The goal of this research is to formulate the enhanced concept of community resilience by the incorporation of recent theoretical and computational capabilities from structural system reliability. The fundamental theory builds upon the integrity of a structure in terms of its role in the physical, social and economic well-being of the community it serves. This leads to an approach to structural design based on the inherent tradeoffs for societal investment in structures in terms of their sustainability for intended functions, and the consequent resilience of the community. Inherent in this concept is the sensitivity of the performance of structures to such issues as low probability, high consequence events, and even black swans. To accomplish this overarching goal, structural system reliability algorithms will be utilized to investigate the sensitivity of structures to multiple failure modes and compound load combinations, leading to an evaluation of their impact on overall integrity and functionality. This will form the basis for the relationship between resilience and life cycle evaluation, the building block of sustainability. Alternative measures of uncertainty will be incorporated into models of structure and community resilience as time-varying processes, enabling the development of efficient approaches to holistic infrastructure.
This research will directly enhance the relationship between structural reliability and long-term sustainability, increasing the effectiveness of investments in the built environment, especially in a life cycle context. Potential consequences due to occupancy, environmental degradation, and both natural and intentional hazards ultimately impact community development decisions regarding growth patterns, land use restrictions, building codes and mitigation plans. The results of this research will lead to more robust, resilient and efficient built environments. New educational paradigms will also result, with interdisciplinary approaches to civil systems and hazards from both engineering and sociological perspectives. The project will also provide advanced training to graduate students on the topic of systems reliability, resilience and sustainability through their involvement in the project research activities.
Mathematical theories of structural reliability have been developed over the past several decades, and are now available as a practical resource that has led to a firm probabilistic basis for structural building codes and standards. One of the major mathematical advances that has enabled this is the ability to handle structural systems with a high degree of dependent failure modes. In this project, the concepts have been extended and for the first time adapted for a system that consists of a community portfolio of buildings, rather than a single building of multiple modes. The application considered in this research is the seismic safety of a community of buildings, and the approach adopted is a widely-used and proven method of linearization called the First Order Reliability Method (FORM). The prediction of future losses from earthquake events and other natural hazards is of importance to community developers, insurance entities, political organizations and many others in hazard-prone regions. Often, this risk assessment is preferred at a regional level as many private and public entities are concerned with the impact of an earthquake on a suite of buildings, as opposed to that for a single site. Assessing risk at a regional level is more complicated than doing so for individual sites due to the correlation that exists between the performances of spatially distributed buildings within a single hazard. This spatial correlation has been shown to be vital for characterizing potential loss at a regional level; however, it is often neglected in existing loss estimation methodologies. Unlike existing loss estimation tools that evaluate loss based on expected values or with the use of simulation, the proposed method evaluated the distribution of potential losses analytically while incorporating the spatial correlation, and is also computationally efficient. In addition, sensitivity measures are computed using FORM to prioritize cost-effective retrofit strategies within a building portfolio. The proposed method is applied to a selected San Francisco building inventory to estimate total structural and nonstructural repair cost in the form of loss exceedance curves. Sensitivity measures are used to prioritize building types that yield the most reduction in regional risk per dollar of retrofit. In addition to quantifying losses, the proposed framework is extended to assess the seismic resilience for the San Francisco building portfolio. Sensitivity measures are computed relative to changes in system resilience for each dollar allocated to pre-disaster retrofit and to increasing post-disaster restoration efficiency. The study also investigates the extension of the proposed FORM-based approach to assess the cumulative hazard-induced risk for regions subjected to multiple hazards. In this extended study, FORM is used to compute the distribution of loss for Charleston County, South Carolina, specific to potential earthquake and hurricane wind hazards. The approach provides an analytical and efficient tool for quantifying hazard risk at a regional level. By more effectively quantifying hazard-induced loss, resilience and sensitivities within a portfolio system, information is provided to improve hazard risk assessments and support more efficient risk management decision making. A second area of investigation in this study involves situations where methods of uncertainty quantification other than probability might be more appropriate. Specifically, situations involving natural hazards, and earthquake in particular, are often descried by linguistic variables that involve a high degree of subjective judgment and imprecision. Research over the past few decades into the role of uncertainty has led to development of new uncertainty theories, distinct from the theory of probability. These theories provide mathematical models different from probability theory, to represent and manipulate uncertainty. Fuzzy set theory relaxes the notion of classical set theory and provides a mathematical model for handling information that is vague, ambiguous, or 'fuzzy' in nature. Other contributions include imprecise probabilities, possibility theory, and evidence theory. Together these form an area termed Generalized Information Theory (GIT), aimed at formally recognizing and systematically dealing with the nature and scope of uncertainty, and is association with partial knowledge. In this project selected ideas of GIT, in particular fuzzy set theory and possibility classification concepts, have been investigated from the practical structural engineering viewpoint. They have been explained in the context of their adoption by the structural engineering community as alternatives to probability theory for those situations in which they might be more appropriate. A detailed application has been provided for the 'tagging' of buildings after an earthquake. This is the situation in which rapid assessment of the condition of buildings following an earthquake is made to determine whether individual buildings fall into the category of safe for immediate occupancy (green tag), safe after relatively minor repairs (yellow tag), or unsafe until major repair or replacement (red tag). The advantages of the broader tools of uncertainty is shown to be advantageous in certain situations.
