{"id":{"repo_id":"uiuc","oai_identifier":"oai:www.ideals.illinois.edu:2142/116120"},"canonical_url":"https://search.dev.ndltd.org/etd/uiuc/oai:www.ideals.illinois.edu:2142/116120","repository":{"repo_id":"uiuc","name":"University of Illinois - Urbana-Champaign","base_url":"https://www.ideals.illinois.edu/oai-pmh"},"display":{"title":"Hierarchical framework for flood evacuation: Optimization of evacuation order policies with consideration of human behavior","abstract":"Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2024-08-01","abstract_html":"Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2024-08-01","abstract_has_math":false,"creators":["Bianchini, Michael Joseph"],"institution":"University of Illinois at Urbana-Champaign","degree_name":"M.S.","degree_level":"Thesis","degree_discipline":"Civil Engineering","degree_department":null,"school":null,"contributors":["Cai, Ximing"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2022,"date_issued":"2022-08","date_published":"2022-08","updated_at":"2026-07-22T22:24:55Z","subjects":["flood evacuation","decision-support tool","MATSim","agent-based modeling (ABM)","reinforcement learning (RL)","machine learning (ML)","integrated modeling","human behavior","consequence estimation","flood risk","flood severity"],"languages":["en","eng"],"rights":["Copyright 2022 Michael Bianchini"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/2142/116120","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Cai, Ximing"]},{"key":"dc:creator","label":"Author","values":["Bianchini, Michael Joseph"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2022-08","2022-07-20"]},{"key":"dc:type","label":"Dc Type","values":["text","Thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Civil Engineering"]},{"key":"thesis:degree_level","label":"Degree Level","values":["Thesis"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M.S."]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["University of Illinois at Urbana-Champaign"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["flood evacuation","decision-support tool","MATSim","agent-based modeling (ABM)","reinforcement learning (RL)","machine learning (ML)","integrated modeling","human behavior","consequence estimation","flood risk","flood severity"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["en","eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Copyright 2022 Michael Bianchini"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["https://hdl.handle.net/2142/116120"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2024-08-01","The student, Michael Bianchini, accepted the attached license on 2022-07-19 at 13:18.","The student, Michael Bianchini, submitted this Thesis for approval on 2022-07-19 at 13:36.","This Thesis was approved for publication on 2022-07-20 at 09:58.","DSpace SAF Submission Ingestion Package generated from Vireo submission #18380 on 2022-11-15 at 21:40:36","Recent flood disasters and assessment of available modeling tools showcase the outstanding need for more effective flood emergency management. At the core of the problem is management failure to interpret and successfully act on hydrologic model results. The consequence is unnecessarily deadly floods, which have occurred recently around the world. A promising solution is the development of a decision-support tool built on the most advanced modeling applications to make actionable flood response recommendations from real-time hydrologic model results. Such models integrate flood, human, and traffic factors to uncover emergent properties of flood response. To leverage these integrated models for decision-support, optimization techniques are required. In this study, we develop an integrated modeling framework and apply a reinforcement learning (RL) algorithm to iteratively optimize the timing and location of evacuation order issuances from emergency management to residents. The core models include an agent-based human behavioral model (opinion dynamics model) to evaluate household evacuation decisions and an agent-based traffic model (MATSim) to simulate evacuation routes and associated consequences. A stochastic optimization model developed by Yi et al. (2016) is extended to include suggested evacuation orders and is applied to characterize the emergency manager’s objectives when issuing evacuation orders. The inclusion of both mandatory and suggested evacuation orders is a novel advancement in modeling techniques. Also, modifications to the MATSim application to include driving reactions to flooding represent a significant model improvement. We conclude the methodology with a proposed procedure to use the models to develop a rapid tool for real-time decision support. Tool development is ongoing as part of the National Science Foundation (NSF) funded Urban Flooding Open Knowledge Network (UF-OKN) project, and results are expected to be published soon. Tool development requires a large dataset of flood inundation model results; however, the tool is simple to apply in real-time because the decisions are predetermined. Readily available inputs of lead time to flood arrival and forecasted return period are the only two requirements. The performance of the RL algorithm and efficacy of different evacuation order strategies are compared in a demonstration study in the City of Wilmington, North Carolina. Five strategies are considered for evacuation: a phased evacuation with both suggested and mandatory orders; a phased evacuation with suggested orders only; a phased evacuation with mandatory orders only; a uniform evacuation with suggested orders only; and a uniform evacuation with mandatory orders only. Of the five strategies, a combination of suggested and mandatory orders with a phased evacuation strategy performed the best. Strategies with phased evacuations are shown to either reduce or maintain flood consequences compared to all strategies with uniform evacuations. Strategies with only suggested orders produce gradual evacuation responses with high evacuation efficiency but with concomitant risks at home and on the road. The least efficient evacuation responses resulted from strategies with only mandatory orders because of traffic bottlenecks that developed during the evacuation process. However, both phased and uniform evacuations with mandatory orders all resulted in zero expected loss of life. These results provide support for further development of the rapid decision-support tool. The proposed tool has demonstrated advantages over existing tools, which either ignore flooding risks, do not consider suggested orders, or consider uniform strategies only. Continued tool development is recommended to overcome limitations of the study. Improvements include consideration of more realistic transportation model assumptions, advancements to the human behavioral model, and data collection to calibrate the modeling environment."]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Hierarchical framework for flood evacuation: Optimization of evacuation order policies with consideration of human behavior"]}]}],"canonical_facts":{"dc:contributor":["Cai, Ximing"],"dc:creator":["Bianchini, Michael Joseph"],"dc:date":["2022-08","2022-07-20"],"dc:description":["Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2024-08-01","The student, Michael Bianchini, accepted the attached license on 2022-07-19 at 13:18.","The student, Michael Bianchini, submitted this Thesis for approval on 2022-07-19 at 13:36.","This Thesis was approved for publication on 2022-07-20 at 09:58.","DSpace SAF Submission Ingestion Package generated from Vireo submission #18380 on 2022-11-15 at 21:40:36","Recent flood disasters and assessment of available modeling tools showcase the outstanding need for more effective flood emergency management. At the core of the problem is management failure to interpret and successfully act on hydrologic model results. The consequence is unnecessarily deadly floods, which have occurred recently around the world. A promising solution is the development of a decision-support tool built on the most advanced modeling applications to make actionable flood response recommendations from real-time hydrologic model results. Such models integrate flood, human, and traffic factors to uncover emergent properties of flood response. To leverage these integrated models for decision-support, optimization techniques are required. In this study, we develop an integrated modeling framework and apply a reinforcement learning (RL) algorithm to iteratively optimize the timing and location of evacuation order issuances from emergency management to residents. The core models include an agent-based human behavioral model (opinion dynamics model) to evaluate household evacuation decisions and an agent-based traffic model (MATSim) to simulate evacuation routes and associated consequences. A stochastic optimization model developed by Yi et al. (2016) is extended to include suggested evacuation orders and is applied to characterize the emergency manager’s objectives when issuing evacuation orders. The inclusion of both mandatory and suggested evacuation orders is a novel advancement in modeling techniques. Also, modifications to the MATSim application to include driving reactions to flooding represent a significant model improvement. We conclude the methodology with a proposed procedure to use the models to develop a rapid tool for real-time decision support. Tool development is ongoing as part of the National Science Foundation (NSF) funded Urban Flooding Open Knowledge Network (UF-OKN) project, and results are expected to be published soon. Tool development requires a large dataset of flood inundation model results; however, the tool is simple to apply in real-time because the decisions are predetermined. Readily available inputs of lead time to flood arrival and forecasted return period are the only two requirements. The performance of the RL algorithm and efficacy of different evacuation order strategies are compared in a demonstration study in the City of Wilmington, North Carolina. Five strategies are considered for evacuation: a phased evacuation with both suggested and mandatory orders; a phased evacuation with suggested orders only; a phased evacuation with mandatory orders only; a uniform evacuation with suggested orders only; and a uniform evacuation with mandatory orders only. Of the five strategies, a combination of suggested and mandatory orders with a phased evacuation strategy performed the best. Strategies with phased evacuations are shown to either reduce or maintain flood consequences compared to all strategies with uniform evacuations. Strategies with only suggested orders produce gradual evacuation responses with high evacuation efficiency but with concomitant risks at home and on the road. The least efficient evacuation responses resulted from strategies with only mandatory orders because of traffic bottlenecks that developed during the evacuation process. However, both phased and uniform evacuations with mandatory orders all resulted in zero expected loss of life. These results provide support for further development of the rapid decision-support tool. The proposed tool has demonstrated advantages over existing tools, which either ignore flooding risks, do not consider suggested orders, or consider uniform strategies only. Continued tool development is recommended to overcome limitations of the study. Improvements include consideration of more realistic transportation model assumptions, advancements to the human behavioral model, and data collection to calibrate the modeling environment."],"dc:format":["application/pdf"],"dc:identifier":["https://hdl.handle.net/2142/116120"],"dc:language":["en","eng"],"dc:rights":["Copyright 2022 Michael Bianchini"],"dc:subject":["flood evacuation","decision-support tool","MATSim","agent-based modeling (ABM)","reinforcement learning (RL)","machine learning (ML)","integrated modeling","human behavior","consequence estimation","flood risk","flood severity"],"dc:title":["Hierarchical framework for flood evacuation: Optimization of evacuation order policies with consideration of human behavior"],"dc:type":["text","Thesis"],"thesis:degree_discipline":["Civil Engineering"],"thesis:degree_level":["Thesis"],"thesis:degree_name":["M.S."],"thesis:institution_name":["University of Illinois at Urbana-Champaign"]},"updated_at":"2026-07-22T22:24:55Z"}