⚡ electrical engineering

Reachability-based Time-domain Distance Protection

This paper proposes a reachability-based time-domain distance protection method that models transmission line faults as a full RLC network, derives efficient two-dimensional reduced-order models for apparent voltage and current, and defines instantaneous fault tests using set-based state estimation to overcome computational intractability.

Original authors: Joshua A. Taylor, Nathan Baeckeland, Alejandro D. Domínguez-García

Published 2026-08-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Joshua A. Taylor, Nathan Baeckeland, Alejandro D. Domínguez-García

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

The electrical grid is a vast, interconnected machine that delivers power from generators to homes and businesses, but it is also a system that must react instantly to survive. When a fault occurs—a short circuit caused by a fallen tree, a lightning strike, or equipment failure—the flow of electricity becomes chaotic and dangerous. To prevent widespread blackouts or fires, the grid relies on devices called distance relays. These are the guardians of the transmission lines, constantly watching the voltage and current flowing through their specific section of wire. Their job is to decide, within a fraction of a second, whether a problem is happening on their line or somewhere else, and if it is on their line, to cut the power before damage spreads. Traditionally, these guardians have worked by taking a snapshot of the electricity, converting it into a steady, rotating wave pattern, and measuring its resistance. This method is reliable but takes time, often requiring the system to wait for a full cycle of the electrical wave to pass before making a decision. In a world increasingly filled with solar panels and wind turbines, which behave differently than the massive traditional generators of the past, this delay can be a liability.

A team of researchers has proposed a new way for these guardians to think, one that skips the waiting game entirely. Instead of converting the raw, messy data of voltage and current into a smooth wave pattern, they suggest looking at the raw data itself as it happens in time. The core idea is to treat the electrical network not as a static puzzle to be solved, but as a dynamic system with a set of possible futures. Imagine a relay standing at one end of a wire, knowing that a fault could happen anywhere along that wire with any level of resistance. The researchers realized that the measurements the relay could possibly see during such a fault form a specific, bounded region of possibilities. They call this a "reachable set." If the relay measures a value that falls outside this region, it knows a fault has occurred. If the value falls inside, the system is behaving normally. This approach, known as set-based state estimation, allows the relay to ask a simple question: "Is what I am seeing right now consistent with the laws of physics for a healthy line, or does it match the pattern of a fault?"

The challenge with this idea is that calculating these regions of possibility for a complex network is incredibly difficult, like trying to map every possible path a river could take through a mountain range in real-time. The network involves dozens of buses, lines, and sources of power, creating a high-dimensional problem that is too slow for a device that needs to act in milliseconds. To solve this, the researchers built a simplified, two-dimensional model for each type of fault. Instead of tracking the entire network, they focused only on the apparent voltage and current that the relay actually sees. By reducing the problem to these two key numbers, they could create a fast, efficient test that runs directly on the raw time-domain data. This new method does not require the relay to solve complex equations or wait for a full cycle of electricity to pass. Instead, it checks if the current measurement fits the unsolved differential equation that describes the fault.

The researchers tested this new method using a computer simulation of a standard electrical grid, specifically a model based on the IEEE 14-bus test system, which includes five modern inverters that convert power from renewable sources. They simulated a fault where two phases of the line touched each other, occurring at the midpoint of the line. They ran the test with three different levels of resistance: a very low resistance of 0.01 ohms, a medium resistance of 1 ohm, and a high resistance of 10 ohms. In the simulation with the highest resistance, the relay detected the fault in just one-eighth of an electrical cycle. In the medium resistance case, where some of the inverters had to limit their current output, detection happened in about three-eighths of a cycle. Even in the most difficult low-resistance scenario, where all the inverters were pushing their limits, the relay identified the fault just after one full cycle. These results, derived from electromagnetic transient simulations, suggest that the method can detect faults significantly faster than traditional approaches, which typically wait for 1.5 cycles.

The significance of this work lies in its ability to integrate complex network information without sacrificing speed. By keeping the analysis entirely in the time domain and using a reduced-order model, the researchers have created a framework that is both mathematically rigorous and computationally practical. The method does not rely on the assumption that the power sources are perfectly stable or that the network is simple; it accounts for uncertainty in the location of the fault and the behavior of the power sources. While the current study is a proof of concept based on simulations, it demonstrates that it is possible to define instantaneous fault tests using unsolved differential equations. The authors note that future work will need to refine these tests to ensure they do not mistakenly trip for faults on neighboring lines and to determine the most accurate way to model the behavior of modern inverters. For now, the study offers a promising path toward a grid that can protect itself faster and more intelligently, using the raw language of time and change rather than the slower language of steady waves.

Technical Summary: Reachability-based Time-domain Distance Protection

Problem Statement
Traditional distance relays detect transmission line faults by computing voltage and current phasors to determine apparent impedance, a process that typically requires 1.5 cycles. While time-domain schemes offer faster detection (potentially within a quarter cycle) and better performance with inverter-based resources (IBRs), they face significant challenges in modeling the full network. Specifically, a relay lacks complete information about the states and sources elsewhere in the network, leading to uncertainty regarding fault location, resistance, and remote infeed currents. Furthermore, computing reachable sets and performing set-based state estimation for high-dimensional network models is generally intractable, making rigorous time-domain fault testing difficult to implement efficiently.

Methodology
The authors propose a framework that reformulates time-domain fault detection using reachability and set-based state estimation.

Network Modeling: The power system is modeled as a lumped parameter RLC circuit. The network dynamics are represented as a Linear Time-Invariant (LTI) system that changes structure upon the occurrence of a fault (modeled as a resistor). The state vector includes line currents, nodal voltages, and internal states of synchronous machines (SGs) and IBRs.
Incremental Modeling: To mitigate the need for precise knowledge of source values, the authors utilize an incremental model. By subtracting measurements from one cycle prior ( $t - \delta$ ), the model eliminates dependence on the specific operating point of the sources, assuming they remain periodic shortly after the fault. The resulting dynamics depend on the network structure and the pre-fault voltage at the fault point, rather than the instantaneous source values.
Set-Based State Estimation: The set of all possible measurements a relay can observe during a fault is defined as a "reachable set." A fault test is formulated as checking whether a new measurement is consistent with this reachable set (i.e., whether the set-based state estimate remains non-empty).
Reduced-Order Approximation: Recognizing that computing high-dimensional reachable sets is computationally prohibitive, the authors construct a reduced-order, two-dimensional model for each fault type. The states of this reduced model are the apparent voltage and current seen by the relay. This model is derived via structured model reduction (using singular perturbation).
Instantaneous Fault Tests: Instead of solving the differential equations to find reachable sets, the authors define "instantaneous tests." These tests check if the measured apparent voltage and current are consistent with the unsolved differential equation of the reduced-order model. This is achieved by evaluating the model error (residual) over a set of possible fault parameters (location and resistance) and uncertainty bounds. The test passes if zero lies within the convex hull of the possible model errors.

Key Contributions

Novel Formulation: This is the first work to explicitly apply reachability and set-based state estimation to time-domain distance protection.
Model Reduction: The paper introduces a structured model reduction technique that transforms the high-dimensional network problem into a tractable two-dimensional problem involving only the relay's apparent voltage and current.
Computational Efficiency: By shifting from set intersection operations to checking the consistency of measurements against unsolved differential equations (instantaneous tests), the approach aligns with the speed requirements of protection systems.
Handling Uncertainty: The method incorporates uncertainty in fault location, resistance, and source drift (modeled as zonotopes) without requiring real-time communication or full network state knowledge.

Results
The proposed instantaneous tests were validated using a Simulink model based on the IEEE 14-bus test system, featuring five grid-forming (GFM) inverters with various control methods (droop, virtual synchronous machine, and dispatchable virtual oscillator control).

Simulation Setup: A phase-to-phase fault was simulated on a transmission line with varying fault resistances (0.01 $\Omega$ , 1 $\Omega$ , and 10 $\Omega$ ).
Detection Speed:
- High Resistance (10 $\Omega$ ): Detection occurred within 1/8 of a cycle, with no current limiters active and IBRs behaving as voltage sources.
- Medium Resistance (1 $\Omega$ ): Detection occurred at approximately 3/8 of a cycle, with some current limiters active.
- Low Resistance (0.01 $\Omega$ ): Detection occurred just after one full cycle, coinciding with the activation of all IBR current limiters.
Performance: The results demonstrate that the method can effectively detect faults across different resistance levels and IBR behaviors, with detection times varying based on the system's dynamic response to the fault.

Significance and Claims
The paper claims that its primary significance lies in integrating network information into time-domain distance protection without sacrificing computational speed. By leveraging reachability theory and model reduction, the authors provide a rigorous alternative to heuristic phasor-based methods or time-domain schemes that ignore network topology.

The authors are modest in their claims, noting that the current implementation is "overreaching" (detecting faults on the line and potentially neighboring lines) and that the work serves as a "proof of concept." They acknowledge that future work is needed to develop "underreaching" tests, refine IBR modeling (current vs. voltage sources), and reduce dependency on specific network models to improve robustness against topology changes. The paper does not claim to replace existing standards immediately but rather offers a new theoretical foundation for faster, more robust time-domain protection.

Technical Summary: Reachability-based Time-domain Distance Protection

More like this