Smart Charging Strategy Cuts Costs and Balances Grid Load
As electric vehicles (EVs) continue their rapid ascent in the global automotive market, the challenge of managing their impact on power grids has become increasingly critical. With projections indicating that EVs could account for up to 70% of China’s vehicle fleet by 2050, the strain on residential electricity networks is expected to grow significantly. Uncontrolled charging behaviors—especially during peak hours—threaten grid stability, increase electricity costs for consumers, and reduce overall system efficiency. In response, researchers from Nanjing University of Information Science & Technology and Wuxi University have developed a novel optimization strategy that not only reduces peak-to-valley load differences but also lowers user charging costs while boosting profitability for charging infrastructure operators.
Led by Professor Li Peng and Associate Professor Yu Bin, the team has introduced an advanced operational framework for energy storage-enabled charging stations, leveraging an improved multi-objective particle swarm optimization (IMOPSO) algorithm. Their findings, published in the Journal of Nanjing University of Information Science & Technology, present a comprehensive solution that aligns economic incentives with grid stability, offering a scalable model for smart urban energy management.
The research addresses a growing concern: as more households adopt EVs, unregulated charging patterns amplify the existing peaks in daily electricity demand. This phenomenon, known as “peak load amplification,” can lead to transformer overloads, voltage instability, and degraded power quality in residential areas. Moreover, when users charge during high-tariff periods, their electricity bills rise, creating a financial disincentive for EV adoption. Traditional solutions often focus on either grid-side regulation or consumer pricing mechanisms, but few integrate both perspectives effectively.
Li Peng and his colleagues propose a holistic approach that treats the residential power network—including base load and charging infrastructure—as a single, intelligent system. At the heart of this system is a dynamic scheduling model that coordinates the charging and discharging cycles of EVs and integrated energy storage units. The goal is twofold: minimize the fluctuation between peak and off-peak loads (known as peak-to-valley difference), and simultaneously optimize two competing objectives—reducing user charging expenses and maximizing revenue for charging station operators.
To achieve this balance, the team designed a time-of-use (TOU)-driven control strategy. Under this model, energy storage units within charging stations are charged during low-demand, low-price periods (typically overnight), and then discharge during high-demand, high-price windows. This process, commonly referred to as “valley filling and peak shaving,” helps flatten the overall load curve, reducing stress on the grid and lowering operational costs.
What sets this study apart is the sophistication of its optimization engine. The researchers enhanced the standard multi-objective particle swarm optimization (MOPSO) algorithm by introducing adaptive adjustments to key parameters such as inertia weight and learning factors. These modifications allow the algorithm to maintain a better balance between exploration and exploitation during the search process, avoiding premature convergence to suboptimal solutions—a common pitfall in complex, multi-dimensional optimization problems.
Additionally, the team implemented a dynamic position-splitting mechanism that disperses overly clustered particles in the search space, enhancing population diversity and improving the algorithm’s ability to locate the true Pareto front—the set of optimal trade-off solutions in a multi-objective context. This innovation significantly improves convergence speed and solution accuracy, particularly in high-dimensional, non-linear environments typical of real-world energy systems.
The performance of the IMOPSO algorithm was rigorously tested against benchmark functions, including Sphere, Schwefel’s 1.2, Rastrigin’s, Ackley’s, Hartman’s, and Shekel’s functions. Results showed that IMOPSO consistently outperformed both standard PSO and conventional MOPSO in terms of convergence speed and solution quality. Notably, it achieved global optimality in fewer iterations and produced a more evenly distributed Pareto front, indicating superior handling of multi-objective trade-offs.
In practical application, the model was tested in a simulated residential community served by a 110kV substation with a 30 MW capacity. The neighborhood, consisting of 1,500 households, had an EV penetration rate of 16%, with 60% of vehicles requiring daily charging. Each EV was assumed to have a 60 kWh battery capacity, and the charging infrastructure included energy storage units with a total capacity of 1,050 kWh.
Under unoptimized conditions, the community’s peak load reached 15.4 MW, with a minimum valley load of 7.8 MW, resulting in a peak-to-valley difference of 7.6 MW. After applying the IMOPSO-based scheduling strategy, the peak load dropped to 12.3 MW, while the valley load rose slightly to 8.9 MW, reducing the peak-to-valley difference to 3.4 MW—a remarkable 55% reduction. This outcome surpasses the performance of the original MOPSO algorithm, which achieved only a 30% reduction, representing a 36% improvement in optimization efficiency.
From an economic standpoint, the benefits are equally compelling. The optimized strategy reduced total user charging costs from 4,901.71 yuan to 3,808.37 yuan—a savings of 1,093.34 yuan, or approximately 22.3%. For charging station operators, the profit from energy arbitrage—buying low during off-peak hours and selling high during peak periods—reached 861 yuan per day. This dual benefit underscores the model’s ability to create a win-win scenario for all stakeholders: the grid gains stability, users save money, and infrastructure providers increase revenue.
The success of the strategy hinges on precise coordination of charging and discharging schedules. During off-peak hours (23:00–08:00), the system prioritizes charging the on-site energy storage units using low-cost electricity. During mid-peak periods (08:00–11:00 and 13:00–17:00), the storage units remain idle, and EVs are powered directly from the grid. During high-demand periods (11:00–13:00 and 17:00–23:00), the stored energy is discharged to meet EV charging demand, thereby reducing reliance on the grid during expensive tariff windows.
This dynamic allocation is made possible by real-time data collection and predictive modeling. The system continuously monitors base load patterns, EV arrival and departure times, battery state of charge (SOC), and user charging preferences. Based on this information, it calculates the optimal charging window for each vehicle, ensuring that SOC targets are met without violating grid constraints or exceeding user-specified time limits.
One of the key innovations is the integration of user choice into the optimization process. Drivers can opt for either “ordered” or “unordered” charging. Unordered charging prioritizes speed, allowing immediate full-power charging regardless of cost. Ordered charging, on the other hand, defers charging to off-peak or mid-peak periods in exchange for lower rates. The system provides users with a cost comparison, enabling informed decisions. In practice, most users are expected to choose the ordered option due to the significant savings—ranging from 20% to 30%—offered by the optimized schedule.
The scalability of the model is another strength. While tested in a single residential community, the underlying principles can be applied to larger urban districts, commercial fleets, or public charging networks. By aggregating multiple charging stations into a virtual power plant (VPP), the strategy could contribute to broader grid services such as frequency regulation, voltage support, and renewable energy integration.
Moreover, the model’s adaptability makes it suitable for different tariff structures and EV adoption rates. The researchers tested various scenarios, including higher vehicle penetration (up to 200 cars) and varying SOC requirements (from 50% to 90%). In all cases, the IMOPSO algorithm maintained robust performance, consistently reducing peak loads and user costs. However, the study notes that as demand approaches the physical limits of the charging infrastructure, the marginal benefits of optimization diminish. This highlights the importance of co-planning EV infrastructure deployment with grid upgrades.
Despite its success, the study acknowledges certain limitations. The current model assumes a fixed energy storage capacity of 1,050 kWh and does not account for the capital and maintenance costs associated with such systems. Future work will explore cost-benefit analyses across different storage configurations, aiming to identify the most economically viable setups for various community sizes and EV adoption levels.
Additionally, the model relies on accurate forecasting of user behavior and load patterns. In real-world deployment, uncertainties such as unexpected vehicle arrivals, changes in driving habits, or extreme weather events could affect performance. Integrating machine learning techniques for demand prediction and adaptive control could further enhance the system’s resilience.
The implications of this research extend beyond technical innovation. It offers a policy-relevant framework for urban planners, utility companies, and EV service providers seeking to manage the transition to electrified transportation. By demonstrating that intelligent charging can simultaneously improve grid reliability, reduce consumer costs, and generate new revenue streams, the study provides a compelling argument for investing in smart charging infrastructure.
In an era where sustainability and energy efficiency are paramount, the work of Li Peng, Yu Tianyang, Yu Bin, Zhou Chengwei, and Meng Wei represents a significant step forward. Their IMOPSO-based optimization strategy not only addresses the immediate challenges of EV integration but also lays the groundwork for a more resilient, responsive, and equitable energy future.
As cities worldwide grapple with the dual imperatives of decarbonization and digital transformation, solutions like this will be essential. The ability to harmonize the needs of individual consumers with the stability of the broader power system is no longer a luxury—it is a necessity. With continued refinement and real-world validation, this smart charging approach could become a cornerstone of next-generation urban mobility.
Li Peng, Yu Tianyang, Yu Bin, Zhou Chengwei, Meng Wei. Optimized operation strategy for energy storage charging piles based on improved multi-objective particle swarm optimization. Journal of Nanjing University of Information Science & Technology, 2024. DOI: 10.13878/j.cnki.jnuist.20220627002