Framework for Risk Assessment of the Electrical Power Grid under Extreme Weather Conditions

This paper presents a framework to assess the vulnerability of the electrical power grid (EPG) to extreme weather events. The paper presents a methodology based on the Extra-Trees classifier and historical weather data to identify the EPG assets that are most likely to be affected in future extreme weather conditions under various climate change scenarios. The developed methodology considers the EPG different asset classes (lines, towers, poles, transformers, substations...) and identifies the weather parameters that are most relevant to their vulnerability. The paper presents results concerning wind speed, wind gusts, soil type, and altitude, which are used to train a model that predicts the probability of an asset being damaged based on the future weather parameters. The methodology was developed has been applied to a dataset of historical events in Portugal, from the major Portuguese DSO, thus assessing the future vulnerability of the EPG under three different scenarios of climate change. The developed methodology is a successful tool, that would not only help prevent occurrences of faults/failures in the Electrical Power Grid and its recovery from these occurrences, but also to have a better perception of a geographically safe future expansion of infrastructures. In this way it contributes to a continuous, non-faulty EPG operation, fulfilling society's demands by generating maps that identify the most vulnerable areas for each future climate scenario.