Date of Award

5-2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mechanical and Civil Engineering

First Advisor

Jean-Paul Pinelli

Second Advisor

Nezamoddin Nezamoddini-Kachouie

Third Advisor

Nakin Suksawang

Fourth Advisor

Albert M. Bleakley

Abstract

As the population and properties in the state of Florida increase, there is an increase in hurricane losses. Hurricane winds and debris can impact the exterior components of a building envelope. The envelope breaches and defects of these components provide a pathway to hurricane wind-driven rain into the building, which causes interior and contents damage. Loss models, such as the Florida Public Hurricane Loss Model (FPHLM), estimate hurricane insured losses for portfolios of buildings. The vulnerability component of the FPHLM has a personal residential module (PR), for single family homes, and a commercial residential module (CR-LR), for apartment buildings. They are separate modules and although they share similar damage estimation procedures, the CR-LR module is more advanced and uses more recent methodologies. State-of-the-art vulnerability models use a component approach to model damage to exterior components while interior and contents damage are estimated with heuristic equations, which extrapolate these damages from exterior components damage. An exception is the FPHLM CR-LR vulnerability model which includes a novel interior and contents damage methodology, developed in 2019 by the author of this dissertation. The component and physics-based approach converts wind tunnel tests results to estimate the distribution of water ingress inside a building, its propagation among interior and contents components, and the subsequent damage from the amount of water each component absorbs. This interior and contents damage model is further improved in this dissertation. The PR and CR-LR vulnerability models of the FPHLM are then merged into a Unified Personal and Commercial Hurricane Vulnerability Model (UPCM). The PR component of the UPCMM benefits from the newer and more advanced interior and contents damage methods from CR-LR. The larger set of claims for PR allows for better validation and calibration of the interior and contents models, which shall benefit the CR-LR component of the UPCM. The UPCM includes an updated weighted vulnerability scheme based on updated county-based buildings statistics from a previous study for PR, which is now extended to CR-LR. The dissertation also includes a preliminary validation on this new UPCM. Finally, the dissertation discusses how the uncertainty attached to some of the key sets of input variables of the vulnerability component of the FPHLM affect the overall model vulnerability outputs. The input data sets include: the estimates of exterior damage and defects, the wind-driven rain (WDR), the rain-admittance factors (RAF) and surface run-off coefficients (SRC), the coefficients defining the water ingress distribution inside the building, and the water absorption capacities of interior and contents components. In addition, the dissertation considers the impact of the variation of these set of variables with the direction of the maximum wind speed. A sensitivity study of the impact of some stochastic as well as deterministic input variables complemented the uncertainty study and showed that the vulnerability model is relatively insensitive to large variations of individual input variables. The analyses showed that at low wind speeds the uncertainty from WDR and RAF and SRC had the highest impact on both the model coefficient of variation (COV) and mean, and the model was most sensitive to WDR and to a lower instance to RAF and SRC. The analyses show that at moderate wind speeds the exterior damage components are responsible for most of the uncertainty in the model and have the highest contribution to the overall uncertainty. Some of the estimates of external damage such as roof damage have also some of the highest relative impacts on the expected vulnerabilities. At high wind speeds where most if not all components are completely damaged with almost complete certainty, the uncertainty from any of the model processes and input variables has little to no influence on the model uncertainty, which is close to zero at high wind speeds. In summary, this dissertation presents a novel interior and contents vulnerability damage model, which is part of a unified vulnerability model that covers both personal and commercial residential buildings. Moreover, sensitivity and uncertainty analyses result in a better understanding of the impact of different input variables on the mean and uncertainty range of non-linear vulnerability models.

Share

COinS