TL;DR
This paper proposes a two-stage operational strategy for Virtual Power Plants (VPPs) to optimize their participation in day-ahead and real-time electricity markets. By integrating distributed energy resources (DERs) and employing a risk-aware optimization approach, the VPP can maximize profits while effectively managing uncertainties and reducing wind power curtailment.
What Problem Does This Paper Address?
The increasing integration of renewable energy sources like wind power into electricity grids presents challenges. While these resources offer environmental benefits, their intermittent nature introduces significant uncertainty into power generation. This variability can lead to costly imbalances between supply and demand, and potentially result in wind power curtailment (i.e., deliberately reducing wind turbine output to avoid grid instability or market saturation). Virtual Power Plants (VPPs) are emerging as a solution to aggregate and manage these distributed energy resources (DERs) – including generators, storage, and flexible loads – to act as a single, dispatchable entity in electricity markets. However, effectively scheduling a VPP, especially when considering the inherent risks associated with renewable energy and market price fluctuations, remains a complex problem. This paper specifically addresses how a VPP can optimize its operations across both day-ahead (predictive) and real-time (responsive) markets, while explicitly accounting for the financial risks associated with profit variability and the costs of curtailing wind power.
Methods
The paper introduces a two-stage coordinated operational strategy for VPPs.
- Day-Ahead (DA) Market: This stage focuses on maximizing the VPP's expected profit while simultaneously minimizing risk. The authors employ a mean-variance Markowitz theory approach to quantify and manage risk. This means the optimization not only aims for high profits but also considers the standard deviation of those profits, effectively balancing potential gains against potential losses. The problem is formulated as a mixed-integer nonlinear programming problem.
- Real-Time (RT) Market: In this stage, the VPP responds to actual conditions and deviations from the DA schedule. The objective here is to minimize the imbalance cost (the cost of rectifying discrepancies between scheduled and actual power) and the wind power curtailment penalty cost. This stage uses a deterministic nonlinear programming approach.
A key methodological contribution is the explicit inclusion of network constraints and unit commitment decisions for DERs, making the model more practical than simplified linear approaches. The optimization problem is solved using a Differential Evolution (DE) algorithm for the DA stage and an Interior Point Optimizer (IPOPT) for the RT stage.
Datasets
The study utilizes:
- Nord Pool historical market data (2015): This data is used to model correlated scenarios for market price, wind speed, and demand.
- Vestas V27 wind turbine power curve: This is used to calculate wind power output based on wind speed.
- Real 12-bus distribution system: This system, located at the CSIRO energy center in Newcastle, Australia, serves as the testbed for the proposed VPP scheduling approach.
Key Insights and Ideas
- Aggregated VPPs reduce risk: By coordinating multiple DERs, a VPP can effectively share and mitigate risks that individual DERs would face alone. This leads to a more stable and predictable profit profile.
- Risk-aware scheduling is crucial: Ignoring the financial risks associated with volatile electricity markets can lead to suboptimal decisions. Incorporating risk measures like Value-at-Risk (VaR) helps VPPs make more conservative and robust trading decisions.
- Two-stage optimization balances foresight and responsiveness: The DA stage provides a strategic plan based on forecasts, while the RT stage allows for agile adjustments to real-time conditions, minimizing costly deviations.
- Wind power curtailment can be managed: The model explicitly penalizes wind power curtailment, incentivizing the VPP to find ways to utilize wind energy more effectively, either through storage or by selling it in the market.
- The VPP acts as a buffer: In the aggregated VPP model, resources like BESS and interruptible loads are more frequently utilized to cover the risks and fluctuations of other DERs (especially wind), leading to higher imbalance costs for these specific components but overall better system performance.
Key Results
- Increased Profits: The proposed integrated VPP scheduling approach (Case 2) resulted in significantly higher net profits compared to an uncoordinated strategy (Case 1). For example, the total net profit for Case 2 was $4737.30, compared to $3002.00 for Case 1.
- Reduced Risk: The standard deviation of profits and the Value-at-Risk (VaR) were lower in the integrated VPP approach, indicating a more stable and less risky operation. For instance, the VaR-95% for Case 2 was $3485.60, compared to $2506.20 for Case 1.
- Reduced Wind Power Curtailment: The integrated VPP approach led to a reduction in wind power curtailment (WPC) from 1132.65 kWh (Case 1) to 956.36 kWh (Case 2).
- Sensitivity Analysis: The study demonstrated that higher risk aversion leads to more conservative trading, increased energy traded in the DA market, and more wind power curtailment. Conversely, increased uncertainty leads to higher imbalance costs and more energy traded in the RT market.
Limitations and Future Work
- Computational Complexity: While the DE algorithm is effective, solving complex mixed-integer nonlinear programming problems can still be computationally intensive, especially with a large number of DERs and scenarios.
- Scenario Generation: The accuracy of the results relies heavily on the quality of the generated scenarios for uncertainties. The paper uses ARIMA and Monte Carlo simulations, but more advanced forecasting techniques could be explored.
- Future Work: The authors suggest exploring more sophisticated forecasting techniques and investigating the impact of increased renewable energy penetration on VPP operations. Further research could also focus on extending the model to include other market services like ancillary services.
Summary updated