⚡ electrical engineering

On the Impact of Operating Points on Small-Signal Stability: Decentralized Stability Sets via Scaled Relative Graphs

This paper introduces a decentralized frequency-domain framework based on Scaled Relative Graphs to characterize how operating points influence small-signal stability in converter-dominated power systems, enabling individual converters to independently evaluate their feasible stability regions through geometric tests.

Original authors: Eder Baron-Prada, Adolfo Anta, Florian Dörfler

Published 2026-03-17

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Eder Baron-Prada, Adolfo Anta, Florian Dörfler

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a modern power grid not as a giant, rigid machine, but as a bustling dance floor filled with dancers. In the old days, the grid was led by massive, heavy-footed "synchronous generators" (like old-school steam engines) that kept the rhythm steady. Today, those heavy dancers are being replaced by agile, fast-moving "converters" (like solar panels and wind turbines) that can change their steps in milliseconds.

While these new dancers are great for the environment, they are tricky. If they don't coordinate perfectly, they start tripping over each other, causing the whole floor to shake (oscillations) or even collapse (instability).

This paper presents a new way to ensure everyone stays on the dance floor without crashing, specifically focusing on how the dancers' current mood (operating point) affects their ability to stay stable.

Here is the breakdown of the paper's solution using simple analogies:

1. The Problem: The "Guess and Check" Trap

Traditionally, engineers tried to figure out if the grid was safe by running millions of computer simulations. They would ask: "What if the wind turbine is spinning at 50% speed? What if it's at 51%? What if the sun is bright but the battery is low?"

The Analogy: Imagine trying to find the safe spots on a trampoline by jumping on every single square inch. If you have 10 dancers, and each can stand in 6 different spots, you have to check 60 million combinations. It takes forever, and you might still miss a dangerous spot in between.
The Issue: As more renewable energy is added, the number of combinations explodes. We can't simulate every possibility, so we often don't know exactly where the "danger zone" begins.

2. The Solution: The "Personal Safety Bubble"

Instead of simulating the whole grid at once, the authors propose a decentralized approach. They give each converter its own "Personal Safety Bubble."

The Analogy: Instead of checking if the whole dance floor is safe, we give each dancer a map. This map tells them: "As long as you stay within this specific shape (your bubble), you will be safe, no matter what the other dancers do."
The Magic: The authors found that for these converters, the relationship between their settings (like how much power they are sending) and their stability is actually quite simple and predictable (mathematically "affine"). This allows them to draw the boundary of the safety bubble using simple straight lines, rather than complex curves.

3. The Tool: "Scaled Relative Graphs" (SRG)

To draw these bubbles, they use a mathematical tool called Scaled Relative Graphs (SRG).

The Analogy: Imagine every converter has a "personality" that changes slightly depending on how hard it's working.
- The SRG is like a 3D radar screen that shows the converter's personality at every possible speed (frequency).
- The grid itself is like a giant, invisible wall.
- The Rule: If the converter's radar screen (SRG) touches the grid's wall, the system becomes unstable (chaos!). If the radar screen stays strictly away from the wall, the system is safe.
- The authors developed a way to calculate exactly where this "touching point" happens for any setting the converter might choose, creating a clear "Do Not Enter" zone on the dancer's map.

4. The Result: A "Feasible Set"

The paper calculates a Feasible Set for each converter.

The Analogy: Think of this as a traffic light system for the power plant operators.
- Green Zone: You can set your power output here. The math guarantees you won't cause a crash.
- Red Zone: Do not go here. Even a tiny nudge could send the whole grid into a wobble.
- Yellow Zone: You are getting close to the edge; proceed with caution.

Because the math is "decentralized," Converter A doesn't need to know exactly what Converter B is doing. Converter A just checks its own map. If everyone stays in their own Green Zone, the whole grid is safe.

5. Why This Matters

Speed: Instead of running millions of simulations (which takes hours or days), this method calculates the safety boundaries almost instantly using simple geometry.
Flexibility: It works for different types of converters (those that just follow the grid's rhythm and those that create their own rhythm).
Safety: It gives operators a clear, visual guide on how much they can push the system before it breaks.

Summary

In short, this paper moves us away from "guessing" if the power grid is stable by running endless simulations. Instead, it gives every power converter a personalized, mathematically proven safety map. As long as the operators keep the converters within their specific "Green Zones," the entire grid can dance safely, even as the wind blows and the sun shines.

1. Problem Statement

Modern power grids are increasingly dominated by power electronic converters (Grid-Following and Grid-Forming) replacing synchronous generators. This shift introduces significant small-signal stability challenges due to strong interactions between converters and network impedances.

The Core Issue: Stability margins are highly sensitive to operating points (setpoints like active/reactive power, controller gains, and grid impedance). Small variations can trigger oscillations or resonance.
Limitations of Current Methods:
- Centralized Analysis: Traditional eigenvalue analysis or Nyquist criteria require simulating the entire system. As the number of converters grows, the operating point space becomes high-dimensional (the "curse of dimensionality"), making exhaustive simulation computationally intractable.
- Existing Decentralized Methods: Current decentralized certificates (e.g., passivity, small-gain) typically assume Linear Time-Invariant (LTI) models with fixed operating points. They fail to capture the parameter dependence of converter dynamics, offering limited insight into robustness across varying operating ranges.
Goal: Develop an analytical tool that characterizes stability margins across a range of operating points without exhaustive scenario enumeration, enabling decentralized verification.

2. Methodology

The authors propose a decentralized frequency-domain framework based on Scaled Relative Graphs (SRG), extended to handle Linear Parameter-Varying (LPV) systems.

A. System Modeling

LPV Representation: Converter dynamics are modeled as LPV systems where the admittance matrix $Y(\gamma, s)$ $Y (γ, s)$ depends affinely on a parameter vector $\gamma$ $γ$ (representing operating points like $i_{d0}, i_{q0}$ $i_{d 0}, i_{q 0}$ ).
- Grid-Following (GFL): Admittance is naturally affine in operating currents.
- Grid-Forming (GFM): Non-affine dependencies are approximated using series expansions (e.g., shifted geometric expansions) to fit the affine LPV form.
Closed-Loop Structure: The system is modeled as a feedback loop between the aggregate converter admittance and the grid admittance (after Kron reduction).

B. Scaled Relative Graph (SRG) Extension

The paper extends SRG theory (originally for LTI systems) to LPV systems:

Frequency-wise SRG: For a transfer function $H(j\omega)$ , the SRG is a set in the complex plane representing the gain and phase shift of the operator.
Chord Property & Approximations: To simplify computation, the authors use:
- Tight Chord Approximation: The convex hull of the SRG.
- Disk Approximation: A conservative outer bound (a closed disk) that satisfies the chord property, used to bound the grid admittance.
Decomposition: Leveraging the affine structure $Y(\gamma, s) = Y_0(s) + \sum \gamma_k Y_k(s)$ , the SRG of the parameter-dependent system is bounded by the Minkowski sum of the SRGs of the individual terms ( $Y_0$ and $Y_k$ ).

C. Decentralized Stability Sets

The core theoretical contribution is Theorem 2, which derives a decentralized stability condition:

Instead of checking the global system, each converter $i$ independently evaluates a stability margin $\rho_i(\gamma_i, j\omega)$ .
The condition ensures that the SRG of the converter does not intersect with the scaled SRG of the grid (specifically, $-\tau \text{SRG}(\tilde{Y}_{grid})$ ) at any frequency.
Result: This yields a set of linear inequalities in the parameter space ( $\gamma_i$ ). The solution to these inequalities defines the Decentralized Stability Set ( $S_i$ ), representing the feasible operating region for that specific converter.

3. Key Contributions

LPV-SRG Framework: First extension of SRG analysis to Linear Parameter-Varying systems, allowing stability analysis across continuous operating ranges rather than fixed points.
Decentralized Stability Sets: A method to compute closed-form geometric characterizations (sets of linear inequalities) for the feasible operating region of each converter independently.
Unified Treatment: The framework applies to both Grid-Following (GFL) and Grid-Forming (GFM) converters, despite their different control architectures and mathematical properties.
Scalability: The approach avoids the exponential growth of computational cost associated with centralized eigenvalue analysis as the number of converters increases.

4. Results and Validation

The framework was validated using two case studies:

A. 4-Node System (GFL and GFM)

Setup: One GFM and one GFL converter connected to an external grid.
Findings:
- The GFL converter exhibited a narrow, diamond-shaped feasible region, indicating high sensitivity to operating points.
- The GFM converter had a much larger feasible region, limited mainly by physical current constraints.
- Time-Domain Validation:
  - Stable Case: Operating points inside the calculated set resulted in damped oscillations and stable steady-state after a load step.
  - Unstable Case: Operating points outside the set (specifically for the GFL) resulted in divergent oscillations, confirming the accuracy of the predicted stability boundaries.

B. Modified IEEE 14-Node System

Setup: Four GFL converters at different nodes with varying Phase-Locked Loop (PLL) bandwidths ( $f_{pll}$ ).
Findings:
- The feasible stability set expanded as the PLL bandwidth decreased.
- This confirms the known sensitivity of GFL converters to grid strength; higher PLL bandwidths require stiffer grids to maintain stability.
- The decentralized approach successfully identified these differences without needing a full system simulation for every point.

5. Significance and Limitations

Significance:
- Operational Safety: Provides operators with explicit "safe zones" for converter setpoints, preventing instability before it occurs.
- Computational Efficiency: Reduces the complexity from $O(N_\gamma)$ (exponential with parameters) to a pre-computation step plus a simple evaluation of linear inequalities.
- Corrective Action: Enables real-time adjustment of setpoints to remain within certified stable regions.
Limitations & Conservatism:
- The method introduces conservatism due to:
  1. Disk approximation of the grid SRG.
  2. Chord approximation of the converter SRG.
  3. Lack of Geographical Information: The decentralized nature ignores the specific electrical location of converters. In networks with highly non-uniform nodal strength, this may yield more conservative (smaller) stability sets compared to centralized analysis.
Future Work: The authors plan to incorporate geographical information to reduce conservatism and apply the method to black-box models and dynamic grid topologies.

Conclusion

This paper presents a robust, scalable, and mathematically rigorous framework for assessing small-signal stability in converter-dominated grids. By transforming the stability problem into a geometric set estimation task using SRGs, it overcomes the dimensionality barriers of traditional methods, offering a practical tool for ensuring the secure operation of future renewable-rich power systems.