EV Smart Charging Strategy Cuts Grid Losses, Boosts Stability
A groundbreaking new strategy for managing electric vehicle (EV) charging and discharging is showing significant promise in addressing the growing strain that EVs place on urban power grids. As the global transition to electric mobility accelerates, the uncontrolled and random charging behavior of millions of EVs is creating serious challenges for distribution networks, including voltage fluctuations, increased power losses, and grid instability. A team of researchers from Jiaxing Hengchuang Electric Power Equipment Co., Ltd. has proposed an innovative “scheduled charging and discharging” model designed to transform this chaotic interaction into a coordinated, grid-supportive service. Their findings, published in the May 2024 issue of Zhejiang Electric Power, demonstrate that by leveraging advanced optimization techniques, EVs can not only be charged efficiently but can actively participate in stabilizing the grid, effectively turning a potential problem into a valuable asset.
The research, led by Cao Hangtao, along with colleagues Zhang Yong, Jiang Ning, and Lü Bin, directly confronts one of the most pressing issues in the modern energy landscape: the integration of high volumes of distributed, variable, and often unpredictable loads like EVs into aging distribution infrastructure. The “double carbon” goals set by many nations have catalyzed a surge in EV adoption. Projections cited in the study suggest that within the next decade, EV sales in China could grow by over 40% annually, potentially reaching a fleet of 300 million vehicles by 2040. This massive fleet represents an enormous mobile energy storage capacity, estimated at 200 billion kWh. However, if this charging demand is left unmanaged, it could lead to a scenario where the grid is overwhelmed during peak hours, causing blackouts and requiring costly infrastructure upgrades.
The current paradigm, often referred to as “dumb charging,” is fundamentally flawed. EV owners typically plug in their vehicles when they return home in the evening, coinciding with the peak demand period for household electricity. This concentrated load can cause voltage to drop at the end of distribution lines, increase resistive losses in power lines, and force utilities to rely on expensive and polluting peaker plants. While the concept of Vehicle-to-Grid (V2G) technology has been around for years, offering the potential for EVs to discharge power back to the grid, its real-world implementation has been limited. Most existing V2G strategies are based on reactive, real-time market signals or user preferences, which can be highly unpredictable. The aggregated charging and discharging requests from users are often random and do not align with the system’s operational needs, making it difficult for grid operators to plan and manage effectively.
The research team’s “scheduled charging and discharging” model offers a proactive alternative to this reactive chaos. The core idea is simple yet powerful: introduce a reservation system. Instead of charging at random, EV owners would pre-schedule their charging or discharging sessions with a charging station, providing information on their desired time window and energy requirements. This transforms the EV load from a stochastic variable into a predictable, dispatchable resource.
The proposed system operates through a multi-step process facilitated by modern communication networks like 5G or WiFi. It begins with the grid’s central dispatch center, which analyzes the forecasted load profile for the following day. Based on this forecast, the center identifies periods of low demand (valley periods) and high demand (peak periods) and sets corresponding time-of-use electricity prices. These peak and off-peak price signals are then communicated to all participating EV charging stations. An EV owner, planning their next day’s travel, would use a mobile app or web portal to submit a reservation request to their nearest charging station. This request would specify their vehicle’s arrival and departure times and whether they need to charge (typically during off-peak valley periods) or are willing to discharge (typically during on-peak periods).
The charging station acts as an aggregator, collecting and consolidating all reservation requests from its users. It then forwards this aggregated demand profile to the central dispatch center. This is where the sophisticated optimization takes place. The dispatch center runs a complex mathematical model that considers not only the scheduled EV demand but also the output from renewable distributed generation (RDG) sources like solar photovoltaic (PV) panels and wind turbines (WT) that are also connected to the same distribution network. The goal of this model is to minimize the total active power losses in the network and to keep voltage levels across all nodes within safe, stable limits. By treating the EV charging stations and the inverters of the PV and WT systems as controllable assets that can provide both active and reactive power, the model can find an optimal operating point for the entire system.
This is a significant advancement over previous approaches. Most prior research on V2G or smart charging focused solely on managing the active power flow—the actual energy being consumed or supplied. However, the Jiaxing team’s model is holistic, incorporating the coordinated optimization of both active and reactive power. Reactive power, though not consumed as energy, is essential for maintaining voltage stability. By allowing EV charging stations and renewable generators to also adjust their reactive power output, the system can actively correct voltage sags or swells, further enhancing grid resilience. The optimization variables in their model include the active and reactive power output of each EV charging station, as well as the reactive power output of the PV and WT systems. The primary objective is to minimize the sum of the active power losses across all the network’s branches over a 24-hour period.
Solving such a complex, non-linear, and non-convex optimization problem is a formidable computational challenge. The underlying physics of power flow in a distribution network are governed by equations that are inherently quadratic and difficult to solve to a guaranteed global optimum. The researchers addressed this by employing a sophisticated mathematical technique known as Second-Order Cone Relaxation (SOCR). This method transforms the original, intractable problem into a convex Second-Order Cone Programming (SOCP) problem, which can be solved efficiently and reliably using powerful commercial solvers like Gurobi. This choice of methodology is critical, as it ensures that the solution found is the true optimal one, avoiding the pitfalls of heuristic algorithms that might get stuck in sub-optimal local minima and produce inconsistent results.
To validate their theoretical model, the researchers conducted a comprehensive simulation study using a modified version of the standard IEEE 33-node test system, a benchmark network widely used in power system research. They enhanced this model with real-world geographical data from a region in Hangzhou, China, making the simulation more realistic. The network was outfitted with three EV charging stations located at strategic nodes, two wind turbine installations, and two solar PV farms. The simulation assumed 500 EVs, each with a 50 kWh battery, making reservations through the proposed system. The daily load and renewable generation profiles were based on actual forecast data for the region.
The results of the simulation were compelling. The study compared three distinct scenarios. The first, a baseline scenario, assumed no EVs, wind, or solar power—only the conventional “base load” of homes and businesses. The second scenario integrated EVs, wind, and solar, but managed them in a simpler way, only optimizing the active power of the EVs and ignoring the reactive power capabilities of any of the distributed energy resources. The third and final scenario implemented the full “scheduled charging and discharging” model with coordinated active and reactive power optimization.
The performance was evaluated using three key metrics: the daily peak-to-valley load difference, the total daily active power loss (ζ), and the system-wide voltage deviation (Г). The results clearly demonstrated the superiority of the proposed model. Compared to the baseline scenario, the second scenario, which only managed EV active power, reduced the peak-to-valley load difference by 40.60%, cut power losses by a remarkable 85.19%, and reduced voltage deviation by 70.74%. This alone shows the immense benefit of simply managing EV charging times.
However, the full model (Scenario 3) delivered even greater improvements. When compared to Scenario 2, it achieved an additional 57.00% reduction in power losses and a further 6.86% reduction in voltage deviation. This translates to a total power loss reduction of over 94% compared to the baseline. The “scheduled” EVs effectively performed a “peak shaving and valley filling” operation. During the off-peak valley hours (8 AM to 3 PM), the aggregated EVs charged their batteries, absorbing excess power and helping to stabilize voltage. During the peak hours (5 PM to 11 PM), the aggregated EVs discharged power back to the grid, reducing the load on the main supply and preventing voltage drops. The visualization of the system’s equivalent load curve showed a much flatter, more manageable profile after the integration of the scheduled EVs.
The impact on voltage stability was particularly noteworthy. In the baseline scenario, voltage levels across the network, especially at the far ends of the feeder, were often at the lower limit of the acceptable range (0.95 p.u.), indicating a stressed system. The second scenario improved this, but some nodes still operated near the upper (1.05 p.u.) or lower limits. In contrast, the third scenario, powered by the full optimization model, kept the voltage at nearly all 33 nodes remarkably close to the ideal 1.00 p.u., with a very uniform distribution. This level of voltage control is crucial for the reliable operation of sensitive electronic equipment and for preventing damage to grid infrastructure.
This research presents a compelling blueprint for the future of EV integration. It moves beyond the simplistic notion of EVs as just another load and instead positions them as intelligent, flexible grid resources. The “scheduled charging and discharging” model provides a practical framework for achieving this. It gives EV owners the predictability and control they desire—knowing when and how much their car will be charged—while simultaneously providing immense value to the grid. For utilities, this means reduced operational costs from lower power losses, deferred investments in new infrastructure, and a more stable and reliable network. For society, it means a smoother, more sustainable transition to electric mobility.
While the model shows great promise, the researchers are the first to acknowledge its limitations. Their simulation assumed perfect information and compliance from EV owners, which may not reflect real-world behavior where users might change their plans or “default” on their reservations. The model also simplified the user’s decision-making process, focusing only on proximity to a charging station and not accounting for traffic conditions or personal schedule changes. Furthermore, the study did not incorporate hard constraints on the physical capacity of specific grid components, which would be essential for real-world deployment.
Despite these limitations, the work represents a significant leap forward. It provides a robust, mathematically sound, and demonstrably effective solution to a critical problem. As EV adoption continues its exponential growth, the need for smart, coordinated management systems like the one proposed by Cao Hangtao and his team will become not just beneficial, but essential. Their research offers a clear path toward a future where millions of electric vehicles are not a threat to the grid, but a cornerstone of a smarter, more resilient, and more efficient energy system.
Zhang Yong, Jiang Ning, Lü Bin，Jiaxing Hengchuang Electric Power Equipment Co., Ltd., Zhejiang Electric Power, DOI: 10.19585/j.zjdl.202405005