Data-driven framework for automated simulation of wind and windborne debris effects for hurricane regional loss estimation

Abstract

Hurricanes are a major driver of disaster losses in the United States; these losses can be minimized through the development of open-source scientific workflows that conduct site-specific, building-specific and component-level loss assessments across entire regions. While today's open-source workflows provide parcel-level granularity, common use of non-specific building characterizations limits the fidelity of these representations, particularly with respect to site-specific effects of wind and windborne debris (WBD) hazards. In response, this paper presents a new framework that utilizes publicly available data to enable modeling and analysis capabilities necessary to (1) capture site-specific, building-specific exposure to wind and WBD hazards and (2) propagate these demands through the building envelope's load path. The proposed data-driven framework is applied to an office building located in Florida's Bay County, the landfall site of Hurricane Michael. Leveraging a new building data model for regional loss estimation, this case study demonstrates the framework's use of open data to automate population of site models, WBD missile environment, wind pressure distributions, and component capacities to enable fault tree analyses for target and source buildings. The proposed approach can be utilized to identify hazard- and building-specific, component-level damage consistent with actual damage observed in Hurricane Michael.