⚡ electrical engineering

Pre-Fault Voltage Discrimination and Time-Domain Protection for Distribution Networks with Inverter-Based Resources

This paper proposes a pre-fault voltage discrimination strategy integrated with time-domain protection to enhance fault detection speed and reliability in distribution networks with inverter-based resources, addressing the limitations of traditional overcurrent protection through offline and real-time hardware-in-the-loop validation.

Original authors: Junyuan Zhao, François Bouffard, Géza Joós

Published 2026-06-10

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Junyuan Zhao, François Bouffard, Géza Joós

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 the electrical grid as a vast, high-speed highway system. For decades, the "traffic cops" (protective relays) that keep this system safe have been trained to look for specific signs of trouble: a massive surge in traffic volume (high current) or a specific pattern of movement (frequency changes).

However, the highway is changing. We are adding more and more "smart cars" (Inverter-Based Resources, or IBRs, like solar panels and batteries). Unlike old-school power plants that act like heavy trucks capable of roaring to a halt and pushing back hard when an accident happens, these smart cars are gentle. If a crash occurs, they don't generate a massive surge of energy to signal the traffic cop. They barely flinch.

The Problem:
Because these smart cars don't create a big "surge," the old traffic cops can't see the accidents. They might think everything is fine when, in reality, a short circuit (a crash) has just happened. This is dangerous because the system needs to react instantly to prevent widespread blackouts.

The Solution: A New Kind of Traffic Cop
The authors of this paper propose a new strategy called Pre-Fault Voltage Discrimination (PVD) combined with Time-Domain Protection. Here is how it works, using simple analogies:

1. Listening for the "Snap" instead of the "Roar"

Traditional protection waits for a loud "roar" (a big spike in current). The new method listens for the "snap" of a traveling wave.

The Analogy: Imagine a long rope. If you yank it suddenly, a wave travels down the rope. Even if you are gentle, the moment you yank it, the rope changes shape instantly.
The Science: When a fault happens, it creates a tiny, ultra-fast "wave" of voltage change that travels at the speed of light. The new algorithm doesn't wait for the wave to get big; it detects the instant the wave starts moving by looking at how fast the voltage is changing (the "snap"). This allows it to spot a fault in microseconds, much faster than waiting for the current to build up.

2. The "Before and After" Check (Pre-Fault Voltage)

The biggest challenge is that sometimes a normal event (like flipping a switch to turn on a light or charging a transformer) can look a little bit like a fault. The algorithm needs to tell the difference between a "false alarm" and a real crash.

The Analogy: Think of a security guard checking a person's ID.
- Old Method: "If you look scary (big current), I stop you." (Fails if the intruder is small).
- New Method: "I check your ID before you entered. If you were a normal person (normal voltage) and then suddenly you changed your behavior drastically compared to that baseline, I know something is wrong."
The Science: The algorithm compares the size of the "snap" (the fault wave) against the voltage that was there just before the event.
- If the "snap" is huge compared to the background, it's a fault.
- If the "snap" is small, it might just be a normal switch flip. The system then applies a tiny, split-second delay to double-check, ensuring it doesn't panic over a harmless event.

3. The "Transformer Inrush" Trap

One specific "false alarm" is when a transformer (a device that changes voltage) is turned on. It sucks in a huge burst of energy that looks like a short circuit.

The Analogy: Imagine a vacuum cleaner. When you first turn it on, it makes a loud, chaotic noise as the motor spins up. A fire alarm might mistake this for a fire.
The Solution: The new algorithm uses a mathematical "pattern match." It knows that a real fault creates a smooth, predictable electrical curve. A transformer turning on creates a messy, "non-linear" curve (like the chaotic vacuum noise). By checking if the electrical signal fits a smooth pattern, the system can ignore the transformer startup and only trip for real faults.

4. The Safety Net

What if the fault happens exactly when the voltage is at zero (the "quietest" moment)? The "snap" might be too faint to hear.

The Analogy: If a car crashes in a soundproof room, you might not hear the crash.
The Solution: The system has a backup plan. If the "snap" detector is silent, it falls back to a traditional "overcurrent" check (looking for a big surge). While this is slower, it ensures that no fault is missed, even the quietest ones.

The Results

The authors tested this new "traffic cop" in two ways:

Computer Simulations: They created a virtual power grid with solar panels and batteries and simulated crashes, switch flips, and transformer startups. The new system caught every crash instantly and ignored every false alarm.
Hardware-in-the-Loop (C-HIL): They connected their actual computer code to a real-time simulator that acts like a physical power grid. Even with the limitations of real hardware, the system reacted in milliseconds—far faster than traditional methods.

In Summary:
This paper presents a smarter way to protect power grids filled with solar and battery systems. Instead of waiting for a loud "roar" of electricity that these new systems don't produce, the new method listens for the tiny, instant "snap" of a fault wave. It uses a "before-and-after" comparison to ignore false alarms and a pattern check to ignore transformer startups, ensuring the grid stays safe and fast without tripping unnecessarily.

Technical Summary: Pre-Fault Voltage Discrimination and Time-Domain Protection for Distribution Networks with Inverter-Based Resources

Problem Statement
The increasing integration of Inverter-Based Resources (IBRs) in distribution networks poses a significant challenge to traditional phasor-based overcurrent protection (e.g., ANSI 51). Unlike synchronous machines, IBRs lack the capacity to source high short-circuit currents, often resulting in fault currents that are insufficient to trigger conventional overcurrent relays. Furthermore, traditional frequency-domain relays require data windows of one to one-and-a-half cycles for accurate signal estimation, leading to detection delays that are suboptimal for modern grid requirements. While time-domain protection utilizing traveling waves (TWs) has been explored for transmission systems, its application in distribution networks is hindered by two main factors: the lower voltage levels (resulting in smaller TW amplitudes) and the difficulty in distinguishing fault-induced TWs from those generated by normal switching operations and transformer inrush currents.

Methodology
The paper proposes a time-domain protection algorithm centered on a Pre-Fault Voltage Discrimination (PVD) strategy. The methodology decomposes fault evolution into three stages: the arrival of the first fault wavefront, the reflection of TWs, and the transition to steady-state lumped parameter behavior.

The core algorithm operates through a three-step process:

Traveling Wave Detection (Step I): Instead of relying solely on amplitude thresholds, which can be unreliable for low-magnitude TWs, the algorithm detects the arrival of a TW by monitoring the rate of change of voltage ($dV/dt$). A TW is flagged if the instantaneous change in voltage exceeds a threshold derived from the maximum historical $dV/dt$ during steady-state operation. This approach is independent of the absolute TW amplitude.
Fast Tripping for Non-Switching Events (Step II): Once a TW is detected, the algorithm calculates the ratio of the TW amplitude ( $V_{TW}$ $V_{T W}$ ) to the pre-fault voltage ( $V(t-1)$ $V (t - 1)$ ), termed the Traveling Wave Pre-fault Voltage ratio (TWPV).
- If the TWPV is high (typically $\ge 1$ ) and the polarity of the TW is opposite to the pre-fault voltage, the event is classified as a fault, and a trip signal is issued immediately.
- If the pre-fault voltage is near zero (low TWPV scenario), the algorithm applies a verification delay to avoid false tripping due to numerical errors or small signal noise.
Inrush and Switching Discrimination (Step III): For events that trigger the initial TW detection (Step I) but fail the TWPV criteria (Step II), the algorithm performs a nonlinearity check to distinguish between faults and transformer inrush.
- The algorithm models the expected current response of a linear RL circuit using a least-squares fit over a short time window ( $\Delta \tau_3$ ).
- It calculates the Sum of Squared Errors (SSE) between the measured current and the estimated linear model.
- Transformer inrush currents, being highly nonlinear, produce a significantly higher SSE compared to the linear response of a fault. If the SSE is below a threshold, the event is confirmed as a fault; otherwise, it is rejected as an inrush or switching event.
- A backup Overcurrent Supervision element is included for cases where TW detection fails (e.g., faults initiating exactly at voltage zero-crossing), though the paper notes that exact zero-voltage faults are practically impossible.

Key Contributions

Pre-Fault Voltage Discrimination (PVD): The paper introduces PVD as a strategy to decouple fault detection (via $dV/dt$) from fault discrimination (via TWPV). This allows for rapid detection of faults even when TW amplitudes are small due to low pre-fault voltages or IBR limitations.
Time-Domain Inrush Discrimination: The authors propose a time-domain method to distinguish transformer inrush from faults using SSE analysis of the current waveform, avoiding the need for long data windows required by traditional harmonic analysis.
Unified Framework: The approach unifies fault analysis across different time spans (from the initial wavefront to steady state) to enhance both speed and security without requiring prior knowledge of specific network parameters like line impedance.

Results
The proposed algorithm was validated through two primary methods:

Offline Simulation (EMTP): Tests were conducted on a modified IEEE 34-node test feeder with grid-connected PV units. The algorithm successfully detected all simulated fault types (including high-resistance and low-angle faults) within tens of microseconds to a few milliseconds. Crucially, it correctly rejected all normal switching events, including feeder energization/de-energization and PV transformer energization (inrush), demonstrating high security.
Controller Hardware-in-the-Loop (C-HIL) Real-Time Simulation: The algorithm was implemented on a control card connected to a real-time simulator of an islanded microgrid with a battery storage system (GFM mode). Using a sampling rate of 90.9 kHz (lower than commercial transmission relays), the system demonstrated:
- Dependability: Faults with inception angles as low as $4^\circ$ were detected within 10 ms (via overcurrent backup) or faster (via TW elements).
- Security: Normal switching operations, including grid-connection events and load changes, did not trigger false trips.
- Speed: The proposed method significantly outperformed a simulated ANSI 51 inverse time overcurrent relay, which required 25 ms to over 120 ms for detection, whereas the proposed method detected faults in under 1 ms for high-angle faults and under 10 ms for low-angle faults.

Significance
The paper claims that the proposed PVD strategy and time-domain algorithm address the critical gap in protecting IBR-rich distribution networks. By leveraging the rate of change of voltage rather than absolute amplitude, the method overcomes the limitation of low fault currents typical of IBRs. The integration of a time-domain nonlinearity check provides a robust mechanism to distinguish faults from transformer inrush without the latency of harmonic analysis. The authors assert that this approach offers a faster, more secure, and dependable alternative to traditional overcurrent protection, validated through both offline and real-time hardware simulations. Future work is identified as extending the method to multi-ended and meshed networks.

1. Listening for the "Snap" instead of the "Roar"

2. The "Before and After" Check (Pre-Fault Voltage)

3. The "Transformer Inrush" Trap

4. The Safety Net

The Results

More like this