Abstract
Power utilities are continuously under high pressure to ensure the best performance of their grid. Nevertheless, power outages continue to be periodically observed. This paper assesses the applicability and implications of the Three-Phases method for optimized dataset selection in dynamic risk analysis, through a case study focusing on vegetation along power lines—a major hazard in power grid management. The case study comprises 17 different real-world datasets originating from 12 different types of data sources. We estimate how these datasets can inform eight parameters related to the physical configuration—one of the three dimensions impacting the probability of tree falls on power lines. The results provide two main take-aways: (1) datasets initially considered as less valuable for risk analysis can end up being the most relevant ones; (2) the potential of knowledge of a dataset needs to be assessed parameter per parameter. The results demonstrate that the Three-Phases method is a step toward traceable, data-driven, and dynamic risk analyses of power grids, resulting in a more reliable management of those large-scale infrastructures.