Treatment of Uncertainty for Windstorm Risk Assessment

The special collection on Treatment of Uncertainty for Windstorm Risk Assessment is available in the ASCE Library (https://ascelibrary.org/page/ajrua6/treatment_uncertainty_windstorm_risk).

Wind is an inherently uncertain process. This uncertainty is due to the various factors (e.g., terrain) that affect the wind, and the foundation for assessing the wind impacts on the built and natural environment (i.e., wind engineering) incorporates this uncertainty as a core component. Subsequently, a challenge of the analysis of the risks posed by wind (e.g., structural motion, wind damage, debris, deaths, and injuries) is to determine new and improved methods of treating uncertainties that emanate from wind and its associated impacts. By their very nature, extreme wind events generate large uncertainties due to their rarity and the difficulty of capturing vital information near the ground due to the pre- and postwindstorm conditions. These extreme events, however, cause a large proportion of wind-induced losses. As an example, the recent 2017 hurricanes that impacted the United States—Harvey, Irma, and Maria—caused $250 billion in damages to the United States and its territories (NOAA 2018); however, many significant questions remain as to the full extent of the impacts (e.g., Santos-Lozada and Howard 2018).

These recent wind events and others highlight the need to rigorously treat uncertainties for windstorm risk assessment. Because of the significant impacts and also the well-accepted linear framework (Davenport 1983) for assessing wind hazards, the study and treatment of wind-related uncertainties is a multidisciplinary problem. Topics of interest for this special collection include, but are not limited to, measurements and analysis from wind tunnel studies; poststorm field surveys; forensic engineering; analytical or numerical studies; remote sensing and full-scale studies; component to system to community level analysis (including the built and natural environment); comparison studies for single or multiple events; studies across different windstorm types; probabilistic and/or physical wind hazard (or multihazard) characterization, simulation, and modeling; wind vulnerability assessment and modeling; risk and reliability assessments; risk mitigation or reduction techniques and/or technologies; climate change impacts; human response and behavior; interdependencies or connections between infrastructure; and resilience identification and quantification.

This special collection is therefore dedicated to highlighting the ongoing research on the treatment of uncertainties for windstorm risk assessment. In the development of this collection, the guest editors looked for work that had the ability to bridge the gap between the range of scientific and engineering disciplines involved in windstorm risk assessment. A hope of the author is that the articles introduced here will help to pave the way for an open dialogue between the various interested sectors and to encourage collaboration between individuals and groups within these sectors. A particular focus of this special collection is to include contributions that seek to rigorously identify, attribute, and quantify uncertainties in the processes leading to windstorm losses as well as taking the additional step of proposing methods of risk mitigation and reduction based on their results.

In short, this special collection highlights and addresses some of the needs in the windstorm research community. The collection consists of 11 papers, each of which provides a different viewpoint of how uncertainties are treated in wind-related problems. Announcing each paper in order of this linear wind risk assessment framework, Duthinh et al. (2017b) developed a model that is able to estimate peaks from time series using a two-dimensional Poisson process approach in which the data are thresholded. Thresholding provides additional data for analysis that are considered extreme, which can improve the reliability of results. This type of method can be used for all types of windstorm risk assessment (e.g., wind hazard, wind loading). Three papers deal specifically with analysis of wind loading. Simiu et al. (2017) outlined a flexible approach for the use of wind load factors as applied to the wind tunnel method as outlined in ASCE 7-16 (ASCE 2017). Wind load factors are based on uncertainties in loading, which include the extreme wind climate, terrain effects, building aerodynamics, and structural response. Duthinh et al. (2017a) and Gierson et al. (2017) focused on the analysis of pressure coefficients ($C_p$) on low-rise buildings with specific emphasis on $C_p$ values associated with components and cladding (C&C). These studies used different methods to estimate $C_p$ values; however, generally it was found that the negative $C_p$ (suction) specified by ASCE 7-10 (ASCE 2010) for gable roofs of low-rise buildings can be too small. For walls, both positive and negative pressures can also be underestimated by current ASCE specifications.

From information on wind loading, practitioners and researchers can derive information on the response of structures to wind loading. Judd (2018) took an interesting look as this concept as applied to a 10-story steel frame building, taking a page from performance-based seismic engineering. Wind tunnel load records were modified to emulate the nonstationary effects of windstorms and response was estimated via finite-element analysis. Nonlinear response history analyses were used to calculate story drifts and roof and floor accelerations, and through fragility analysis, Monte Carlo simulations found damage to cladding and structural components, especially nonductile beam connections, was predicted.

Fragility analysis is a tool that is used to model the interrelation between wind speed, wind loading, structural response, and subsequent damage and is the focus of two papers in this special collection. Gavanski and Kopp (2017) investigated the behavior of roof-to-wall connections (RTWCs) in wood-frame housing, commonly damaged in extreme windstorms, through a fragility framework. They found that roof shape and the number and type of RTWCs are critical indicators of resistance to wind loading and that the treatment of the wind speed to damage relationship in the Enhanced Fujita (EF) Scale needs reassessment. Roueche et al. (2018) investigated the epistemic uncertainties of empirical fragility functions in postdisaster damage assessments considering both intensity measures (e.g., wind speed) and damage metrics (e.g., percent of roof damage). Using the 2011 Joplin, Missouri, tornado as a demonstration, Roueche et al. found that uncertainty in the intensity measure (wind speed for windstorms) contributed the most to the overall uncertainty of the fragility assessment.

Naturally progressing from analysis of wind damage (and all preceding steps) is the estimation of losses from said damage. Three papers in this collection address losses in windstorms as their primary focus. Caracoglia (2018) developed a unified framework that enables the estimation of both dynamic structural response and intervention costs in the case of nonstationary winds, such as thunderstorm downbursts by means of a generalized power-law cost function. This work necessarily included examination of the interdependence among wind load, damage, and costs, and the associated uncertainties. Prompted by the ever-increasing exposure along US coastlines, Gulati et al. (2017) investigated the probable maximum loss (PML) of residential and commercial building stock in South Florida when subjected to hurricanes using the Florida Public Hurricane Loss Model (FPHLM). They fit both parametric and nonparametric models to estimate PML, and found, in general, the parametric models provided better fits. Done et al. (2018) also investigated structures in Florida and the possible changes in losses due to the implementation of the Florida Building Code (FBC). They analyzed seven historical hurricanes that impacted Florida during 2004 and 2005 and found that wind speed, duration, and directional change were significantly correlated with insured loss. A multiple regression analysis found that homes built after the implementation of the FBC experienced significantly lower losses than homes built before the FBC implementation.

Finally, one last component of windstorm risk assessment was added in the last study of this special collection. Dong and Li (2017) looked beyond damage and loss as well as individual structures and attempted to quantify resilience of an entire residential community to the impact of a hurricane. This quantification of resilience included uncertainty quantification and the recovery time of a community to a hurricane event. If resilience goals of a community were identified, it was found that a given community could meet these goals, provided the community set individual resilience goals for each residence in its building stock.