Probabilistic Estimation Model for Seismic Physical Portfolio Loss of A Water Supply Pipeline System

Abstract

Losses due to hazards are expected. A study estimating the damages and losses was conducted, however, the damages per mesh were estimated as independent to one another. Spatially distributed lifeline systems, such as water supply pipelines, are interconnected, and seismic spatial variability affects the damages across the region; thus, spatial correlation of damage spots is an important factor to account for uncertainty in the target area for portfolio loss estimation. Researchers opt for simulations to predict and estimate possible losses; however, these simulations assume each damage behaves independently and is uncorrelated. The researchers assumed the damages behave in a Poisson Distribution, to avoid over-dispersion and eliminate negative losses in estimations. There are limited ways to Simulate Correlated Poisson Distributed values, thus NORTA (NORmal To Anything) is used. Including NORTA accounts for the correlation for Poisson Distributions. From the simulated losses at different scenarios and correlations, a probabilistic model based on the generalized linear model and central limit theory was produced to estimate the losses without simulations. d. Once the probabilistic model is obtained, the seismic risk index of a spatially distributed water supply pipeline system, such as the PML or NEL can be easily estimated. The proposed model can also be used in other lifeline systems; though additional investigation is needed for confirmation. The proposed method is then tested using an artificial field and a location-based scenario. This would aid in pre-disaster planning and would require only a few steps and time.