Show simple item record

dc.rights.licenseAl consultar y hacer uso de este recurso, está aceptando las condiciones de uso establecidas por los autores.es_CO
dc.contributor.advisorSánchez Silva, Edgar Mauricio 
dc.contributor.advisorDueñas Osorio, Leonardo Augusto
dc.contributor.advisorSong, Junho
dc.contributor.advisorAgarwal, Jitendra
dc.contributor.advisorBaker, Jack
dc.contributor.authorGómez Castro, Camilo Hernando 
dc.date.accessioned2018-09-27T16:39:18Z
dc.date.available2018-09-27T16:39:18Z
dc.date.issued2014
dc.identifier.urihttp://hdl.handle.net/1992/7819
dc.description.abstractThis thesis presents a framework for risk-informed decision support for infrastructure networks operation, referred to as the Complex Distributed Agent Network (CoDAN) framework. CoDAN has been designed to recognize and model infrastructure systems as the result of interacting physical, natural, and social systems, and is built upon the following principles: - different decision problems occur at different system description, or resolution, levels; - decentralization schemes respond to each specific system and problem; - control processes are embraced by distributed decision units to maintain and improve network indicators of performance and risk; and - socio-economic and organizational aspects affect infrastructure performance as much as technical aspects. To operationalize the stated principles, CoDAN is comprised of the three following phases: - Phase I ("Knowing the system") explores network properties (e.g., relative importance of network components) and embraces a human-machine interactive process to define relevant sub-systems in the context of a decision problem, supported on supervised and unsupervised clustering algorithms; - Phase II ("Who's in charge") develops resource allocation models in terms of subsystems, signaling which of those are appropriate for autonomous use of resources in the context of problems such as maintenance or disaster relief throughout the network; - Phase III ("Behavior under control") defines agents associated to key decision problems of autonomous sub-systems, and provides them with the capabilities to control variables of interest (e.g., connectivity, flow reliability) and pursue a target system statees_CO
dc.description.abstractPhase III is further divided into three modules. - first, the modeling module describes deterioration and shocks across a set of sub-networks that are associated to key decision problems; - second, the assessment module estimates current and future indicators of performance and risk in the network based on state-of-the-art algorithms (e.g., for reliability computation); - and third, the intervention module supports (sub)-optimal decisions about maintenance actions to maintain and improve performance or recover from disrupting events. The integration of CoDAN Phases enables the construction and use of an agentbased model in which decisions and policy about infrastructure can be studied. Subsystems (found in Phase I) act in a decentralized yet coordinated way (defined in Phase II) to execute local control processes (devised in Phase III) aimed at maintaining performance and service levels in a cost-effective way. Agents represent decision-makers in key sub-systems whose behavior depends on aging, hazards, and maintenance actions defined by the agents. Thus, the overall behavior of infrastructure networks is the result of physical processes, natural events, and interrelated decisions from multiple parties. These concepts are discussed, implemented, and illustrated with several examples throughout the document and a step-by-step analysis of the Chilean electricity supply system and its recovery after the 2010 earthquake using the CoDAN framework. The implementation and quantitative analysis of the CoDAN framework provide evidence of the benefits in terms of decision-support and computational efficiency of detecting relevant sub-systems in infrastructure and incorporating such information into optimal resource allocation problems, as well as into performance assessment and optimal maintenance scheduling problems, as demonstrated through the case study of the Chilean networkes_CO
dc.description.abstractThe distributed analysis of infrastructure networks provides insight about the effects of local decisions on global behavior, while enabling the analysis of sub-systems of tractable size. CoDAN's socio-technical modeling of infrastructure networks opens many possibilities for future work, aimed at exploring different decision-making settings, including cooperative/competitive parties, inefficient or corrupt entities, principal-agent models that capture regulatory policies, among others. Moreover, because of CoDAN's modularity, several extensions and improvements can be pursued, namely: devising formal large-scale and multi-objective approaches to complement the optimization in the intervention module; incorporating more sophisticated and efficient models for progressive deterioration and shocks, which fit specific realistic scenarios; expand the scope of the risk management problem beyond reliability (e.g., integrate robustness, resiliency, consequence analysis). Finally, a catalogue of realistic case studies for validation and adjustment of CoDAN modules is to be developed, which will allow to devise practical guides to use CoDAN and provide reference for further researches_CO
dc.format.extent164 hojases_CO
dc.language.isoenges_CO
dc.publisherUniandeses_CO
dc.sourceinstname:Universidad de los Andeses_CO
dc.sourcereponame:Repositorio Institucional Sénecaes_CO
dc.titleRisk-informed decision support for infrastructure network operation - the Complex Distributed Agent Network (CoDAN) frameworkes_CO
dc.typedoctoralThesises_CO
dc.publisher.programDoctorado en Ingenieríaes_CO
dc.rights.accessRightsopenAccess
dc.subject.keywordAnálisis de sistemas - Investigacioneses_CO
dc.subject.keywordAnálisis cluster - Investigacioneses_CO
dc.subject.keywordToma de decisiones - Investigacioneses_CO
dc.subject.keywordRedes de información - Investigacioneses_CO
dc.publisher.facultyFacultad de Ingenieríaes_CO
dc.type.versionpublishedVersion
dc.description.degreenameDoctor en Ingenieríaes_CO
dc.description.degreelevelDoctoradoes_CO


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record