Network Traffic Management under Disaster Conditions by Hediye Tuydes PhD Candidate Northwestern University
Motivation Most Common Disasters Natural Earthquakes Hurricanes Floods Volcanic eruptions Fires Tsunamis etc. Non-natural (Man-made) Accidental Intentional Meltdowns or Terrorist attacks malfunctions Nuclear power plant Chemical factories Hazardous material spills During transportation On-site accidents Changes in the traffic conditions
Motivation Transportation issues under disaster conditions Pre-disaster During-disaster Post-disaster Administrative actions Data collection Infrastructure assessment Disaster scenario analysis Emergency actions preparedness Mitigation actions Disaster assessment Traffic network assessment Emergency response Evacuation preparation Evacuation deployment Immediate recovery actions Post-disaster traffic management Short-term recovery actions Long-term recovery actions Major traffic management problems Evacuation Emergency operations Post-disaster commuter traffic management Prioritization of recovery operations
Motivation Objectives 1. To develop models that account for disasterspecific problems 2. To analyze and obtain insights on disaster-related problems 3. To develop an integrated framework that simultaneously accounts for interrelated problems
Why an integrated approach? Simultaneous traffic problems Interactions and interdependencies among Overlapping traffic network totally or partially Varying priorities in response as disaster conditions evolve Possible increase in overall efficiency
Selected traffic management problems Evacuation massive operation take place before or after disaster determination of evacuation time determination of disaster region & demand human response to evacuation order availability of network information critical in case of post-disaster evacuations Emergency operations smaller in scale higher priorities for search&rescue accessibility and connectivity within in the disaster zone can be affected negatively from congestions due to evacuation Recovery Operations immediate actions may be needed during disaster to ease evacuation and emergency operations mostly take place after the disasters includes construction material or debris transportation may conflict with the post-disaster commuter traffic can cause further reduction in the available network capacity due to reserved routes Post-disaster commuter traffic may start right after the disaster gradually increase in volume eventually converge to pre-disaster volumes and patterns can require updating as the network and traveler characteristics change
Disaster problems and introduced models SCENARIO PLANNING DECISIONS Planning problem Pre - disaster Historical data NETWORK RE-DESIGN Logistics Infrastructure Design Technology Real-time data Disaster traffic management problem TRAFFIC ASSIGNMENT Disaster DATA WAREHOUSE Technology Immediate recovery actions Evacuation & Emergency Operations Real-time data Repair and reconstruction works Post - disaster * Post-disaster traffic assignment NETWORK RE-DESIGN Commuter Traffic management & Recovery Action Prioritization TRAFFIC ASSIGNMENT * after the rescue operations are ended
Disaster problems and introduced models (continued) Event Detection Event profile Populati on at risk Potential disaster? no Local management yes Response Disaster management Assessment of possible damage & risk transportation need Post-disaster actions Event monitoring & information gathering Safety measures Precautions warning Pre-disaster evacuation? Define shelters yes Immediate actions Evacuation preparations Disaster traffic assignment Divert other traffic flows Emergency operations preparations Immediate recovery actions Recovery Traffic monitoring & control Commuter traffic assignment Debris removal Reconstruction jobs Recovery actions Repair jobs
Proposed methodology Modeling issues Nature of the traffic problems demand, network and flow characteristics Simultaneous consideration of sub-problems evacuation with emergency flow post-disaster evacuation with repair prioritization commuter traffic with repair scheduling Evolving nature of disaster conditions disaster and collected information disaster response Network re-design and capacity reversibility Problem definition better utilization of the existing capacity by some operational or design changes without increasing the total network capacity Capacity reversibility (contra-flow) a solution for network re-design problem relaxation of network design constraints implementation criteria: no disconnection of the network parts no blocking traffic flow components increase in the overall efficiency Implementation issues for modeling network preparation traffic flow model selection modification
Proposed methodology i * I * i I (a) (b) (c) a) Two-way street segment in b) regular CTM representation c) suggested capacity distribution with reversibility
Example scenarios Hypothetical traffic network N Arterial Shelter Hurricane direction Highway Highway Arterial Side streets Traffic network Number of lanes per link Link speed limit (ft/s) Link saturation flow (veh/h/lane) Cell representations 3 100 2,160 2 50 1,800 1 50 1,800 Cell number 1-10 77-78,81-82, 85-86,89-90 11-76,79-80, 83-84, 87-88 Cell jam density (N it ) (veh/cell) Cell saturation flow (Q it ) (veh/10secs/cell) Cell δ it ratio 120 18 1 40 12 1 20 6 1
Example scenarios (continued) corresponding CTM network (Examples 1-3) 78 77 82 81 86 85 76 75 93 79 80 83 84 94 87 88 89 90 46 48 52 54 58 60 64 66 70 72 45 47 51 53 57 59 63 65 69 71 97 44 43 49 50 55 56 61 62 95 67 68 73 74 2 4 6 8 10 1 3 5 7 9 96 11 15 12 16 22 27 21 28 34 33 14 18 20 24 26 30 32 13 17 19 23 25 29 31 91 35 36 92 42 41 38 40 37 39 cells 91-95 origin points cells 96-97 destination points 96 end of disaster zone on the highway 97 shelter
Example scenarios (continued) Example 1: Hurricane evacuation Comparative results System travel time (vehicle-hours) Base Case : 8.52 Re-designed : 6.44 Reduction in the total travel time 24% The latest arrival at sink cell 96 (shelter) Base Case : t = 270 th second Re-designed : t = 170 th second at sink cell 97 (end of risk zone on the highway) Base Case : t = 270 th second Re-designed : t = 200 th second N Arterial Shelter Hurricane direction Highway
Example scenarios (continued) Example 1: Hurricane evacuation Comparative results At Sink Cell 96 200 Cumulative number of arriving vehicles 150 100 50 0 0 10 20 30 40 T(x10secs) (a) base case re-design At Sink Cell 97 Cumulative number of arriving vehicles 120 90 60 30 Figure 6. 0 0 10 20 30 40 T(x10secs) (b) base case re-design Distributions of arrival of evacuated vehicles (a) at the shelter, (b) at the highway exit
Example scenarios (continued) Example 2: Sensitivity analysis Comparative results System travel time (vehicle-hours) Base Case : 27.60 Re-designed : 18.44 Reduction in the total travel time 32% The latest arrival at sink cell 96 Base Case : t = 440 th second Re-designed : t = 460 th second at sink cell 97 Base Case : t = 240 th second Re-designed : t = 280 th secon
Example scenarios (continued) Example 2: Sensitivity analysis Comparative results At Sink Cell 96 Cumulative number of arriving vehicles 400 300 200 100 0 0 10 20 30 40 50 60 T(x10secs) (a) base case re-designed case At Sink Cell 97 Cumulative number of arriving vehicles 240 180 120 60 0 0 10 20 30 40 50 60 T(x10secs) (b) base case re-designed case Distributions of arrival of evacuated vehicles (a) at the shelter, (b) at the highway exit
Example scenarios (continued) Example 3: A demand management study Problem and mathematical model evacuation of the region before a pre-defined target time Models 2a-2b for base-case and re-designed case Origin-destination assignment and demand distribution the target evacuation time is t 340 +10 seconds for every O-D pair Comparative results System travel time (vehicle-hours) Re-designed : 18.44 The earliest departure at sink cell 96 Re-designed : t = 210 th second at sink cell 97 Re-designed : t = 170 th second
Example scenarios (continued) Example 3: A demand management study Comparative results 200 150 Sink Cell 96 Arrivals at sink cells Departures from source cell 91 Departures from source cell 92 Departures from source cell 93 Departures from source cell 94 Departures from source cell 95 Cumulative number of vehicles 100 50 0 15 20 25 30 35 40 T(x10secs) Cumulative number of vehicles 200 150 100 50 Sink Cell 97 (a) Distributions of optimal demand distribution curves from given origins to (a) the shelter, (b) eastbound exit point on the highway 0 15 25 35 45 T(x10secs) (b)
Example scenarios (continued) Example 4: Evacuation and emergency operations Problem and mathematical model evacuation of the region possible emergency demand between selected points simultaneous consideration with equal weights Models 1a-1b for base-case and re-designed case Origin-destination assignment and demand distribution evacuation demand same as in Example 1 additional emergency O-D pairs with unit flow during t = [50,240] Cells 98-99 emergency origin points Cell 100 emergency destination point corresponding CTM network (Example 4) 78 77 82 81 86 85 76 75 93 79 80 83 84 94 87 88 89 90 46 48 52 54 58 60 64 66 70 72 98 45 47 51 53 57 59 63 65 69 71 97 44 43 49 50 55 56 61 62 95 67 68 73 74 2 4 6 8 10 99 1 3 5 7 9 96 11 15 12 16 22 27 21 28 34 33 14 18 20 24 26 30 32 13 17 19 23 25 29 31 91 35 36 92 1 0 0 42 41 38 40 37 39
Example scenarios (continued) Example 4: Evacuation and emergency operations Comparative results System travel time (hours) Base Case : 9.11 (evacuation flows : 8.55 emergency flows : 0.56) Re-designed : 7.07 (evacuation flows : 6.52 emergency flows : 0.56) Reduction in the total travel time: 22 % Reduction in the total evacuation travel time: 24 % The latest arrival (without emergency flows) at sink cell 96 Base Case : t = 270 th second (t = 270 th second) Re-designed : t = 200 th second (t = 170 th second) at sink cell 97 Base Case : t = 310 th second (t = 270 th second) Re-designed : t = 200 th second (t = 200 th second)
What is next?
1. Developing evacuation models DTA models for evacuation with capacity reversibility DTA models for evacuation and emergency operations Network connectivity requirements for developed evacuation models 2. Alternative approaches Analytical evacuation models with capacity reversibility Simulation-based models for large-scale networks Contact : Hediye Tuydes h-tuydes@northwestern.edu