Electric Vehicles Help Cut Grid Losses and Emissions in New Study
A groundbreaking study led by Cheng Hulin from State Grid Hebei Electric Power Co., Ltd., in collaboration with researchers from Yanshan University, presents a novel strategy to reduce power grid losses and carbon emissions through coordinated optimization of transmission and distribution networks. The research, titled “Research on the Low-Carbon Strategy of Coordinated Loss Reduction of Transmission and Distribution Network Considering the Temporal-Spatial Distribution of EV Charging and Discharging,” introduces a framework that leverages the spatial and temporal flexibility of electric vehicle (EV) charging and discharging to enhance grid efficiency and support renewable energy integration.
Published in Electrical Measurement & Instrumentation (Vol. 61, No. 11, November 15, 2024), the study addresses one of the most pressing challenges in modern power systems: how to manage the growing penetration of distributed renewable energy sources while minimizing technical losses and operational costs. With solar and wind power becoming increasingly dominant in the energy mix, grid operators face new complexities in maintaining stability, ensuring economic dispatch, and reducing environmental impact. Traditional methods that treat distribution networks as passive loads or transmission grids as fixed sources are no longer sufficient in this dynamic environment.
Cheng Hulin and his team propose a three-layer hierarchical optimization model that redefines the interaction between transmission and distribution systems. At the core of their approach is the recognition that EVs are not just consumers of electricity but active participants in grid management. By strategically scheduling when and where EVs charge and discharge, the researchers demonstrate that it is possible to shape load profiles, flatten demand peaks, and improve power flow distribution—ultimately reducing network losses and increasing photovoltaic (PV) utilization.
The first layer of the proposed framework operates at the transmission level, where the primary objective is cost minimization. This includes minimizing generation costs from conventional units, penalties for curtailed renewable output, and demand response expenses, while maximizing revenue from electricity sales. A key innovation lies in the incorporation of transmission network losses into the calculation of clearing prices—the rate at which electricity is bought and sold in wholesale markets. Unlike conventional models that assume fixed or negligible losses, this study explicitly accounts for how power dissipation across long-distance lines affects pricing signals. As a result, the clearing price becomes a dynamic variable influenced by real-time grid conditions, creating a feedback loop with downstream distribution networks.
This bidirectional interaction is crucial. When the transmission system sends a price signal to distribution networks, it reflects not only supply and demand but also the physical cost of delivering power over distance. In turn, distribution networks respond by adjusting their internal dispatch strategies, including how much power they purchase from the main grid and how they manage local resources such as rooftop solar panels, microturbines, and EV fleets. These adjustments then feed back into the transmission system’s loss calculations and future pricing, forming a closed-loop optimization process that enhances overall system efficiency.
At the distribution level, the second layer of the model focuses on minimizing local operational costs, including fuel consumption for backup generators, penalties for solar curtailment, EV charging expenses, and—critically—network losses within the distribution feeder. Distribution grids, especially radial ones like the IEEE 33-node system used in the simulations, are particularly vulnerable to high losses due to their topology and the concentration of loads far from substations. The farther electricity travels, the more energy is lost as heat in conductors, a problem exacerbated during peak hours.
To tackle this, the researchers introduce a third layer: an EV spatial scheduling mechanism. While many existing studies optimize only the timing of EV charging—encouraging off-peak charging to avoid congestion—this work goes further by optimizing the location of charging and discharging activities across the distribution network. This spatial dimension is often overlooked, yet it has profound implications for voltage profiles, line loading, and ultimately, technical losses.
The spatial scheduling layer uses the time-optimal EV dispatch results from the upper two layers as input and then determines the best nodes (connection points) within the distribution network for EVs to plug in. The goal is to minimize total distribution losses by aligning EV charging with nodes closer to the substation (where voltage is higher and impedance is lower) and positioning discharging activities near heavily loaded downstream nodes. This strategic placement helps balance the load, reduce current flow in overloaded branches, and maintain voltage stability—all of which contribute to lower resistive losses.
For example, in the simulation based on the IEEE 33-node system, the optimized solution shows that EVs are predominantly charged at nodes 18–22 and 29–32 during nighttime hours when electricity prices are low and demand is minimal. These nodes were selected because they have relatively light existing loads, allowing them to absorb additional EV demand without causing thermal overloads. Conversely, during daytime hours, EVs discharge primarily at high-load nodes such as 5, 10, 31, and 32, effectively acting as distributed energy resources that supply power locally, reducing the need to draw current from distant sources and thereby cutting down on line losses.
The results are compelling. Compared to a traditional isolated dispatch approach—where transmission and distribution systems operate independently and EVs follow uncontrolled charging patterns—the proposed coordinated strategy reduces total system costs by nearly $7,000 per day in the test case. More impressively, it cuts transmission network losses significantly, as evidenced by a noticeable reduction in clearing prices across all 24 hours. Lower clearing prices indicate reduced marginal costs of supply, which are directly tied to transmission losses. In practical terms, this means less wasted energy and lower carbon emissions per unit of electricity delivered.
At the distribution level, the benefits are equally significant. The average network loss rate in the test distribution feeder drops from 0.24% under conventional operation to 0.21% with the new strategy—a seemingly small difference that translates into substantial energy savings when scaled across millions of nodes in real-world grids. Moreover, the improved load management enables higher PV penetration, increasing the solar consumption rate from 95.4% to 98.1%. This reduction in curtailment not only improves economic returns for solar owners but also maximizes the environmental benefits of clean energy generation.
One of the most notable aspects of this research is its computational efficiency. Solving a fully coupled transmission-distribution-EV optimization problem is notoriously difficult due to the high dimensionality and nonlinearities involved. Traditional iterative methods require frequent data exchange between subsystems, leading to long computation times and potential convergence issues. To overcome this, the team employs the Concurrent Subspace Optimization (CSSO) algorithm, a parallel computing technique that decomposes the global problem into smaller, independently solvable subproblems.
In this approach, the transmission and each distribution network are treated as separate subspaces. Each subsystem performs its own optimization using local data and shared boundary conditions (such as interconnection power flows and prices). Instead of waiting for real-time updates from other subsystems, they use response surface approximations—built using RELU neural networks—to predict how changes in their decisions will affect the broader system. These approximations allow for parallel computation, drastically reducing the number of iterations needed to reach a global optimum.
The performance gains are dramatic. While a conventional Hybrid Gradient Descent (HGD) method took over 1,100 seconds to converge, the CSSO-based solution achieved optimality in just 112 seconds—more than an order of magnitude faster. This speed-up is critical for practical implementation, where grid operators must make scheduling decisions in near real-time, especially in systems with high renewable volatility and responsive demand.
Beyond technical achievements, the study has important implications for policy and market design. It demonstrates that EVs can play a central role in grid decarbonization if properly integrated into system operations. Rather than being seen as a source of uncertainty and stress, EVs can become a valuable flexibility resource—provided that market signals and control architectures are designed to harness their full potential.
The findings also highlight the need for closer coordination between transmission system operators (TSOs) and distribution system operators (DSOs). Historically, these entities have operated in silos, with limited information sharing and misaligned incentives. The proposed framework suggests that by establishing standardized interfaces—such as shared pricing mechanisms and bilateral power schedules—it is possible to create a more cohesive, efficient, and resilient power system.
From a consumer perspective, the strategy aligns well with economic incentives. By charging at night and discharging during the day, EV owners can take advantage of time-of-use pricing, earning revenue through vehicle-to-grid (V2G) services while contributing to grid stability. In the simulation, the aggregated EV fleet earns approximately $412 per hour in arbitrage and ancillary service benefits, making participation financially attractive even without direct subsidies.
The research also underscores the importance of infrastructure planning. To fully realize the spatial benefits of EV scheduling, utilities may need to invest in smart charging stations equipped with communication capabilities, dynamic pricing interfaces, and location-aware control systems. Furthermore, regulatory frameworks should evolve to allow DSOs to actively manage distributed resources like EVs, potentially through aggregators or virtual power plants.
While the current study focuses on a simplified test system, the principles are scalable and applicable to real-world urban and suburban grids. As EV adoption continues to accelerate—projected to exceed 200 million vehicles globally by 2030—the ability to coordinate their charging behavior will become a cornerstone of sustainable energy systems.
Future work could extend the model to include other distributed energy resources such as home batteries, heat pumps, and industrial loads, creating a holistic framework for integrated demand-side management. Additionally, incorporating uncertainty modeling for renewable generation and EV availability would enhance the robustness of the solution under real-world conditions.
In conclusion, the study by Cheng Hulin, Wu Jian, Zhang Jing, Li Yanlin, Shi Benbo, and Lu Zhigang offers a comprehensive and computationally efficient solution to one of the most complex challenges in modern power systems. By integrating the temporal and spatial dimensions of EV flexibility into a coordinated transmission-distribution optimization framework, the researchers have opened a new pathway toward lower losses, reduced emissions, and higher renewable integration. Their work stands as a testament to the transformative potential of intelligent, data-driven grid management in the era of electrified transportation.
Cheng Hulin, Wu Jian, Zhang Jing, Li Yanlin, Shi Benbo, Lu Zhigang, State Grid Hebei Electric Power Co., Ltd. and Yanshan University, Electrical Measurement & Instrumentation, DOI: 10.19753/j.issn1001-1390.2024.11.016