330 likes | 534 Views
Evacuation Demand. CE 4745 – Natural Hazards and the Built Environment Spring 2004. Why Do We Want to Estimate Evacuation Demand?. To be able to “recreate” (or mimic) evacuation travel under alternative scenarios. With this ability we can: Estimate impact of different storm scenarios
E N D
Evacuation Demand CE 4745 – Natural Hazards and the Built Environment Spring 2004
Why Do We Want to Estimate Evacuation Demand? • To be able to “recreate” (or mimic) evacuation travel under alternative scenarios. • With this ability we can: • Estimate impact of different storm scenarios • Test alternative policies and strategies • Identify optimum contingency plans
Sample setting for development of policies and strategies road d1 d2 d3 t1 t3 d t2 Zone 1 Zone 2 Zone 3 The load on the road network is dependent on the dynamic loading rates at each zone, the relative timing (sequencing) of the loading among zones, and the relative location of the zones. dd4
Examples of policies • Reverse laning • Where? • When to initiate and close? • Evacuation orders: • Type • Timing • Coordination with others
Example of Strategies • Phased evacuation • Dynamic routing • Suppression of shadow evacuation through effective public announcements • Dynamic information systems • Development of contingency plans
Factors Motivating Evacuation • 1. Risk of flooding: • High risk – elevation < 10 foot above sea level • Moderate risk – elevation 10-15 feet above sea level • Low risk – elevation > 15 feet above sea level • Evacuation rates in high risk areas are often 3 times those in low risk areas. • People in low risk areas may not need to evacuate at all – those that do are shadow evacuees.
Factors Motivating Evacuation • 2. Evacuation Orders: • Precautionary or voluntary evacuation order • Recommended evacuation • Mandatory evacuation • Dependent on means of dissemination • Of those who hear a mandatory evacuation order, over 80% have evacuated in the past. • Of those who do not hear, less than 20% have evacuated in the past
Factors Motivating Evacuation • 3. Housing: • Mobile home dwellers are more likely to evacuate than persons in other home types. • People in high-rise buildings are less likely to evacuate than those in regular houses, all else being equal.
Factors Motivating Evacuation • 4. Storm Threat Information: • The National Hurricane Center issues storm advisories (storm watches and storm warnings). • Storm watches are issued when a storm is expected to make landfall within 36 hours. • Storm warnings are issued when a storm is expected to make landfall within 24 hours.
Factors Motivating Evacuation • 5. Storm severity: • High correlation with evacuation orders and flooding. • Few studies have been conducted following weak storms, so information on low storm severity is sparse.
Factors Influencing Decision to not Evacuate • Protect property from storm • Protect property from looters • Fulfill obligation to employer • Sometimes, peer pressure from neighbors • < 5% said they did not have transportation
Historical Development • Three-mile Island nuclear accident (threatened meltdown) in 1979 introduced interest in modeling evacuation. • Interest spread to other events such as chemical spills, hurricanes, and wildfires. • Current interest is in security of transportation infrastructure and evacuation from the aftermath of terrorist attacks.
Existing Hurricane Evacuation Models Simulation models Analytical models NETVAC (MIT, 1981) UTPP (PBS&J, 1985) DYNEV (KLD, 1982) DTA (Janson, 1985) MASSVAC (VP, 1985) ETIS (PBS&J, 2000) HURREVAC (COE, 1994) OREMS (ORNL, 1999) TransModeler (Caliper, 2000)
Modeling the Decision to Evacuate • Existing models: Participation rate type • Category and speed of storm • Flooding potential • Tourist occupancy • Proportion of mobile homes Logistic regression type
Participation Rate Models • Cross-classification type models
Logistic regression model of Hurricane Andrew Evacuation (2)
Logistic regression model of Hurricane Andrew Evacuation (3)
Participation Rate Model of Hurricane Andrew (PBS&J model of S.W. Louisiana
Time of Departure • Response rates based on: Past evidence Stated intentions Functions chosen using professional judgment Estimates based on expected rate of diffusion of warning messages
Observed Mobilization Evacuation start time, Hurricane Andrew, 1992, Louisiana
Mobilization Start Times • Evacuation start times, Hurricane Andrew, 1992, Louisiana
Trip Distribution • Professional judgment based on past evacuation patterns: • Default dispersion factors for each county or evacuation zone • Spreadsheet-based model • Spatial interaction model such as the Gravity model
Trip Distribution • Common factors determining destination: • Relatives and friends (50-70%) • Hotels/motels (15-25%) • Public shelters (5-15%)
Trip Assignment • Route selection paradigms: • Myopic behavior • User or System Optimal behavior • Combined myopic and imposed behavior • Imposed behavior according to evacuation plan
Trip Assignment • Common methods: • Microsimulation • Static User Equilibrium • Emerging methods • Dynamic traffic assignment
Crucial areas for research • Spatial and temporal data: • Route choice • Destination • Departure time • Clearance time • Volumes and speeds • Real-time data • Dynamic traffic assignment • Large networks