Shared Energy Storage: How EVs Are Becoming Virtual Batteries
The electric vehicle (EV) revolution is no longer just about replacing internal combustion engines. A new frontier is emerging where EVs transcend their role as mere transportation and evolve into dynamic, mobile energy assets. This transformation is not science fiction; it’s a rapidly advancing reality grounded in sophisticated energy management systems. Researchers at Shanghai Jiao Tong University are pioneering a model where parked EVs, often seen as a drain on the grid, are instead being harnessed as a vast, distributed network of “virtual batteries.” This innovative approach, detailed in a recent publication in the journal Automation of Electric Power Systems, promises to reshape the economics of renewable energy integration and address one of the most persistent challenges in the clean energy transition: energy storage.
The high cost of large-scale battery installations has long been a bottleneck for widespread renewable adoption. Solar and wind power are inherently intermittent, generating energy when the sun shines and the wind blows, which doesn’t always align with peak demand periods. To balance this supply and demand, massive investments in stationary battery storage are required, representing a significant financial barrier. The research led by Zhu Jianan, a graduate student, along with Professor Ai Qian and Dr. Li Jiamei, proposes a paradigm shift. Instead of relying solely on expensive, centralized storage, they advocate for a “generalized shared energy storage” system that aggregates the potential of existing, underutilized resources—specifically, the batteries of parked EVs and the thermal inertia of climate-controlled buildings.
This concept, known as Vehicle-to-Grid (V2G), has been discussed for years. The novelty of the Shanghai Jiao Tong team’s work lies in its comprehensive economic and operational framework. They don’t just suggest that EVs can feed power back to the grid; they have developed a detailed, mathematically robust model for how a commercial operator can manage this complex ecosystem to maximize value for all parties involved: the EV owners, the energy users, and the operator themselves. The core idea is to treat the collective battery capacity of a fleet of EVs as a single, virtual energy storage unit, which can be leased and dispatched just like a traditional power plant.
The researchers’ model begins with the fundamental challenge of uncertainty. Unlike a stationary battery with a known state of charge, an EV’s availability, its current charge level, and the owner’s driving schedule are highly unpredictable. An EV owner might need to leave for work earlier than expected, making their car unavailable for a scheduled discharge. This uncertainty has been a major deterrent for utilities and grid operators considering large-scale V2G programs, as it introduces risk and complicates planning. Previous optimization methods have struggled to balance this risk. Stochastic optimization relies on probabilistic forecasts, which can be inaccurate. Robust optimization, while safe, often leads to overly conservative and expensive solutions by planning for the worst-case scenario, which may never occur.
To overcome this, the team employed a cutting-edge technique called Distributionally Robust Optimization (DRO). This method is a sophisticated blend of the two, designed to handle uncertainty when the exact probability distribution of future events is unknown. Instead of assuming a specific distribution (like a normal bell curve), DRO considers a “fuzzy set” of all possible distributions that are reasonably close to the historical data. The distance between these possible distributions is measured using a mathematical concept called the Wasserstein distance, which is particularly adept at capturing the full shape and variability of complex, real-world data. By optimizing for the worst-case outcome within this entire set of plausible distributions, the DRO model provides a solution that is both economically efficient and highly reliable. It avoids the pitfalls of being too aggressive (like stochastic models) or too timid (like traditional robust models), striking a crucial balance between cost and risk.
The practical implications of this model are profound. In their case study, the researchers simulated a shared energy storage operator serving a mixed community of residential, commercial, and industrial users, all with their own renewable generation (solar and wind). They compared four scenarios: no storage, users owning their own batteries, a shared system with only physical batteries, and their proposed “generalized” system that includes EVs and thermal loads as virtual storage. The results were striking. When users owned their own storage, the total battery capacity required was massive—over 3,400 kilowatt-hours—and the cost savings were modest, around 5-10%. A shared physical-only system reduced the total capacity needed by nearly 30% and cut user energy costs by 30-40%, demonstrating the power of aggregation. However, when virtual storage from EVs and building climate systems was incorporated, the required physical battery capacity plummeted by another 40%, down to just 1,425 kilowatt-hours. This represents a massive capital cost saving for the operator and, by extension, lower service fees for the end-users.
This dramatic reduction is achieved through intelligent load shifting. The model doesn’t require EVs to discharge large amounts of power at peak times. Instead, it focuses on flexibility. For example, if an EV owner arrives at work at 9 a.m. with a 50% charged battery and plans to leave at 6 p.m., the “virtual battery” model defines a safe operating window. The car’s battery can be charged from 50% up to its maximum, or discharged down to a minimum level that still ensures the owner has enough charge to get home. This creates a buffer of usable energy. The shared operator can then use this buffer to charge the car during periods of low electricity prices or high renewable output (e.g., midday solar peak), and discharge it during high-price periods (e.g., evening peak demand). The same principle applies to thermal loads. By slightly adjusting the temperature in a building’s air conditioning system within the occupants’ comfort range, the system can “store” coolness during the day and reduce its power draw during peak hours, effectively acting as a thermal battery.
A critical component of this model is the concept of “satisfaction compensation.” For this system to work, EV owners and building occupants must be willing participants. The researchers recognize that asking someone to potentially delay their car’s charging or accept a minor temperature fluctuation has a cost—namely, a potential impact on their convenience and comfort. To ensure fairness and incentivize participation, the model includes a compensation scheme. The operator pays a fee to the virtual storage providers (EV owners and building managers) based on factors like the duration of their participation, the amount of flexibility they provide, and the degree to which their comfort is affected. For EVs, this includes a compensation for battery degradation from extra cycling. For thermal loads, it includes compensation for temperature deviations from the setpoint. This transforms the relationship from one of extraction to one of partnership. The EV owner is not just a passive node on the grid; they become an active participant in the energy market, earning revenue for the service their parked car provides.
The study found that the level of this compensation is a key economic lever. If the compensation is set too high, the operator’s costs rise, potentially making the entire service unprofitable. If it’s set too low, few people will opt in, and the virtual storage pool remains small. The researchers’ model allows operators to find the optimal balance. Their simulations showed that even with reasonable compensation payments, the overall savings from reduced physical storage needs and more efficient energy arbitrage far outweigh the costs, leading to a net profit for the operator and lower energy bills for consumers.
The success of this model hinges on trust and clear communication. For an EV owner, the idea of a third party managing their car’s battery can be daunting. What if the car isn’t charged when they need to leave? What if the battery wears out faster? The researchers’ model directly addresses these fears by building in strict constraints. The algorithm guarantees that the car will be charged to the owner’s desired level by their specified departure time. The compensation for battery degradation is a direct acknowledgment of this risk, ensuring the owner is not left bearing the cost. This focus on user experience is essential for widespread adoption. It moves the conversation beyond pure technology to one of user-centric service design.
The broader impact of this research extends far beyond a single academic paper. It provides a concrete, data-driven blueprint for a future where our energy infrastructure is more decentralized, resilient, and efficient. As EV adoption continues to skyrocket—projected to reach hundreds of millions globally in the coming decades—the collective battery capacity of these vehicles will dwarf that of all stationary storage combined. Harnessing even a fraction of this potential could eliminate the need for billions of dollars in new battery factories and power lines. It democratizes energy storage, turning every EV owner into a potential energy provider.
Furthermore, this model enhances grid stability. By aggregating thousands of small, flexible resources, the shared operator can respond to grid signals with incredible speed and precision, helping to balance frequency and voltage in real-time. This is a service traditionally provided by large, fossil-fuel-powered power plants. By replacing these with clean, distributed resources, the grid becomes not only greener but also more agile and less prone to large-scale failures.
The path to implementation, however, is not without hurdles. The technical infrastructure for bidirectional charging (V2G) is still nascent and not standard on most EVs or charging stations. Cybersecurity is a paramount concern, as any system that allows remote control of a vehicle’s battery is a potential target for hackers. Regulatory frameworks need to be established to define the rights and responsibilities of all parties, from the EV owner to the utility company. The business model itself needs to be proven at scale.
Despite these challenges, the research from Shanghai Jiao Tong University represents a significant leap forward. It moves the discussion from the theoretical possibility of V2G to a practical, economically viable system. It demonstrates that the solution to our energy storage problem may not be to build more giant batteries, but to better manage the ones we already have—those sitting in our driveways and garages. The humble parked EV, once seen as a static object, is poised to become a cornerstone of a smarter, more sustainable energy future.
The implications for automakers are also significant. This research underscores the need for them to design their vehicles not just for driving, but for energy management. Features like advanced battery management systems, secure V2G communication protocols, and user-friendly apps that allow owners to set their charging preferences and monitor their energy earnings will become key differentiators. The car of the future is not just a mode of transport; it is a mobile energy hub, a financial asset, and a vital node in a new, decentralized energy network. The work of Zhu Jianan, Ai Qian, and Li Jiamei provides a compelling vision of how this future can be built, one virtual battery at a time.
Zhu Jianan, Ai Qian, Li Jiamei, School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Automation of Electric Power Systems, DOI: 10.7500/AEPS20230619009