Accurate and efficient resilience assessment of coastal electric power distribution networks

Weather events such as storm induced wind, surge, and wave have imposed significant damage to coastal power infrastructure.Although many studies have investigated the performance of coastal infrastructure, there remains a critical gap in computationally efficient probabilistic resilience assessment of coastal distribution systems.This study addresses this issue by first developing a probabilistic wind-surge-wave model, using 10,000 years of synthetic storm dataset.The dataset comprises approximately 3,000 storms.The storms are categorized based on their maximum wind speed.As storm events follow a Poisson Point Process, the time between storm events can be estimated using the exponential distribution.To reduce computational costs, 200 storm samples were selected from the 3,000-storm ensemble such that the CDF of the model remains identical to the original dataset and the storms' spatial distribution is kept uniform.A Sequential Monte Carlo (SMC) simulation was then adopted to simulate storm events over a 70-year period.For the 200-storm ensemble, wind, surge, and wave values are estimated via ADCIRC+SWAN platform.To further decrease the computational costs while ensuring the accuracy of the model, an Artificial Neural Network (ANN) model was trained on failure probabilities estimated by Monte Carlo (MC) simulations, to generate a set of efficient high-fidelity parametrized fragility models for wood and prestressed concrete (PC) poles.To assess the efficiency of the framework, a power distribution