A Benchmark for the Application of Distributed Control Techniques to the Electricity Network of the European Economic Area

Alessandro Riccardi, Luca Laurenti, Bart De SchutterarXiv 2024 · DOI · arXiv

Introduction: The Challenge of Modern Grid Stability

Maintaining a stable grid frequency, known as Load Frequency Control (LFC), is becoming increasingly complex as the global energy mix shifts toward renewable energy sources (RESs). Traditional power plants utilize large rotating masses that provide natural inertia, resisting sudden frequency changes. In contrast, modern wind and solar installations are "inertia-less" and highly intermittent. While Energy Storage Systems (ESSs) can mitigate these fluctuations, coordinating them across a continent-sized network presents a massive computational and control challenge.

Researchers frequently propose Distributed Model Predictive Control (DMPC) as a solution, where individual regions optimize locally while coordinating with neighbors. However, the field has lacked standardized, large-scale benchmarks. Most studies rely on simplified models that fail to capture the multi-national complexity of real-world grids. The European Economic Area Electricity Network Benchmark (EEA-ENB) was developed to fill this gap, providing a realistic, open-source framework for evaluating distributed control strategies.

The EEA-ENB Framework

The EEA-ENB models 26 interacting electrical areas, representing countries within the European Economic Area. Each area is treated as an "equivalent machine" that aggregates its local inertia and generation capacity. The network topology is based on actual geographical distances and tie-line strengths between countries.

The EEA-ENB network topology, representing 26 interacting electrical areas in the European Economic Area.

The dynamics of each area are captured using linearized discrete-time equations for power angle, frequency deviation, and energy storage levels. The benchmark is designed to be modular, allowing for extensions such as Piecewise Affine (PWA) dynamics for batteries or detailed hydroelectric storage models. The internal structure of a single area $i$ and its interaction with the controller is illustrated below.

Block diagram of the internal dynamics and control structure for a single area $i$ within the benchmark.

Data Integration and Baseline Performance

The benchmark is built on real-world data from the ENTSO-E (European Network of Transmission System Operators for Electricity) transparency platform. It includes 24 hours of actual load requests and renewable generation data from January 1, 2022. By providing both actual measurements and day-ahead forecasts, the framework allows researchers to test controller robustness against prediction errors.

To provide a performance ceiling, the authors implemented a Centralized MPC strategy using the Gurobi optimizer. This "gold standard" achieved frequency regulation within strict operational limits (±0.04 Hz). However, simulating 24 hours of operation required over 206 hours of compute time on a high-end processor. This result quantifies the "optimality gap" and highlights why distributed control is essential for real-time grid management: while distributed controllers may perform slightly below the centralized baseline, they offer the necessary speed for practical application.

Conclusion and Future Directions

The EEA-ENB provides a standardized environment for "apples-to-apples" comparisons between different distributed control techniques, from traditional PID loops to advanced Reinforcement Learning agents. While the current version uses a single-day snapshot, future iterations will incorporate seasonal data and deregulated market dynamics. The benchmark is available under an MIT license to encourage community-driven research into stabilizing the future European grid.

Summary updated

BibTeX
@misc{riccardi2024benchmark,
  author = {Alessandro Riccardi and Luca Laurenti and Bart De Schutter},
  title = {{A Benchmark for the Application of Distributed Control Techniques to the Electricity Network of the European Economic Area}},
  year = {2024},
  eprint = {2403.14372},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2403.14372},
}
Figure