Towards Critical Clearing Time Sensitivity for DAE Systems with\n Singularity

Chetan Mishra, Chen Wang, Xin Xu, Virgilio CentenoarXiv 2020 · DOI · arXiv

TL;DR

This paper provides a mathematical framework for calculating how the Critical Clearing Time (CCT) of a power system responds to changes in system parameters. Specifically, it tackles the challenging scenario where stability is lost because the system hits a "singular surface"—a mathematical point of no return that triggers immediate voltage collapse.

What Problem Does This Paper Address?

In power systems engineering, the Critical Clearing Time (CCT) is the maximum time a fault (like a short circuit) can persist before the system becomes irrecoverably unstable. Traditionally, researchers calculated CCT sensitivity—how much more "buffer time" you gain by changing a setting like generator excitation—by assuming the system fails by drifting toward an unstable equilibrium point.

However, real-world power grids are modeled as Differential-Algebraic Equations (DAEs). In these systems, stability can be lost in a much more violent way: the system trajectory can hit a singular surface. At this surface, the algebraic constraints (like voltage balance) lose their causal relationship with the dynamic states (like rotor angles). This is a precursor to total voltage collapse.

The mathematical challenge is that at this singular point, standard sensitivity equations "blow up" to infinity, making it historically difficult to derive a clean formula for how CCT changes with respect to controllable parameters.

Methods

The authors derive a new expression for CCT sensitivity by focusing on the geometry of the DAE state space. Their approach involves three main steps:

  1. Trajectory Sensitivity: They calculate how the system's state at the moment of fault clearing changes as parameters vary.
  2. Boundary Sensitivity: They characterize how the singular surface itself moves in response to parameter changes.
  3. The "Left Eigenvector Trick": To solve the problem of sensitivities going to infinity at the singular surface, the authors employ a clever mathematical workaround. Since the Jacobian matrix of the algebraic constraints ($\partial g / \partial y$) becomes singular at the boundary (having a zero eigenvalue), they pre-multiply the sensitivity equations by the corresponding left eigenvector. This effectively "cancels out" the infinite terms, leaving behind a solvable, finite expression for CCT sensitivity.

Datasets

The paper does not use large-scale empirical datasets. Instead, it validates the derivation using a standard one-bus one-machine (Single Machine Infinite Bus) model. This is a common benchmark in power systems research used to provide clear visual insights into state-space trajectories and stability boundaries.

Key Insights and Ideas

  • Singularity as a Physical Boundary: The paper treats the singular surface not just as a numerical failure point, but as a fundamental component of the stability boundary that must be explicitly modeled.
  • Geometric Intuition: By visualizing the "nose" of the algebraic constraint surface (Figure 4 in the paper), the authors show how decreasing generator excitation ($E$) makes the surface "steeper," driving the system toward singularity faster and thus reducing CCT.
  • Preventive Control: The derived sensitivity formula allows operators to identify which "knobs"—such as Automatic Voltage Regulator (AVR) settings or power dispatch levels—are most effective at "pushing" the singular surface away from the system's path, thereby increasing the safety margin.

Key Results

  • Mathematical Validation: The authors successfully derived a closed-form expression for CCT sensitivity for DAE systems encountering singularities.
  • Numerical Accuracy: In simulations, the CCT estimates produced by the sensitivity formula (represented as dotted tangent lines in Figures 3 and 5) perfectly matched the slope of the actual CCT curves obtained through exhaustive time-domain simulations.
  • Parameter Impact: The results confirmed that increasing generator excitation ($E$) increases CCT, while increasing mechanical power loading ($P_m$) decreases CCT, providing a linear approximation of exactly how much these parameters matter.

Limitations and Future Work

  • Specific Instability Mode: This work specifically addresses cases where the system hits the singular surface during the fault-on phase. It does not yet cover "post-fault" voltage collapse, where the system clears the fault but hits a singularity later during the recovery phase.
  • Scalability Testing: While the authors state the method is "easily scalable" because trajectory sensitivity can be parallelized, the paper only demonstrates the method on a very small, 2-state test system. Testing on large-scale, multi-machine grids remains a necessary next step.
  • Future Extensions: The authors plan to derive sensitivity expressions for other instability phenomena, such as those involving periodic orbits or post-clearing collapses.

Summary updated

Abstract

Standard power system models are parameter dependent differential-algebraic\nequation (DAE) type. Following a transient event, voltage collapse can occur as\na bifurcation of the transient load flow solutions which is marked by the\nsystem trajectory reaching a singular surface in state space where the voltage\ncausality is lost. If a fault is expected to cause voltage collapse, preventive\ncontrol decisions such as changes in AVR settings need to be taken in order to\nget enhance the system stability. In this regard, the knowledge of sensitivity\nof critical clearing time (CCT) to controllable system parameters can be of\ngreat help. The quasi-stability boundary of DAE systems is more complicated\nthan ODE systems where in addition to unstable equilibrium points (UEP) and\nperiodic orbits, singularity plays an important role making the problem\nchallenging. The stability boundary is then made up of a number of dynamically\ndistinct components. In the present work, we derive the expression for CCT\nsensitivity for the phenomenon where the critical fault-on trajectory\nintersects the singular surface itself which is one such component forming the\nstability boundary. The results are illustrated for a small test system in\norder to gain visual insights.\n
BibTeX
@misc{mishra2020towards,
  author = {Chetan Mishra and Chen Wang and Xin Xu and Virgilio Centeno},
  title = {{Towards Critical Clearing Time Sensitivity for DAE Systems with\textbackslash\{\}n Singularity}},
  year = {2020},
  eprint = {2002.08999},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2002.08999},
}

Loading related content...

Figure