Studying Tree Trimming Effects on Power Grid Resilience Using Weather and Outage Models

In many areas of the world, trees are the top cause of outages in electric distribution systems, and widespread power outages may occur during storms due to the interaction between weather, vegetation, and power lines. Vegetation management is therefore performed by electric utilities along transmission and distribution lines to reduce the impact of storms on the electric system and maintain high levels of power delivery reliability. Different vegetation management techniques are performed by electric utilities, such as a cyclic tree trimming of trees around power lines (usually performed every 4 years), or a more invasive form of tree trimming (called by Eversource-Connecticut enhanced tree trimming (ETT)), consisting in a complete removal of trees and brushes around power lines. The effectiveness of this more invasive, but potentially more effective, tree trimming technique, is often opposed by local citizens due to the reduction of vegetation in urban areas. ETT has been evaluated during storms using statistical techniques and an outage prediction model (OPM). This presentation focuses on the OPM-based evaluation of the reduction of weather-related outages due to ETT. For such evaluation, 144 historical storms have been simulated using the WRF v3.7 model in hindcast mode. Numerical weather prediction simulations are aggregated with satellite data related to leaf area index, land cover soil, and infrastructure datasets, to recreate the environmental conditions occurred during storms. Machine learning models for associating environmental variables to power outages are therefore trained and validated. Historical outages are therefore re-predicted using such models, and modelling results are compared with actual outages for various tree trimming conditions. This frameworks allows to consider changing storm intensity thanks to the association of outages with weather simulations through the outage model.