Smart Charging Strategy Turns EVs into Grid Allies
As electric vehicles (EVs) surge in popularity across China, their integration into the power grid presents both opportunity and challenge. While EVs offer a promising pathway toward decarbonization, their uncoordinated charging behavior threatens to destabilize power systems already grappling with the intermittency of renewable energy. A new study by researchers at State Grid Zhejiang Electric Power Co., Ltd. in Ningbo offers a breakthrough solution—transforming EVs from grid stressors into intelligent, distributed energy storage units through a novel aggregation control strategy.
Published in Power Demand Side Management, the research led by Xia Shizhe, Wang Aoqun, and Cai Menglu introduces a data-driven approach that leverages clustering algorithms and behavioral modeling to align EV charging with grid needs—without compromising user satisfaction. The findings, validated in the real-world setting of Lingfeng Zero-Carbon Evolution Park in Ningbo, demonstrate a dramatic reduction in supply-demand imbalances, paving the way for a more resilient and sustainable power system.
The rapid rise of EVs in China is undeniable. In 2022, they captured 25.6% of the automotive market, with production and sales increasing by nearly 95% year-on-year. This surge is a cornerstone of national climate strategy, but it brings significant strain on distribution networks. When thousands of EV owners plug in after work, typically between 6 p.m. and 9 p.m., they create a sharp spike in demand. This evening peak coincides with the decline of solar generation, creating a widening gap between energy supply and consumption—what grid operators call the “source-load imbalance.”
If left unchecked, this imbalance forces utilities to rely on fossil-fueled peaker plants or invest heavily in grid-scale battery storage, both costly and carbon-intensive solutions. The study highlights that unmanaged EV charging can increase the peak-to-valley difference in load by over 35%, and raise the variance of supply-demand gaps by more than 36%. In the case of the Lingfeng industrial park, integrating EV charging without control increased the maximum source-load gap from 36.7 MW to 52.9 MW—a 44% jump that could jeopardize grid stability.
But the research team saw not a problem, but a latent resource. “Every EV is a mobile battery,” said Xia Shizhe, lead author and senior engineer at Ningbo Power Supply Company. “With the right signals and controls, we can harness their collective storage potential to support the grid, especially during critical hours.”
The key lies in aggregation—managing a large fleet of EVs as a single, coordinated entity. This is where the innovation begins. Instead of imposing rigid charging schedules, which would frustrate users and reduce adoption, the team developed a strategy rooted in user behavior and machine learning.
The first step was to model the chaos. Using Monte Carlo simulation, the researchers generated thousands of realistic driving and charging scenarios for different EV types—private cars, taxis, and public buses—based on real-world statistical data. They factored in average speeds, trip durations, daily mileage, and the psychological tendency of drivers to charge when battery levels drop to around 15–40%, a range that balances range anxiety with practicality.
The simulation revealed a predictable pattern: two primary charging waves. The first, a midday fast-charge pulse around 2–3 p.m., often used by taxis and commercial fleets during breaks. The second, a much larger and more problematic wave, begins at 6 p.m. and extends into the night, driven by private vehicle owners returning home. This evening surge directly opposes the solar generation curve, which peaks at noon and drops to near zero by 7 p.m.
Armed with this data, the team turned to machine learning to identify optimal intervention points. They employed an enhanced version of the k-means++ clustering algorithm—a method known for its efficiency in handling large datasets. The goal was to segment the 24-hour day into distinct periods based on load characteristics: peak, shoulder, off-peak, and critical imbalance zones.
But here’s where the study diverges from conventional approaches. Rather than clustering only the base load or only the EV load, the researchers performed a dual clustering analysis—on both the total charging demand and the source-load difference. This allowed them to pinpoint “overlapping clusters”—time windows where high EV charging demand coincides with high grid stress.
The results were revealing. The most critical overlap occurred in the evening, between 6 p.m. and 9 p.m., where both EV charging and base load peak, while renewable output plummets. A secondary, though less severe, overlap appeared in the early morning, around 6:30 to 10 a.m., where residual charging demand meets rising industrial and commercial loads.
“This overlap is where we can achieve the highest impact with the least user disruption,” explained Wang Aoqun, one of the co-authors. “By focusing our control efforts only on these high-stress periods, we avoid unnecessary interference during times when the grid can easily absorb EV loads.”
The proposed strategy is twofold: price incentives and vehicle-to-grid (V2G) discharge. During the overlapping off-peak periods—specifically the midday lull when solar generation is high and demand is low—the system offers discounted charging rates. This encourages EV owners to charge earlier, using surplus renewable energy that might otherwise be curtailed.
The pricing mechanism is dynamic and transparent. It adjusts in real time based on the current grid load, with rates increasing during peak stress periods and decreasing during surplus periods. The algorithm ensures that price changes are proportional to grid conditions, avoiding extreme fluctuations that could alienate users.
For the most critical hours—those overlapping peak periods—the strategy goes a step further. Here, the researchers advocate for V2G technology, where EVs not only draw power from the grid but can also feed it back. By offering financial incentives, the system encourages EV owners with sufficient battery charge to discharge a portion of their energy back into the grid during peak demand.
This is not a call for all EVs to become mini power plants. The V2G component is targeted and limited. It applies only to vehicles that are parked, have adequate state of charge, and have opted into the program. The discharge is carefully managed to ensure that the vehicle retains enough energy for its next trip, preserving user confidence and mobility.
The success of such a strategy hinges on user acceptance. A control system that constantly overrides driver preferences will fail, regardless of its technical merits. The Ningbo team prioritized charging satisfaction from the outset. Their model does not ban charging during peak hours; it makes it less economically attractive while offering better alternatives. It respects user autonomy while guiding behavior toward collective benefit.
To test the strategy, the researchers applied it to the Lingfeng Zero-Carbon Evolution Park, a real-world industrial zone with 30 MW of existing solar capacity and plans for further renewable expansion. The site hosts 5,000 private EVs, 1,000 taxis, and 250 electric buses—representing a diverse and demanding fleet.
The outcomes were striking. After implementing the aggregation control strategy, the maximum source-load gap dropped from 52.85 MW to 27.66 MW—a reduction of nearly 48%. The peak-to-valley difference fell by 41.8%, and the overall variance in supply-demand imbalance decreased by 38.5%. Remarkably, these metrics not only improved from the uncontrolled EV scenario but also outperformed the baseline grid without any EVs.
“This means we’re not just mitigating the negative impact of EVs—we’re turning them into a net positive for grid stability,” said Cai Menglu, another co-author. “They’re actively helping to smooth the load curve and integrate more renewable energy.”
The implications extend far beyond Ningbo. As cities across China and the world electrify transportation, the lessons from this study offer a scalable blueprint. The use of clustering algorithms allows the strategy to adapt to different regions, load profiles, and EV adoption rates. The focus on overlapping periods ensures efficiency, while the combination of pricing and V2G provides flexibility.
Utilities can deploy such systems through smart charging platforms, integrated with existing demand response programs. For EV owners, participation can be seamless—built into navigation apps, charging networks, or utility billing systems. The financial incentives, though modest per vehicle, become significant at scale, creating a new revenue stream for EV owners while reducing system-wide costs.
Moreover, the strategy aligns with broader energy transition goals. By shifting EV charging to times of high renewable output, it increases the utilization of clean energy and reduces curtailment. By enabling V2G during peaks, it defers the need for new fossil-fueled plants and reduces carbon emissions. And by stabilizing the grid, it enhances reliability for all consumers.
The research also underscores the importance of data and modeling in modern grid management. Monte Carlo simulations capture the randomness of human behavior, while machine learning extracts actionable insights from complex load patterns. Together, they form a powerful toolkit for proactive grid planning.
One of the study’s strengths is its grounding in real-world conditions. The Lingfeng park is not a theoretical testbed but an active industrial zone with real energy flows and real user behaviors. The validation in this environment lends credibility to the findings and increases their transferability.
Still, challenges remain. Widespread V2G adoption requires compatible vehicles, bidirectional chargers, and robust communication infrastructure. Consumer trust must be built through transparency and fair compensation. Regulatory frameworks need to evolve to support dynamic pricing and grid services from distributed resources.
But the technical foundation is now clearer. The Ningbo study demonstrates that with the right algorithms and incentives, EVs can be a cornerstone of the new power system—one built on “source, grid, load, and storage.” In this vision, EVs are not just consumers of electricity but active participants in grid balancing, energy storage, and carbon reduction.
The road to a zero-carbon future is not just about replacing internal combustion engines with electric motors. It’s about reimagining the relationship between transportation and energy. This research shows that when EVs are intelligently managed, they become more than vehicles—they become vital nodes in a smarter, cleaner, and more resilient energy network.
As EV adoption continues to accelerate, the insights from this work will be increasingly relevant. Grid operators, policymakers, automakers, and charging network providers all have a role to play in scaling these solutions. The technology exists. The data supports it. The economic and environmental benefits are clear.
The transformation of EVs from passive loads to active grid assets is no longer a distant dream—it is a practical reality, being shaped today in places like Ningbo, one smart charging decision at a time.
Xia Shizhe, Wang Aoqun, and Cai Menglu, State Grid Zhejiang Electric Power Co., Ltd., Ningbo, China. Published in Power Demand Side Management, DOI: 10.3969/j.issn.1009-1831.2024.04.010