Three-Player Energy Game Cuts Grid Emissions—and Rewrites EV’s Role in China’s Carbon Market
A new dispatch model pioneered in China is turning electric vehicles from passive chargers into active grid assets—while cutting regional carbon costs by over 37 percent.
In a quiet corner of eastern China, three regional energy hubs—each feeding factories, data centers, and suburbs with electricity and natural gas—began trading energy not just like utilities, but like strategic partners. One had surplus solar at noon; another had idle gas-fired capacity overnight; the third, a fleet of office commuters’ EVs returning home with 80 percent battery left. Previously, each managed its own load, bought power from the local distribution grid at fixed rates, and paid carbon penalties when fossil output spiked. Now, guided by real-time pricing and a shared optimization algorithm, they trade electricity and synthetic gas—produced via power-to-gas from captured CO₂—across shared lines and pipelines. The result: a 37.9 percent drop in net carbon trading costs, a 16.2 percent rise in wind-solar utilization, and a surprising new role for EVs—not just as demand, but as dispatchable storage and carbon-offset generators.
This isn’t speculative futurism. It’s the outcome of a master–slave game model developed by researchers at Nanjing Institute of Engineering and Aalborg University, tested on a modified IEEE 14-node distribution network with three interconnected Regional Integrated Energy Systems (RIES). The architecture treats the Active Distribution Network (ADN) not as a passive conduit, but as a price-setting leader. The RIES coalition—comprising co-located electric, thermal, and gas assets, plus EV fleets, batteries, and carbon-capture units—acts as the response-optimizing follower. Through this two-layer game, the ADN sets time-varying buy/sell electricity tariffs; the RIES alliance jointly minimizes its total energy and carbon cost, leveraging multi-energy sharing, demand response, and flexible resources. Finally, gains from cooperation are split fairly among RIES members using Nash bargaining—ensuring no participant is worse off than going it alone.
The implications extend far beyond academic elegance. As China’s national carbon market tightens (coverage now includes over 2,200 power plants, representing ~4.5 billion tons of CO₂ annually), regional energy operators face mounting pressure to decarbonize without sacrificing reliability or margin. This model offers a blueprint: transform siloed assets into a collaborative, price-responsive bloc—where EVs, carbon capture, and synthetic fuel aren’t add-ons, but core dispatch levers.
At the heart of the innovation is a redefinition of flexibility. Traditionally, demand response meant nudging industrial users to shift loads by a few hours, or throttling air-conditioning in commercial buildings. Batteries provided short-duration buffering. EVs? Largely ignored in system planning—or treated as an unpredictable surge threat to evening grids.
Here, flexibility is multi-vector, multi-temporal, and co-optimized.
Consider the EV integration layer. Rather than modeling EVs as static load profiles, the team built a dynamic charging window for each vehicle—based on arrival/departure times, minimum required state-of-charge, and maximum charge/discharge rates. Crucially, they embedded EVs directly into the carbon accounting framework. Every kWh discharged back to the grid—via vehicle-to-grid (V2G) during evening peaks—is assigned a negative carbon equivalent. Why? Because that kWh offsets higher-emission generation elsewhere: coal peakers, imported grid power, or even gas turbines operating above their efficient range. In one simulation peak at 21:00, EV discharging from RIES-1 alone reduced local CO₂ emissions by 1.8 tons per hour—equivalent to taking 85 gasoline sedans off the road for that hour.
Even more striking is how the model coordinates EV action across the coalition. When RIES-2 faces a gas-demand spike—but low local electricity surplus—it can request “energy imports” from RIES-1 in the form of electricity (to run its own P2G unit) or synthetic methane (produced earlier by RIES-3 using excess midday solar and captured CO₂). EVs in RIES-1 charge during midday; discharge into the grid at 19:00; part of that power flows to RIES-3’s P2G plant, which converts it—plus CO₂ captured from RIES-2’s gas turbine—into pipeline-ready methane. That gas then flows back to RIES-2, displacing a purchase from the city mains (whose upstream emissions are high). The loop closes: clean electrons, captured carbon, and EV mobility jointly enable a net-negative carbon transaction.
The carbon-trading layer makes this economically rational. Under the baseline allocation method used in China’s ETS, each RIES receives free allowances proportional to its historical output—but faces penalties for exceeding them. The model quantifies four offset streams:
- Conventional unit reductions: Lower output → fewer emissions → lower penalty.
- Grid import displacement: Charging EVs or P2G from clean ADN imports (e.g., upstream hydro/wind) yields lower embedded CO₂ than self-generation using coal/gas.
- EV V2G credits: Discharged energy is treated as avoided emissions—calculated as the difference between grid-average intensity and zero for battery discharge.
- Direct CCS savings: Every ton of CO₂ captured and used in P2G (instead of vented) counts as a negative emission—and avoids the purchase of carbon permits.
In Scenario 4—the full Stackelberg game with flexibility—total carbon trading costs across the three RIES fell to negative $22.5 thousand, meaning the coalition earned carbon credits overall. Compare that to Scenario 1 (no flexibility, no cooperation): a net cost of $129.3 thousand. That swing—over $150K in avoided/earned carbon value over 24 hours—is enough to fund significant infrastructure upgrades.
Price signals are the conductor of this orchestra. The ADN doesn’t just broadcast a flat “off-peak/peak” tariff. It computes dynamic buy and sell prices—constrained only by its own purchase cost from the upstream transmission grid and a ceiling on average selling price (to prevent gouging). The optimal result? A highly responsive curve that mirrors—but amplifies—the underlying cost volatility.
During grid valley hours (00:00–08:00), the ADN sets low purchase prices (as low as $34/MWh) and moderate sell prices ($52/MWh). This encourages RIES units to:
- Minimize local gas turbine run-time (avoiding both fuel cost and carbon liability),
- Maximize EV and battery charging (storing low-cost electrons),
- Run P2G at full capacity (electricity is cheap; synthetic gas displaces expensive, high-carbon pipeline imports later).
At 14:00–18:00, as solar wanes and industrial demand climbs, the ADN purchase price spikes ($98/MWh), but sells only at $78/MWh—effectively rewarding RIES for exporting surplus. RIES respond by:
- Discharging EVs and batteries into the grid,
- Coordinating gas-load sharing (e.g., RIES-1 ships synthetic methane to RIES-2, which reduces its gas turbine load),
- Using captured CO₂ to buffer emissions from unavoidable generation.
The effect is twofold: load flattening for the ADN (reducing its need for costly spinning reserves or upstream imports), and profit generation for RIES—even as they cut emissions.
Critically, the model proves cooperation isn’t optional. In Mode 2 (coalition without Nash bargaining), two RIES saw cost reductions—but the third increased its operating cost by $261 versus going solo. Rational actors would defect. Only with Nash-based profit sharing—where surplus is redistributed proportionally to each member’s marginal contribution to the coalition’s gains—do all three achieve lower costs than independence. Trust is engineered into the math.
Why does this matter to global investors and policymakers?
First, it demonstrates scalable decarbonization without overbuilding renewables. China added 216 GW of solar in 2023 alone—but curtailment remains stubbornly high in regions with weak grid integration. This model doesn’t require more panels or turbines; it unlocks value from existing assets by making them talk to each other. The study shows wind-solar curtailment fell 14.3 percent in the cooperative scenario—not through new wires, but through smarter dispatch.
Second, it repositions EVs as system assets, not liabilities. With over 20 million EVs on Chinese roads (and 10 million added in 2024), grid planners worldwide fear “the 7 p.m. cliff”—when millions plug in simultaneously. This work proves that with the right pricing and coordination, EVs can smooth that cliff. Every V2G-capable EV becomes a 60–100 kWh distributed battery—deployable precisely when the grid needs it most. The business case shifts: fleet operators can earn revenue from grid services; utilities avoid $billions in peaker-plant investments.
Third, it reveals a hidden synergy between carbon capture and sector coupling. Most CCS projects focus on point-source capture (e.g., cement, steel), with CO₂ stored geologically. Here, CO₂ is valorized—turned into revenue-generating gas. The P2G unit isn’t a cost center; it’s a flexibility engine and carbon sink. When electricity is cheap, it consumes surplus power and CO₂ to make fuel. When gas prices spike, it sells that fuel back. The carbon credit provides a floor; the energy arbitrage provides upside.
This approach aligns with EU “renewable and low-carbon fuels” definitions and U.S. 45Q/45V tax credits—opening export potential for Chinese technology packages.
Implementation hurdles remain. Real-world EV participation requires standardized V2G protocols, consumer incentive programs, and regulatory clarity on ownership of discharged energy. Gas pipeline operators may resist third-party synthetic methane injection without purity guarantees. And while the model uses hour-ahead dispatch (practical for today’s ADN control systems), future versions must handle intra-hour flexibility—especially as inverter-based resources dominate.
Yet pilot signals are encouraging. State Grid’s Suzhou Industrial Park already runs a multi-energy microgrid with P2G and EV aggregation. Shenzhen’s “Virtual Power Plant” platform has enrolled over 1,200 commercial buildings and 30,000 EVs for demand response. The leap to cross-entity, multi-energy, game-theoretic coordination is incremental—not revolutionary.
For investors, the takeaway is clear: the next wave of grid-edge value won’t come from bigger batteries or longer transmission lines. It will come from orchestrating heterogeneity—turning EV fleets, waste CO₂, idle gas pipes, and rooftop solar into a responsive, self-balancing ecosystem. The technology exists. The economics now check out. And in a world racing toward net-zero, the ability to cut emissions while improving margins isn’t just attractive—it’s essential.
The era of static energy planning is over. The game has begun.
GAO Ruiyang¹, WANG Xinbao², GAO Xian³, WANG Fang¹, BIAN Haihong¹, XU Dongli¹
¹ School of Electric Power Engineering, Nanjing Institute of Engineering, Nanjing 210000, China
² NR Electric Co., Ltd., Nanjing 211100, China
³ AAU Energy, Aalborg University, Aalborg 9220, Denmark
Southern Power System Technology, Vol. 18, No. 2, February 2024
DOI: 10.13648/j.cnki.issn1674-0629.2024.02.009