Smart Charging Network: How EVs Are Reshaping Urban Mobility and Grid Stability
The rise of electric vehicles (EVs) is no longer just a shift in automotive technology—it is a transformation of how cities manage energy, mobility, and infrastructure. As EV adoption accelerates globally, the interdependence between transportation networks and power grids has become a critical focus for researchers, utilities, and urban planners. A groundbreaking study published in Electrical Engineering explores a novel approach to optimizing EV charging station deployment by integrating mixed traffic flow dynamics with power distribution systems. Led by Gu Yingbin and his team from Chaozhou Power Supply Bureau of Guangdong Power Grid Co., Ltd., the research presents a comprehensive dual-layer planning model that redefines how charging infrastructure can be strategically designed to minimize societal costs while enhancing both traffic efficiency and grid reliability.
For years, EV charging station planning has largely focused on individual user needs or isolated grid impacts. Most early models treated charging stations as standalone facilities, often ignoring the complex interactions between driver behavior, road congestion, and electricity demand. However, as EVs become a significant portion of urban traffic, this siloed approach is no longer sufficient. The reality is that drivers do not make routing decisions based solely on battery level—they consider travel time, traffic conditions, charging availability, and even electricity pricing. At the same time, the surge in EV charging load affects voltage stability, power losses, and generation costs within distribution networks. This bidirectional relationship demands a holistic framework that captures the coupling between transportation and power systems.
Gu Yingbin’s team addresses this challenge head-on by developing a bi-level optimization model that simultaneously considers traffic assignment and optimal power flow. Their work stands out not only for its technical rigor but also for its practical relevance in real-world urban environments. Unlike previous studies that assumed homogeneous EV fleets or idealized network conditions, this research explicitly accounts for mixed traffic—where internal combustion engine vehicles (ICEVs) and EVs coexist on the same roads. This distinction is crucial because, despite rapid EV growth, most cities still operate with a majority of gasoline-powered vehicles. Ignoring ICEVs would lead to inaccurate traffic flow predictions and suboptimal charging station placement.
At the heart of the model is the concept of generalized social cost—a metric that aggregates multiple dimensions of system performance, including travel time, energy consumption, infrastructure investment, and grid operation expenses. The upper layer of the model aims to minimize this total cost by determining the optimal capacity of charging stations across the network. Capacity, in this context, refers to the power output per charging unit installed along key road segments. The lower layer then evaluates the consequences of these capacity decisions by solving two interdependent problems: the Traffic Assignment Problem (TAP) under user equilibrium conditions and the Optimal Power Flow (OPF) problem in the distribution grid.
The TAP component models how drivers choose their routes based on perceived travel costs, which include both time and monetary expenses such as charging fees. The model incorporates Wardrop’s user equilibrium principle, which assumes that no individual driver can reduce their travel cost by unilaterally changing routes. This behavioral assumption leads to a stable traffic pattern where all used paths between origin-destination pairs have equal and minimal cost. For EVs, the cost function includes not only travel time but also the price of electricity at charging stations, which varies depending on location and time of day. By embedding charging cost into route choice, the model captures the feedback loop between electricity pricing and traffic distribution.
On the power system side, the OPF problem ensures that the increased demand from EV charging does not compromise grid security. The researchers use a branch flow model with second-order cone relaxation to accurately represent power losses, voltage drops, and thermal limits in radial distribution networks. This level of detail is essential because unlike transmission systems, distribution networks are more sensitive to localized load increases. A poorly placed fast-charging station could cause voltage violations or overload transformers, leading to equipment damage or service interruptions. By integrating OPF into the planning process, the model prevents such scenarios by ensuring that every proposed charging station operates within safe electrical boundaries.
What makes this dual-layer framework particularly powerful is its iterative nature. The upper layer proposes a set of charging station capacities, which the lower layer uses to simulate traffic flows and power system responses. These results are then fed back to the upper layer to assess the resulting social cost. If the cost is too high or constraints are violated—such as exceeding the investment budget or failing to cover grid operation costs—the model adjusts the capacities and repeats the process. This feedback mechanism allows the solution to converge toward a globally optimal configuration that balances competing objectives.
To solve this computationally intensive problem, the team employs a surrogate model-based algorithm. Traditional optimization methods like gradient descent are ill-suited for this task because the relationship between charging station capacity and social cost is highly nonlinear and discontinuous. Instead, the algorithm builds an approximate model—called a surrogate—that mimics the behavior of the full simulation but requires far less computational effort. Using this surrogate, the optimizer can rapidly explore thousands of potential solutions, identifying promising candidates before validating them with the exact model. This hybrid approach significantly reduces computation time without sacrificing solution quality, making it feasible to apply the method to large-scale urban networks.
The effectiveness of the proposed strategy is demonstrated through a case study involving a ring-shaped transportation network connected to a radial distribution system. The test system includes eight candidate charging station locations, each linked to specific nodes in the power grid. Realistic parameters such as road capacity, free-flow travel time, electricity tariffs, and vehicle demand profiles are incorporated to reflect actual operating conditions. A typical daily traffic demand curve shows peak periods during morning and evening commutes, while the electricity price follows a time-of-use structure with peak, off-peak, and super-peak rates.
Two scenarios are compared: one using the optimized capacities derived from the model (Case 1), and another using a uniform average capacity across all stations (Case 2). The results reveal a clear advantage for the optimized approach. In Case 1, the total generalized social cost is approximately $444,140 per day, compared to $448,090 in Case 2—a savings of nearly $4,000 daily. While this may seem modest, when scaled to a metropolitan area with hundreds of charging stations, the cumulative savings could reach millions of dollars annually. More importantly, the reduction comes not from cutting corners but from smarter resource allocation.
Breaking down the components of social cost, the optimized scenario achieves lower traffic congestion, as evidenced by a reduction in total vehicle-hours traveled. This improvement stems from better alignment between charging station capacity and actual demand patterns. For instance, stations located on high-traffic corridors are assigned higher power ratings (350 kW), enabling faster service and reducing queue times. In contrast, stations in less congested areas are sized smaller (120 kW), avoiding unnecessary capital expenditure. This differentiated sizing strategy ensures that infrastructure investment is concentrated where it delivers the greatest benefit.
From a grid perspective, the optimized plan maintains voltage stability even during peak load conditions. Analysis shows that voltage deviations at all nodes remain below 7%, well within acceptable limits. This outcome highlights the importance of coordinated planning—without considering power system constraints, a seemingly efficient charging station layout could destabilize the local grid. Furthermore, the model ensures that revenue from electricity sales covers the operational costs of the distribution network, satisfying a key financial constraint.
One of the most intriguing findings from the study is the impact of EV charging behavior on marginal electricity prices. As the proportion of EVs increases, so does the average node marginal price—the cost of supplying one additional unit of electricity at a given location and time. This upward trend reflects the added strain on generation and transmission resources during peak charging periods. However, the study also suggests that this challenge can be turned into an opportunity through vehicle-to-grid (V2G) integration. When EVs are allowed to discharge back into the grid during high-price periods, they effectively act as distributed energy storage, helping to flatten demand peaks and reduce overall system costs.
This insight underscores a broader shift in how we view EVs—not merely as consumers of electricity, but as active participants in energy markets. With smart charging algorithms and dynamic pricing signals, EV owners can be incentivized to charge when renewable generation is abundant and discharge when demand is high. Such flexibility enhances grid resilience, reduces reliance on fossil-fuel peaking plants, and supports the integration of wind and solar power. The research by Gu Yingbin and colleagues lays the groundwork for this future by demonstrating how charging infrastructure can be designed to facilitate bidirectional energy flows.
The implications of this work extend beyond technical modeling. For policymakers, it provides a data-driven framework for evaluating public investments in EV infrastructure. Rather than relying on rules of thumb or political considerations, cities can now assess proposed charging station projects based on their contribution to overall societal welfare. For utility companies, the model offers a tool for proactive grid planning, allowing them to anticipate future load growth and upgrade infrastructure before reliability issues arise. And for automakers and charging network operators, the insights can inform business strategies, such as where to deploy ultra-fast chargers or how to structure pricing plans to influence driver behavior.
Moreover, the study highlights the importance of cross-sector collaboration. Traditionally, transportation agencies and electric utilities have operated independently, each focused on their own domain. But the electrification of transport blurs these boundaries, requiring new forms of coordination and data sharing. For example, traffic management centers could provide real-time congestion data to utilities, enabling them to predict charging demand spikes. Conversely, grid operators could share locational marginal prices with navigation apps, allowing drivers to factor electricity costs into their route choices. Such integration would create a truly intelligent mobility ecosystem.
Looking ahead, several extensions of this research are possible. One direction is to incorporate uncertainty in demand forecasts, renewable generation, and driver behavior using stochastic or robust optimization techniques. Another is to include battery degradation costs in the objective function, encouraging charging patterns that prolong EV battery life. Additionally, the model could be expanded to account for different types of charging technologies—such as conductive, inductive, or battery swapping—and their respective impacts on traffic flow and grid load.
Another promising avenue is the integration of microgrids and distributed energy resources (DERs) into the planning framework. Future charging stations may not only draw power from the main grid but also generate it locally through solar canopies or store it in stationary batteries. By co-optimizing the placement of chargers, photovoltaic panels, and energy storage systems, planners could create self-sustaining hubs that enhance local resilience and reduce transmission losses.
In conclusion, the research by Gu Yingbin, Huang Peifeng, Wang Juan, Tang Lize, and Huang Shuqiang represents a significant advancement in the field of sustainable urban infrastructure. By bridging the gap between transportation and power systems, their dual-level optimization model offers a powerful tool for designing EV charging networks that are not only technically sound but also economically efficient and socially beneficial. As cities around the world strive to decarbonize their transportation sectors, this kind of integrated thinking will be essential for building resilient, equitable, and low-carbon mobility systems.
The study demonstrates that the future of urban transportation is not just electric—it is interconnected, intelligent, and optimized. With the right planning tools and collaborative frameworks, we can ensure that the EV revolution delivers on its promise of cleaner air, quieter streets, and more reliable energy for all.
Gu Yingbin, Huang Peifeng, Wang Juan, Tang Lize, Huang Shuqiang, Chaozhou Power Supply Bureau of Guangdong Power Grid Co., Ltd., Electrical Engineering, DOI: 10.19426/j.cnki.cn11-4746/tm.2024.07.003