Hybrid Interconnection of Iterative Bidding and Power Network Dynamics for Frequency Regulation and Optimal Dispatch

Tjerk Stegink, Ashish Cherukuri, Claudio De Persis, Arjan van der Schaft, Jorge Cortés

TL;DR

This paper proposes a novel hybrid algorithm that seamlessly integrates the discrete, iterative nature of electricity market bidding with the continuous dynamics of power grid frequency. By developing a robust theoretical framework and a time-triggered implementation, the authors demonstrate that this approach can achieve both efficient economic dispatch and stable grid frequency, even with strategic, self-interested generators.

What Problem Does This Paper Address?

Modern power grids face a growing challenge: integrating intermittent renewable energy sources while maintaining grid stability and economic efficiency. Traditionally, economic dispatch (ensuring the cheapest power generation meets demand) and frequency regulation (keeping the grid's frequency at its nominal value) have been handled by separate, hierarchical control layers operating on different timescales. This separation becomes problematic with renewables, as their variability impacts both layers simultaneously. This paper tackles the problem of designing a unified system where strategic generators, who don't reveal their true costs, can participate in a real-time market. The goal is to achieve an "efficient Nash equilibrium" – a state where generators are maximizing their profit while the overall system cost is minimized and the grid frequency is stable, all within a practical, implementable framework.

Methods

The core of the paper lies in bridging the gap between continuous-time power system dynamics and discrete-time market interactions.

  • Continuous-Time Dynamics: The paper starts by modeling the power network's frequency evolution using standard swing equations.
  • Iterative Bidding Game: A game-theoretic model is established where generators submit bids to an Independent System Operator (ISO). The ISO aims to minimize costs and regulate frequency, while generators aim to maximize profit.
  • Hybrid Algorithm: The key innovation is a "multi-rate hybrid algorithm." This algorithm features:
  • A fast inner loop for iterative bidding, where generators adjust their bids based on received power setpoints and current network frequency.
  • A slower outer loop for market clearing and updating power generation setpoints sent to the physical grid.
  • Robustness Analysis: The authors develop a novel LISS-Lyapunov function to analyze the stability and robustness of the interconnected system. This function is crucial for proving that the system is stable even with small disturbances or mismatches between the continuous and discrete dynamics.
  • Time-Triggered Implementation: Based on the theoretical analysis, a practical "time-triggered" implementation is proposed. This means the bidding and market clearing events don't necessarily happen at fixed intervals, but are triggered by the system's state, making it more adaptable.

Datasets

The simulations were performed on the IEEE 14-bus power network. This is a standard benchmark system used in power systems research, allowing for comparison with existing studies. The paper adapts this benchmark by assigning generators and loads to specific buses to match the model's assumptions.

Key Insights and Ideas

  • Bridging the Timescale Gap: The most significant insight is the successful theoretical and practical integration of fast, discrete market dynamics with slow, continuous physical grid dynamics. This is a crucial step towards more adaptive and efficient power grids.
  • Strategic Bidding and Stability: The paper demonstrates that even when generators are strategic and don't reveal their true costs, a well-designed iterative bidding mechanism can lead to an efficient outcome where the system operates optimally and stably.
  • LISS-Lyapunov Function for Hybrid Systems: The development and application of a LISS-Lyapunov function for this specific hybrid system is a major theoretical contribution. It provides a powerful tool for analyzing the stability of complex, interconnected systems with mixed continuous and discrete dynamics.
  • Time-Triggered is Practical: The time-triggered implementation highlights that real-world systems don't need perfectly synchronized, fixed-interval updates. Adaptable triggering mechanisms can maintain stability and efficiency, which is more realistic for digital communication networks.

Key Results

  • Asymptotic Convergence: The proposed hybrid algorithm is proven to asymptotically converge to an equilibrium state. This equilibrium corresponds to an efficient Nash equilibrium (generators maximize profit, overall cost is minimized) and zero frequency deviation (grid frequency is at its nominal value).
  • Robustness: The system is shown to be robust to additive disturbances, meaning it can tolerate small external influences without losing stability.
  • Simulation Validation: Simulations on the IEEE 14-bus system demonstrate that the time-triggered hybrid system successfully regulates frequency and converges to optimal power generation levels and bids, even after changes in load or generator costs. The simulations also show that faster update rates generally lead to better performance, but stability is maintained within certain bounds.

Limitations and Future Work

  • Local Stability: The current analysis primarily focuses on local stability, meaning convergence is guaranteed for initial conditions close to the equilibrium. Extending this to global stability would be a significant advancement.
  • Simplified Model: The paper uses a simplified model of the power network and generator behavior. Future work could incorporate more complex aspects such as:
  • Elastic demand: Allowing demand to adjust based on price.
  • Generator bounds: Considering physical limitations on generation capacity.
  • Power flow constraints: Modeling the physical flow of electricity through transmission lines.
  • Distributed Implementation: While the bidding is decentralized, the ISO plays a central role. Future research aims to develop fully distributed and opportunistic self-triggered implementations where no central authority is needed.
  • Data-Driven Approaches: The authors suggest exploring data-driven optimization algorithms to further enhance the system's adaptability and performance.

Summary updated

Abstract

This paper considers a real-time electricity market involving an independent system operator (ISO) and a group of strategic generators. The ISO operates a market, where generators bid prices at which they are willing to provide power. The ISO makes power generation assignments with the goal of solving the economic dispatch problem and regulating the network frequency. We propose a multirate hybrid algorithm for bidding and market clearing that combines the discrete nature of iterative bidding with the continuous nature of the frequency evolution in the power network. We establish sufficient upper bounds on the interevent times that guarantee that the proposed algorithm asymptotically converges to an equilibrium corresponding to an efficient Nash equilibrium and zero frequency deviation. Our technical analysis builds on the characterization of the robustness properties of the continuous-time version of the bidding update process interconnected with the power network dynamics via the identification of a novel local input-to-state Lyapunov function. Simulations on the IEEE 14-bus system illustrate our results.
BibTeX
@article{stegink2019hybrid,
  author = {Tjerk Stegink and Ashish Cherukuri and Claudio De Persis and Arjan van der Schaft and Jorge Cortés},
  title = {{Hybrid Interconnection of Iterative Bidding and Power Network Dynamics for Frequency Regulation and Optimal Dispatch}},
  journal = {IEEE Trans. Control. Netw. Syst.},
  year = {2019},
  url = {https://openreview.net/forum?id=G8QsTDoxkV},
}
Figure