Office of Emergency Management Mr. Andrew Mark Group 6: Paul Antonios, Tamara Dabbas, Justin Fung, Adib Ghawi, Nazli Guran, Donald McKinnon, Alara Tascioglu Quantitative Capacity Building for Emergency Evacuations of Manhattan, NYC: Manhattan Evacuation Simulation Application (MESA) - A time-oriented application for modeling capacity- constrained scenarios as a network model
23
Embed
Group 6: Paul Antonios, Tamara Dabbas, Justin Fung, Adib Ghawi, Nazli Guran, Donald McKinnon, Alara Tascioglu Quantitative Capacity Building for Emergency.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Office of Emergency Management Mr. Andrew Mark
Group 6: Paul Antonios, Tamara Dabbas, Justin Fung, Adib Ghawi, Nazli Guran, Donald McKinnon, Alara Tascioglu
Quantitative Capacity Building for Emergency Evacuations of Manhattan, NYC: Manhattan Evacuation Simulation Application (MESA) - A time-oriented application for modeling capacity-constrained scenarios as a network model
“An efficient evacuation of Manhattan is impossible.”
-Joel Friedman, P.E., Chief Engineer of NYC Department of Transportation
Our Client: Office of Emergency Management
OEM NYC is responsible for planning & preparing for any emergency that the city might encounter:
• Increase awareness of imminent emergency• Proper preparation procedures• Coordinating with all agencies for:• Maximum safety for people of NYC• Minimum damage to affected areas• Dependable emergency response and recovery
Our Goal
OEM challenge: the ability to evacuate all people from necessary areas in time, using the limited resources and methods of transportation available.
Thus, our goal is to supply our client with an application that will simulate the evacuation of Manhattan, New York City.
Initial HypothesisInitial formulation and methods of transportation are available
A Linear Programming Formulation:
Objective Function: Minimize maximum evacuation time of populations
Subject to: Constrained Capacities, modes of transport available
• Roadway Vehicles (Cars, taxis, buses…)• Subways & Railways • Ferries• Planes• Ships & Boats (Both public and private• Bikes & skateboards• Walking • Hot air balloons
Data CollectionWhat data do we need? Where did we get it from?
• Neither of them gives the optimal solution as an output
• Finds the shortest path based on the path of all previous census tracts chosen with the minimal paths for all source-end node combinations
• Source nodes increase the flow over the path that is chosen to be optimal by the initial population of the node.
Different Models Floyd-Warshall and CCRP
Floyd-Warshall’s Algorithm:
Advantages:• Runs in n3 time• For each source node, first it calculates the shortest distance between all node pairs
Disadvantages:• Assumes a static weight over each edge• Does not update the flow over edges throughout the execution of the algorithm
CCRP
Advantages:• Has a relatively better run time due to its super source node
Disadvantages:• Super source node cannot hold specific data about each source node• It takes one order• Requires maximum capacity for all nodes and edges• Demands constant weight inflow characteristics for all nodes and edges
MESA Algorithm
Different Models
MESA algorithm:
Advantages:• Considers multiple orders and chooses the best one • Calculates the weight over each edge• Updates the flow over as it iterates through the current optimal path of
each source node
Disadvantages:• Difficult to implement compared to CCRP
Basic Formula:
• Time = function (people that want to use the route, people that are already using the route, the capacity of the route, adjustment factor for assumed congestion)
• Dynamically optimizes the evacuation route for every tract
Model Assumptions
• Each entity emanating from a source node (a person) is assumed to be equal.
• Each person(s) will evacuate and comply with all evacuation instructions.
• The structure of Manhattan – the speed over the available roads, bridges, tunnels, and subways, does not change throughout the course of the evacuation.
• After reaching an exit point, people automatically and ‘perfectly’ dissipate.
• All Edges can be traversed in both directions.
Application Development
Three types of nodes:
1. Source nodes2. End nodes3. Intermediary nodes
Edges represent traversable routes between nodes.
Explanation of the Algorithm
GOAL: Map source nodes to exit nodes; effectively moving a body of persons within a community district to optimal exit points via the most efficient route.
Four slightly different algorithms for four different objectives:
1. Minimize the total elapsed exit time2. Minimize the average exit time for each starting location3. Maximize the total elapsed exit time (worst-case)4. Maximize the average exit time for each starting location
(worst-case)
MESAQuick look at the application
Data Analysis & Sample SimulationsTotal Evacuation Time
Data Analysis & Sample SimulationsEffect of Excluding the Subway Transportation System
Data Analysis & Sample SimulationsEffect of Excluding Bridges and Tunnels South of 14th Street
Data Analysis & Sample SimulationsEffect of Increasing Population of Financial District by 50%
Data Analysis & Sample SimulationsEffect of Increasing Subway System Efficiency
Future Research1. Capacity Improvement
Roadway Network
Implement Evacuation Circles:
Continually moving routes with outbound pickups in Manhattan and drop-offs outside Manhattan.
Subway Network
•Decrease stops through Manhattan and increase continuous travelling•Have subway act as a shuttle on/off island
Ferries & other waterborne vehicles network
All private boats should be utilized in conjunction with military and public vehicles
Future Research 2. Algorithmic Improvements
• Model NYC instead of just Manhattan
• Determine more accurate capacities for all transportation routes
• Simulate stochastic events
• Implement directed edges and time-delayed events
• Addressing MESA assumptions:• Uniform evacuees• Refugee-effect on neighboring geographies