A Proportional-Integral Model for Fractional Voltage Tripping of Distributed Energy Resources

Milos Katanic, Gregor Verbic, John Lygeros, Gabriela Hug

Introduction

The global transition toward renewable energy has led to power grids saturated with small-scale, inverter-based distributed energy resources (DERs), such as rooftop solar. Unlike traditional synchronous generators, these inverters operate under "Voltage Ride-Through" (VRT) logic. This logic dictates that inverters must disconnect (trip) to protect themselves if grid voltage deviates beyond specific thresholds.

Voltage Ride-Through (VRT) characteristics across different generations of Australian grid codes (2005, 2015, and 2020).

A significant challenge for grid operators is "fractional tripping." During a local fault, inverters across a region experience varying voltage levels and are governed by different grid code "vintages" (e.g., older vs. newer standards). If a substantial fraction of solar power trips offline simultaneously, it can jeopardize grid stability and potentially trigger widespread blackouts. Current industry-standard models, such as the DER_A model, often rely on simple linear approximations that fail to capture these complex, time-dependent behaviors accurately.

The PI Fractional Tripping Model

To address these limitations, researchers from ETH Zurich and the University of Sydney have proposed a "gray-box" model inspired by classical control theory: the PI Fractional Tripping Block.

The proposed PI Fractional Tripping Block architecture, featuring proportional and integral branches for deactivation and reactivation.

The core innovation is the use of a Proportional-Integral (PI) structure to model the fraction of tripped DERs:

  1. Proportional Deactivation: Captures immediate tripping caused by rapid hardware protection mechanisms.
  2. Integral Deactivation: Models delayed tripping by "integrating" voltage deviations over time. This reflects the reality that inverters are more likely to trip the longer a disturbance persists.
  3. Integral Reactivation: Accounts for "momentary cessation" and the subsequent recovery behavior mandated by modern grid codes like Australia's AS 4777.2:2020.

By tracking the minimum or maximum voltage experienced during an event, the model incorporates a "memory" of the disturbance, which is crucial for predicting cumulative tripping effects.

Experimental Validation

The researchers validated their model using a high-fidelity synthetic dataset, as high-resolution real-world data for mass-tripping events is limited. The simulation environment consisted of an IEEE 9-bus transmission system connected to ten parallel CIGRE 18-bus distribution grids, totaling 510 individual solar inverters. These inverters were modeled with randomized VRT characteristics based on three generations of Australian standards (2005, 2015, and 2020).

The model's parameters were tuned using Particle Swarm Optimization (PSO), ensuring a robust fit to the synthetic data without requiring complex gradient calculations.

Key Results and Insights

The proposed DER_A_PI model demonstrated superior performance compared to existing benchmarks:

  • High Accuracy: The model achieved a Mean Absolute Error (MAE) of approximately 1.0% for undervoltage events. In contrast, the standard DER_A model exhibited an error of 17.1%, and the more recent DERAEMO1 update showed a 6.4% error.
  • Generalization: The PI model maintained its accuracy during "out-of-sample" tests involving fault types not seen during the training phase, indicating it is not merely overfitting to specific scenarios.
  • Explainability: Unlike "black-box" machine learning approaches, every parameter in the PI model has a physical interpretation (e.g., specific voltage thresholds or timer constants), making it a practical tool for utility engineers.

Conclusion and Future Work

The PI Fractional Tripping model provides a computationally efficient yet highly accurate method for grid stability assessments. However, some limitations remain. The current study focused solely on voltage disturbances; future work will aim to incorporate frequency-based tripping. Additionally, the researchers plan to develop methods for tuning parameters based on known grid code distributions and feeder characteristics, reducing the reliance on heavy simulation data. Finally, exploring the interaction between tripping DERs and dynamic loads, such as air conditioning units, remains an important area for further investigation.

Summary updated

BibTeX
@inproceedings{katanic2025proportional,
  author = {Milos Katanic and Gregor Verbic and John Lygeros and Gabriela Hug},
  title = {{A Proportional-Integral Model for Fractional Voltage Tripping of Distributed Energy Resources}},
  booktitle = {Sustainable},
  year = {2025},
  doi = {10.1016/j.segan.2025.101854},
}
Figure