Gain and Phase: Decentralized Stability Conditions for Power Electronics-Dominated Power Systems

Linbin Huang, Dan Wang, Xiongfei Wang, Huanhai Xin, Ping Ju, Karl Henrik Johansson, Florian Dörfler

Introduction

As the global energy landscape transitions from synchronous generators to power electronics-dominated power systems (PEDPS), traditional stability analysis methods are reaching their limits. The shift from physical inertia to complex control loops in thousands of heterogeneous inverters introduces a "curse of dimensionality" that makes global state-space modeling computationally prohibitive.

A recent study, "Gain and Phase: Decentralized Stability Conditions for Power Electronics-Dominated Power Systems," proposes a decentralized framework to address these challenges. By integrating the small gain theorem with the small phase theorem, the authors provide a scalable method to ensure grid stability without requiring a full system solution.

Modeling and Methodology

The core of the proposed approach lies in the Mixed Small Gain-Phase Theorem. This theorem posits that a system remains stable if, at any given frequency, it satisfies either a gain condition or a phase condition.

The feedback interaction between device dynamics and network dynamics forms the basis for the stability analysis.

The modeling process involves:

  1. Device Dynamics: Representing each converter (Grid-Following or Grid-Forming) through an admittance model $Y_{C,i}(s)$.
  2. Network Dynamics: Using a reduced Laplacian matrix via Kron reduction to represent the grid.
  3. System Partitioning: A significant innovation is treating Grid-Forming (GFM) converters as part of the "equivalent network." This strengthens the network model, making it easier for Grid-Following (GFL) converters to satisfy stability criteria.

The Mixed Small Gain-Phase Approach

Stability is evaluated by checking the interaction between the "size" (singular values) and "phases" of the converter and grid responses.

The numerical range $W(A)$ and its phase bounds $\bar{\phi}(A)$ and $\underline{\phi}(A)$ define the phase conditions for matrix-based stability analysis.
  • Gain Condition: If the converter's response magnitude is smaller than the grid's "strength," the system is stable.
  • Phase Condition: Even with high gain, stability is maintained if the phases of the converter and grid do not align to create positive feedback. The authors use the numerical range of matrices to define these phase bounds more broadly than traditional passivity methods.

Key Insights

  • Reduced Conservatism: Unlike passivity-based methods that require phases within $\pm 90^\circ$, this approach allows for wider phase spreads, provided the total phase "width" remains under $180^\circ$.
  • Decentralized Design: Manufacturers can verify the stability of their specific inverters using only local dynamics and basic grid parameters (like gSCR and R/X ratios), facilitating plug-and-play integration.
  • GFM as a Stabilizer: The framework mathematically confirms that GFM converters act as stabilizers by increasing the "network gain," thereby lifting the stability threshold for other devices.

Validation and Results

The theory was validated using several benchmarks, including a single-converter system and a modified IEEE 68-bus system.

  • Accuracy: In single-converter tests, the method predicted stability limits with minimal conservatism (within 0.045 pu of time-domain simulations).
  • Oscillation Detection: The analysis correctly identified an 11.3 Hz oscillation in the 68-bus system.
  • Remediation: The framework successfully guided control adjustments, such as reducing PLL bandwidth, to move the system out of unstable "gray areas."

Conclusion and Future Work

While currently limited to small-signal (linearized) stability, this decentralized framework offers a powerful tool for modern grid management. Future research aims to extend this to large-signal stability and automate the synthesis of optimal controller parameters using reduced-order models.

Summary updated

BibTeX
@article{huang2024gain,
  author = {Linbin Huang and Dan Wang and Xiongfei Wang and Huanhai Xin and Ping Ju and Karl Henrik Johansson and Florian Dörfler},
  title = {{Gain and Phase: Decentralized Stability Conditions for Power Electronics-Dominated Power Systems}},
  journal = {IEEE Transactions on Power Systems},
  year = {2024},
  doi = {10.1109/tpwrs.2024.3380528},
}
Figure