform.
Despite these challenges, the three-layer charging optimization method represents a major step forward. It moves the conversation from simply managing EV charging to intelligently orchestrating it as a key component of a larger, integrated urban system. As the world continues its transition to electric mobility, strategies like this will be essential for ensuring that the benefits of EVs are realized without creating new problems. By balancing the needs of the grid, the road, and the driver, this research offers a practical and powerful solution for the challenges of the 21st-century city.
Wang Haodong, Yu Jiangtao, Zheng Di, China Jiliang University, Modern Electronics Technique, DOI: 10.16652/j.issn.1004⁃373x.2024.10.030