New Model Optimizes EV Charging Network Planning for Efficiency and Reliability
As electric vehicle (EV) adoption continues to accelerate globally, the need for a robust, efficient, and user-centric charging infrastructure has never been more critical. While the number of EVs on the road is growing at an unprecedented rate, the development of charging networks has often lagged behind, creating bottlenecks that hinder user convenience and, by extension, broader market growth. A recent study published in Power System Protection and Control introduces a novel, data-driven approach to EV charging network planning that simultaneously addresses two of the most pressing challenges: maximizing service coverage and minimizing the distance drivers must travel to find a charger.
Led by Professor Zhang Xinsong and his team from the School of Electrical Engineering at Nantong University, the research presents a multi-objective optimization model that breaks away from traditional, single-focus methodologies. Instead of prioritizing either traffic flow capture or charging distance in isolation, the new framework integrates both metrics, offering a more holistic and practical solution for urban planners and energy policymakers. The study, titled “Electric vehicle charging network planning considering captured traffic flows and charging driving distance,” leverages advanced computational techniques to account for real-world uncertainties, such as the variable initial charge levels of EV batteries, which directly influence driver behavior and network performance.
The core innovation of the model lies in its dual optimization objectives. The first objective is to maximize the minimum value of captured traffic flow. This is a strategic shift from simply maximizing average traffic capture. By focusing on the worst-case scenario—the lowest possible traffic flow the network might capture under uncertain conditions—the model ensures a high level of service reliability. This approach is rooted in robust optimization principles, designed to make the charging network resilient against the inherent unpredictability of driver behavior and vehicle states. As Zhang Xinsong explains, “A charging network should not only perform well on average but must also guarantee a baseline level of service in any given situation. Our model ensures that even in the most challenging conditions, a significant portion of potential EV traffic can be served.”
The second objective is to minimize the average charging driving distance. This directly translates to user convenience and service efficiency. Long detours to find a functioning charger are a major source of “range anxiety” and user dissatisfaction. By optimizing for shorter average travel distances to the nearest charging station, the model enhances the overall user experience, making EV ownership more attractive and practical. This dual focus—on both system-wide service capacity and individual user efficiency—represents a significant advancement in the field.
A key challenge in accurately modeling traffic flow capture is the uncertainty surrounding an EV’s initial state of charge (SOC) when it begins a journey. An EV starting with a full battery has a much longer potential range than one starting with a half-charged battery, which in turn affects whether it will need to charge during a trip and, consequently, whether it will be “captured” by a particular charging station. Previous models often treated this initial range as a fixed value, a simplification that can lead to inaccurate planning. The Nantong University team addressed this by employing the Monte Carlo simulation (MCS) method. This statistical technique runs thousands of simulations, each time randomly sampling the initial SOC based on a realistic probability distribution. By doing so, the researchers were able to map out the full range of possible traffic flow capture scenarios, revealing a significant spread between the minimum and maximum values. This probabilistic analysis provided the critical foundation for the first optimization objective, ensuring the planning model is based on a realistic understanding of system uncertainty.
The research also introduces a sophisticated constraint known as an “opportunity constraint” for the charging driving distance. Instead of imposing a rigid, non-negotiable limit on how far an EV can travel to reach a charger, the model uses a probabilistic threshold. For example, it can ensure that there is a 95% probability that an EV needing a charge will find a station within 80 kilometers. This flexibility is crucial for real-world planning. It allows for a more balanced and cost-effective solution, acknowledging that while most drivers should have easy access, a small fraction of edge cases can be tolerated without compromising the network’s overall performance. This approach moves away from overly conservative designs that might require an excessive number of stations, towards a more efficient and economically viable network.
To solve this complex, non-linear, multi-objective problem, the team utilized the Non-dominated Sorting Genetic Algorithm II (NSGA-II). This powerful evolutionary algorithm is specifically designed for such challenges, where multiple, often conflicting, goals must be balanced. Rather than producing a single “best” solution, NSGA-II generates a Pareto optimal solution set. This set contains a range of planning options, each representing a different trade-off between the two objectives. For instance, one solution might offer a slightly shorter average driving distance but a lower minimum traffic capture, while another might capture more traffic in the worst case but require drivers to travel a bit farther on average. This gives decision-makers the flexibility to choose a plan that best aligns with their specific priorities, whether that is maximizing user convenience, ensuring broad service coverage, or finding a balanced compromise.
The model was rigorously tested on a 25-node transportation network, a standard benchmark in such studies. The results were compelling. When compared to a randomly placed or conventionally planned network of four charging stations, the optimized network produced by the model significantly increased the minimum captured traffic flow, demonstrating its superior reliability. Furthermore, the average charging driving distance was minimized, confirming its efficiency. The study also conducted a sensitivity analysis, examining how key parameters affect the final outcome. For instance, lowering the confidence level of the opportunity constraint from 95% to 90% increased the number of viable solutions in the Pareto set, highlighting the trade-off between service reliability and planning flexibility. More importantly, the research showed that increasing the number of charging stations leads to diminishing returns. While adding stations initially provides large gains in both service coverage and reduced travel distance, the marginal benefit of each new station decreases as the network becomes more dense. This insight is invaluable for budget-constrained municipalities, as it helps identify the optimal point of investment before the cost of building new stations outweighs the benefits.
The implications of this research extend far beyond the academic realm. For city planners, it provides a powerful, evidence-based tool for designing future-proof charging infrastructure. For utility companies, it offers a way to anticipate and manage the load from EV charging more effectively by understanding where and when charging demand is likely to occur. For EV manufacturers and fleet operators, it can inform vehicle deployment strategies and route planning. The model’s focus on real-world uncertainties and practical trade-offs makes it a highly relevant and applicable framework.
The study also contributes to a broader shift in how we think about energy infrastructure. Traditionally, power systems and transportation systems have been planned and operated in silos. This research exemplifies the emerging field of “electrified transportation systems,” where these two domains are deeply intertwined. A charging station is not just a power outlet; it is a node in a complex network that influences traffic patterns, urban development, and energy consumption. By taking a systems-level approach, Zhang Xinsong and his colleagues are helping to build the integrated, intelligent infrastructure that a sustainable transportation future demands.
The work is particularly timely given the global push for decarbonization. As countries strive to meet their climate goals, the transition to electric mobility is a cornerstone strategy. However, this transition cannot succeed without a corresponding transformation of the supporting infrastructure. Poorly planned charging networks can create congestion, increase user frustration, and ultimately slow down EV adoption. This new model provides a clear path forward, ensuring that the charging network is not an afterthought but a well-optimized, reliable, and user-friendly system from the outset.
One of the strengths of the paper is its clear acknowledgment of the limitations of prior research. Many existing models focus solely on either maximizing the number of vehicles that can be charged (a traffic-flow-centric view) or minimizing the distance to the nearest charger (a spatial-distance-centric view). The authors correctly identify that these single-objective approaches are inherently limited. A network that captures a lot of traffic but forces drivers to make long detours is inefficient. Conversely, a network with many closely spaced chargers might minimize travel distance but could be underutilized and economically unsustainable. By combining these two objectives, the new model achieves a more balanced and realistic outcome.
The use of the FRLM (Flow Refueling Location Model) as a foundation for calculating captured traffic flow is also a sound choice. This well-established model considers the network of travel paths and the vehicle’s driving range, providing a more accurate picture of where charging demand will arise than simple population-based or distance-based models. By integrating FRLM with the Monte Carlo simulation for SOC uncertainty, the researchers have created a highly sophisticated and realistic simulation environment.
The practical value of the Pareto optimal solution set cannot be overstated. In the real world, decision-makers are rarely faced with a single, clear-cut answer. They must balance competing interests: budget, land use, environmental impact, and political priorities. The Pareto set provides a menu of options, each with its own performance profile. This allows for a more transparent and informed decision-making process. Stakeholders can see the explicit trade-offs and choose a solution that best fits their community’s unique needs and constraints.
Furthermore, the finding of diminishing returns with increased station density is a crucial piece of economic insight. It suggests that there is a “sweet spot” for investment in charging infrastructure. Beyond this point, spending more money yields progressively smaller improvements in service. This can help prevent wasteful spending and ensure that public and private funds are used as efficiently as possible. It also suggests that future investments might be better directed towards improving the speed of charging (e.g., more DC fast chargers) or enhancing the user experience (e.g., better payment systems and apps) rather than simply building more stations in already well-served areas.
In conclusion, the research by Zhang Xinsong, Zhu Chenxu, Li Daxiang, and Luo Laiwu from Nantong University represents a significant step forward in the science of EV charging network planning. By developing a model that simultaneously optimizes for service reliability and user efficiency, while rigorously accounting for real-world uncertainties, they have provided a powerful new tool for building the sustainable transportation systems of the future. Their work is a prime example of how advanced computational methods can be applied to solve complex, real-world problems with tangible societal benefits.
Electric vehicle charging network planning considering captured traffic flows and charging driving distance by Zhang Xinsong, Zhu Chenxu, Li Daxiang, and Luo Laiwu from the School of Electrical Engineering, Nantong University, published in Power System Protection and Control, DOI: 10.19783/j.cnki.pspc.231537