Assessing Power System Resilience to Floods: A Geo-Referenced Statistical Model for Substation Inundation Failures: A Geo-Referenced Statistical Model for Substation Inundation Failures

Floods can cause widespread and prolonged power outages by inundating substations. However, assessing the inundation failure of substations to improve power system resilience is challenging, as there may not be sufficient historical
data to capture the impact of flooding on substations at a given location, due to their evident temporal and spatial variability. To tackle this gap in knowledge, this paper proposes a georeferenced statistical model that is not constrained by historical flooding data. The geo-referenced model embeds a hydrological
model that uses established digital rainfall and topography data to simulate flood depths at the location of selected substations and calculate associated inundation risks. The stochastic inundation profiles are used within a Monte Carlo simulation to model substation failures and explore options to improve power system resilience (e.g., asset elevation). The proposed model is
demonstrated using a case study from Bintulu, Malaysia, where real empirical data was used to identify the breaking points of the power system and estimate probability density functions of energy not supplied caused by inundated substations. The simulation results show that substation failures can abruptly
lead to significant energy not being supplied, whereas elevating the substation to withstand an additional 20 cm flood depth will significantly delay flood impacts, and effectively improve system resilience. The proposed methodology and key findings will enable system planners and operators to understand when and where the power system will experience energy losses under unpredictable extreme floods and help them to decide on the most effective resilience enhancement strategies.