⚡ electrical engineering

Multi-Time Scale Optimal Dispatching of Virtual Power Plants under Carbon-Green Certificate Interconnected Trading

This paper proposes a multi-time-scale optimal dispatching framework for Virtual Power Plants that integrates a carbon-green certificate interconnected trading mechanism to prevent market risks, employs a two-layer robust model for day-ahead planning, and utilizes distributed model predictive control for intra-day adjustments, thereby achieving low-carbon, economic, and reliable operation under renewable energy uncertainties.

Original authors: Mingchen Huang, Renping Zhu, Hongzhe Li, Xiaohui Yang, Haojie Liu, Jiajing Xu

Published 2026-07-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Mingchen Huang, Renping Zhu, Hongzhe Li, Xiaohui Yang, Haojie Liu, Jiajing Xu

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a Virtual Power Plant (VPP) not as a giant factory, but as a super-smart, digital "energy chef" running a massive kitchen. This chef has to juggle ingredients that show up unpredictably (like wind and solar power, which are as fickle as the weather) while trying to keep the lights on for everyone and save money. But here's the twist: this chef is playing in a market with two different rulebooks—one for "Green Certificates" (proof you used clean energy) and one for "Carbon Trading" (paying for pollution).

The paper by Mingchen Huang and team at Nanchang University suggests that if you don't connect these two rulebooks carefully, you might accidentally get paid twice for the same clean energy or end up paying for pollution you didn't actually create. They call this "double-counting" and "cross-subsidization," and it's like a chef getting paid for a salad they didn't make, while the real salad maker gets nothing.

The Problem: The "Double-Dip" Trap

The authors argue that current market rules have a glitch. If a VPP sells its clean energy to get a Green Certificate, it shouldn't also claim that same energy is "zero carbon" to get a reward in the Carbon Trading market. If it does, it's like a student getting extra credit for a test they already passed, then getting a scholarship for the same test. This "institutional arbitrage" tricks the system, inflating the apparent environmental benefits while actually causing more chaos in the grid.

The paper explicitly rules out the idea that we can just ignore the tiny amount of carbon emitted when making and shipping the solar panels and wind turbines. Many current systems pretend these machines are "zero carbon" from birth to death. The authors say no way: even green machines have a "carbon footprint" from their lifecycle. Ignoring this leads to a false sense of security and unfair subsidies.

The Solution: A Two-Step Dance

To fix this, the team proposes a new "Multi-Time Scale" strategy. Think of it as a two-step dance routine for the energy chef:

Step 1: The Day-Ahead Plan (The "Robust" Forecast)
The day before, the chef has to plan the menu. But since they don't know exactly how much wind or sun they'll get, they can't just guess. The paper suggests using a "Two-Layer Robust Optimization" model.

The Metaphor: Imagine the chef is planning a picnic. A normal plan assumes the weather will be perfect. A "robust" plan assumes it might rain, or the sun might hide, so the chef packs extra umbrellas and tents just in case.
The Twist: The paper introduces a special "conservatism knob" (a robustness adjustment coefficient). If the chef is feeling nervous about the weather, they turn the knob up to pack more umbrellas. If they feel confident, they turn it down to save space. This allows the plan to be flexible based on how risky the day feels.
The Result: In their simulations, this method helped avoid a massive "adjustment cost" later. By planning for the worst-case scenario upfront, they saved 15,791 CNY in extra costs that would have been needed to fix mistakes later.

Step 2: The Intra-Day Adjustment (The "Distributed" Dance)
Once the day starts, the weather might change faster than expected. The chef needs to tweak the menu in real-time.

The Metaphor: Instead of one head chef shouting orders to the whole kitchen (which gets chaotic and slow), the paper suggests using "Distributed Model Predictive Control" (DMPC). Imagine the kitchen is broken into small teams (the grill, the salad station, the dessert team). Each team talks to its neighbors and adjusts its own cooking instantly without waiting for a boss.
The Speed: This distributed approach is much faster. The paper's simulations show that while the old "centralized" method took about 5.023 minutes to solve a single scheduling puzzle, the new DMPC method did it in 3.658 minutes. This speed is crucial because the grid changes every 5 minutes; if the chef is too slow, the food burns.

The Big Win: Cutting the "Greenwashing"

When the team tested their full system (the new trading rules + the two-step dance), the results were clear in their simulations:

They stopped the "double-dipping" revenue of 45,449 CNY (money that was counted twice but didn't actually exist).
They cut the extra adjustment costs by 15,791 CNY.
Most importantly, after fixing the math, the actual carbon emissions dropped by 1,014 kg.

The paper concludes that by connecting the Green Certificate and Carbon Trading markets properly, and by using a smart, flexible, and fast control system, Virtual Power Plants can actually be low-carbon and economical. Without these fixes, the system risks becoming a game of "greenwashing" where everyone looks green, but the planet doesn't actually get any cleaner.

In short, the authors suggest that if we want a truly green future, we need to stop counting the same clean energy twice, admit that even green tech has a carbon cost, and give our energy chefs a faster, smarter way to react to the weather.

Technical Summary: Multi-Time Scale Optimal Dispatching of Virtual Power Plants under Carbon-Green Certificate Interconnected Trading

Problem Statement
As "dual-carbon" market mechanisms advance, Virtual Power Plants (VPPs) are increasingly integrating high-proportion renewable energy. However, the linkage between Carbon Trading Mechanisms (CTM) and Green Certificate Trading (GCT) introduces systemic risks, including "double-counting" of emission reductions and "cross-subsidization" between markets. These institutional gaps can lead to market arbitrage, where VPPs pursue excessive renewable energy consumption for dual subsidies, inadvertently inflating emission reduction claims and destabilizing system operations. Furthermore, the random and fluctuating nature of renewable sources and loads creates significant uncertainty, rendering traditional static day-ahead optimization vulnerable to forecast deviations. Existing robust optimization methods often rely on overly conservative worst-case scenarios, compromising economic efficiency, while centralized control architectures struggle with the computational complexity and single-point failure risks associated with large-scale distributed resources.

Methodology
The paper proposes a comprehensive multi-time-scale optimal dispatching framework under a Carbon-Green Certificate Interconnected Trading Mechanism (C-GITM). The approach is structured into two primary stages:

Day-Ahead Scheduling (Robust Optimization with Adjustable Conservatism):
- C-GITM Framework: A trading mechanism is established to prevent double-counting and cross-subsidies. It incorporates full-life-cycle carbon accounting for renewable energy (Levelized Cost of Carbon Emission, LCCE) and differentiates carbon emission factors based on generation technology. This ensures that ownership of emission reduction rights is transferred correctly during certificate trading, preventing institutional arbitrage.
- Two-Layer Robust Model: A bi-level robust optimization model is constructed to handle source-load uncertainty. The outer layer minimizes market incentive costs, while the inner layer minimizes operating costs under the worst-case uncertainty scenario.
- Adjustable Conservatism: A robustness adjustment coefficient ( $\xi$ ) is introduced to dynamically tune the model's conservatism based on operator risk preferences. This allows the system to balance economic efficiency and security margins.
- Solution Algorithm: The model is solved using a Decomposed Column and Constraint Generation (DC&CG) algorithm, which decouples the problem into a master problem and sub-problems to reduce computational complexity.
Intra-Day Scheduling (Distributed Model Predictive Control):
- Forecasting: A Genetic Algorithm-Back Propagation (GA-BP) neural network is employed to optimize the initial weights and thresholds of the prediction model, continuously updating with real-time data to improve ultra-short-term forecast accuracy.
- Hierarchical Rolling Optimization: A Distributed Model Predictive Control (DMPC) strategy is adopted. The system is decomposed into collaborative units (upper-layer for thermal/cooling loads and lower-layer for electrical power) to manage the different dynamic response characteristics of various loads.
- Real-Time Adjustment: Unlike traditional methods that rely on a fixed day-ahead baseline, this approach uses real-time operational data as the dynamic reference benchmark. This mechanism synchronously updates dispatching instructions to correct prediction deviations and mitigate cumulative errors.

Key Contributions

C-GITM System: The paper introduces a trading framework that accounts for the full life-cycle carbon emissions of renewable energy and establishes clear ownership transfer rules for Green Certificates (GC). This effectively solves the issues of double-counting and cross-subsidies, preventing market arbitrage that drives excessive renewable energy consumption plans.
Novel VPP Scheduling Method: A fair market incentive mechanism is integrated into a day-ahead two-layer robust optimization model with adjustable conservatism. This adapts to varying risk preferences across different operational scenarios.
DMPC-Based Intra-Day Optimization: The study implements a DMPC algorithm for real-time adjustment. This distributed architecture reduces computational load, avoids single-point failures, and effectively responds to significant source-load fluctuations by utilizing a receding-horizon optimization strategy.

Results
Simulation results based on a VPP integrating combined cooling, heating, and power (CCHP) systems demonstrate the efficacy of the proposed strategy:

Economic and Environmental Performance: Compared to a scenario allowing cross-subsidies and double-counting (Scenario 5.7), the proposed C-GITM strategy (Scenario 5.8) reduced repeated-calculation benefits by 45,449 CNY and additional adjustment expenses by 15,791 CNY. Crucially, after accounting for these factors, the actual carbon emissions decreased by 1,014 kg.
Robustness vs. Determinism: The bi-level robust optimization method (Scenario 5) significantly reduced intra-day adjustment costs compared to deterministic optimization (Scenario 3) by proactively considering worst-case uncertainty scenarios, thereby lowering the need for emergency resource inputs.
Algorithm Efficiency: The DMPC rolling optimization model demonstrated superior performance over traditional MPC. The average time for a single optimization step was reduced from 5.023 minutes (MPC) to 3.658 minutes (DMPC), a reduction of approximately 27%. This ensures the model meets the strict 5-minute scheduling cycle constraints required for real-time operation.
Tracking Accuracy: The DMPC-based strategy showed a closer match between intra-day operational trajectories and day-ahead reference curves, effectively balancing local operational efficiency with global planning objectives.

Significance
The paper claims that the proposed multi-time-scale dispatching strategy enables VPPs to achieve low-carbon and economic operation while mitigating uncertainties from both supply and demand sides. By establishing a fair market incentive framework that eliminates double-counting and cross-subsidies, the method prevents the "greenwashing" of emission reductions and ensures that market incentives drive genuine low-carbon transitions. Furthermore, the integration of adjustable robustness and distributed control provides a resilient scheduling system capable of handling the high volatility of renewable energy integration without compromising operational stability or economic efficiency.

The Problem: The "Double-Dip" Trap

The Solution: A Two-Step Dance

The Big Win: Cutting the "Greenwashing"

More like this