Decentralized Privacy-Preserving Distributionally Robust Restoration of Electricity/Natural-Gas Systems Considering Coordination of Pump Storage Hydropower and Wind Farms

Nima Nasiri, Saeed Zeynali, Sajad Najafi Ravadanegh, Sylvain Kubler, Yves Le TraonIEEE Access 2024 · DOI

Introduction

When a massive blackout occurs, "blackstarting" the power grid is a race against time. Modern power systems increasingly rely on Gas-Fired Units (GFUs) because they can ramp up quickly to provide critical initial power. However, this creates a deep interdependence between the electricity grid and the natural gas network (NGN). If the gas network is congested or lacks sufficient pressure, GFUs cannot provide the power needed for restoration, potentially stalling the entire process.

Traditional restoration models often assume a centralized operator with full visibility into both networks. In reality, power and gas networks are managed by separate entities that cannot share internal topological data due to privacy and security concerns. Furthermore, the increasing penetration of volatile wind power introduces uncertainty that traditional models struggle to balance: being too optimistic leads to instability, while being too conservative leaves consumers in the dark longer than necessary.

A Decentralized, Privacy-Preserving Framework

To address these challenges, this research proposes a bi-level decentralized optimization model coordinated via the Analytical Target Cascading (ATC) algorithm. This framework allows the Transmission System Operator (TSO) and the Gas Network Operator (GNO) to solve their respective optimization problems independently.

The proposed decentralized restoration framework coordinating TSO and GNO via the ATC algorithm.

As shown in the framework above, the operators only exchange "boundary" information—specifically, the amount of gas required versus what can be supplied—rather than sharing sensitive internal network maps. This preserves data privacy while ensuring the two systems remain synchronized during the restoration timeline, from the initial blackout to the final restoration of loads.

Methods: Distributionally Robust Optimization

To handle the inherent uncertainty of wind power, the authors employ a moment-based Distributionally Robust Optimization (DRO) approach. Unlike standard stochastic methods that require an exact probability distribution, or robust methods that plan for the absolute worst-case scenario, DRO uses mean and variance to find a reliable "middle ground." This ensures the restoration plan is robust against wind fluctuations without being overly expensive or conservative.

Additionally, the model incorporates linepack technology—utilizing the gas pipes themselves as temporary storage—and dedicated Gas Storage Systems (GSS). This provides the necessary flexibility to fuel GFUs during the high-demand restoration phase before gas producers can fully ramp up production.

Key Insights and Results

The framework was validated using the IEEE 57-bus power system and the Belgian 20-node gas network. Key findings include:

  • The "Gas Gap": Ignoring natural gas constraints leads to an 11.92% overestimation of restorable load. Without accounting for gas network limits, a power-only model might promise a restoration that the fuel supply cannot support.
  • Synergy of Wind and Pumped Hydro (PSH): While wind is volatile, coordinating it with Pumped Storage Hydropower acts as a stabilizer. This coordination allowed for a 21.34% reduction in unserved loads compared to traditional methods.
  • Impact of Storage: Utilizing linepack and GSS reduced unserviced loads by an additional 2.09% and 1.98%, respectively, by smoothing out fuel supply bottlenecks.
  • Computational Feasibility: Despite the complexity of the bi-level decentralized approach, the models solved in 321 to 2,896 seconds, making the method feasible for operational planning during a crisis.

Conclusion

This research demonstrates that a decentralized, privacy-preserving approach can effectively manage the complex interdependencies of modern energy systems during a blackout. By leveraging distributionally robust optimization and coordinating diverse resources like wind, gas storage, and pumped hydro, operators can restore power more reliably and quickly. Future work will explore extending this framework to the "last mile" of the grid, integrating local resources like microgrids and electric vehicle fleets into the restoration process.

Summary updated

BibTeX
@article{nasiri2024decentralized,
  author = {Nima Nasiri and Saeed Zeynali and Sajad Najafi Ravadanegh and Sylvain Kubler and Yves Le Traon},
  title = {{Decentralized Privacy-Preserving Distributionally Robust Restoration of Electricity/Natural-Gas Systems Considering Coordination of Pump Storage Hydropower and Wind Farms}},
  journal = {IEEE Access},
  year = {2024},
  doi = {10.1109/access.2024.3354891},
}
Figure