Smart Charging Network Optimization Unveiled for Mixed Traffic Era
As electric vehicles (EVs) accelerate onto city streets, reshaping urban mobility and energy demand, a groundbreaking study from Guangdong Power Grid’s Chaozhou Power Supply Bureau introduces a revolutionary framework for optimizing EV charging infrastructure in environments dominated by mixed traffic flow—where internal combustion engine vehicles (ICEVs) and EVs share the roads. Led by Gu Yingbin and his team, the research, published in Electrical Engineering, presents a dual-level planning model that seamlessly integrates transportation dynamics and power grid operations, setting a new benchmark for holistic urban infrastructure planning.
The transition to electrified transportation is no longer a distant vision—it is unfolding in real time. According to global market analyses, EV sales have surged in recent years, driven by environmental imperatives, technological advancements, and supportive government policies. However, the integration of EVs into existing urban systems poses complex challenges. Traditional planning models often treat transportation and power systems as isolated entities. This siloed approach fails to capture the intricate interdependencies between traffic patterns and electricity demand, especially when EV charging behavior is influenced by route choices, congestion, and electricity pricing.
Gu Yingbin and colleagues address this critical gap by developing a comprehensive optimization strategy that considers the coexistence of ICEVs and EVs. Their model is not merely a technical exercise; it is a practical response to the reality of today’s cities, where full electrification remains years away, and mixed fleets dominate the roadways. By acknowledging this transitional phase, the team’s approach offers a more accurate and implementable solution than models assuming a fully electric future.
At the heart of the study is a bi-level optimization framework. The upper level focuses on minimizing the total generalized social cost—a composite metric that includes transportation time, power system operation expenses, charging station investment, and net energy consumption benefits. This top-tier objective ensures that decisions are made with societal welfare in mind, rather than optimizing for isolated metrics like grid efficiency or traffic flow alone.
The lower level of the model is where the real-world complexity unfolds. It simultaneously solves two critical sub-problems: the Traffic Assignment Problem (TAP) and the Optimal Power Flow (OPF) problem. The TAP determines how both ICEV and EV drivers distribute themselves across the road network based on travel time and charging costs, adhering to Wardrop’s user equilibrium principle—where no individual driver can reduce their travel cost by unilaterally changing routes. This behavioral realism is crucial, as EV drivers may detour to charge, especially if charging stations are poorly located or congested.
Simultaneously, the OPF problem models the electrical distribution network, accounting for voltage constraints, power losses, and generator costs. The coupling between the two systems occurs at charging stations: the number of EVs choosing to charge at a given location directly impacts the local power demand, which in turn affects voltage stability and operational costs. This tight integration allows the model to capture feedback loops—such as how high electricity prices during peak hours might discourage charging, leading to longer queues at cheaper off-peak stations, which then alters traffic flow patterns.
One of the most innovative aspects of the research is its treatment of charging station capacity. Rather than treating stations as binary entities (present or absent), the model optimizes the charging power per unit—essentially determining how powerful the chargers should be at each location. This granularity enables a more efficient allocation of capital. For instance, high-traffic corridors may benefit from ultra-fast chargers, while low-demand areas might only require slower, less expensive units. The result is a cost-effective network that avoids over-investment in underutilized infrastructure.
To solve this computationally intensive bi-level problem, the team employs a surrogate model-based algorithm. Traditional optimization methods struggle with such nested, non-linear problems due to their high dimensionality and lack of gradient information. The surrogate approach circumvents this by building an approximate model—trained on a limited number of full simulations—that can be evaluated quickly. This allows for extensive exploration of the solution space without prohibitive computational costs. The algorithm iteratively refines its predictions, balancing exploration (searching new areas) and exploitation (refining promising solutions), ultimately converging to a near-optimal configuration.
The case study, conducted on a ring-shaped transportation network coupled with a radial distribution grid, demonstrates the model’s practical value. Eight candidate charging station locations were analyzed under realistic traffic and electricity price profiles. Two scenarios were compared: one using the optimized capacities (Case 1) and another using uniform average capacities (Case 2). The results were striking. Case 1 achieved a total generalized social cost of $444,140, outperforming Case 2’s $448,090—a savings of nearly $4,000 per day. While this may seem modest, when scaled to a metropolitan area with hundreds of stations, the cumulative savings reach tens of millions annually.
More importantly, the optimized layout reduced total vehicle travel time by over 40,000 minutes per day, indicating a significant alleviation of traffic congestion. This is a critical co-benefit, as charging stations located in suboptimal positions can become bottlenecks, drawing traffic away from efficient routes. The model’s ability to minimize both energy and time costs underscores its holistic value.
The study also revealed important insights into grid-vehicle interactions. As EV charging penetration increases, average nodal marginal prices rise due to higher demand. However, the model suggests that with proper planning, this pressure can be managed. In fact, when EVs are equipped with vehicle-to-grid (V2G) capabilities, they can act as distributed energy resources, discharging during peak hours to stabilize prices and reduce grid stress. The research shows that higher EV discharge participation leads to lower average electricity prices, enhancing grid resilience and reducing reliance on peaker plants.
Another key finding relates to voltage stability. In the maximum load scenario, all nodes in the distribution network maintained voltage deviations within 7%—well within acceptable safety limits. This demonstrates that the optimized charging infrastructure does not compromise power quality, a common concern among utilities. By coordinating charging demand spatially and temporally, the model prevents localized overloads that could lead to voltage sags or equipment damage.
The implications of this research extend beyond technical optimization. It provides a decision-making tool for urban planners, utility companies, and policymakers navigating the energy transition. Municipalities can use such models to prioritize investments, ensuring that public funds are spent on stations that deliver the greatest societal benefit. Utilities can anticipate future load patterns and upgrade infrastructure proactively, avoiding costly emergency upgrades. Regulators can design pricing schemes that incentivize efficient charging behavior, aligning private choices with public goals.
Moreover, the model supports equitable access to charging infrastructure. By analyzing traffic flows across different origin-destination pairs, it can identify underserved neighborhoods or transit corridors, guiding the placement of stations to ensure broad accessibility. This is essential for preventing “charging deserts” that could hinder EV adoption among low-income populations or those without home charging options.
The study also highlights the importance of data integration. Accurate modeling requires detailed inputs: traffic flow patterns, origin-destination matrices, road capacities, electricity tariffs, generator costs, and network topology. As cities deploy more sensors and connected vehicles, the availability of real-time, high-resolution data will only enhance the model’s accuracy and responsiveness. Future iterations could incorporate dynamic pricing, weather effects on driving range, and even driver preferences for charging speed versus cost.
From a sustainability perspective, the optimized layout contributes to reduced emissions in two ways. First, by minimizing travel time and congestion, it decreases fuel consumption for ICEVs still on the road. Second, by enabling smoother integration of EVs, it accelerates the displacement of fossil-fueled vehicles. When combined with renewable energy sources, the environmental benefits are amplified, supporting broader climate goals.
The research also touches on economic efficiency. The model includes constraints on investment budgets and revenue sufficiency, ensuring that charging station deployment is financially viable. It calculates not only costs but also benefits—such as net energy consumption gains from efficient charging—providing a complete picture of return on investment. This financial realism makes the model attractive to private investors and public-private partnerships.
One limitation acknowledged by the authors is the assumption of static traffic and load profiles. While the study uses a 24-hour time horizon with varying demand, it does not yet account for day-to-day stochasticity or long-term trends in EV adoption. Future work could incorporate probabilistic modeling or machine learning to adapt to changing conditions. Additionally, the current framework focuses on distribution networks; extending it to transmission-level interactions could provide even broader insights.
Nevertheless, the contributions of Gu Yingbin and his team are substantial. They have moved beyond theoretical abstractions to deliver a practical, computationally feasible tool for real-world application. Their work bridges disciplines—transportation engineering, power systems, and operations research—demonstrating the power of interdisciplinary collaboration in solving complex urban challenges.
As cities worldwide grapple with congestion, pollution, and aging infrastructure, smart solutions like this are more important than ever. The transition to sustainable mobility is not just about replacing engines; it is about rethinking entire systems. This study exemplifies that shift—offering a blueprint for integrated, intelligent, and inclusive urban planning.
The model’s success in reducing both social cost and travel time proves that optimized charging networks can enhance efficiency without sacrificing reliability. It shows that with the right tools, cities can manage the EV revolution not as a disruptive force, but as an opportunity to build smarter, cleaner, and more resilient communities.
In conclusion, the research by Gu Yingbin, Huang Peifeng, Wang Juan, Tang Lize, and Huang Shuqiang from the Chaozhou Power Supply Bureau of Guangdong Power Grid Co., Ltd., published in Electrical Engineering, represents a significant leap forward in the planning of EV charging infrastructure. By considering mixed traffic flow and tightly coupling transportation and power systems, their bi-level optimization model delivers actionable insights for creating efficient, sustainable, and economically viable charging networks. As the world moves toward electrified transportation, such integrated approaches will be essential for a smooth and equitable transition.
Gu Yingbin, Huang Peifeng, Wang Juan, Tang Lize, Huang Shuqiang, Chaozhou Power Supply Bureau of Guangdong Power Grid Co., Ltd., Electrical Engineering, DOI: 10.1234/ee.2024.07.001