Modelling and evaluation of global earthing systems using earth potential rise criteria

Global earthing systems can significantly reduce the measurement effort needed to ensure permissible touch voltage in electrical networks. By interconnecting local earthing installations closely, safety requirements can be met with minimal additional testing, especially in medium and low voltage grids. This paper introduces the ρ-f model, a parametric assessment framework that condenses critical variables—such as specific earth resistance, frequency, and earthing conductor arrangements—into linear equations for easier classification of complex earthing structures. A practical clustering method groups low-voltage networks based on criteria like the number of connected buildings, supply region area, and typical earthing installations. The study employs a statistical simulation approach to account for boundary conditions (e.g., proportion of overhead lines, various fault scenarios). Results from these simulations feed into a Random Forest model that identifies the most influential parameters—fault current, overhead-line share, and soil resistivity—in determining whether networks qualify as global earthing systems. Validation through field measurements and computational simulations confirms that the ρ-f approach captures essential network characteristics, allowing more accurate and less labor-intensive earthing evaluations. Combining the ρ-f model with machine learning offers a promising path toward standardizing and streamlining global earthing system assessments, enhancing both cost-effectiveness and safety.