This study shows the potential of an alert system addressed to the prediction of failures for underground distribution lines for the Milan urban area. In the summer season, one of the most widespread causes of failure in urban areas is due to a degradation of the insulating material that constitutes the joints between the underground cables. Causes are due both to the age of the joint and to the repeated stress conditions to which the electrical component can be subjected related to the abnormal and frequent increases in cable temperature during summer heat waves. In particular, the high electrical load for cooling, the high soil temperatures and the low moisture content prevent the heat dispersion of the cable (soil drying-out phenomenon). The predictors of such a system are atmospheric and soil meteorological variables that show a clear correlation with the daily failure rate. Another important variable in this forecasting system is the electrical load which represents a considerable stress source for underground power lines. In this alert system, weather forecasts come from the ECMWF IFS model, while load and fault prediction systems are based on a Random Forest machine learning method. These systems are trained with forecast predictors related to atmospheric and soil variables and using as predictands the time series of electrical load and failure provided by the distribution company of Milan. The warning system for the city of Milan is therefore defined by linking together the weather forecast to the load and fault forecasting systems. The performance of the alert system was evaluated by means of a k-fold cross validation, training the forecast system on the time series excluding one year at a time and using it each time as a test set. The system demonstrates the ability to identify quite satisfactorily the main failure events associated with summer heat waves for the period 2010-2019, albeit with an underestimation of failure peaks. This alert system can be a valuable support for the management of the distribution network in urban areas. In fact, the reliability of the electricity service in the perspective of resilience requires planner tools but also operational strategies to face this problem. This tool makes it possible to prepare sufficiently in advance all the measures necessary to mitigate the effects and reduce the time necessary to restore the energy supply service.
How to cite: Bonanno, R., Lacavalla, M., and Sperati, S.: An alert system for faults critical conditions on underground distribution lines for Milan urban area, EMS Annual Meeting 2022, Bonn, Germany, 5–9 Sep 2022, EMS2022-108, https://doi.org/10.5194/ems2022-108, 2022.