Loading...

Modeling Interdependent Infrastructure Systems Using Bayesian Network for Emergency Management

Doctor Arastoo, Maral | 2020

388 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 53845 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Mahsoli, Mogtaba
  7. Abstract:
  8. This thesis proposes a probabilistic framework to increase situational awareness about serviceability of infrastructure systems, and extend and distribution of losses in the posthazard state of communities. Maintaining situational awareness is the first and foremost step in prioritizing search and rescue operations and organizing resources. This is a challenging task due to the uncertainties that exist in infrastructure systems’ performance after occurrence of a hazardous event. These uncertainties gradually diminish as further information is received. Therefore, a probabilistic framework is the necessary element for disaster management. This framework must be able to update initial beliefs about community status using different information sources. The other challenge in attaining situational awareness is the need to model the interdependencies between different infrastructure systems. The interdependencies can either make the system redundant and therefore lower the failure probability or give rise to cascading failure. In this study, Bayesian network is used to model hazard, infrastructure systems and the interdependencies among them. Bayesian network uses Bayes’ rule to update the failure probability of infrastructure systems based on available observations while taking into account the interdependencies. On the other hand, near-real-time inference algorithms are developed to update prior probabilities in a Bayesian network. These unique features make Bayesian network a great tool for emergency management. The proposed framework consists of five major modules i.e., hazard, damage, functionality, component serviceability and infrastructure system serviceability. In the hazard module, seismic sources, earthquake magnitude, location and time of earthquake occurrence, earthquake intensity, and permanent ground displacement due to liquefaction and surface faulting at the location of system components are modeled. In the damage module conditional probability distribution of the damage state of each infrastructure component is modeled. A damaged component may either be partially functional or unfunctional. Conditional probability distribution of a component’s functionality conditioned on its damage state is modeled in the functionality module. Functionality of a component if solely a function of its own damage, so neither dependencies nor interdependencies are playing a role in determining the functionality level of a component. In the serviceability module, first the dependencies and interdependencies between and within different infrastructure systems are modeled. Then, serviceability of each component is modeled as a conditional probability distribution conditioned on functionality of the component itself and other components that influence its serviceability. In the last module, i.e., infrastructure serviceability, the serviceability of infrastructure systems at different regions of the city are modeled as a conditional probability distribution conditioned on serviceability of components. Bayesian network fuses pieces of information received from different information sources in order to alleviate uncertainties in damage, functionality and serviceability of infrastructure systems. These information sources include reconnaissance teams, physical sensors, aerial images, crowdsourcing, social sensors, et cetera. The framework is showcased by a comprehensive application to a community comprising a portfolio of 125000 occupants, 17000 buildings, an electric power system, a water system, a transportation network and a healthcare system subject to seismic hazard. The proposed framework may be used as a decision support system in emergency management by increasing situational awareness and decreasing uncertainties about damage, functionality and serviceability of infrastructure systems. For example, having the posterior probability of hospitals’ serviceability levels provided by Bayesian network, the framework can determine the best hospital to transfer the injured victims to
  9. Keywords:
  10. Infrastructure ; Probabilistic Modeling ; Disaster Management ; Bayesian Network ; Interdependency ; Interdependent Infrastructure Systems ; Situational Awareness

 Digital Object List

 Bookmark

No TOC