Modelling correlated damages of residential building portfolios under tropical cyclone wind loads

Abstract

Performing probabilistic damage assessment for a community under a tropical cyclone event needs to consider the collective damage of individual structures within the community, which involves modeling the spatial correlations of hazard demands and structural capacities between individual structures. However, how to model these two kinds of spatial correlations and how they influence the damage assessment remain unclear. In this paper, focus is given to the roof sheathing damage of a residential building portfolio consisting of multiple wooden residential buildings under the wind loads of a tropical cyclone event. Two methods are used to predict the damage of individual buildings based on their hazard demands: one is using a direct Monte Carlo Simulation in which structural (capacity) parameters of different buildings are treated as correlated. This method provides accurate results but needs a lot of information. Another approach is to probabilistically predict the damage state of each building based on its hazard demand using its fragility functions. The relative importance of the correlations of hazard demands and structural capacities is investigated. It is demonstrated that the correlations of damage states are strongly dependent on hazard demands. Finally, a method is developed to simulate correlated damage states of a building portfolio given hazard demands, through incorporating the hazard-dependent correlations with fragility functions using Gaussian Copula.