Electronic Thesis/Dissertation


A Resilience Analysis of Buildings Critical Systems Using Bayesian Networks Open Access

Downloadable Content

Download PDF

Despite the efforts in developing frameworks to analyze and measure resilience, there is lack of consensus on the development and definition of metrics and limited quantitative analysis approaches. The resilience of buildings’ critical systems has not been assessed quantitatively, thus impacting the readiness and recovery of buildings during systems performance variations caused by natural, human-made, and technological disturbances. The quantitative analysis of systems-specific (non-networked) resilience can contribute to optimize the lifeline systems resilience and utilization during disturbances. This research is proposing a causal model using Bayesian Networks to estimate the operational resilience of buildings’ critical systems. Besides the gaps in resilience research, this bottom-up approach also considers the gaps in building sciences and facilities and maintenance management domains where static models provide qualitative information about systems conditions to decision makers, despite the abundant quantitative data collected during the life cycle of the assets. The probabilistic graphical model using Bayesian Networks provides causal models using systems failure and maintenance performance data to estimate the resilience of the systems based on internal and external disturbances. Internal disturbances are a function of reliability, availability, and maintainability attributes of the buildings’ critical systems and external disturbances are characterized by natural and technological disasters. Data for this research was collected from generic databases of buildings operating in the USA. Data was aggregated from multiple databases and analyzed to estimate the parameters needed to quantify the systems’ resilience. A stochastic point process was used to analyze failure data and continuous distribution models for maintainability metrics and external disturbances data analysis to estimate prior probabilities. Although this model measures resilience of buildings systems, it can be applied to other engineering systems and domains. Results will provide facility, maintenance and engineering managers a quantitative decision support tool for a data-driven and effective informed decision-making process to support organizations’ mission and business continuity.

Author Language Keyword Date created Type of Work Rights statement GW Unit Degree Advisor Committee Member(s) Persistent URL