Skip to content United States Department of Transportation - Federal Highway Administration Go to FHWA website Feedback

Managing Pedestrians During Evacuation of Metropolitan Areas > Table of Contents

2  Literature and Case Study Review

2.1    Methods for Identifying Literature

The research applied several approaches to find examples of evacuations and written reports outlining evacuation characteristics in urban settings, in particular.  A professional librarian conducted a conventional literature search to find written items relating to urban evacuation of pedestrians and efficient movement of emergency vehicles into a disaster area, augmenting existing resources available to the researchers.  Several different computerized databases were searched to determine the state of the literature, including newspapers (Dialog and ProQuest databases), transportation journals (TRIS database), and psychological journals (PsycINFO).   Researchers also sought open-source “after-action reports” from a few major incidents, drawing upon the Lessons Learned Information Sharing (LLIS) site maintained by the Department of Homeland Security (DHS) as well as other sites.  In addition to these resources, the FHWA provided draft reports of several case studies of major disaster or emergency events in metropolitan areas, collected from the perspective of their effects on transportation operations.  While researchers found that the search term “pedestrian evacuation” yielded mainly studies of the evacuations from buildings, planes, ships, and trains, this literature for the most part is not addressed here. 

2.2    Major Findings from the Literature Review

The purpose of the literature review was to examine what is already known about pedestrian evacuation in U.S. urban areas.  In short, the research found that there is little mention of pedestrians in any of the literature specifically on emergencies and disasters. It appears to be considered axiomatic that people evacuate in private cars.  There is a growing literature based on research on evacuation from large buildings, but a dearth of literature that addresses what happens to building evacuees once they reach street level.  This very limited record of people evacuating on foot may be either an artifact of the literature encountered or an actual characteristic of most emergency evacuations in the United States.  With attention called to pedestrians in at least two recent urban emergency situations, future research is likely to examine the issue more closely.  

2.3    Sources of Information on Evacuation

2.3.1   Newspaper and Internet Sources

The news article database, Dialog, covers 40 major newspapers. Starting with 1994 very few newspaper articles were found that were relevant to the primary topic of interest—major “no-notice" evacuations in urban areas.  Newspaper articles on evacuation typically focused on hurricane and hazardous material evacuations, which either provided adequate warning to evacuation officials or only affected a very small, often rural, area.  Several articles highlighted the need for effective evacuation planning, citing the political and logistical difficulties experienced when creating an evacuation plan.2  One newspaper feature article described significant crowd disasters at nightclubs, stadiums, and arenas.3  Club and arena disasters generally occur when a large number of people rush to escape an enclosed facility and crush persons against the exit, or via crowd stampedes.  The few recommendations mentioned are limited to enforcement of existing occupancy safety laws.  An extensive set of brief descriptions of disastrous crowd stampedes in foreign countries was found on the website of a firm that provides consulting and facility design for events involving large numbers of participants.4 

2.3.2   Transportation Research on Evacuation Planning

A search for relevant literature in the transportation database identified examinations of evacuation planning and of new methods for designing evacuation routes.  The evacuation planning literature in this database generally was concerned with evacuation plans for a large area when either facing a foreseeable natural disaster, such as a hurricane, flood, or volcanic eruption5, or following a destructive seismic event or hazardous materials release6.  The majority of the articles are concerned with evacuation of a large area via passenger vehicles, so the recommendations are not directly relevant to pedestrian evacuation.  However, the basic principles are similar as far as providing smooth channels for efficient egress from a disaster area.  For instance, evacuation management issues related to hurricane evacuations are relevant to other large scale evacuations considered in homeland security planning.7  First, efforts need to be made to reduce the occurrence of overreaction and panic to limit evacuation travel demand and therefore decrease the possibility of overcrowding an evacuation route.  Second, transportation officials need to maintain good communication and data exchange to ensure smooth emergency management.  This can include law enforcement and other emergency services, as well as access to ITS technologies and local radio stations.

Another area of literature reviewed focused on the use of emerging technologies for evacuation plan development.  Technologies include geographical information systems (GIS), database management systems (DBMS), census and other descriptive population data, and dynamic route planning.8  GIS is highlighted as the essential tool for efficient evacuation route planning, especially when linked to population and land use data.  For example, spatiotemporal land usage data – information on where the population of a given area is typically located during different times of the day – can be very useful when creating dynamic evacuation routes for hazardous material releases and other localized disasters.9  GIS has also been used to characterize the road network and identify bottlenecks and other sources of flow problems to determine optimal traffic planning during passenger vehicle-based evacuations.10

2.3.3   Psychological Research

Physics models have been used by psychologists to characterize pedestrian flow of sets of individuals exiting a room, moving through corridors, and moving around obstructions in both normal and panic situations.11  Such modeling is used to characterize mass behavior of pedestrians, including reduced attention, ignoring side exits, and “herding.”  This modeling indicates that in constricted areas “faster is slower” – that is, as desired velocity increases, actual flow velocity typically decreases due to increased forces as pedestrians push toward an exit.  Crushing fatalities occur during room egress due to increased forces as the crowd pushes toward the exit.  The model shows that placing a barrier close to the exit can distribute the equilibria of these forces and serve to increase exit efficiency.  Similarly, placing columns along a walkway can effectively segregate flow into “lanes.”  Finally, the model illustrates that a temporary widening in a narrow corridor can actually reduce flow velocity with high pedestrian density due to bottlenecking where the corridor returns to it normal width.  The observations from these studies have relevance for the design of pedestrian facilities in cities that could contribute to safer pedestrian evacuation.

Several articles examined human behavior under the stress of emergencies.12  These suggest four cognitive stages that people experience when responding to a stressful emergency event: 1) defining the situation; 2) devising a plan; 3) carrying out the plan; and, 4) looking back to assess whether the action solved the problem.  People first attempt to gather available information about the situation to help them decide what action to take.  This information includes both the present environmental status as well as referring to their past experiences in dealing with emergencies.  Environmental information ambiguity is common in emergency situations.  Typical reactions of people when information is ambiguous include: ignoring the situation; looking for information by asking questions; or moving toward the danger area to investigate.  These reactions consume precious time and can greatly affect the outcome of the evacuation. 

Social scientists who study disaster behavior hold that it is a very common misperception that people confronted with a threatening situation will engage in irrational and non-functional behavior often referred to as “panic flight.”  Much of what might appear as random and non-functional behavior to an outside observer of a group of people when they become alert to a danger is actually highly rational because it is related to first ascertaining the status of other family members or friends before engaging in appropriate evacuation behavior.  Numerous researchers and reviewers of research conclude that irrational panic behavior on the part of the public when warned of an imminent danger is not something emergency managers will need to contend with.13  Nonetheless, the specter of panic makes for dramatic movies and high newspaper readership, so it continues to be perpetuated by the media.14  Research findings do contain evidence that when community officials and emergency managers believe people will flee in panic and be hurt if alerted to a serious danger, they may withhold critical information that could help people avoid injury or death by enabling them to act more quickly to take appropriate protective actions.15 

2.3.4   Evacuation as Policy

During the early 1980’s the newly created Federal Emergency Management Agency was mandated to carry out a civil defense policy of crisis relocation for population protection in the event of a nuclear attack.16 Issues related to relocating the population of communities thought to be potential targets were examined.  One study examined more than twenty case studies of city evacuations in different countries and for various reasons, in order to address the efficacy of long-term evacuation of cities as a civil defense strategy.  Crisis relocation planning was viewed negatively by many communities, which objecting to the possibility of being designated as host communities. The policy was never fully implemented.17  However, FEMA fostered “all hazard” planning and temporary evacuation for population protection from natural hazards became more common. As pre-event evacuations became more common in the face of impending natural disasters, research was focused on evacuation warning response. 

2.3.5   Evacuation Decision Making

Research on decision-making suggests that people may only consider a few options, rather than considering all available information.  Under time pressure, people tend to give priority to familiar options and disregard unfamiliar options that are perceived as more likely to increase danger.  For example, emergency exits are often disregarded because building users have no experience in using them and therefore are unsure about where they lead.  This theory explains why people evacuating from buildings tend to use the main entrance as an exit rather than emergency exits.

The goal during evacuation is to keep stress low to enhance people’s performance in coping with the situation. Excessive stress reduces decision-making abilities and creates increased opportunity for irrational (panic) behavior.  Information reduces both confusion and time spent denying the situation.  Three information characteristics have an impact on interpretation of a situation: quality, quantity, and relevance.  People need to know specific information regarding the situation as well as the correct course of action.  Information should be provided as early as possible to get people moving toward safety.  Delays in warning may lead to situations that negate the ability for people to make good decisions and thus fatalities occur.

An analysis of the 1987 King’s Cross Underground (London subway) station evacuation tragedy provided several important observations about how to improve evacuation success including the effectiveness of different types of emergency evacuation instructions.18  For example, with respect to type of information provided to evacuees from the subway station the research indicated that announcements with specific information and instructions were more effective than announcements consisting of instructions or personnel simply directing evacuees toward exits.  The study also found that evacuees are more likely to follow the instructions of uniformed officials (e.g., police and firemen) than subway workers, due to their perceptions of the authority and confidence of the policemen.  In a similar study, researchers found that evacuations were faster and more effective when evacuation officials led one or two evacuees from the situation than if they simply shouted instructions and used gestures toward the exits.19  Thus, it is critical that officials assisting evacuees have good information about the situation and safe evacuation routes, in order to provide appropriate instructions.

Several reviews of the large body of social science research on disasters include the findings on warning compliance and decision making of individuals in disaster evacuation. Important factors in an effective evacuation process include specific information about the source of the danger and urgency of evacuation, specific instructions regarding what to do, and the use of uniformed and authoritative personnel to give instructions. 20  Non-residents (e.g., tourists and business travelers) who find themselves faced with emergency evacuation will be unfamiliar with place names, routes, distances, and facilities, making them particularly dependent on hotels and public safety authorities for assistance.21 

Timely public evacuation is related to individuals defining themselves as being in danger and believing that leaving the area is appropriate.22  An examination of the factors related to how quickly people begin to evacuate, found that the steepness of the mobilization curve (i.e., what proportion leave within how many minutes) is related to the perceived severity and timing of the impending threat.23  This suggests that, for events that have occurred with no warning, the perceived severity of the threat may determine how many people are evacuating simultaneously.

Researchers, especially those who have studied public response to warning and evacuations instructions, acknowledge that there is not as yet direct evidence as to what might be the public response in the United States in the face of an incident created through the malicious use of an agent in the category of weapons of mass destruction (WMD).  Two examples were found of scholars who have made extrapolations from what is known about the behavior of the public in past disasters to what might reasonably be expected in the event of a terrorist incident involving some chemical, biological, or radiological weapon.  Both are optimistic that what is currently known about the reactions of the public and the need for interagency coordination for complex responses during disasters are good predictors of behavior in the face of not-yet-experienced disaster agents.24  They add the caveat that aggressive public education to create public familiarity with new hazards and appropriate responses is necessary.

2.3.6   Sheltering-in-Place

In situations of inadequate time for a very large population to clear an area in the face of, for example, an airborne release of a dangerous substance, “sheltering-in-place” can be employed as a protective action.  This action should only be used if adequate protection is possible.  However, having people sheltering in place, that is, staying where they are once the period of danger is in effect, reduces the numbers of persons who become part of an evacuation stream or who have to be provided public transportation.  Even though some surveys have found fairly large proportions of people saying they would be willing to shelter in place, or reporting that they will shelter in place “the next time” to avoid the traffic congestion and hassle of evacuation, there is little research to indicate what proportion actually will shelter rather than leave.25  This means that the proportion of the population at risk that will stay or go cannot be easily predicted.  Activities related to managing evacuation need to be maintained, even if the protective action of sheltering-in-place is recommended or ordered. 

2.3.7   Modeling Pedestrian Crowd Behavior

Pedestrian behavior has been the subject of empirical studies since the 1960s. The major goal of most of these studies was to provide information to develop design guidelines for pedestrian facilities (such as sidewalks) and to aid in the design and construction of facilities that are safe, efficient, and pleasant to use. These studies have used manual logging of observations, video cameras, time lapse photography, and ultraviolet/infrared sensors to study the way pedestrians normally behave.26  Other studies have focused on the evacuation of buildings and other structures (e.g., ships).  The proceedings from the first International Conference on Pedestrian and Evacuation Dynamics, held in 2001,27 and subsequent conferences in 2003 and 2005 include presentations from across several disciplines, including architecture, civil, naval and fire safety engineering, physics, computer science, and mathematics.  None of the presentations address issues of crowd dynamics that occur when people evacuating a large building become part of the sidewalk and roadway traffic congestion

Crowd modeling and crowd management technology are used increasingly by consultants to address practical issues regarding pedestrian planning and design in terms of both crowd safety and emergency evacuation. Specific pedestrian flow models have been developed to simulate the flow of large groups of pedestrians, often those associated with large-scale evacuations from large buildings/offices/sports stadiums, or conveyances like airplanes and ships or evacuation strategies for multi-building sites.28  Highly sophisticated modeling approaches such as these may be appropriate for planning for evacuation in areas with very high pedestrian densities likely to be created as workers evacuate from large office buildings. 

2.4    Case Studies of Transportation Operations in No-Notice Events

The perspective of the transportation agencies is represented in the series of case study reports sponsored by the DOT/FHWA, Office of Operations, titled The Effects of Catastrophic Events on Transportation System Management and Operations. The draft case studies were focused on the types and timelines of transportation disruptions that occurred in conjunction with several catastrophic events and how they were resolved.  The two studies in New York City contain some observations of the phenomenon of large numbers of people leaving the city on foot in each of the events. This was treated more as one of the disruptions to transportation than as a focus for analysis.  The overall set of case studies, many still in draft when reviewed for this study, includes the following:

Revised versions of the reports on the Northeast Blackout, one for the Great Lakes Region and one for New York City, have since been released, as well as a companion report titled Comparative Analysis which considers all six cases.29

The most information on pedestrian behavior was found in the reports describing the two major incidents in New York City—the 9/11/2001 attack on the World Trade Center and the Northeast Blackout in 2003.  These reports described instances of pedestrian evacuees encroaching on roadways or presenting a potential problem for emergency response vehicles. Transportation authorities were concerned about large numbers of pedestrians as well as vehicles using the same bridges, because the authorities did not consider this situation safe for pedestrians and also felt that the heavy congestion associated with both cars and pedestrians on these bridges might hinder the progress of emergency responders trying to get into the city over the bridges. Recommendations included approaches to keep pedestrians and vehicles separated. The two major approaches described are to:  assign certain roads and tunnels or bridges to only vehicles and others to pedestrians; and, assemble the pedestrians in a staging area near a traffic choke-point, where they can be put on buses to be taken across a bridge or through a tunnel as part of the vehicle traffic.

There were major situational differences between the 2001 attack on the World Trade Center and the regional blackout in New York City two years later. In the 2001 event, the main impact scene was a fixed location; the people streamed out of the affected office buildings during approximately two hours and went off in all directions.  For the most part, private vehicles were not available to evacuees from around the World Trade Center complex.  Thousands of people left Lower Manhattan on foot.  A large proportion of them were occupants of office buildings, so it is assumed their preferred destination was their home.  Many apparently walked several miles to reach their residences or other preferred destinations. For others, their evacuation was carried out in part by walking and in part by taking various forms of transportation. For example, many evacuees walked south to the ferry terminal to catch a ferry or other boat to cross the river, or walked north to an operating subway station or to staging areas for bus transportation that were eventually established.

The other event where thousands of people left the city on foot was associated with a large scale power outage that affected all or part of several northeastern states.  When power went out in New York City, it became difficult for most people to continue working. Tens of thousands of workers left their office buildings all across the city in a relatively short time. Since the cause of the situation was known and not viewed as life threatening, the major impetus for the surge of traffic and pedestrians away from the business areas was more akin to a homeward commute when the workday is over than an urgent need to leave a hazardous area.  A high percentage of the office building occupants probably had taken mass transit to work, but the total loss of power meant neither the subways nor commuter trains were operating. The alternatives were to walk, find a bus, or catch a ferry. The vehicle traffic was gridlocked for several hours because traffic signals were not operating and public safety officials were engaged in such activities as rescuing people from stalled elevators rather than directing traffic. The studies of both events made the same recommendations: plan for ways to keep pedestrians separate from cars at the major chokepoints (e.g., bridges), and establish and publicize transportation hubs where pedestrians can assemble in order to catch public transportation out of the city.

When the two New York City case studies are compared to the others, three observations stand out as related to situations associated with pedestrians as part of the evacuation mix. 

2.5    Emergency Response After-Action Reports

Several emergency response after-action reports were located for events that involved some type of evacuation. Best practice in emergency response calls for reports to be prepared following any exercise or response event, in order to identify where improvements in the response actions can be made. When information is handled as “lessons learned” and not for assigning blame and punishing mistakes, after-action reports can provide details about and insights into problems associated with an event, how the response was managed, and how adaptations were made or could have been made to the basic plan. However, after action reports typically do not find wide distribution because of concerns that the information will be misused.

After-action reports most typically are prepared by the emergency responders and focus mainly on activities at the scene of the impact, including fire fighting, search and rescue, medical care and access control.  These emergency response reports contain very little about what the non-responders were doing, with the exception of those who are badly injured and are thus an object of a first response function.  Transportation operations and evacuation modes outside the “hot zone” of the response activities are seldom addressed in these reports. Nonetheless, some ideas about what people did once they exited a building or area, for example, can be pieced together from the details about response.

After action reports from several major emergencies perpetrated by terrorist actions and with no warning were reviewed.  These are:  a vehicle bomb detonation in the basement under the World Trade Center Tower 1 and the Vista Hotel in New York City in 1993; the vehicle bomb explosion in front of the Oklahoma City federal building in 1995; the Sarin gas releases in the Tokyo subway in 1995, and the deliberate crash of an airplane into the Pentagon on September 11, 2001.  A report prepared by city agencies on the blackout response in New York City in 2003 also was reviewed. These reports provide insights into what it was like at the scene and to some extent, the actions of people evacuating from the impacted area. 

A search on the internet identified several accounts, mainly newspaper articles, about the bomb at the Olympic Games site in Atlanta and the recent Madrid, Spain, train bombing. However, the articles focused on likely perpetrators and the number of injured. No information was provided on whether or how the survivors left the area. The literature review had been prepared before the suicide bombs in the London Subway or the levee breaches around New Orleans during Hurricane Katrina in 2005.

2.5.1   New York City World Trade Center Bombing, 1993

This set of articles30 was prepared mainly by fire service people who had carried out different functions under the incident command, and was originally published in a fire service journal. The evacuation of the Vista Hotel and Towers 1 and 2 at the World Trade Center (WTC) involved an estimated 50,000 people (about 25,000 from each of the two towers), as well as several hundred from the 850-room hotel. At that time, it was characterized as the “largest building evacuation in history.”

The evacuation is described as a self-evacuation, because the tower occupants began to evacuate without being told, apparently as a reaction to the uncertainty created by the information about an explosion in the basement.  A few started, and more and more joined in. Early in the course of the response, the fire department’s assessment of the situation determined that such a building evacuation was unnecessary.  However, the responders did not think the evacuation could be stopped without confusing and frightening people. The fire department concentrated initially on getting the fire out in the garage, in order to limit the spread of the smoke through the buildings. Fire personnel who went up in the towers found people needing assistance and medical care, including women in labor. Several hundred people were trapped in the various elevators. According to the report, it was eleven hours before everyone was out of the towers. In contrast, the Vista Hotel management had ordered guests to evacuate immediately and then checked each floor to ensure it was clear. The 800-room hotel evacuation took about 10 minutes, aided by the fact the hotel was at its lowest occupancy period for the day.

The report indicates that many people who came down out of the hotel and towers stayed in the lower levels once they got there, relieved to be down and safe and not wanting to go any further. Officials had determined the danger was over and it was bitter cold outside.  It is likely that many of the hotel guests were unfamiliar with the area and transportation options. The presence of emergency response vehicles around the area of the explosion probably prohibited taxis from coming very close to the hotel and towers.

Some EMS teams performed triage in the large areas of the lobby and mezzanine even though the air was still smoky. They prepared the injured for transport as rapidly as they could. Others worked with the injured up in the towers.  The EMS teams tagged injured victims for transport to area hospitals by police car, school bus, or mutual aid ambulance. The ambulances of the NYC EMS teams generally remained parked at the scene while the EMS squads worked inside with patients. Hospital records indicated that just over 400 people had gone to hospitals for care, although not necessarily immediately upon leaving the scene.  Many of them had made their way home, the adrenalin had worn off, and then they had gone to the hospital because they started to feel bad. The hospital closest to the scene received about 200 self-referrals into the hospital over the first 12 hours. No information in any of the reports indicated how many came on foot.

The WTC complex has about 16 exits, so people were able to leave the area and disperse in several different directions.  Many people left their offices without their coats so were inappropriately dressed to spend an extended time outside the building in the freezing weather. Hundreds gathered in a large food court area across the street. The EMS teams conducted quick assessments at several exits and set up to provide treatment in the food court location.  According to one report, there were six fatalities and 1,042 injuries associated with the bombing and evacuation.  No information is given on the actions of the thousands of evacuees who did not stay at the scene.  No mention is made of the status of the subway service. 

The bombing and response activities were generally limited to a small area of the city immediately around the WTC. An indication of the limited scope of the event compared to the overall metropolitan context is the fact that the EMS received about 3,000 other calls for service across the city that same day, unrelated to the WTC bombing scene.

These observations come from the perspective of the emergency responders so they don’t extend to what was happening beyond the event scene.  The observations indicate that large numbers of people stayed in the vicinity at least for a time but don’t say anything about the impact on city traffic of the probably tens of thousands of others who left the area.  Notable observations for consideration in urban evacuation planning are the following:

2.5.2   Oklahoma City Murrah Federal Building Bombing, 1995

This after action report31 is related to the emergency response, and provides details on what happened to the building occupants specifically. The noise and smoke from the explosion triggered the self-dispatch of several fire units to the scene within minutes. Other responders were officially sent to the scene after about ten minutes, as the magnitude of the event became evident. In addition to the quick arrival of eleven fire units, there was also an immediate convergence onto the scene of workers from other downtown buildings who tried to help with the search and rescue activities.

The Murrah Building had about 650 occupants at the time of the blast. A major portion of the structure was destroyed and more than 100 occupants perished. The blast also caused damage to other buildings nearby, and injuries to people from flying glass from blown out windows and other debris. The report mentions only the building occupants leaving the scene who were injured; no mention is made of the non-injured and whether they were able to use their vehicles to leave the area. Mention is made of the transit system being brought into the response effort in order to evacuate the residents of a nearby apartment building, and efforts to maintain service along regular routes near the scene as well as deliver supplies and take exhausted responders back to their stations to rest. Evacuation of buildings in the downtown area that were not directly affected by the blast was at the discretion of the management; no mention is made of how many evacuees might have been involved.

Eleven fire units arrived at the scene within minutes, followed by another ten units in about ten minutes. The fire units followed established procedure in taking different routes to the scene, and took up positions around the building. The report does not indicate whether any of the fire units encountered traffic congestion. Many units reported coming across “walking-wounded” as they approached the area, and stopped to see if assistance was needed before proceeding to the scene. Since a hospital is located only six blocks from the scene, it is likely that these injured pedestrian evacuees were headed to it for treatment.

Emergency medical services (EMS) units also came upon the injured pedestrians. Some units set up triage sites well away from the Murrah building, somewhat distant from those with the worst injuries. Many injured stopped for treatment at the triage sites and were transported to various hospitals. About 200 injured persons were treated and transported to hospitals under the direction of EMS during the first day. Final reports put the number of injured treated at hospitals at about 425, and those treated by private physicians at around 200. Thus, many sought treatment at a hospital, sometimes several hours after the explosion.

Initial actions by law enforcement included establishment of a traffic perimeter to deflect non-emergency vehicle traffic away from the response scene. The highway patrol diverted traffic from certain exits of the Interstate, which most likely was still crowded with morning commuter traffic. No mention is made in the report of traffic congestion beyond the response perimeter or of large numbers of non-injured pedestrian evacuees. A mention is made of how well the incident command managed to keep the street area right around the scene from becoming congested, giving the responders room to work. This was attributed to the fact that many of those in the incident command had attended a course at the Emergency Management Institute (EMI) run by FEMA, and had received instruction about the need to control congestion at the immediate scene. The report also notes the large number of fire and EMS units that arrived to offer mutual aid, although not requested, and the difficulty the incident commander had keeping track of all of them.

The evacuation challenges of this event contrast sharply with those of the evacuation of the WTC towers and adjoining hotel in New York City noted earlier. The number of building occupants evacuating from the area was probably less than a thousand. A high proportion of the survivors were injured by the bomb blast, suffering from lacerations and glass shards in their eyes. The weather was pleasant. Mass transit was not likely the dominant mode of transportation for the building occupants. No information is provided on the availability of personal vehicles for use in the evacuation. However, it is known that many of the less severely injured left the scene on foot, but were picked up and transported under EMS direction to hospitals for further medical care.

Two notable observations from this report are the following:

2.5.3   Tokyo Subway Sarin Gas Attack, 1995

The Sarin gas attack was planned to exploit the confined nature of a crowded subway system. Thus, the transportation system itself was an element of the attack. The author of the report32 examined the response to the attack and identified several organizational characteristics of the emergency response system as causes inappropriate actions that were taken initially, and discusses what has been done to improve emergency response procedures for the transit system.

The attack provided one of the first opportunities for emergency responders to work in an environment in which a weapon of mass destruction (WMD), in this case a nerve gas, had been released. In the worst case, the nerve gas release could have resulted in many more casualties and the complete closure of a subway system that carries millions of passengers every day. Since the nerve gas was released at multiple locations throughout the subway system, a need for emergency response arose quickly at or near the release locations as people became sick. The potential first responders, the transit workers on the subway, viewed the receptacles and leaking gas as just another mess to be cleaned up. The subway trains continued to run, and people continued to board the train at each station even though something was apparently wrong when  a few people were noticed staggering out of the subway train at various stations and making their way with obvious difficulty to the street level. More than an hour had passed before the calls about sick people from fifteen or more subway stations were recognized as related and as cause for emergency response. When the emergency responders arrived at the various subway stations, they rushed down to see what was going on. They did not understand what was happening, and they had neither the training nor the equipment necessary to deal with an agent like Sarin gas.

The report suggests that large numbers of subway riders left the train and went up to the street level following the gas release. The sick staggered about or lay in the street, where other passengers and by-standers tried to help, and eventually medical care and transportation was provided for several hundreds of victims. The authorities probably knew within two hours that the agent was Sarin, but withheld the information from the public. Later that same day the subway system was restored to full operations. People who had witnessed the scene at a subway station were confused and frightened, and thousands sought medical care throughout the next few days. The report did not indicate that there was any evacuation with the exception of people leaving the trains for the street.

One observation from this report relevant to pedestrian evacuation is the following:

2.5.4   Pentagon Attack, September 11, 2001

This lengthy after-action report33 provides more than two hundred recommendations for ways to improve emergency response to catastrophic events in Arlington County. However, it is like the other reports in that it contains sparse details on the process of evacuation of the building occupants. The Pentagon complex covers 29 acres and employs a workforce of about 23,000. The complex includes an underground station of the Washington Metro system.

Almost immediately after the plane struck the west side of the Pentagon, the Defense Protective Services for the Pentagon called for a full evacuation and assisted people to exit out the north side of the building. The evacuation of the building was accomplished in about an hour.  About 20,000 people left the scene within the first 90 minutes or so after impact.  The report notes that the parking lots emptied of cars as the evacuated employees who had driven to work left the area.. A few hundred people stayed to assist with the rescue and treatment of other employees. Casualties included 139 killed by the impact and about 100 injured who were treated and transferred to hospitals. Within two hours EMS activities could be re-directed to other areas.

A contributing factor to the extraordinary traffic congestion reported in and around the county was the release of government employees from work as information became available about the attack on the World Trade Center in New York. The plane had not yet hit the Pentagon when a portion of the evacuees from the District started their evacuation toward Arlington County. By the time the Pentagon was hit, all roads, bridges, and highways in the area were well over capacity. For people evacuating the Pentagon in private cars, the only option was to join the crawling traffic. Had the attack on the Pentagon occurred at the same time as that in New York, the transportation and traffic managers undoubtedly would have made different decisions for the evacuation. For example, they might have directed the traffic leaving the District away from the routes most needed for emergency responders to get to the Pentagon and evacuated employees to leave in their cars. Or they might have simply closed the highway closest to the Pentagon.  As it was, the unanticipated surge of drivers leaving the District in reaction to the news about the WTC attack initially took any route they chose.  This eliminated options for selecting some routes for specific uses related to the situation at the Pentagon.

The report does not mention pedestrians specifically as part of the congestion. Some members of the response organizations were delayed two or three hours by the traffic as they attempted to report to the response scene. However, dozens of mutual aid units were able to reach the impact site at the Pentagon as the employees from the Pentagon were evacuating from that direction. Aid units encountered pedestrians in the Pentagon parking lot where they provided medical care to employees who hailed them. By 1:00 p.m., there were an estimated 3,000 persons at the scene, acting in response roles.

A characteristic of this event was the merging of two evacuation streams: one composed of government employees who were leaving work at their own discretion and the other composed of Pentagon employees leaving a facility that had sustained an attack. The subway could not be used in the evacuation because service at the Pentagon Station was cut off temporarily, for security reasons. The bus transfer point normally serving the Pentagon could not be used once the highway exits serving that area were temporarily closed, also for security reasons. Thus, anyone who had used mass transit to get to work had to find some other mode for the evacuation. Law enforcement personnel directing traffic for routing the Pentagon evacuees away from and emergency workers toward the scene had fewer options because of the surge of traffic already evacuating from the District.

2.5.5   Northeast Blackout, New York City, August 2003

This report34 was prepared at the request of the Mayor of New York City and addresses ways the city could be better prepared for an emergency like the widespread, long blackout that occurred in August 2003. The New York City Emergency Response Task Force members included representatives from the private and non-public sectors and City officials. The Task Force gathered information to determine how six broad areas of city operations, one of which was Transportation, were affected by the blackout. The power outage shut down subway and commuter rail service completely. As a result, the rest of the transportation system was overtaxed. Surface traffic congestion included vehicles in gridlocked intersections and pedestrians streaming onto bridges. The traffic on Manhattan Island, where about 2 million people may be found on any week day in the business district at the southern end, came to a complete stop. An important factor in the evacuation was the interest on the part of the businesses owners and workers in the city to reach a safe place before nightfall.

Since the blackout occurred in the early afternoon, the evacuation resulted in a major increase in traffic congestion earlier than a normal rush hour, but one that lasted three to four hours. The demand on the ferry system for transportation out of the city was about 140,000 more passengers than usual. There were long wait times and a lack of crowd control. On the roadways, the traffic congestion caused by drivers leaving the business districts created difficulties for essential employees coming in to deal with emergency problems. Dissemination of traffic information through the broadcast media, normally provided by two major, independent traffic monitoring centers, was severely hampered by the power outage. There was no central source for reporting and disseminating all transportation information during this large-scale out-flow of traffic across jurisdictional boundaries.

The Task Force made thirty-five recommendations for improving the city’s ability to respond during an emergency, with five recommendations related specifically to transportation. One recommendation was to link the Metropolitan Transportation Authority and the Port Authority of New Jersey to the City’s Joint Traffic Operation Center, to make all transportation information available in one command center. This integration would support a more regional approach and facilitate coordination during large-scale evacuations. Another recommendation was to conduct transportation planning to maximize the use of all available transportation, including school buses, ferries, private buses, taxis, liveries and commuter vans, in conjunction with designated transportation hubs for moving people out of the city or back to the city to report for work. The Task Force recommended that an inventory and assessment be made of all available waterborne resources, both public and private, that could be called upon to meet short term surges in demand for pedestrian transportation across the rivers.

The Task Force also recommended that the city develop an emergency transportation plan that would optimize the use of facilities and expedite high-priority trips. Such a plan would consider implementation of emergency roadway restrictions to facilitate emergency vehicle access inbound to Manhattan, lane reversals to facilitate evacuation, and dedication of certain facilities to pedestrians, mass transit vehicles, essential personnel, and critical deliveries.

This evacuation event is notable for its scope. The base population contributing to the number of evacuees was at least 2 million, a substantial proportion of whom worked in the city, but lived elsewhere. The circumstance triggering the evacuation was not in one or two fixed locations, but everywhere; it had a sudden onset and an uncertain end point. Many people who wanted to be somewhere else by nightfall began to evacuate shortly after the power went off. A substantial amount of capacity of the heavily used mass transit system was incapacitated, stranding an unusually large number of people without transportation. Even though the borough of Manhattan is relatively small in area, evacuation was hampered by the presence of physical barriers (water, bridges, tunnels) in all directions which created choke points for the evacuation traffic.

For most New Yorkers, who are familiar with the circumstances of traffic congestion and limited routes, this was more along the lines of a particularly difficult commute than an evacuation. There was no need for first responders who would treat people suffering from injury or shock, as there had been with the attack on the World Trade Centers two years previously. Still, the emergency response demonstrated the need for better transportation planning to facilitate future emergency evacuations, especially under such adverse conditions as the loss of appreciable capacity of the mass transit system.

2 Dixon, 2001; Rocco, et al., 2001; Saavedra, 2001; Eisler, Bayles, & Vergano, 2001; Van Bronkhorst, 1999.

3 Bayles & Hampton, 2003.

4 Crowd Dynamics, Ltd.

5 Pal, Graettinger, & Triche, 2003.

6 Petrucelli, 2003.

7 Wolshon, 2004.

8 Campos, da Silva, & Netto, 2000; Li & Wang, 2004; Pal, Graettinger, & Triche, 2003; Alam, & Goulias, 1999; Abkowitz, & Meyer, 1996.

9 Alam & Goulias, 1999.

10 Abkowitz, & Meyer, 1996.

11 Helbing, et al., 2002.

12 Lang & Lang, 1961; Drabek, 1969; Sugiman & Misumi, 1988; Sorensen, 1991; Proulx, 1993; Riad, Norris, & Ruback, 1999; Pacheco-Costello, Maguire, & Chang, 2002; Saloma, et al., 2003.

13 Quarantelli, 1981; Keating, 1982; Drabek, 1986; Zelinsky and Kosinski, 1991; Tierney, et al., 2001; Feinberg and Johnson, 2001; Fischer, 1998.

14 Mitchell, et al., 2000; Fischer, 1998; Walters, 1988.

15   Fischer, 1998; Scanlon, 1991.

16 Perry, 1982.

17 May and Williams, 1986, Chap. 8.

18 Proulx & Sime, 1991; Proulx, 1991; Proulx, 1993.

19 Sugiman & Misumi, 1988.

20 Drabek, 1986; Lindell & Perry, 1991; Lindell & Perry; 1992; Tierney, Lindell & Perry, 2001.

21 Drabek, 1995; 1996; 2000.

22 Fitzpatrick & Mileti, 1991.

23 Sorensen, 1991.

24 Fischer, 1998; Perry & Lindell, 2003. 

25 Lindell and Perry,1992:242-244.

26 Willis, et al., 2002; Helbing, et al., 2002.  Helbing, et al., (2002) cites 153 references to American and European pedestrian and evacuation studies.

27 Schreckenberg & Sharma, 2002.

28 E.g., Crowd Dynamics Limited.

29 These reports are available on the FHWA's web site for Emergency Transportation Operations (ETO).

30 United States Fire Administration. (n.d.) 

31 City of Oklahoma, 1996.

32 Pangi, R. 2002.

33 Titan Systems Corporation (n.d.)

34 New York City Emergency Response Task Force. 2003.

Previous  |  Table of Contents  Next

FHWA Home | Feedback | FHWA Privacy Policy

United States Department of Transportation - Federal Highway Administration