Smart Charging Strategy Cuts EV Costs by 30%
A groundbreaking study from the Xining Power Supply Company of State Grid Qinghai Electric Power Company has unveiled a new optimization method that significantly reduces electric vehicle (EV) charging costs while enhancing grid stability. The research, led by engineers Zhao Jifang and Zhu Xiaoming, introduces a convex optimization-based charging scheduling model designed to reshape how EVs interact with urban power networks. Published in Microcomputer Applications, the findings offer a scalable solution to one of the most pressing challenges in the energy transition: managing the growing strain that mass EV adoption places on distribution systems.
As cities worldwide accelerate their shift toward electrified transportation, the surge in EV ownership is creating unforeseen stress on local power grids. Most EV owners charge their vehicles at home during evening hours, leading to concentrated spikes in electricity demand. This pattern, often referred to as “peak loading,” can overload transformers, degrade voltage quality, and increase operational costs for utilities. In extreme cases, it risks destabilizing entire residential networks, particularly in older infrastructure not designed for such dynamic loads. The problem is not hypothetical—it is already emerging in regions with high EV penetration, from California to Scandinavia. Without intelligent management, the dream of sustainable mobility could be undermined by its own success.
Zhao and Zhu’s work directly addresses this challenge by reimagining EV charging not as a passive, user-driven activity, but as a coordinated, system-wide optimization problem. Their approach hinges on the principle that charging does not need to happen instantly upon plug-in. Instead, by intelligently scheduling when and how fast each vehicle charges, it is possible to flatten demand curves, avoid costly peak periods, and ultimately reduce the total cost of energy delivery. The innovation lies in the mathematical framework used to achieve this: convex optimization.
Convex optimization is a branch of applied mathematics that deals with minimizing or maximizing functions under constraints, where the solution space is “convex”—meaning any line segment between two feasible solutions remains within the feasible region. This property ensures that any local minimum is also a global minimum, making the problem computationally tractable and reliable. In the context of EV charging, the objective function is the total cost of energy consumed by all vehicles over a 24-hour period. This cost is not fixed; it varies with time through real-time pricing mechanisms that reflect the actual cost of electricity generation and distribution at any given moment. During peak hours, when demand is high, electricity is more expensive. During off-peak hours, such as late at night, it is cheaper.
The model developed by Zhao and Zhu treats each EV as a flexible load that can be shifted within a user-defined time window—typically from arrival at home to departure the next morning. Each vehicle has a specific energy requirement based on its initial state of charge and the desired final charge level. The algorithm determines the optimal charging power profile for every vehicle, ensuring that all reach their target charge by their departure time, while collectively minimizing the total cost. Crucially, the model incorporates both the direct cost of electricity and the impact of EV load on the overall system, including network losses and voltage deviations.
The constraints are carefully designed to reflect real-world limitations. Charging power cannot exceed the vehicle’s maximum capacity or the circuit’s safety limits. The total energy delivered must meet the driver’s needs. And the system must remain within safe voltage ranges to prevent equipment damage and ensure power quality. By formulating these conditions as linear constraints and the cost function as a convex quadratic expression, the researchers transform a complex, nonlinear problem into one that can be efficiently solved using standard optimization tools.
To validate their model, the team conducted a detailed simulation based on the IEEE 4-node feeder system, a standard benchmark in power systems research. The setup represents a typical urban distribution network with 18 residential customers connected through a 63 kVA transformer. In this scenario, every household owns an EV with a 20 kWh battery and a 3 kW charging capacity. The baseline load—covering lighting, appliances, and HVAC—follows a realistic daily profile with a pronounced peak between 4:00 PM and 9:00 PM. Real-time pricing is modeled as a linear function of total demand, with higher rates during peak periods to reflect the true cost of supply.
Two charging scenarios were compared. In the first, representing uncontrolled behavior, every EV begins charging at maximum power as soon as it is plugged in, regardless of the time or electricity price. This mimics the current reality for most EV owners who prioritize convenience over cost or grid impact. In the second scenario, the convex optimization algorithm dynamically schedules charging, adjusting power levels across vehicles and time slots to minimize total cost.
The results were striking. Under uncontrolled charging, the peak system load reached 97.8 kW, exceeding the transformer’s 63 kVA capacity—a clear sign of potential overloading and voltage instability. The average load was 49.9 kW, yielding a load factor of 51%. The total charging cost for all vehicles amounted to 256.2 yuan (approximately $35 USD). Voltage levels dipped as low as 0.78 per unit, well below the acceptable threshold of 0.9, indicating poor power quality and risk of equipment malfunction.
In contrast, the optimized scenario transformed the system’s behavior. Peak load was reduced to 56.5 kW—a 42.2% decrease—bringing it safely within the transformer’s limits. The load factor improved to 88%, reflecting a much flatter and more efficient demand profile. Total charging cost dropped to 179.8 yuan, a savings of 76.4 yuan or 29.7%. Perhaps most importantly, the minimum voltage rose to 0.88 per unit, and the average voltage increased to 0.92, indicating a significant improvement in power quality and grid resilience.
These outcomes demonstrate that intelligent charging is not just about saving money for consumers—it is about preserving the integrity of the grid itself. By shifting load away from peak periods, the optimization effectively performs “valley filling” and “peak shaving,” smoothing out the demand curve and reducing the need for costly infrastructure upgrades. Utilities benefit from lower stress on transformers and feeders, while consumers enjoy lower bills and more reliable service. The environment gains as well, since a stable grid can better integrate renewable energy sources like solar and wind, which are inherently variable.
The study also highlights the role of financial incentives in shaping user behavior. While the model assumes full participation in the scheduling program, in practice, such coordination would likely require some form of compensation or pricing signal to encourage adoption. Time-of-use (TOU) tariffs, where electricity prices vary by hour, are already in place in many regions and serve as a natural mechanism to align consumer choices with system needs. The convex optimization framework can be integrated into utility-controlled charging platforms or smart home energy managers, allowing users to set their preferences—such as departure time and minimum charge level—and let the algorithm handle the rest.
One of the strengths of this approach is its scalability. The computational complexity of convex optimization grows predictably with the number of vehicles, making it suitable for deployment in neighborhoods, cities, or even entire regions. Unlike heuristic methods that may get stuck in suboptimal solutions, convex optimization guarantees the best possible outcome given the constraints. This reliability is essential for grid operators who must ensure safety and reliability at all times.
Moreover, the model is adaptable. It can be extended to include vehicle-to-grid (V2G) capabilities, where EVs not only draw power from the grid but also feed it back during peak demand, acting as distributed energy storage. It can incorporate renewable generation forecasts, allowing EVs to charge when solar or wind output is high, further reducing reliance on fossil fuels. And it can be combined with building energy management systems to optimize the use of on-site solar panels, batteries, and other loads.
The implications of this research extend beyond technical performance. It represents a shift in mindset—from viewing EVs as a problem to be managed, to seeing them as a resource to be harnessed. With the right policies and technologies, millions of parked cars can become a vast, distributed battery network, providing flexibility and stability to the grid. This vision is central to the concept of the “smart grid,” where digital intelligence enables a two-way flow of energy and information, creating a more efficient, resilient, and sustainable energy system.
Zhao and Zhu’s work contributes to a growing body of research exploring the intersection of transportation and energy. Previous studies have examined game-theoretic models, machine learning approaches, and decentralized control strategies for EV charging. What sets this study apart is its rigorous mathematical foundation and clear demonstration of tangible benefits in a realistic setting. By using a standard test system and well-defined metrics, the authors provide a benchmark against which other methods can be compared.
The findings also have policy relevance. As governments set ambitious targets for EV adoption, they must also invest in the enabling infrastructure—not just charging stations, but the software and market mechanisms that allow those stations to operate efficiently. Regulatory frameworks should encourage dynamic pricing, data sharing, and interoperability between vehicles, chargers, and utilities. Public awareness campaigns can help drivers understand the benefits of flexible charging, turning passive consumers into active participants in the energy transition.
Looking ahead, the next frontier may lie in real-world implementation. While simulations are valuable, the true test of any model is its performance in the field. Pilot programs involving hundreds or thousands of EV owners could validate the algorithm’s effectiveness, measure user satisfaction, and identify practical barriers to adoption. Integration with existing utility systems, cybersecurity considerations, and data privacy protections would all need to be addressed.
Another area for future research is the impact of heterogeneous user behavior. Not all drivers have the same flexibility. Some may need their vehicles fully charged immediately, while others can wait. Some may prioritize cost savings, while others value convenience above all. A robust scheduling system must account for these differences, perhaps through tiered service levels or incentive structures that reward flexibility.
Battery degradation is another factor that could be incorporated into the model. While modern EV batteries are designed for long life, frequent fast charging and deep cycling can accelerate wear. An advanced optimizer might balance cost savings against battery health, recommending slower, gentler charging when possible to extend vehicle lifespan.
In conclusion, the research by Zhao Jifang and Zhu Xiaoming offers a compelling solution to one of the most critical challenges in the electrification of transport. By applying convex optimization to EV charging, they have demonstrated a path toward lower costs, higher efficiency, and greater grid stability. Their work is a testament to the power of interdisciplinary thinking, combining insights from power systems engineering, operations research, and economics to tackle a complex real-world problem.
As the world moves toward a zero-emission future, the integration of EVs into the energy system will only become more important. Studies like this one provide the technical foundation for that integration, showing that with the right tools, the transition can be not only sustainable but also economically beneficial for all stakeholders. The road to a cleaner, smarter energy future is being paved—one optimized charge at a time.
Zhao Jifang, Zhu Xiaoming, Xining Power Supply Company of State Grid Qinghai Electric Power Company. Optimal Charging Scheduling Model of Electric Vehicle in Distribution Networks Based on Convex Optimization. Microcomputer Applications. DOI: 10.1007-757X(2024)07-0164-04