A Polynomial-Time Method for Testing Admissibility of Uncertain Power Injections in Microgrids

Cong Wang, Eleni Stai, Jean‐Yves Le BoudecarXiv 2018 · DOI · arXiv

Abstract

We study the admissibility of power injections in single-phase microgrids, where the electrical state is represented by complex nodal voltages and controlled by nodal power injections. Assume that (i) there is an initial electrical state that satisfies security constraints and the non-singularity of load-flow Jacobian, and (ii) power injections reside in some uncertainty set. We say that the uncertainty set is admissible for the initial electrical state if any continuous trajectory of the electrical state is ensured to be secured and non-singular as long as power injections remain in the uncertainty set. We use the recently proposed $\mathcal{V}$-control and show two new results. First, if a complex nodal voltage set $\mathcal{V}$ is convex and every element in $\mathcal{V}$ is nonsingular, then $\mathcal{V}$ is a domain of uniqueness. Second, we give sufficient conditions to guarantee that every element in some power injection set $\mathcal{S}$ has a load-flow solution in $\mathcal{V}$, based on the impossibility of obtaining load-flow solutions at the boundary of $\mathcal{V}$. By these results, we develop a framework for the admissibility-test method; this framework is extensible to multi-phase grids. Within the framework, we establish a polynomial-time method, using the infeasibility check of convex optimizations. The method is evaluated numerically.

Introduction

In modern microgrids, power injections are increasingly volatile due to the integration of intermittent renewable energy sources like solar and wind, alongside unpredictable loads such as electric vehicle charging stations. While controllers set target power levels, actual physical injections fluctuate within an "uncertainty set." Ensuring that these fluctuations do not lead to voltage violations or grid collapse is the core challenge of admissibility.

This paper introduces a rigorous, polynomial-time method to verify if a given uncertainty set of power injections is admissible. By combining $\mathcal{V}$-control theory with semi-definite programming (SDP) relaxations, the authors provide a framework that guarantees a microgrid remains within safety limits and maintains voltage stability.

Methodology: The Two-Step Framework

The proposed approach leverages $\mathcal{V}$-control theory to bypass the non-linear complexities of standard power flow equations. The process is formalized into a two-step algorithm:

  1. Step 1: Constructing the Safe Zone. The algorithm first constructs a "safe" voltage set $\mathcal{V}$ that is both convex and non-singular. This set is defined by physical security constraints, such as bounds on branch and nodal currents.
  2. Step 2: Admissibility Testing. The uncertainty set $\mathcal{S}^{\text{uncertain}}$ is then tested against this safe zone. This is framed as an infeasibility check of an optimization problem. To ensure computational efficiency, the authors employ a sparsity-exploiting hierarchy of SDP relaxations, converting a difficult non-convex problem into a series of convex ones solvable in polynomial time.
The two-step framework for testing the admissibility of uncertain power injections.

Key Theoretical Insights

The method's reliability rests on two primary theoretical pillars:

  • Domain of Uniqueness: The authors prove that if a voltage set $\mathcal{V}$ is convex and non-singular, it serves as a domain of uniqueness. This ensures that for every power injection in the uncertainty set, there is exactly one corresponding stable voltage state.
  • Boundary Logic: Instead of checking every point in an infinite uncertainty set, the method focuses on the topological boundary of $\mathcal{V}$. If it can be proven that a solution cannot exist on the "edge" of the safe zone, and at least one solution exists inside it (the initial state), then all solutions for that uncertainty set must remain trapped within the safe interior.
Complex voltage trajectories at PQ buses staying within the safe, convex set V under uncertainty.

Numerical Evaluation and Results

The method was evaluated using standard benchmarks, including a 3-bus meshed grid, a modified IEEE 13-Bus system, and the CIGRE North American LV Distribution Network.

  • Tightness: Numerical tests demonstrated that the method is remarkably "tight." In the meshed grid example, the safety check correctly identified the limit where nodal voltages reached 0.9506 p.u., just above the 0.95 p.u. safety threshold.
  • Efficiency: The sparsity-exploiting SDP approach ensures the method scales polynomially. For the complex CIGRE network, the safety check was completed in under 15 seconds, making it viable for real-world grid management.

Conclusion

This work provides a mathematically rigorous yet computationally efficient tool for microgrid operators. While the method is inherently conservative due to its reliance on sufficient conditions and convex relaxations, it offers a 100% guarantee of safety and stability within the verified uncertainty bounds. Future work aims to extend this framework to handle topological uncertainties, such as line switching events, and broader classes of grid components.

Summary updated

Abstract

We study the admissibility of power injections in single-phase microgrids, where the electrical state is represented by complex nodal voltages and controlled by nodal power injections. Assume that (i) there is an initial electrical state that satisfies security constraints and the non-singularity of load-flow Jacobian, and (ii) power injections reside in some uncertainty set. We say that the uncertainty set is admissible for the initial electrical state if any continuous trajectory of the electrical state is ensured to be secured and non-singular as long as power injections remain in the uncertainty set. We use the recently proposed V-control and show two new results. First, if a complex nodal voltage set V is convex and every element in V is nonsingular, then V is a domain of uniqueness. Second, we give sufficient conditions to guarantee that every element in some power injection set S has a load-flow solution in V, based on impossibility of obtaining load-flow solutions at the boundary of V. By these results, we develop a framework for the admissibility-test method; this framework is extensible to multi-phase grids. Within the framework, we establish a polynomial-time method, using the infeasibility check of convex optimizations. The method is evaluated numerically.
BibTeX
@misc{wang2018polynomial,
  author = {Cong Wang and Eleni Stai and Jean‐Yves Le Boudec},
  title = {{A Polynomial-Time Method for Testing Admissibility of Uncertain Power Injections in Microgrids}},
  year = {2018},
  eprint = {1810.06256},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1810.06256},
}
Figure