China’s Integrated Energy Systems Embrace EVs to Cut Costs and Emissions
By integrating electric vehicles into next-generation energy grids, Chinese researchers have demonstrated a planning framework that simultaneously reduces operating costs, slashes carbon emissions, and enhances energy autonomy—key milestones for a nation racing toward its 2030/2060 climate targets.
At the heart of this breakthrough is a multi-objective, bi-level optimization model developed by a team at Shanghai University of Electric Power. Their approach reimagines how integrated energy systems (IES) interact with demand—shifting from the traditional “supply follows load” paradigm to a dynamic “source-load interaction” model. This evolution is not merely theoretical; it’s engineered to accommodate the surging presence of electric vehicles (EVs) and their potential as flexible grid assets.
The stakes are high. As China adds millions of EVs to its roads each year, uncoordinated charging threatens to strain the grid, especially during evening peak hours. Yet, if intelligently managed, these same vehicles can become mobile energy reservoirs—absorbing surplus renewable power during the day and feeding it back when needed. The Shanghai team’s model formalizes this vision, embedding EV charging stations directly into the IES architecture and co-optimizing their behavior with heating, cooling, and electricity loads.
Unlike prior studies that treated cost or emissions in isolation, this research introduces a triad of competing objectives: minimize total economic cost, reduce carbon output, and maximize system autonomy—defined as the share of energy generated and stored locally, rather than imported from external grids or gas networks. The tension among these goals is real: greater autonomy and lower emissions typically demand higher capital investment. But the model doesn’t force a single “best” answer. Instead, it generates a Pareto frontier of 100 technically viable configurations, allowing planners to choose based on policy priorities or budget constraints.
Central to the system’s intelligence is its dual-layered demand response strategy. First, it activates “interruptible” and “transferable” loads across electricity, heating, and cooling domains. During supply shortages, non-essential thermal or electrical loads can be curtailed or shifted in time—actions compensated via market mechanisms. Second, and more innovatively, it deploys a dynamic electricity pricing scheme that responds in real time to the balance between renewable generation and total equivalent load.
When solar and wind output exceeds local demand, the algorithm automatically lowers charging tariffs at EV stations, incentivizing drivers—or their smart chargers—to draw power immediately. This isn’t static time-of-use pricing; it’s a live feedback loop that aligns EV behavior with renewable availability. In simulations, this strategy significantly flattened demand peaks and increased renewable utilization without compromising user convenience.
The technical implementation is equally sophisticated. To handle the inherent uncertainty of wind and solar output, the team employed sequence operation theory (SOT) to convert probabilistic constraints into deterministic equivalents—a method that preserves system reliability while avoiding overly conservative (and costly) reserve margins. For load forecasting, they trained a hybrid generative adversarial network (GAN) on historical data, conditioning scenarios by month to capture seasonal variations in heating and cooling demand. This yielded high-fidelity, diverse load profiles that better reflect real-world volatility than traditional clustering methods.
The bi-level structure separates long-term planning from short-term operations. The upper level decides equipment capacities—gas turbines, absorption chillers, battery and thermal storage units—over a 20-year horizon, optimizing for the three core objectives. The lower level simulates daily operations, co-optimizing the IES and EV charging station dispatch under the installed capacities and current pricing signals. This two-way coupling ensures that investment decisions are grounded in realistic operational performance.
The results are compelling. In a comparative analysis across four scenarios, the full source-load interaction model (featuring both dynamic pricing and multi-energy demand response) outperformed all alternatives. It reduced total economic costs by 5.9–12.1% compared to baseline cases, cut carbon emissions by 8.5–17.9%, and improved system autonomy by 3.0–5.2 percentage points. Notably, it achieved these gains while requiring less installed capacity in key equipment categories—such as gas turbines and electric chillers—than simpler models, proving that smarter coordination can substitute for brute-force infrastructure.
Seasonal case studies further validate the approach. In summer, excess solar power drives electric chillers and charges cold storage units during midday, displacing evening grid imports. In winter, combined heat and power units operate efficiently overnight, with thermal storage capturing waste heat for morning demand. Throughout the year, EVs act as shock absorbers: charging aggressively when renewables are abundant and curtailing or even discharging during tight supply periods.
The computational engine behind this innovation is a hybrid solver combining NSGA-II, a robust evolutionary algorithm for multi-objective problems, with CPLEX, a commercial-grade optimizer for mixed-integer linear programs. Benchmarks against alternative solvers (MOPSO and NSGA-II with hybrid heuristics) confirmed that this pairing delivers superior solution quality and faster convergence—critical for real-world deployment where time and accuracy are at a premium.
From a policy perspective, the implications are profound. As China advances its “dual carbon” strategy, local governments and state-owned utilities need actionable tools to design low-carbon, resilient energy hubs—particularly in industrial parks, university campuses, and new urban districts where integrated systems are most viable. This model provides a blueprint that balances economic pragmatism with environmental ambition.
Moreover, it reframes EVs not as burdens but as strategic assets. With over 20 million EVs already on Chinese roads—a number expected to triple by 2030—their collective battery capacity represents a vast, distributed energy resource. Harnessing this potential requires precisely the kind of coordinated planning this research enables.
Critically, the study acknowledges limitations. It assumes a single ownership structure for the IES and EV charging stations, whereas real-world deployments may involve multiple stakeholders with conflicting incentives. Future work could extend the model to include game-theoretic interactions among investors, operators, and consumers—a necessary step toward market-ready solutions.
Still, the core contribution stands: a technically rigorous, computationally efficient framework that demonstrates how deep integration of EVs and flexible loads can transform integrated energy systems from passive consumers into active, adaptive participants in the clean energy transition.
For global observers, the lesson extends beyond China’s borders. As nations worldwide electrify transport and decarbonize grids, the coupling of EVs with multi-energy systems will become increasingly vital. The Shanghai team’s work offers a replicable methodology—one that prioritizes not just cost or carbon, but system resilience through autonomy. In an era of energy volatility and climate urgency, that triad may well define the next generation of smart infrastructure.
Author: Dongdong Li, Lulu Wang, Wei Wang, Shunfu Lin, Bo Zhou
Affiliation: College of Electrical Engineering, Shanghai University of Electric Power, Shanghai 200090, China
Journal: Power System Technology, Vol. 48, No. 2, February 2024
DOI: 10.13335/j.1000-3673.pst.2023.0185