TRANSIT OPERATIONS UNDER EMERGENCY CONDITIONS AMAR S. CHAUKAR VIRGINIA P. SISIOPIKU, COMMITTEE CHAIR ROBERT W. PETERS WILBUR A. HITCHCOCK A THESIS

Transcription

1 TRANSIT OPERATIONS UNDER EMERGENCY CONDITIONS by AMAR S. CHAUKAR VIRGINIA P. SISIOPIKU, COMMITTEE CHAIR ROBERT W. PETERS WILBUR A. HITCHCOCK A THESIS Submitted to the graduate faculty of The University of Alabama at Birmingham, in partial fulfillment of the requirements for the degree of Master of Science BIRMINGHAM, ALABAMA 2007

2 TRANSIT OPERATIONS UNDER EMERGENCY CONDITIONS AMAR S. CHAUKAR CIVIL ENGINEERING ABSTRACT The events of September 11 th and the recent terrorist attacks on transit systems around the world have increased interest in protection and emergency preparedness for the transit systems. Emergency preparedness planning for transit systems is necessary to prevent distractions and effectively respond to major man-made or natural disasters. In this thesis a large-scale dynamic simulation/assignment model was developed and used to evaluate the impact of emergency conditions on transit operations in the Birmingham, AL, area. The Birmingham regional model was developed using the Visual Interactive System for Transport Algorithms. After the base model was tested and calibrated, a major traffic incident at a key network location was assumed and scenarios were developed to consider the impact of full and partial road closures and, traffic diversions using Variable Message Signs on traffic and transit operations and transit vehicle rerouting to assist evacuations. This study confirms the utility of traffic simulation modeling as a means for testing and evaluating emergency preparedness plans that serve the needs of various modes, including transit. The results also indicate that transit vehicles can serve the evacuation needs of captive users under emergency situations, thus contributing to the safe, efficient evacuation of people from affected areas. They can also assist the quick transportation of emergency responders to the affected site. The study findings are expected to assist re- ii

3 lated emergency agencies in better planning their responses and selecting implementation strategies with a greater potential for success. iii

4 ACKNOWLEDGMENTS I am thankful to Dr. Virginia P. Sisiopiku, my advisor at the University of Alabama at Birmingham and chair of my thesis committee, for her kindness, support, and invaluable guidance during my research. I would also like to thank her for providing me with financial assistance and the opportunity to do this research. I truly appreciate my thesis committee members, Dr. Robert W. Peters and Dr. Wilbur A. Hitchcock, for their advice, suggestions, and critical review of the draft of this thesis. I would like to thank Mr. Curtis Barrett for preparing the model and for his timely help in tackling technical problems throughout my research. This was invaluable in the completion of this research. I am also thankful to Mr. Abdul Muqeet Abro and Mr. Nataraj N. Khade for their help and assistance throughout the project. iv

5 TABLE OF CONTENTS Page ABSTRACT... ii ACKNOWLEDGMENTS... iv LIST OF TABLES... vii LIST OF FIGURES... viii LIST OF ABBREVIATIONS... x CHAPTER 1 INTRODUCTION Study Objective and Scope Background METHODOLOGY Study Approach Test Bed Selection and Incident Site VISTA APPROACH AND FEATURES Simulation Model Selection Considerations of Model Level of Detail VISTA Approach Model Development Master Roadway Network Birmingham Sub-Network, Schedule, and Bus Stops Signal Timing Plans for Selected Corridors Model Validation DESCRIPTION OF SCENARIOS Incident Management Case Study Scenario 1: Base Case no incident v

6 4.1.2 Scenario 2: Emergency Conditions full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision Scenario 3: Emergency Conditions full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, information about incident provided to all drivers Scenario 4: Emergency Conditions partial lane blockage (2 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision Scenario 5: Emergency Conditions full lane blockage, information provision via VMS before junction of I-65 and University Boulevard, user Response 50% Transit Emergency Response Scenarios RESULTS AND ASSESSMENTS Incident Management Case Study Scenario 1: Base Case no incident Scenario 2: Emergency Conditions full lane blockage, no information provision Scenario 3: Emergency Conditions full lane blockage, information provision to all users Scenario 4: Emergency Conditions two lane blockage, no information provision Scenario 5: Emergency Conditions full lane blockage, information Provision, VMS before junction of I-65 and university Boulevard, user response 50% Comparison of Incident Impacts on Transit Operations under Scenarios 1 through Transit Emergency Response Case Study Results for Emergency Response at Brookwood Village Mall Results for Emergency Response at Riverchase Galleria Results for Emergency Response at Wal-Mart Supercenter at Lakeshore Drive Results for Emergency Response at The Summit CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH Summary and Conclusions Recommendations for Future Research LIST OF REFERENCES vi

7 LIST OF TABLES Table Page 1 Summary of Incident Scenarios Incident Case Study Scenario 1 Base Case Incident Case Study Scenario 2 Emergency Conditions Full Lane Blockage No Information Provision Incident Case Study Scenario 3 Emergency Conditions Full Lane Blockage Information Provision to all Users Incident Case Study Scenario 4 Emergency Conditions Two Lane Blockage No Information Provision Incident Case Study Scenario 5 Emergency Conditions Full Lane Blockage Information Provision with VMS Bus Travel Time from Central Station to Brookwood Village Mall Comparison of Routes for Emergency Site at Brookwood Village Mall Bus Travel Time from Central Station to Riverchase Galleria Comparison of Routes for Emergency Site at Riverchase Galleria Bus Travel Time from Central Station to Wal-Mart Supercenter at Lakeshore Drive Comparison of Routes for Emergency Site at Wal-Mart Super center at Lakeshore Drive Bus Travel Time from Central Station to The Summit Comparison of Routes for Emergency Site at The Summit vii

8 LIST OF FIGURES Figure Page 1 Map of the Master Roadway Network The Incident Location Location of Riverchase Galleria Shopping Mall Location of The Summit Shopping Mall Location of Brookwood Village Mall Location of Wal-Mart Supercenter at Lakeshore Drive Master Roadway Network The Sub-Network of the Birmingham Regional Network Centerpoint Bus Route Highway 31 South Bus Route Graymont-Ensley Bus Route Hollywood-Brookwood Mall Bus Route Bessemer Bus Route Altadena Bus Route Observed Counts vs. Simulation Counts The VMS Location Alternative Route Using University Boulevard Average Travel Time of Buses for the 5 Incident Case Study Scenarios Optimum Path from Central Station to Brookwood Village Mall viii

9 LIST OF FIGURES (Continued) Figure Page 20 Optimum Path 1 from Central Station to Riverchase Galleria (Route ) Optimum Path 2 from Central Station to Riverchase Galleria (Route ) Optimum Path 3 from Central Station to Riverchase Galleria (Route ) Optimum Path from Central Station to Wal-Mart Supercenter at Lakeshore Drive. (Route ) Optimum Path from Central Station to The Summit Shopping Mall. (Route ) 49 ix

10 LIST OF ABBREVIATIONS ALDOT BJCTA DTA DUE GIS ITS MOE O-D RPCGB STD SO TAZs TRANPLAN VISTA VMS VMT CORSIM Alabama Department of Transportation Birmingham Jefferson County Transit Authority Dynamic Traffic Assignment Dynamic User Equilibrium Geographic Information Systems Intelligent Transportation Systems Measures of Effectiveness Origin-Destination Regional Planning Commission of Greater Birmingham Standard Deviation System Optimal Traffic Analysis Zones Transportation Planning Software Visual Interactive System for Transport Algorithms Variable Message Sign Vehicle Miles Traveled Corridor Simulation x

11 1 CHAPTER 1 INTRODUCTION The first few hours following any large-scale disaster present a complex organizational demand and a unique managerial problem. As the type of disaster changes from tornado or hurricane to toxic chemical spill or terrorist threat, so does the set of responding organizations and the specific tasks they confront. Despite obvious differences in the demand structure generated by the unique circumstances of large scale emergencies, the studies completed in the last two decades have validated a generalized approach. The concept of emergency management has emerged as a partial response to a longrecognized need for improved hazard and management including emergency responses. Transit systems can play an important role in case of emergencies, and preparedness plans should consider the availability of transit to assist in evacuation. Natural and man-made emergencies vary widely in size, location, cause, and effect and often threaten public safety, health, and welfare (1). Public transit agencies have a history of providing assistance during emergency situations and performing vital services, such as the evacuation of victims and the transportation of emergency personnel. In the aftermath of major disasters, public transit systems have often played a key role in maintaining the mobility of residents and assisting repair and recovery workers in accessing affected sites (2). For example, during the events of 9/11, timely decision making by transit operators and the rapid evacuation of transit facilities in the immediate vicinity of the World Trade Center avoided any casualties to transit riders as a result of the collapse

12 2 of the two towers. Hundreds of thousands of people were safely evacuated from lower Manhattan after tunnels were deemed safe for operation. Transit operators also brought emergency responders and emergency aid to the World Trade Center. In a similar fashion in Washington, D.C., the shutdown of the federal government following the strike on the Pentagon on 9/11 clogged roads, and the Metro system became the transportation mode of choice, evacuating several hundred thousand people from Washington and northern Virginia in a few hours. Emergency personnel arrived at the site by Metro and regional bus system buses, which stayed on the scene in support of the response and recovery (3). The experience in both New York and Washington indicated that existing evacuation plans had many weaknesses, however. This was also the case in New Orleans where, in response to Hurricane Katrina emergency plans relying on transit for the evacuation of residents, many of whom were completely dependent on transit, failed completely. Lessons learned from recent transit evacuation experiences indicate that preparedness planning, configuration and capacity of evacuation routes and a consideration of transit system capabilities are vital to the success of an evacuation strategy. Therefore, it becomes important to understand the ability of a transit system to accommodate the evacuation of people to or from critical locations in times of emergency through the study of available transit system assets and local conditions on a case-by-case basis. This study considers hypothetical incident and emergency scenarios at key locations in the Birmingham, AL, region, along with responses using the local bus system.

13 3 1.1 Study Objective and Scope This study focuses on the application of Dynamic Traffic Assignment (DTA) modeling to emergency management so as to analyze the potential impact of emergencies on existing transit operations due to emergencies and to assess the effectiveness of responses using transit. The impact of incidents of varying degrees of severity is evaluated in a study area. Intelligent Transportation Systems (ITS) technologies are used for the dissemination of information about the incidents, along with carefully designed candidate response plans. A sensitivity analysis of the potential impact of the response plans on transit operations under emergency conditions is then performed. An assessment of the capabilities of transit systems to respond to major emergencies at key locations is done using DTA simulation modeling to find out the best possible path for transit vehicles from the Central Station to the selected emergency sites. To satisfy the study objective, a regional transportation network model is developed and tested for the Birmingham, AL, metropolitan area using the Visual Interactive System for Transport Algorithms Platform (VISTA). More specifically, VISTA is used to analyze the impact of emergencies of various durations and assess the effectiveness of candidate emergency management plans on transit operations in the Birmingham area. Furthermore, relevant ITS technologies, such as alternative designs of Variable Message Signs (VMS) are implemented on the network to assess their impact. The model is also used to determine, the optimal route for transit buses from selected origins to particular destinations for emergency evacuation purposes. This study also showcases the capabilities of DTA modeling and the benefits associated with using the model to improve preparedness plans. This is important because

14 4 the emergency response is often not intuitive, and the implementation of emergency plans involving transit systems, without knowing their capabilities and limitations, might actually worsen the situation. This thesis is organized in six chapters, as follows: Chapter 1 discusses the scope and objectives of the research. Chapter 2 presents the study methodology and provides information on the study approach, the test bed used in the analysis, the data collection process, and the model validation. Chapter 3 describes the features of the simulation model used in the analysis, along with model requirements and functions. Chapter 4 offers a description of the incident and emergency scenarios considered in the study. Chapter 5 discusses the results obtained from the simulation runs. Chapter 6 presents the summary and conclusions drawn from the results, along with recommendations for future research. 1.2 Background Most incidents experienced in the transportation environment are handled by public transportation authorities according to established policies, plans, and procedures; however, because often transit systems lack experience with major emergencies and disasters, the potential benefits of an incident management organization are not always readily apparent. The standard operating procedures of public transportation systems are typically good enough to manage normal conditions and minor emergencies. During day-to-

15 5 day service, public transportation personnel handle minor incidents generally on their own without having to interact with public safety agencies. During a major incident, however, multiple agencies are involved in the response. Unfamiliar and unanticipated tasks arise. The normal flow of information may be interrupted, and unpredictable system activities may occur. Frequently, more equipment and personnel are required to stabilize the scene, and many of the materials needed may not be available locally. The knowledge and experience gained from the management of the incidents is very helpful for the transit agencies. Few categories of emergency events, however, require the involvement of multiple jurisdictions, functional agencies, and emergency responder disciplines. An effective and efficient coordination across a broad spectrum of organizations is the key to quick, successful responses to such emergencies. Communication among agencies, responders, and the public and the technology to facilitate it are critical parts of successful emergency management plans. Examples of such emergencies include the following: Natural disasters, such as floods, earthquakes, hurricanes, tornadoes, fires, droughts, and winter storms. Major accidents, such as chemical spills, industrial accidents, radiological or nuclear incidents, explosions, and utility outages. Civil or political incidents, like unrest or disorder resulting from riots, public demonstrations, and strikes. Designated special events, which may include major sporting events, festivals, the Olympics, and international summit conferences.

16 6 Terrorist or criminal incidents, including chemical, biological, radiological, or nuclear releases, more traditional acts involving explosives and armed assaults, and cyber threats or attacks. During such major emergencies, the capabilities of transportation agencies to mobilize resources are most affected by the decisions and directives of other agencies. Those agencies include law enforcement; fire and emergency medical services; local, regional, and state emergency planning agencies; and local and state government. For better management of emergencies, transit agencies are becoming more actively involved with their local communities in planning and preparing for emergencies. Planning agencies and local and state governments often already include public transportation agencies in their plans without fully understanding their capabilities and limitations and without sufficient provisions for coordination in an emergency. To make the best use of transit systems in emergency response, it is necessary to assess the capabilities and limitations of the transit system prior to the emergency and develop appropriate plans that can optimize the use of available resources. During major emergencies, public transportation systems can provide a number of functions and services that are identified in local emergency operations plans and detailed in transportation system plans and procedures. The functions may include (3): Emergency evacuation of citizens from affected areas. This is coordinated with other public safety agencies and local law enforcement; the local/regional/state emergency operations center; the state department of transportation; and local highway, bridge, and tunnel authorities.

17 7 Τransportation of citizens with disabilities and other citizens who may be unable to reach an evacuation staging area and are often dependent on the public transit system. Evacuation of schools and day-care centers. Also support for managing the reuniting of parents and children in the immediate aftermath of a major event. Evacuation of populations in hospitals, nursing homes, and other community and private facilities that are dependent on public transportation for evacuation. Transportation of emergency workers and responders to and from an emergency staging sites. Transportation of meals, goods, and supplies to emergency sites for victims, and emergency responders. Communications support for emergency responders (using hand-held and onboard vehicle radios, personal digital assistants, cell phones, transportation dispatch facilities, and transportation communications infrastructure). Identification of routes to support the safe transportation of emergency responders, public utilities and support personnel, and essential personnel to an incident site or staging area. Provision of vehicles and equipment to support emergency response and incident stabilization. Provision of public information on agency websites and use of public relations facilities and capabilities to improve public awareness. This study focuses on evaluating the impact of incidents on the transit operations in the network along with the effectiveness of the response actions using ITS technolo-

18 8 gies and a DTA simulation model. The study also used the DTA capabilities of VISTA to find the quickest route for the transportation of responders and emergency personnel to the emergency site from the Central Stationduring hypothetical emergency scenarios. The VISTA model can be refined and used in real time to support decisionmaking during the course of an actual emergency, although beyond the scope of this study.

19 9 CHAPTER 2 METHODOLOGY 2.1 Study Approach The overall approach of this study is to use the VISTA, a simulation based DTA model, to support decision making for emergency management. As DTA can capture the dynamic process of an incident and the emergency response, it is particularly appropriate for studying short-term planning applications, such as evaluating various incident management and emergency management programs through infrastructure improvements. In this study, VISTA is employed to assess the impact of designed incident scenarios on transit operations in the study network and evaluate the effectiveness of candidate incident management plans. In addition, for emergency response purposes, the DTA capabilities of VISTA were used to determine the optimum routes from the Central Station to the emergency sites. Furthermore, the implementation of ITS technologies to mitigate the effects of incidents and emergencies on the transportation network is evaluated. The research consisted of the following tasks: 1. Select the study test bed and identify the appropriate incident and emergency sites. 2. Select an appropriate simulation-based DTA model. 3. Identify data needs and data sources. 4. Define study boundaries and develop the study test bed. 5. Develop candidate incident and emergency scenarios and corresponding incident and emergency plans and select relevant ITS and control designs.

20 10 6. Perform the simulation, and analyze and summarize the results. 2.2 Test Bed Selection and Incident Site The Birmingham regional traffic network, which consists of the major freeways and arterial roads serving the Birmingham area was selected as the test bed for the case study, as shown in Figure 1. Figure 1 Map of the Master Roadway Network The I-65 freeway, an interstate highway of great importance to the mobility of Alabamians, is a major north-south facility in the Birmingham region. It is also a northsouth route of national significance. I-65 extends as far north as Lake Michigan and connects the city of Birmingham with Nashville (Tennessee) and Indianapolis (Indiana) to

21 11 the north. It also connects Birmingham with Montgomery and Mobile (Alabama) to the south. On the other hand, I-459 and I-20/59 facilities connect the city of Birmingham to the city of Tuscaloosa (Alabama) to the west and the cities of Anniston (Alabama) and Atlanta (Georgia) to the east. A loop is formed by I-459 and I-20/59 around the city of Birmingham and intersects with other important transportation facilities, including I-65, US-11, US-31, US-280, US-75, US-79, and US-78. These major transportation facilities are coded in the simulation test bed, along with selected arterials of major importance. The selected study network is very extensive, and the coding and development of a simulation model of this size and complexity involves significant effort. The northsouth section of I-65 modeled in this study extends from Warrior to Calera and is approximately 53 miles long. The west-east section of I-59 modeled extends from Rickey to Argo and is around 50 miles long. The incident site in this study is on northbound I-65 and is about 1,000 feet upstream of the junction of I-65 and I-20/59, as shown in Figure 2. This interchange has been the site of many crashes, including a fiery explosion of a gasoline tanker truck on Jan. 5, 2002, and a similar tanker crash on Oct. 25, Both accidents destroyed a section of the bridge overpass and required major reconstruction work. Moreover, this location is where two interstate freeways (I-65 and I-20/59) converge; its closure or decreased capacity would cause severe disruption to the mobility of passenger cars, as well as transit buses and commercial vehicles in the region and beyond.

22 12 Figure 2 The Incident Location For the emergency response case, four major shopping malls were selected as emergency sites, and a hypothetical terrorist attack on those malls was considered during those four scenarios. The optimum route for quick emergency response from Central Station to those four locations using transit buses as emergency vehicles is found using the DTA simulation in VISTA. This kind of emergency response could be especially helpful for the transportation of police, and other emergency response personnel. This could also be used to evacuate people who do not own automobile, including handicapped people and the senior citizens dependent on public transport. The emergency sites chosen are large, extremely popular shopping malls in the Birmingham area having huge popularity. As the main purpose of a terrorist attack is massive damage to life and economy, these

23 13 are more than likely targets for terrorist attacks. One of the emergency sites is Riverchase Galleria, a large super-regional shopping mall in the Birmingham suburb of Hoover. The center is not only Alabama s largest enclosed shopping center, but is one of the state s popular tourist attractions, second only to the Gulf Coast beaches. The mall is also very popular in the neighboring states of Florida, Georgia, Mississippi, and Tennessee. The other shopping malls considered in this study are The Summit, Brookwood Village Mall and a Wal-Mart Supercenter, are also very popular among the region and an attack on these malls could have significant impact on the region s life and economy. The locations of those malls are shown below in Figures 3 through 6. Figure 3 Location of Riverchase Galleria Shopping Mall

24 14 Figure 4 Location of The Summit Shopping Mall Figure 5 Location of Brookwood Village Mall

25 Figure 6 Location of Wal-Mart Supercenter at Lakeshore Drive 15

26 16 CHAPTER 3 VISTA APPROACH AND FEATURES 3.1 Simulation Model Selection Considerations of Model Level of Detail Simulation models are generally categorized as macroscopic, mesoscopic, or microscopic on the analysis detail they provide. Macroscopic models treat traffic at a high level of aggregation as traffic flows. The flows are often defined as differential equations, which are similar to those used to describe flows in fluids or gases. Microscopic models, on the other hand, describe traffic at a detailed level on a vehicle-by-vehicle basis. Mesoscopic models are an intermediate step between macroscopic and microscopic models. Such models typically describe traffic entities at a high level of detail but characterize their behaviors and interactions at a lower level of detail. Each model type has its strengths and weaknesses; therefore, deciding which type of model is most suitable usually depends on the type of application (4). In this study, the main required model capabilities included the simulation of incident and emergency conditions, the response of individual drivers to the incidents, the response of emergency vehicles, and the impact of the dissemination of incident information. Furthermore, the model needed to simulate networks that are large enough to take into account not only the direct effects of incidents and the pre-planned management strategies but also the indirectly impacted areas in order to evaluate the impact on transit operations in the network.

27 17 Although macroscopic models have the ability to model large networks efficiently, they typically lack the level of detail needed to model the individual driver s behavior when an incident occurs. In contrast, microscopic models can model incidents in great detail but require careful coding of all network details; therefore, microscopic models are subject to a high level of sensitivity to coding errors and have limitations related to computational ability and network size. Mesoscopic models fill in the gap by modeling the route choices and other important driver behaviors of individual drivers but limiting the level of detail when modeling driver interactions with the infrastructure and other drivers. Desirable features for incident and emergency management; a review of the model s approaches, capabilities, and limitations; and the availability of models and other resources led to the selection of VISTA as the simulation tool for this study VISTA Approach As mentioned earlier, the simulation model selected in this study was VISTA, the main features of which are presented next. Some of the properties that made this model particularly attractive for this study are discussed in further detail. Moreover, VISTA input and outputs handling, as well as validation issues, are addressed. In this study, the VISTA is used to model traffic and transit operations under emergencies. VISTA is a mesoscopic/microscopic model with dynamic traffic assignment capabilities developed at Northwestern University by Professor Ziliaskopoulos. The model integrates spatio-temporal data and models for a wide range of transportation applications, including planning, engineering, and operational ones. More specifically, VISTA uses a mesoscopic simulator called RouteSim and a DTA routine to emulate the

28 18 behavior of individual drivers and the ways in which they distribute themselves in the transportation network. RouteSim is based on an extension of Daganzo's cell transmission model introduced by Ziliaskopoulos and Lee (5). One important feature of VISTA is that it takes into account complex interactions between supply and demand in a transportation network and computes the spatiotemporal path for every vehicle while accounting for real-time driver behavior. This is a great advantage over traditional micro-simulation models such as CORSIM that do not track the movement of individual vehicles but instead split traffic at intersections. Moreover, it allows for the modeling of a variety of ITS options, a feature of great importance given the proliferation of such systems in the last two decades and their importance for incident and emergency management. Another important feature of VISTA for this study is that it can find the best possible route for a vehicle at a particular instant, depending upon the conditions present in the network. This is particularly important in the decisionmaking process as it determines the best emergency response routes on which transit buses can be used. The VISTA system can generate automated statistics per link, movement, or Origin Destination (O-D) path, as well as area-wide statistics. Furthermore, the system is flexible enough to allow the user to conduct parametric analyses by allowing only a percentage of vehicles to change their original paths. This is a useful feature in case of rerouting, a common strategy implemented as part of emergency management in order to minimize the impact of an incident on traffic operations. The principal characteristics of VISTA are as follows (6)(7):

29 19 The travelers behavior is modeled using a DTA model; A universal database model is used, based on a spatial Geographic Information System (GIS) that can be easily interface with other databases/models; An integrated meso-microscopic traffic simulator is used to propagate traffic at each iteration based on the current network loading; It converges to a Dynamic User Equilibrium (DUE) solution; It can model any size transportation network, and It is Internet-based, allowing access for the stakeholders to run the various algorithms, view the results of the models, query the database, and change the database based on each stakeholder s authorization level of each. 3.2 Model Development The data used in the development of the Birmingham model in the VISTA environment are as follows: Master roadway network and Birmingham roadway sub-network. Transit network, schedule, and bus stops. Signal location and timing plans for selected corridors. Automobile O-D trip data. Interstate and expressway detector counts. First, the Transportation Planning Software (TRANPLAN) regional planning model of Birmingham was acquired from the Regional Planning Commission of Greater Birmingham (RPCGB) that provided 2002 O-D trip data. Moreover, transit-related data were obtained from the Birmingham Jefferson County Transportation Authority

30 20 (BJCTA), including transit routes, schedules, and bus stops (8). In addition, detailed signal data were acquired from the Alabama Department of Transportation (ALDOT) and the City of Birmingham. Finally, recent interstate and expressway traffic counts were obtained from ALDOT to determine the variation of demand through observation of 24-hr demand profiles. The following paragraphs describe in detail the data and assumptions made in the development of the Birmingham regional model Master Roadway Network The master roadway network was extracted from the TRANPLAN base regional planning model and added to VISTA. The network includes major freeways, highways, and many important local arterials in Birmingham and surrounding areas. As an indication of the network size, it is noted that the north-south section of I-65 modeled in this study extends from Warrior, AL, to Calera, AL, and is approximately 53 miles long, and the west-east section of I-59 extends from Rickey, AL to Argo, AL and is about 50 miles long. The study network is very extensive, and the coding and development of a simulation model of this size and complexity is a major accomplishment. For selected bus routes of interest, missing nodes, links, and intersections were coded into the network to complete the bus routes. In addition, where signal data indicated the existence of signals, missing intersections were added to the network by creating a new node at the intersection, splitting the existing links at that node and adding cross-streets. This refinement improved the modeling detail along transit routes and thus enhanced the model s accuracy and realism. Figure 7 shows the master roadway network for Birmingham in VISTA

31 21 Figure 7 Master Roadway Network Birmingham Sub-Network, Schedule, and Bus Stops First, the entire master roadway network was run to capture the trips from far areas, as this research performed simulation analysis at a regional level. When the assignment gave reasonable results, a decision was made to extract a sub network for further experimentation in order to limit the software running time. The extracted subnetwork shown in Figure 8 was used thereafter for the simulation analysis.

32 22 Figure 8 The Sub-Network of the Birmingham Regional Network A geographically diverse set of bus routes was also modified, including the following six: 25 Centerpoint. Highway 31 South. 38 Graymont-Ensley. 42 Hollywood-Brookwood Mall. 45 Bessemer. 51 Altadena.

33 23 The BJCTA routes were coded in terms of the network links. Bus stops were added to the bus routes based on the actual BJCTA bus stop locations. In addition, schedules showing bus departure/arrival times at route start, end, and intermediate points were obtained and assigned as time points to observe the schedule adherence. As specific dwell time data were not available, a 15-second dwell time was assumed, in accordance with the Transit Capacity and Quality of Service Manual guidelines (9). The extracted network was considered for traffic demand from 3:00 p.m. to 9:00 p.m. which represented afternoon peak conditions. Figures 9 to 14 show the details of the bus routes in this study. Figure 9 25 Centerpoint Bus Route

34 24 Figure 10 Highway 31 South Bus Route Figure Graymont-Ensley Bus Route

35 25 Figure Hollywood-Brookwood Mall Bus Route Figure Bessemer Bus Route

36 26 Figure Altadena Bus Route Signal Timing Plans for Selected Corridors Most of the signal timing data on selected corridors were acquired from ALDOT and the City of Birmingham. Knowledge of offset times, cycle lengths, phase timing plans, and phase movements was needed to properly model signal operations (10) Model Validation Traffic counts and travel time data were used to validate the model for local conditions. In this study, hourly volumes for interstates and other major facilities were obtained from the ALDOT (11) and the City of Birmingham over six-hour periods (i.e., 3:00 p.m. to 9:00 p.m.), and were compared with the simulation results generated by the base case. For comparison purposes, the observed and simulated results were plotted on the x-axis and y-axis, respectively, and an ideal reference line was drawn diagonally, as

37 27 shown in Figure 15. The graph shows that further calibration of O-D demand is desirable to provide a closer match with traffic counts. Figure 15 Observed Counts vs. Simulation Counts For further model verification, limited field travel time data were also collected by driving through selected routes. The observed travel time data were then compared with the simulation results generated by VISTA in the base case to ensure compatibility. The relative percentage difference (i.e., [(Estimated - Observed) * 100/Observed]) for both the volume and travel time was to be found within ±20%.

38 28 CHAPTER 4 Description of Scenarios Both incident scenarios which may affect transit operations, and emergency scenarios, in which transit vehicles are involved in evacuations, were considered and tested. First, five incident scenarios were developed in this study to represent incidents such as crashes and severe infrastructure failure, and to test corresponding candidate response plans. The objective of those scenarios was to evaluate the impact of incidents of varying severity on transit operations in the study network, as well as the effectiveness of emergency response actions. A number of scenarios were designed to assess the effectiveness of transit in emergency response in a hypothetical situation. The number of lanes closed and the duration of the lane closure reflected the severity of the incident. Response actions included the strategic closing of selected on-ramps upstream of the incident site, resulting in vehicle diversion away from the affected site. During the response, information on the incident occurrence and diversion routes was conveyed by using ITS technologies, such as VMS. In practice, VMS are located at key points on highways and connecting routes to inform travelers about traffic conditions and provide recommendations on appropriate actions. Information conveyed through a VMS message can either be prescriptive, with recommended routes, or descriptive, with travel times/delay information on certain links or paths. During the simulation, whenever a vehicle passes over a link that has a VMS message, a driver may respond to the VMS message with certain probabilities.

39 29 In this study, a VMS was used to convey emergency events information to drivers as part of the incident case study. The VMS was placed on northbound I-65, approximately 1,000 feet upstream of University Boulevard (Figure 16) and advised drivers to follow University Blvd. as an alternate route to avoid the incident location. Simulations were performed to evaluate the impact of employing VMS on transit operations. An extensive review of previous incident management studies and the author s judgment led to the assumption of a VMS compliance rate of 50%. Figure 16 The VMS Location The information conveyed by VMS in this study is prescriptive with recommended routes. VISTA allows the user to specify a compliance rate, which reflects the probability that drivers will follow the VMS recommendation.

40 Incident Management Case Study As part of the incident case study that assessed the impact of an incident on transit operation, the following five scenarios were considered: Scenario 1: Base Case No incident. Scenario 2: Incident Full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision. Scenario 3: Incident Full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, information about incident provided to all drivers. Scenario 4: Incident Partial lane blockage (2 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision. Scenario 5: Incident Full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, information provision to drivers who travel northbound on I-65 via VMS. The VMS is located 1,000 feet upstream of junction I-65 and University Blvd. and advises drivers to follow University Blvd as an alternate route to avoid the incident location. The user response is assumed to be 50% Scenario 1: Base Case no incident This scenario describes network operations under normal (no incident) conditions and provides the baseline for comparisons. In this study, a peak demand period lasting from 3:00 p.m. to 9:00 p.m. was chosen as the analysis period. In order to establish the base case for this period, a set of equilibrium travel times and the planning O-D matrix for the period were obtained.

41 Scenario 2: Emergency Conditions full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision This scenario was designed for the purpose use in a sensitivity analysis by varying the incident duration in 30-min increments under a full four-lane blockage condition and observing the relative changes in model response. The objective of the sensitivity analysis was to illustrate the impact of the emergency on transit bus travel times in the network. The scenario assumed the occurrence of a major traffic incident on northbound I- 65 about 1,000 feet upstream of the junction of I-65 and I-20/59. The model was run with a full four-lane blockage on link 5994 starting at 4:00 p.m. and lasting for 30 min, 60 min, 90 min, and 120 min. One on-ramp upstream of the incident on I-65 was also closed. No information about the incident was provided to drivers, and it was expected that users would follow their habitual paths. Testing and analysis of the model outputs from this scenario are beneficial since such procedures are very useful for studying and designing response plans in a timely, and cost-effective manner Scenario 3: Emergency Condition full lane blockage (all 4 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, information about incident provided to all drivers This scenario was designed for use in a sensitivity analysis by varying the incident duration in 30-min increments under a full four-lane blockage condition and observing the relative changes in travel times of the buses. Similar to Scenario 2, Scenario 3 assumes the occurrence of an emergency on northbound I-65 about 1,000 feet upstream of

42 32 the junction of I-65 and I-20/59. While all other assumptions remain the same, in Scenario 3 all users are informed about the incident location, thus they can optimize their travel time in response to this information available. It is noted, though, that in practice it is expected that only a percentage of travelers will have knowledge about the incident and the best alternative route to choose such that they will benefit from it Scenario 4: Emergency Conditions partial lane blockage (2 lanes), incident duration 30 min, 60 min, 90 min, and 120 min, no information provision This scenario is similar to Scenario 2, except that a less severe emergency is considered, resulting in blockage of two traffic lanes, instead of four. Moreover, no on-ramp closures due to the emergency are considered this time. All other emergency conditions remain the same. The objective of this scenario was to study the impact of the accident on traffic conditions and operations under partial lane closure on the basis of selected Measure of Effectiveness (i.e., travel time) Scenario 5: Emergency Conditions full lane blockage, information provision via VMS before junction of I-65 and University Boulevard, user response 50% The hypothetical emergency conditions in this case are identical to those of Scenario 2, with a full four-lane blockage of varying duration starting at 4:00 p.m. On the other hand, in this scenario information was available to drivers, and a response action was taken by 50% of them. More specifically, in response to the emergency, the travel times of the designed diversion route were provided to drivers through a VMS strategically located upstream of the decision point. The major alternative route included University Boulevard, US-31/280 and I-20/59.

43 33 This scenario was designed to study the impact of information provision and user compliance with diversion recommendations on transit operations under emergency conditions. Moreover, this scenario tested the effectiveness of emergency management response plans in improving transit conditions during severe mobility disruption on a major facility (Figure 17). Figure 17 Alternative Route Using University Boulevard The five incident scenarios are summarized in Table 1. Table 1: Summary of Incident Scenarios Scenario Lanes Blocked Emergency Duration (min) Information Provision % of Driver Rerouting No No , 60, 90, 120 No No , 60, 90, 120 All All , 60, 90, 120 No No , 60, 90, 120 VMS I-65 NB via Univ. Blvd 50%

44 Transit Emergency Response Scenarios The transit emergency response scenario considered blasts at four popular shopping malls in the Birmingham area, along with an emergency response plan that involved the use of transit vehicles to evacuate people from the affected sites. In those scenarios the quickest route from the Central Station to the selected emergency sites was found, using DTA modeling. This response plan could be especially helpful for the evacuation of people who become stranded at the site and are dependent on transit systems for evacuation. To respond quickly to evacuation needs, regular bus service was temporary halted at the Central Station and buses were directed toward the affected sites for the quick transportation of emergency responders. It was assumed that the first transit bus involved in the evacuation departed from Central Station at 4:30 p.m., traveling south toward the Riverchase Galleria Mall. Similarly, for each other location, the first bus departed at 4:30 p.m. A total of 10 buses were used in the evacuation, with a running frequency of 9 minutes for every affected site. As the typical buses in VISTA run only on pre-defined routes and do not change their paths due to traffic conditions, a new vehicle type with the same dimensions of a bus was created. This enabled VISTA to find the best route for those vehicles from Central Station to the selected emergency sites in the hypothetical emergency response scenarios.

45 35 CHAPTER 5 RESULTS AND ASSESSMENTS 5.1 Incident Management Case Study Scenario 1: Base Case no incident Scenario 1 describes network operations under normal (no-incident) conditions and provides the baseline for comparisons. In this study, a peak demand period lasting from 3:00 p.m. to 9:00 p.m. was chosen as the analysis period. In order to establish the base case for this period, a set of equilibrium travel times and the planning O-D matrix for the period were obtained. The details of the baseline case are summarized in Table 2. Table 2: Incident Case Study Scenario 1 Base Case Total No. of No. of Buses Average Travel STD (min) Buses Loaded Time (min) Scenario 2: Emergency Conditions full lane blockage, no information provision This scenario was designed for the purpose of evaluating the impact of incidents of varying severity on transit bus travel time in the network. The severity is altered by varying the incident duration in 30-min increments under a full four-lane blockage condition and observing the relative changes in model response. The objective of the sensitivity analysis in this study was to illustrate the impact of the emergency on transit bus travel times in the network.

46 36 The scenario assumed the occurrence of a major traffic incident on northbound I- 65 about 1,000 feet upstream of the junction of I-65 and I-20/59. The model was run with a full four-lane blockage on link 5994 starting at 4:00 p.m. and lasting for 30 min, 60 min, 90 min, and 120 min. One on-ramp upstream of the incident on I-65 was also closed. No information about the incident was provided to drivers, and it was expected to follow their habitual paths. Testing and analysis of the model outputs from this scenario are beneficial, since such procedures are very useful for studying and designing response plans in a timely, and cost-effective manner. The results from the analysis are summarized in Table 3. Table 3: Incident Case Study Scenario 2 Emergency Conditions Full Lane Blockage No Information Provision Incident Duration (min) No. of Buses Loaded Total Travel Time (hr) Average Travel Time (min) STD (min) The results demonstrate that if emergency situations occur in strategic locations and are not dealt quickly they can affect transit operations in the entire network, at least in networks similar to the Birmingham regional network under study. The results further indicate that the network under consideration is sensitive to the duration of the incidents. A timely clearance of an emergency and the quick restoration of normal traffic conditions

47 37 not only benefit the emergency site and surrounding areas but also the transit operations in the network Scenario 3: Emergency Conditions full lane blockage, information provision to all users This scenario was designed for use in a sensitivity analysis by varying the incident duration in 30-min increments under a full four-lane blockage condition and observing the relative changes in travel times of the buses. Similar to Scenario 2, Scenario 3 assumes the occurrence of an emergency on northbound I-65 about 1,000 feet upstream of the junction of I-65 and I-20/59. While all other assumptions remain the same, in Scenario 3 all users are informed about the incident location, thus they can optimize their travel time in response to this information. It is noted, though, that in practice it is expected that only a percentage of travelers will have knowledge about the incident and their best alternative route. The simulation results for Scenario 3 are presented in Table 4. Table 4: Incident Case Study Scenario 3 Emergency Conditions Full Lane Blockage Information Provision to all Users Incident Duration No. of Buses Total Travel Average Travel STD (min) (min) Loaded Time (hr) Time (min)

48 38 It should be noted that Scenario 3 appears to perform in a remarkably similar way to the no-incident case (i.e., base case). This finding demonstrates that in this particular network that the impact of a major incident on the transit operations network is minimal, should users take advantage of the information provided. This might suggest great potential for the deployment of various ITS information systems in this area Scenario 4. Emergency Conditions two lane blockage, no information provision This scenario is similar to Scenario 2, except that a less severe incident is considered, resulting in the blockage of two traffic lanes instead of four. Moreover, no onramp closures due to the incident are considered this time. All other incident conditions remain the same. The objective of this scenario was to study the impact of the accident on traffic conditions and operations under partial lane closure on the basis of selected Measure of effectiveness (i.e., travel time). The results from this scenario are shown in Table 5. Table 5: Incident Case Study Scenario 4 Emergency Conditions Two Lane Blockage No Information Provision Incident Duration No. of Buses Total Travel Average Travel STD (min) Loaded Time (hr) Time (min) (min)

49 39 The similarity of the findings in this scenario to those in the base case is not surprising, since the flows on the freeway links are below the capacity of these links. The remaining capacity of the incident link can accommodate the demand in a reasonable manner. The results further justify the common practice of maintaining at least partial capacity of major transportation facilities when possible (instead of shutting down the facilities completely) so that an emergency can be handled relatively efficiently Scenario 5: Emergency Conditions full lane blockage, information provision, VMS before junction of I-65 and University Boulevard, user response 50% The hypothetical incident conditions in this case are identical to those of Scenario 2, with a full, four-lane blockage of varying duration starting at 4:00 p.m. On the other hand, in this scenario, information was available to drivers, and a response action was taken by 50% of them. More specifically, in response to the incident, the travel times of the designed diversion route were provided to drivers through a VMS strategically located upstream of the decision point. The major alternative route included University Boulevard, US-31/280, and I-20/59. This scenario was designed to study the impact of information provision and user compliance with diversion recommendations on transit operations under incident conditions. Moreover, this scenario tested the effectiveness of incident management response plans in improving transit conditions during a severe mobility disruption of a major facility. As shown in Table 6, and under emergency conditions similar to Scenario 2, the network wide travel times were reduced, particularly for long-lasting emergencies (i.e., 120 min) as a result of diversion information availability.

50 40 Table 6: Incident Case Study Scenario 5 Emergency Conditions Full Lane Blockage Information Provision with VMS Incident Duration No. of Buses Total Travel Average Travel STD (min) (min) Loaded Time (hr) Time (min) Comparison of Incident Impacts on Transit Operations under Scenarios 1 through 5 Figure 18 provides a comparison of results from all incident case study scenarios. Average Bus Travel Time for All Five Incident Scenarios 250 Average Bus Travel Time (hr) Base Case Scenario 2 Scenario 3 Scenario 4 Scenario Incident Duration (min) Figure 18 Average Travel Time of Buses for the 5 Incident Case Study Scenarios.

51 41 The overall transit performance is improved by the information provision and corresponding user response, especially for incidents of longer duration (i.e., 120 min). In addition, when all users re-optimize their travel paths in response to the incident, MOEs across the network are largely unaffected by the incident, since the regional network has enough reserve capacity to absorb the diverted traffic with ease. Another interesting observation is that the transit network performance result for a user compliance of 50% (VMS design) for 120-min incident duration is lower than that of Scenario 2 in terms of transit network travel time Transit Emergency Response Case Study In these hypothetical emergency response scenarios, blasts were assumed to have taken place at 4:00 p.m. but not on the same day at the four chosen emergency sites. The regular bus service was halted at Central Station, and 10 buses were sent to the affected sites at a frequency of 9 min for the transportation of emergency responders as well as evacuation of the people stranded at the emergency sites. The first bus left the Central Station at 4:30 p.m., i.e., 30 min after the incident was reported. Transit-related MOEs for the transit emergency response case study were obtained from the critical route delay report as well as the VISTA data tables. The vehicle_path_time and vehicle_path tables from the database were used to calculate travel times and to find the optimized bus routes. The average travel time of buses/emergency vehicles for the routes was used as a MOE. Emergency response travel times in response to an emergency at Brookwood Village Mall are reported in Table 7.

52 42 Table 7: Bus Travel Time from Central Station to Brookwood Village Mall VISTA Path ID VISTA Vehicle ID Departure Time Bus Travel Time (p.m.) (min) : : : : : : : : : : Table 8: Comparison of Routes for Emergency Site at Brookwood Village Mall VISTA Path ID Vehicles following this path , , , , Bus Avg. Travel Time (min) Avg. Speed (mph) , , Results for Emergency Response at Brookwood Village Mall It can be inferred from Table 8 that Route is the most followed route, with 5 buses. Route is shown in Figure 19. An average travel time of 9.1 min and average speed of 31.7 mph was reported along this route, making it the quickest route for the emergency response from Central Station to the emergency site at Brookwood Village Mall.

53 43 Figure 19 Optimum Path from Central Station to Brookwood Village Mall Results for Emergency Response at Riverchase Galleria A similar analysis was performed assuming an emergency at the Riverchase Galleria Mall. The results are summarized in Tables 9 and 10. Table 9: Bus Travel Time from Central Station to Riverchase Galleria VISTA Path ID VISTA Vehicle ID Departure Time Bus Travel Time (p.m.) (min) : : : : : : : : : :

54 44 Table 10: Comparison of Routes for Emergency Site at Riverchase Galleria VISTA Path ID Vehicles following this path Bus Avg. Travel Time (min) Avg. Speed (mph) , , , , , It can be inferred from these results for Routes , , and that average travel times are equal, so these routes are the best for buses for trying to reach the emergency site in this scenario with minimum time as the criterion for comparison with other routes. An average travel time of 16.6 min makes it the quickest route for an emergency response from Central Station to the emergency site at Riverchase Galleria. These results confirm the hypothesis that freeways are the quickest way to get to the destination, as the optimum bus routes included sections of I-65 (Figures 20 through 22).

55 45 Figure 20 Optimum path 1 from Central Station to Riverchase Galleria (Route ) Figure 21 Optimum Path 2 from Central Station to Riverchase Galleria (Route )

56 46 Figure 22 Optimum Path 3 from Central Station to Riverchase Galleria (Route ) Results for Emergency Response at Wal-Mart Supercenter at Lakeshore Drive The third emergency response scenario assumed an emergency at Wal-Mart on Lakeshore Drive. The results from this simulation are summarized in Tables 11 and 12. Table 11: Bus Travel Time from Central Station to Wal-Mart Supercenter at Lakeshore Drive. VISTA Path ID VISTA Vehicle ID Departure Time Bus Travel Time (p.m.) (min) : : : : : : : : : :51 9.5

57 47 Table 12: Comparison of Routes for Emergency Site at Wal-Mart Supercenter at Lakeshore Drive. VISTA Path ID Vehicles following Bus Avg. Travel Time Avg. Speed this path (min) (mph) , , , , , , , As table indicates, route is followed by 6 buses and has the least travel time, 9.3 min, in comparison with the other routes. This is the quickest route between Central Station and the emergency site at Riverchase Galleria. It can be observed from Figure 23 that buses follow I-65 for the larger part of the route, to get to the Wal-Mart Supercenter at Lakeshore Drive. Figure 23 Optimum Path from Central Station to Wal-Mart Supercenter at Lakeshore Drive. (Route )

58 Results for Emergency Response at The Summit The last emergency studied that required transit involvement as part of the response was at the The Summit in Birmingham, AL. The reports from the simulation analysis are presented in Tables 13 and 14. Table 13: Bus Travel Time from Central Station to The Summit VISTA Path ID VISTA Vehicle ID Departure Time Bus Travel Time (p.m.) (min) : : : : : : : : : : Table 14: Comparison of Routes for Emergency Site at The Summit VISTA Path ID Vehicles following this path , , , , , , , , Avg. Travel Time (min) Avg. Speed (mph) From these results it can be concluded that route is followed by 6 buses and is the quickest route to reach the emergency site at The Summit. The average travel speed for the buses on this route is 35.4 mph, compared to 37 mph for alternative route

59 The optimum route for buses from the Central Station to the The Summit is shown in Figure 24.. Figure 24 Optimum Path from Central Station to The Summit Shopping Mall (Route )