Decision Support System for Large-Scale Evacuation Logistics Prof. Dr. Horst W. Hamacher 10. Europäischer Bevölkerungsschutzkongress 9.-10. September 2014 Bonn Fachforum X: Forschung für die zivile Sicherheit Praxisorientierte Forschung für Einsatz- und Rettungskräfte Funded by German Ministry of Research and Technology Outline 1. Evacuation Scenarios a) Types of evacuation b) Evacuation projects in Kaiserlautern 2. Planning Modules a) Selection of Evacuation Area, Collection and Shelter Points b) Identifying Evacuation Network c) Organisation and Control of Evacuation Traffic I. Bus Traffic II. Individual Traffic d) How do deal with conflicting objectives 3. Conclusion Bevölkerungsschutzkongress, Bonn, September 2014-2 - Horst W. Hamacher 1
Evacuation Scenarios Aircraft evacuation All passengers and crew must be evacuated from the aircraft to the ground within 90 seconds (Federal Aviation Administration) Ship evacuation maximum evacuation time is 60 minutes Building evacuation Adherence to evacuation regulations Analysis of existing buildings Design of new buildings Bevölkerungsschutzkongress, Bonn, September 2014-3 - Horst W. Hamacher Evacuation Scenarios Regional evacuation Small emergency planning zone Defined security points Short term evacuation with return Main task: Model pedestrian movement Large-scale evacuation Small and large emergency planning zones Selection of shelters Long term evacuation without immediate return Evacuation logistics Main task: Model evacuation transportation and logistics Bevölkerungsschutzkongress, Bonn, September 2014-4 - Horst W. Hamacher 2
Some Goals of Evacuation Planning Evacuation routes Evacuation times (Lower and upper) Bounds for predictions What if questions Evacuation plans and preparation On- and off-line virtual training Locational planning Logistics for evacuee care Information system User-friendly decision support system Incorporate expert knowledge Acceptance of user community Bevölkerungsschutzkongress, Bonn, September 2014-5 - Horst W. Hamacher Evacuation Project EVACNET+ Funded by NSF (1982-85) Funding for COCO (Center for Optimization and Combinatorics, 1985-91) Francis, Hamacher, Kisko, Tufekci University of Florida, USA, 1982-91 and TUKaiserslautern, 2000 Building evacuation: DSS to check evacuation time for high-rise buildings Regional evacuation: DSS for hurrican warning Bevölkerungsschutzkongress, Bonn, September 2014-6 - Horst W. Hamacher 3
Recent Projects in Kaiserslautern Regional Evacuation: Planning, Control und Adaptation Funded by German Ministry of Research and Technology 2009-2012 Building evacuation (funded by TU Kaiserslautern): Fritz-Walter football stadium Check adherence of evacuation times to FIFA rules Regional evacuation: DSS for evacuating Betzenberg area 48.000 evacuees Bevölkerungsschutzkongress, Bonn, September 2014-7 - Horst W. Hamacher Recent Projects in Kaiserslautern Regional Evacuation: Planning, Control und Adaptation Funded by German Ministry of Research and Technology 2009-2012 Bevölkerungsschutzkongress, Bonn, September 2014-8 - Horst W. Hamacher 4
Recent Projects Unterstützung bei der Katastrophenschutzplanung für kerntechnische Anlagen: Evakuierungsplanung funded by Regional evacuation: Evacuating areas around nuclear power plant in Philippsburg, Germany, Hamacher, Ohst, Ruzika, Seipp Bevölkerungsschutzkongress, Bonn, September 2014-9 - Horst W. Hamacher Recent Projects in Kaiserslautern Urban Security Optimization of Location and Layout for Public Events 2013-2016 (Standort- und Layoutoptimierung für Sichere Öffentliche Veranstaltungen) (StanLay) Hamacher, Kovtaniuk, Maier Bevölkerungsschutzkongress, Bonn, September 2014-10 - Horst W. Hamacher 5
Recent Projects in Kaiserslautern Decision Support System for Large-Scale Evacuation Logistics Joint German-French project 2012-2015 funded by Bevölkerungsschutzkongress, Bonn, September 2014-11 - Horst W. Hamacher Recent Projects in Kaiserslautern Decision Support System for Large-Scale Evacuation Logistics Case studies: Plane crash in the city of Earthquake and Submersion Kaiserslautern on the coast of Nice Long term evacuation of a large number of ecacuees Bevölkerungsschutzkongress, Bonn, September 2014-12 - Horst W. Hamacher 6
Securing and Visualizing data Large-scale evacuation Small and large emergency planning zones Selection of shelters Long term evacuation without immediate return Evacuation logistics Main task: Model evacuation transportation and logistics Is getting data as easy as drawing circles? No!!! Use public data (e.g. Open Street Maps for street networks) confidential data (e.g. census data) and devolop software tool (VisualFlow) Bevölkerungsschutzkongress, Bonn, September 2014-13 - Horst W. Hamacher Evacuation Area Bevölkerungsschutzkongress, Bonn, September 2014-14 - Horst W. Hamacher 7
Evacuation Area (zoomed) Bevölkerungsschutzkongress, Bonn, September 2014-15 - Horst W. Hamacher Evacuation Area (extended) Bevölkerungsschutzkongress, Bonn, September 2014-16 - Horst W. Hamacher 8
Available Collection Points Bevölkerungsschutzkongress, Bonn, September 2014-17 - Horst W. Hamacher Selected Collection Points Bevölkerungsschutzkongress, Bonn, September 2014-18 - Horst W. Hamacher 9
Available Collection Points Is it simple to choose good collection points? How many options does the planner have? In this example 41 available points, 5 selected points 17.985.552 Use mathematical optimization: Locational Planning (Standortplanung) The same holds for the selection of shelters Bevölkerungsschutzkongress, Bonn, September 2014-19 - Horst W. Hamacher Evacuation Area, Collection and Shelter Points Bevölkerungsschutzkongress, Bonn, September 2014-20 - Horst W. Hamacher 10
Individual Shelter Information Bevölkerungsschutzkongress, Bonn, September 2014-21 - Horst W. Hamacher Statistics of Data Bevölkerungsschutzkongress, Bonn, September 2014-22 - Horst W. Hamacher 11
Identify Street Network Start with information from OpenStreetMap (OMP) http://www.openstreetmap.org Safety is guaranteed by comparing with official security maps, and freezing certified versions of maps Bevölkerungsschutzkongress, Bonn, September 2014-23 - Horst W. Hamacher Identify Street Network Bevölkerungsschutzkongress, Bonn, September 2014-24 - Horst W. Hamacher 12
Simplify Street Network VisualFlow_Linearize-Graph Bevölkerungsschutzkongress, Bonn, September 2014-25 - Horst W. Hamacher From Street Network to Math Network Map: Referat Stadtentwicklung, Stadt Kaiserslautern Bevölkerungsschutzkongress, Bonn, September 2014-26 - Horst W. Hamacher 13
From Street Network to Math Network Map: Referat Stadtentwicklung, Stadt Kaiserslautern Bevölkerungsschutzkongress, Bonn, September 2014-27 - Horst W. Hamacher Macroscopic Model Research Topic: Aggregation of networks Bevölkerungsschutzkongress, Bonn, September 2014-28 - Horst W. Hamacher 14
Data Selection in VisualFlow Selection of Evacuation Area, Collection and Shelter Points in demo_evacmapp_01 Bevölkerungsschutzkongress, Bonn, September 2014-29 - Horst W. Hamacher Evacuation Traffic (Bus) in VisualFlow Bevölkerungsschutzkongress, Bonn, September 2014-30 - Horst W. Hamacher 15
Individual Evacuation Traffic Models are suitable for pedestrian movement, and car movement Bevölkerungsschutzkongress, Bonn, September 2014-31 - Horst W. Hamacher Network Model Detail of network Input: Capacities, Travel Times, (Costs) Model: Persons/Cars corresponds to (dynamic) flow units Idealized assumptions to achieve provably bestpossible evacuation Bevölkerungsschutzkongress, Bonn, September 2014-32 - Horst W. Hamacher 16
Dynamic Network Flow 2 1 4 3 Bevölkerungsschutzkongress, Bonn, September 2014-33 - Horst W. Hamacher Dynamic Flows and Time-Expanded Networks 2 1 2 1 2 1 2 1 2 1 2 1 2 1 4 3 3 3 3 3 3 3 4 4 4 4 4 4 time Bevölkerungsschutzkongress, Bonn, September 2014-34 - Horst W. Hamacher 17
FlowLoc Develop models for Evacuation + Location of emergency units + Location of commercial facilities Bevölkerungsschutzkongress, Bonn, September 2014-35 - Horst W. Hamacher Identifying Essential Evacuation Streets Usage of streets in 100 runs of software: >75 % <75% <50% < 25% 0% Bevölkerungsschutzkongress, Bonn, September 2014-36 - Horst W. Hamacher 18
Dealing with Conflicting Objectives Conflicting objectives, e.g. Cost of evacuation (EvacCost) Evacuation time (Evac-Time) Number of busses (Numb-Bus) Apply tools of multicriteria optimization: Find representative set of Pareto optimal solutions Store these solutions in data base Provide search engine for decision makers to find best solution Bevölkerungsschutzkongress, Bonn, September 2014-37 - Horst W. Hamacher Search engine to find best Pareto solution Evac-Time Evac-Cost Numb-Bus Spider-Web Charts (Kasanen, Östermark, and Zeleny, 1989) Bevölkerungsschutzkongress, Bonn, September 2014-38 - Horst W. Hamacher 19
Search engine to find best Pareto solution Evac-Time Evac-Cost Possible evacuation plan taken from data base Numb-Bus Bevölkerungsschutzkongress, Bonn, September 2014-39 - Horst W. Hamacher Search engine to find best Pareto solution Evac-Time Evac-Cost Evacuation plan with improved evacuation cost Numb-Bus Bevölkerungsschutzkongress, Bonn, September 2014-40 - Horst W. Hamacher 20
Search engine to find best Pareto solution Evac-Time Evac-Cost Evacuation plan with improved number of shelters Numb-Bus Bevölkerungsschutzkongress, Bonn, September 2014-41 - Horst W. Hamacher Search Engine in VisualFlow VisualFlow: Search Engine Bevölkerungsschutzkongress, Bonn, September 2014-42 - Horst W. Hamacher 21
Summary To help evacuation planners Evacuation data is visualized number of evacuees distribution over evacuation area available collection shelter points evacuation network Evacuation transportation is optimized optimal selection of collection points and shelters optimal transport from collection to saftety (bus and individual traffic) (Busy) Evacuation routes are identified Conflicting evacuation objectives are visualized spider diagram Results are tested in (international) case studies Bevölkerungsschutzkongress, Bonn, September 2014-43 - Horst W. Hamacher Thank you for your attention more information: hamacher@mathematik.uni-kl.de Bevölkerungsschutzkongress, Bonn, September 2014-44 - Horst W. Hamacher 22