Impacts of Queue Jumpers and Transit Signal Priority on Bus

IMPACTS OF QUEUE JUMPERS AND TRANSIT SIGNAL PRIORITY ON BUS

RAPID TRANSIT

R. M. Zahid Reza

A Thesis Submitted to the Faculty of

College of Engineering and Computer Science

in Partial Fulfillment of the Requirements for the Degree of

Master of Science

Florida Atlantic University

Boca Raton, Florida

August 2012

ACKNOWLEDGEMENTS

I am heartily thankful to my supervisor Dr. Aleksandar Stevanovic for his expertise and circumspective guidance and support all through my graduate studies at the Florida

Atlantic University. I also want to thank Dr. Khaled Sobhan for giving me an opportunity to pursuing higher study and Dr. Evangelos Kaisar for his helpful suggestions and comments during my research work. I would like to expand my thanks to Dr. Milan

Zlatkovic, from the University of Utah whose sincere judgment and recommendations helped me to carry out the study.

Finally, I would like to express my special thanks to my family whose continuous supports and encouragement was constant source of stimulus for this work.

ABSTRACT

Author: R. M. Zahid Reza

Title: Impacts of Queue Jumpers and Transit Signal Priority on Bus Rapid Transit

Institution: Florida Atlantic University

Thesis Advisor: Dr. Aleksandar Stevanovic

Degree: Master of Science

Year: 2012

Exclusive bus lanes and the Transit Signal Priority are often not effective in saturated peak-traffic conditions. An alternative way of providing priority for transit can be queue jumpers, which allows buses to bypass and then cut out in front of waiting queue by getting an early green signal. Utah Transit Authority deployed Bus Rapid Transit system at Salt Lake County, Utah along W 3500 S. This research evaluates the impacts of queue jumpers with TSP on Bus Rapid Transit (BRT) and private vehicular traffic. Four

VISSIM models were developed for analysis: Basic scenario, no TSP with queue jumpers, TSP with no queue jumpers, and TSP with queue jumpers. In TQ scenario travel time was reduced between 13.2-19.82% with respect to basic scenario. At the same time, travel time of private traffic increased vary little 0.38-3.28%. Two TSP strategies: green extension and red truncation are implemented in this research work.

IMPACTS OF QUEUE JUMPERS AND TRANSIT SIGNAL PRIORITY ON BUS RAPID TRANSIT

LIST OF FIGURES ...... x LIST OF TABLES ...... xii 1. INTRODUCTION ...... 1 1.1. Problem Statement ...... 1

1.2. Research Objectives ...... 4

1.3. Research Tasks ...... 4

1.4. Thesis Organization...... 5

2. LITERATURE REVIEW ...... 6 2.1. Overview of Performance of Bus Rapid Transit ...... 6

2.2. Review of Performance of Preferential Treatments ...... 11

2.2.1. Overview of Performance of Transit Signal Priority (TSP) ...... 11

2.2.2. Review of Performance of Exclusive Bus lanes ...... 18

2.2.3. Review of Performance of Queue Jumpers ...... 22

2.3. Overview on Calibration ...... 26

2.4. Summary of the Literature Review ...... 27

3. MICROSIMULATION MODEL DEVELOPMENT ...... 28 3.1. Study Area ...... 28

3.2. Simulation Network ...... 30

3.3. Traffic Control...... 31

3.4. Transit Operations ...... 31

3.5. Preliminary Model building ...... 32

3.6. Scenario Design...... 33

3.7. Queue Jumpers ...... 34

3.8. Queue Jumper Bus Bay ...... 35

3.9. Relocating Bus Stops ...... 39

3.10. Transit Signal Priority Implementation ...... 42

3.11. Queue Jumper Phase Implementation ...... 43

3.12. Summary of Model Development ...... 44

4. CALIBRATION AND VALIDATION OF THE MODEL ...... 45 4.1. Why necessary to calibrate? ...... 45

4.2. Overview of VISSIM ...... 46

4.2.1. VISSIM Car Following Parameters ...... 47

4.2.2. Lane change...... 50

4.2.3. Driving Behavior Parameters for the Model ...... 52

4.3. Model Calibration ...... 53

4.4. Validation ...... 55

vii

4.4.1. Model Validation ...... 55

4.4.2. Summary ...... 57

5. RESULTS AND DISCUSSION ...... 58 5.1. BRT Travel Times ...... 58

5.2. Private Vehicular Travel Times ...... 61

5.3. Bus Travel Times ...... 62

5.4. Overall Comparison of Travel Times...... 65

5.5. Network Performance Evaluation ...... 68

5.5.1. Average Stopped Delay ...... 68

5.5.2. Average Speed ...... 69

5.5.3. Network Wide Comparison of Average Delay ...... 70

5.6. Average Cross Street Delay ...... 70

5.7. Impacts on Intersections on Major Corridor ...... 71

5.8. BRT Time-Space Diagrams ...... 73

5.9. Summary of Overall Result Analysis ...... 74

6. CONCLUSIONS...... 75 6.1. Conclusions ...... 75

6.2. Limitations of the study and future Research Work ...... 76

APPENDIX A ...... 78 APPENDIX B ...... 79 APPENDIX C ...... 80 viii

BIBLIOGRAPHY ...... 81

LIST OF FIGURES

Figure 1 Communication technologies integrated with BRT [Courtesy:

Implementation of BRT Transportation System] ...... 8

Figure 2 Queue jumpers at Charlotte, NC ...... 23

Figure 3 3500 South RT 35 and BRT route with BRT bus stop locations ...... 30

Figure 4 Study Areas along W 3500 S...... 30

Figure 5 Queue jumper lane (Ottawa) (Courtesy: Bus Rapid Transit service design

guideline) ...... 35

Figure 6 Queue jumper bus bay ...... 36

Figure 7 Intersection with or without queue jumpers ...... 38

Figure 8 Different types of bus stops ...... 40

Figure 9 Car following logic (Courtesy: VISSIM manual 5.3) ...... 49

Figure 10 Behavior parameter sets for drivers in VISSIM ...... 53

Figure 11 Model calibration results – traffic movement comparison ...... 54

Figure 12 Model calibration results – traffic movement comparison ...... 55

Figure 13 Model validation results – travel time comparison ...... 56

Figure 14 Model validation results – travel time comparison ...... 57

Figure 15 Travel time comparison of BRT in EB...... 60

Figure 16 Travel time comparison of BRT in WB ...... 60

Figure 17 Travel time comparisons for private traffic in EB ...... 62

Figure 18 Travel time comparisons for private traffic in WB ...... 62

Figure 19 Travel time comparisons for bus in EB ...... 64

Figure 20 Travel time comparisons for bus in WB...... 65

Figure 21 Overall travel time comparisons in WB ...... 67

Figure 22 Comparison of travel time in EB ...... 67

Figure 23 Comparison of network wide stopped delay ...... 68

Figure 24 Comparison of network wide speed ...... 69

Figure 25 Comparison of network wide delay ...... 70

Figure 26 Comparison of cross-street delay ...... 71

Figure 27 Sample BRT time-space diagram ...... 73

LIST OF TABLES

Table 1 List of intersections does not have entrance/ exit queue jumper lane ...... 37

Table 2 List of bus stoppage needs to be shift to queue jumper lane ...... 42

Table 3 : BRT travel times comparison ...... 59

Table 4 Private traffic travel times comparison ...... 61

Table 5 Bus travel times comparison ...... 63

Table 6 Comparison of waiting time and percentage of stops along mina corridor in

WB ...... 72

xii

1. INTRODUCTION

This chapter introduces the research problem and techniques to observe the effectiveness to resolve the problem. Firstly the problem of severe congestion is identified with statistics, secondly a brief idea about how preferential treatment like queue jumper can provide benefits to Bus Rapid Transit (BRT) to overcome congestion problem is mentioned. The next part of the introductory chapter describes the research objectives and tasks. The final part of the chapter describes the organization of the dissertation.

1.1. Problem Statement

Along with economic surge, traffic congestion levels have been increasing day by day due to the excess usage of vehicles. However the road network is not following the similar ratio of increased traffic demand. Therefore congested time is lengthening and so as the travel than in the past. Texas Transportation Institute (TTI) represented congestion trend line by travel time index which is highly significant to comprehend the severity.

This travel time index is the ratio of the average time required to make a trip in rush hours versus making the same trip in free-flow condition for instance value of 1.30 indicates a 20-minute free-flow trip takes 26 minutes in the peak period and TTI reported travel time index from 1982 to 2010 in “2011 Urban Mobility Report and Appendices”.

The ten U. S. metropolitan areas with the greatest congestion, in order of severity, were Washington DC (with a ratio of 1.33 to 1), Seattle, Dallas, New York, Chicago, San

Francisco, Atlanta, San Diego, Miami, Boston, Philadelphia, Phoenix (with a ratio of

1.21 to 1) was identified. This report also identified three reasons behind the traffic congestion regarding population, traffic infrastructure and number of trips. First reason correlates population and vehicles and was expressed as when overall traffic surpasses the capacity that means when many people and lots of freight moving at the same time causes traffic congestion. Second reason focused on the generation of number of trips as when huge number of trip is generated in a short period of time on a small system can create congestion. Third reason was identified as all those non-recurrent events like crashes, vehicle breakdowns, improperly timed traffic signals, special events and hazardous climate [1].

Traffic congestion severely damages the effectiveness and attractiveness of transit vehicles, especially on those that do not use exclusive rights-of-way (ROW). These negative impacts mostly imply travel time increase, increase in delay, bad reliability and on-time performance, bus crowding, increase in passengers‟ bus stoppage waiting times, increase in fuel consumption, wear and tear on vehicles, and health hazard condition for stressed and frustrated passengers, and severe environmental impacts etc. In order to overcome these impacts, transit agencies introduce new, high capacity rapid transit modes, along with transit operational strategies. In the recent years, Bus Rapid Transit

(BRT) has become one of the most common used rapid transit modes. BRT is a flexible, high performance rapid transit mode that uses buses or specialized rubber tired-based

vehicles operating on pavement, and combines a variety of physical, operating and system elements into a permanently integrated system. It is an integrated system which consists of running ways (very often exclusive lanes), special designed rail-like stations, high-capacity low-floor vehicles, services and Intelligent Transportation Systems (ITS).

It provides flexibility of buses and quality of rail transit but construction and operational cost is much lower [2].

Several preferential treatments are adopted to facilitate BRT like Transit Signal

Priority (TSP), exclusive bus lanes i.e., dedicated bus lane, intermittent bus lane etc. TSP is an operational strategy generally implemented on most signalized intersections often for BRT line. TSP facilitates the movements of in-service transit vehicles through signalized intersections and it makes transit faster, more reliable and more cost-effective.

It is often used for regular bus lines, but is most beneficial when combined with BRT systems [3]. However, several studies investigated the relationship between the effectiveness of a TSP application and the surrounding traffic parameters: main corridor volume/capacity ratio, cross street volume/capacity ratio, bus stop location, bus headway, detector location etc. and when volume/capacity ratio is close to 1 along main corridor or cross street, system with or without TSP does not have much difference however system wide disbenefit occurs and have a severe impact on cross street [4, 5]. Moreover exclusive bus lanes like dedicated bus lane, intermittent bus lane is often difficult to justify the use of an exclusive lane for buses during peak hours as it requires one lane completely to serve the bus [6]

A special type of bus-preferential treatment that has the potential of avoiding the shortcomings of both bus lanes and regular TSP is the queue jumper lane. Queue jumper lane is an additional travel lane with signal priority to allow buses to bypass a waiting traffic queue through the use of a right-turn bay. It is also accompanied with special queue jumper phase to pass the intersection ahead of general traffic at mixed lane. Queue jumper lanes do not take a lane away from the general traffic and simultaneously provides the queue-bypassing ability of bus lanes by making use of existing right-turn only lane or curb side bus lane. The queue-bypassing ability of queue jumper lanes also ensures on time arrival of buses at bus stoppage [6].

1.2. Research Objectives

The major objective of this study is to evaluate the impacts of queue jumpers and transit signal priority on BRT and private traffic. Four scenarios: No TSP without queue jumpers (NTNQ), no TSP with queue jumpers (NTQ), TSP with no queue jumpers

(TNQ) and TSP with queue jumpers (TQ) were developed to evaluate the effectiveness of queue jumpers on TSP. Moreover, impacts on cross street, performance of aggregated intersection and network is evaluated. In order to evaluate properly, comparison of several parameters of four scenarios is made for PM peak period from 4 PM -6 PM. The study is using VISSIM microsimulation software.

1.3. Research Tasks

Three major study tasks can be summarized as follows:

 Model building: preliminary model building, calibration and validation of the

model, modeling queue jumpers etc.

 Comparing several traffic parameters: travel times, delay, stopped delay, speed,

cross street delay, intersection performance etc. for transit and private traffic.

 Drawing conclusions based on the impacts on different modes, cross street delay,

aggregated intersection performance, and network performance.

1.4. Thesis Organization

The thesis is divided into six chapters. Chapter 2 provides a comprehensive overview of the previous related research. The literature review concentrates on several topics: performance of BRT and performance of preferential treatments. Performance of preferential treatments is ramified into several parts: overview of TSP and overview of exclusive bus lane i.e. dedicated bus lane, intermittent bus lane and queue jumpers.

Chapter 3 provides a detail explanation of preliminary model building, study area, scenario design, microsimulation platform, signal setting, guidelines of several model structures i.e., formation of queue jumpers, shifting bus stoppages etc. Chapter 4 explains calibration and validation of the model. It provides explanations for adjustments parameters of the microsimulation to obtain a matching with the real filed. Chapter 5 shows the comparison of several traffic parameters for developed scenarios to evaluate the impacts. Finally, Chapter 6 draws conclusions from this research and recommends future work that can be conducted based on this study. This chapter introduces a reader with queue jumpers and its potential benefits from a perspective of travel time improvement, impacts on cross street etc. on other preferential treatment.

2. LITERATURE REVIEW

This chapter presents the findings from the literature review conducted as a part of this study. A wide-ranging literature search is done, and findings are classified in three subtopics. The first subtopic provides a review of performance of Bus Rapid Transit

(BRT). The second subtopic provides a review of the studies in several preferential treatments like Transit Signal Priority (TSP), exclusive bus lanes and queue jumpers etc.

The third subtopic summarizes studies on calibration of simulation models. Finally the last excerpt provides a summary of the literature review.

2.1. Overview of Performance of Bus Rapid Transit

With the development of modern society the traffic demand is increasing everyday so as the usage of number of vehicles. Excess use of vehicles causes congestion which hampered the performance of transit vehicles, which does not have exclusive rights-of- way (ROW). These negative impacts often result in increased travel times, delay, poor reliability, and unpredicted schedule adherence, wear and tear of the vehicle, excess gas emission and fuel consumption, hazardous environmental impacts etc. In order to overcome these negative impacts, transit agencies are trying to introducing new, high capacity rapid transit modes, such as bus rapid transit (BRT), which is facilitated with new transit operational strategies.

Bus rapid transit (BRT) is not a single concept, rather this term covers usage a variety of public transportation systems to provide more efficient service using wide variety of

ROW. Different transit agencies have defined BRT in different way. The FTA defines

BRT as: “Rapid mode of transportation that can combine the quality of rail transit and the flexibility of buses” [8]. BRT Implementation Guideline defined BRT as: “A flexible, rubber-tired rapid-transit mode that combines stations, vehicles, services, running ways, and Intelligent Transportation System (ITS) elements into an integrated system with a strong positive identity that evokes a unique image”[9].

There are many reasons for developing BRT systems, especially in a U.S. context.

Rapid growth of Central business districts (CBDs) demand improving transport capacity and accessibility. Regarding limited road capacity and parking facility of most CBDs

BRT systems can be an important alternative. Additionally BRT can be operated reliably at various ROW for most corridors in U.S. and Canadian cities. For instance, the Ottawa transit way system‟s link to the CBD carries more people in the peak hour than most LRT segments in North America. BRT can be the most cost-effective means compared to rail modes serving a broad variety of urban and suburban environments that foster economic development and transit- and pedestrian-compatible design [10].

BRT is generally facilitated with several communication technologies integrated with

ITS. Wide variety of individual ITS elements are available for inclusion with any BRT service. All ITS technologies require some form of communications technology in order to provide data to other systems and receive commands. Communication technologies can be private radio networks, cellular phone, Wi-Fi, infrared, RFID, inductive loops etc.

Regarding function communications technologies can be classified be of two types: live and deferred. Live data refers to data being sent during normal operations of bus and deferred data refers to data stored on the bus computer and downloaded once the bus is within the garage or at a scheduled time. Communications technologies are inherent to most Intelligent Transportation Systems thus these technologies are a means of achieving benefits. TSP, APCs, Fare Collection, Surveillance systems, fleet management etc. are few examples of integration with ITS systems. Several agencies use ITS integrated with communication technologies for instance BC Transit uses cellular connection to retrieve

APC data and vehicle data remotely, York Regional Transit uses a cellular connection between the buses and the control center for its AVL system etc. [10]

Figure 1 Communication technologies integrated with BRT [Courtesy: Implementation of BRT Transportation System]

Bus Rapid Transit (BRT), one such high-capacity rapid transit mode, has gained popularity and a significant number of BRT lines been deployed in the US in the last few 8

decades [11]. The Metro Orange Line in Los Angeles County, CA, opened in October

2005; one of the first full featured BRT has experienced a big gain in ridership during its first year of operation. This full BRT system included advanced high-capacity vehicles, enhanced stations, improved boarding, frequent and reliable service, dedicated lanes, off- vehicle fare collection, ITS and, improved accessibility of pedestrian and bicycle etc. In only seven months of operation, the line achieved its 2020 goal in ridership gain, which was more than four times greater than the ridership increase projected for the first year.

Moreover, Orange Line has experienced more than twice the increase of rail and almost three times the increase of buses ridership. Significant improvement was informed in community activities as home based trips were increased. About 17% of the ridership gains were new riders, while one third of riders were diverted from cars and travel time reduction is reported by two-third of the riders who previously drove by cars. [12]

The Trans Milenio BRT line in Bogota, Colombia, is one of the leading BRT lines in the world, which carries about 1.4 million passengers per day. This system included high capacity articulated vehicles with multiple doors, high average bus occupancies, exclusive running ways ensuring high commercial speed, high capacity “rail like” stations included level boarding and off-board fare payment, and centralized control of bus operations etc. The implemented BRT features have reduced travel times by more than 32% for transit riders along the corridor, increased public transit travel speeds approximately 9.3 mph to 16.6 mph, and improved safety significantly, was cost effective, and reduced number of collisions along the service corridors from 1060 to 220

(79% reduction) [13].

In West Valley City, Salt Lake County, Utah 10.8 miles long BRT line was launched on July 14, 2008, was the first BRT deployment in that County. Preliminary survey results showed significant improvements in transit operations, with a 33% increase in ridership, reductions of close to 15% in travel times and improved reliability. Dwell times was reduced, mostly due to the new fare collection process and improved accessibility at bus stops. Passenger survey revealed high degree of acceptance specifically in hot weather condition. [14]

98 B-Line in Vancouver, Canada benefitted on travel time savings of approximately

20% reduction in time compared to previous services. Surveys indicated approximately

23% of the users of the 98 B-Line were former car drivers or car passengers who have changed mode to ride transit the shift from auto to transit associated with the 98 B-Line represents a reduction of 8 million vehicle kilometers per year by private automobile.

This system provided better on-time performance and improved service quality. This system was facilitated with real-time vehicle location tracking system, the traffic signal priority measures, changeable message signs in the stations inform riders of the arrival time of the next bus, on-board audio and video displays announce next stops etc. [15]

The cost-effectiveness and relative flexibility of the investment on BRT, twelve Latin

American cities, three Australian cities, seven U .S. cities, eight Asian cities, and eighteen European cities have BRTs in place. Some are complete systems while others are single lines. In reality there are currently more BRT systems under development than existence. Six Latin American cities, two Oceanian cities, one North American city, one

Asian city, one African city and three European cities have BRT under construction. The

dramatic success of BRT propels many cities to plan for BRT system. Sixteen Latin

American cities, one Oceanian city, thirty seven North American cities, twenty four

Asian cities, seven African cities and fifteen European cities have BRT system in planning process (as of March 2007) [16].

2.2. Review of Performance of Preferential Treatments

Transit preferential treatments are a key component to the provision of travel time savings and improved on-time performance for bus and rail systems operating in mixed traffic on urban streets. Major preferential treatments are exclusive bus lanes (i.e., dedicated bus lanes, intermittent bus lanes), TSP, queue jumpers etc. This section includes review of previous works on several preferential treatments.

2.2.1. Overview of Performance of Transit Signal Priority (TSP)

Transit signal priority (TSP) is an operational strategy that facilitates the movement of transit vehicles, either buses or streetcars, through traffic-signal controlled intersections. The reasons behind of TSP implementations are improved schedule adherence and improved transit travel time efficiency alongside minimal impacts to normal traffic operations. Transit priority at signalized intersections has been studied in the United States since the 1970s [17]. In recent years, TSP has been widely implemented by transportation agencies in North America and these growing deployments of TSP across the nation require extensive evaluation studies. A wide number of studies have attempted to evaluate TSP using either empirical, analytical, and/or simulation tools.

Transit Signal Priority can be implemented in a variety of ways including passive, active and adaptive priority treatments as discussed below:

2.2.1.1. Passive Priority

Passive priority is operated based on the database of transit route and ridership patterns, therefore does not require the hardware and software investment to detect or to generate priority request and thereby it can be operated continuously. Passive priority is most effective when and transit routes, passenger loads, schedule, and/or dwell times etc. are known and transit can be operated predictably. In this application, the signal timing plan accounts the average dwell time at transit stops and as the dwell times are highly variable, generally low a cycle length is used. For Improved traffic flow and for the reduction of travel time, operational improvements to signal timing plans, such as retiming, reducing cycle lengths, or coordinating signals on a corridor is required for proper implementation of passive priority. It methods can be operated and implemented at relatively low-cost and easily, since transit detection or communication equipment are not necessary to detect the presence of transit vehicles [18]

2.2.1.2. Active Priority

Active priority strategies provide priority treatment to a specific transit vehicle. It detects the transit vehicle at first and subsequently activate priority request. Various types of active priority strategies are available i.e., green extension, red truncation or early green, actuated transit phase, phase insertion and phase rotation. A green extension strategy extends the green time for the approaching TSP-equipped vehicle and only applies when the signal is green for the approaching TSP equipped vehicle. This is one of

the most effective forms of TSP as a green extension does not require additional clearance intervals. An early green strategy shortens the green time of preceding phases to expedite for the detected TSP equipped vehicle to ensure early green. This strategy applies when the TSP-equipped vehicle approach at red phase. Generally early green and green extension strategies are available together within TSP enhanced control environments but are not applied at the same time. Actuated transit phases are only displayed when a transit vehicle is detected at the intersection. For instance, an exclusive left turn lane for transit vehicles that is left turn phase will be only displayed when a transit vehicle is detected in the lane. Another example would be the use of a special queue jump phase that allows a transit vehicle to cut-out the intersection ahead of general traffic. Phase insertion inserts a special priority phase within a normal signal sequence and is only inserted when a transit vehicle is detected and requests priority for this phase.

An example would be the insertion of a leading left-turn-only phase for transit vehicles entering an off-street terminal on the opposite side of the street. The order of signal phases can also be “rotated” (i.e., phase rotation) to provide TSP. For example, a northbound left-turn phase could normally be a lagging phase as it follows the opposing through signal phase. A northbound left turning bus requesting priority that arrives before the start of the green phase for the through movement could request the left-turn phase and could be served as a leading phase in order to expedite the passage of the transit vehicle. [18]

2.2.1.3. TSP Operating in Real-Time

There are subtle differences between TSP with Adaptive Signal Control Systems and

Adaptive Signal Priority. These are very sophisticated and complex systems. Although an

Adaptive Signal Priority built on top of an adaptive signal control system may offer more benefits however Adaptive Signal Priority does not to be built on top of an adaptive signal control system. The priority strategies include early green, green extension and phase insertion listed etc. TSP with Adaptive Signal Control Systems provides priority while and simultaneously optimize given traffic performance criteria. Adaptive Signal

Control Systems continuously monitor traffic conditions and adjust control strategies.

Adaptive Signal Control Systems accounts person delay, transit delay, vehicle delay, and/or a combination of these criteria. To take advantage of Adaptive Signal Control

Systems TSP typically detects transit vehicle early in order to provide more time to adjust the signals to provide priority without much adverse impacts on general traffic. Adaptive systems combined with TSP requires the regular update of the transit vehicle‟s arrival time, which can vary due to the number of stops and traffic conditions. The updated arrival time is adjusted with the signal timings. Adaptive Signal Priority is a strategy that takes into consideration the trade-offs between transit and traffic delay and allows supple adjustments of signal timing by adapting the movement of the transit vehicle and the prevailing traffic condition. Typically, an adaptive TSP needs to do: 1) accurate prediction of bus time-to-arrival to the intersection in real-time when vehicle is within a specified range 2) detection of traffic system; 3) incorporating a signal control algorithm that adjusts the signals to provide priority while explicitly considering the impacts on the rest of the traffic and ensuring pedestrian safety;4) vehicle to infrastructure 14

communication links; priority request generator(s) (PRG), a priority requests ever (PRS) and a control system with real-time signal timing strategies to facilitate adaptive TSP

[18].

2.2.1.4. Case Study of TSP

Numerous studies have been reported over the years relating TSP. Nonetheless very few studies have been done so far on field evaluation due to its costliness of field implementation. Consequently, most studies have been done based on simulation tool such as VISSIM to evaluate the impacts. Several case studies have shown successful implementation and quantifiable benefits of TSP. Some case studies are mentioned hereby:

 In Tualatin Valley Highway, Portland, active priority strategy: green extension, early

green was implemented for 13 signalized intersections. Bus travel time savings was

reported of 1.7 to 14.2% per trip, 2 to 13 seconds reduction in per intersection delay

[19].

 At Powell Blvd. in Portland for 4 signalized intersections active priority strategy:

early green, green extension was provided. 5-8% reduction of bus travel time as well

as reduction in bus person delay was reported [20].

 At Rainer Avenue in Seattle active priority strategy: green extension, red truncation

was provided for 20 intersections. 5-8% reduction in travel times, 25-34% reduction

in average intersection bus delay and $40,000 passenger benefit per intersection was

reported [21].

 In Tacoma, WA the combination of TSP and signal optimization reduced transit

signal delay about 40% in two corridors and signal coordination alone brought $4.5

million economic benefit [3].

2.2.1.5. Benefits and disbenefit of TSP

The success of the TSP implementation can be varied with the characteristics of traffic environment of deployment site like transit usage, the time of day when used and the characteristics of the transit service etc. Several studies investigated the impacts of traffic parameters on the effectiveness of TSP application. Ngan et al. [5] investigated that performance of TSP was impacted by bus approach volume, cross street volume/ capacity ratio, bus headway, bus stop location, bus detector location, bus arrival time etc.

Bus travel time increased with increase in v/c ratio regardless of TSP. Author reported that TSP application would be most effective under moderate-to-heavy traffic condition however as v/c ratio tends closer to 1.0 system with TSP and without TSP does not have much impact and. Cross street v/c ratio also influence the performance of TSP. Impact is almost same as main street v/c ration. Impact is minimal at low cross street v/c ratios however TSP has a moderate impact on cross street performance with a v/c ratio above

0.8, it has a significant impact with v/c ratios above 0.9 which causes high cross street delay and increases delay recovery cycles. Author examined the impacts of headway on the performance of TSP. Optimal headway was reported 10 minutes, more than this cause fewer TSP requests and limits the benefits of TSP on bus performance and less than the optimal value increases the volume causes increased bus delay. For bus stop position far side bus stop was recommended over near side bus stop as significant portion of the

green extension would be wasted for alighting and boarding of passengers at a nearside bus stop.

Arrival time has impacts on the performance of TSP. Rakha and Zhang [22] informed TSP generally provides benefits to transit vehicles that receive priority by reducing average bus delay, average bus stops and average bus fuel usage etc. but these benefits are highly dependent on the time of arrival of the transit vehicle within the cycle length and the phase of the traffic signal that is requesting priority. TSP impacts are influenced by the demand distribution at a signalized intersection. If the transit vehicle arrived during the early phase it causes minimum disruption to the general traffic while arrival of the transit vehicle at the later phase causes significant system-wide disbenefit.

Author also mentioned TSP performance was influenced by traffic volume. TSP has a marginal system-wide impact for low traffic demands; however, as the demand increases, the system-wide disbenefits of TSP increases. Authors also analyzed impacts of dwell time variance at near-side bus stops. The result showed significant system-wide disbenefits with an increase in bus dwell times for near side bus stop.

Garrow and Machemehl [4] used CORSIM to simulate several green extension measures on the Guadalupe corridor in Austin, Texas. They mentioned that TSP application would be most effective under moderate-to-heavy traffic condition. They reported that the negative impact on the cross-street traffic is “Significant” if the cross- street saturation level is equal or greater than 1.0 with a 10-second green extension and with 10 minutes bus headway. Moreover, the negative impact is “Significant” if the cross-street saturation level is equal or above of 0.9 with a 20-second green extension and

with 10 minutes bus headway in peak hour and under these conditions the cross-street traffic would require 2 to 3 cycles to recover. Authors recommended far side bus stop over nearside bus stop since it would cause a higher delay as significant portion of the green extension would be wasted while passengers board and alight at a nearside bus stop. Moreover, Daniel [23] and Huffman et al. [24] found that TSP implementation is more preferable with far side bus stop is more preferred when TSP is implemented. For near side bus stop the detector should be placed downstream of the bus stop to resolve the problem of higher delay.

2.2.2. Review of Performance of Exclusive Bus lanes

Bus rapid transit (BRT) often operated with a dedicated bus lane (DBL) that delivers comfortable, cost-effective mobility like rail transit. Operationally, BRT includes buses running on exclusive rights-of-way with dedicated stations and pre-boarding fare payments, or operating in mixed traffic lanes on city arterials. The provision of segregated bus and tram lanes has been identified as an efficient means of improving transit reliability and running times when transit is operated at mixed traffic lane.

The analysis of fundamental diagrams and the velocity density profiles shows that when bus lane is dedicated, the flow of buses is far higher than the other two cases, while the flow of cars is the lowest. It indicates that the dedicated bus lane has the advantage to free buses from traffic interference and also has disadvantages in disrupting traffic.

However in practice, the permanent dedication of one lane in the cases with a lower frequency of bus services is very inefficient. Moreover it is not always possible to easily find free space for the dedicated bus lanes, especially in city centers. The DBL strategy is

only appropriate for low traffic flow in a two-lane traffic system. This limitation can be partly overcome by opening the bus lane to general traffic intermittently when the bus lane is not in use by buses [25].

This has led to the new concept of the intermittent bus lane (IBL). An intermittent bus lane (IBL) is a bus-reserved lane that allows private vehicle traffic to use the lane when not in use by the bus. The concept of Intermittent Bus Lane (IBL) was introduced by

Viegas [26] as an innovative approach to achieve bus priority. Normally it is arranged at the rightmost lane. Some kind of variable light signals are placed on the pavement along the line separating that lane from the next. IBL restricts traffic from changing into the bus lane instead of requesting the traffic to change the lane. In order to guarantee that buses and private vehicles do not interact with the system is facilitated with flashing longitudinal light. Vehicles already flowing on the special lane (ahead of the bus) can keep flowing within it or turn left to the other lanes, signal adjustments would be used to flush the queues at traffic signals and clear the way for the bus. These signal adjustments may increase the amount of green time allocated to the arterial at times when the arterial demand is low, bus this could reduce the capacity of side streets and increase delay. But the vehicles on other lanes could not pass through flashing lights into IBL. And when the longitudinal lights are off, the lane becomes a normal one, open to all vehicles accepted on that road [27].

Normally, IBL is integrated with a conventional UTC system. The general structure of the integrated system is similar as a conventional UTC system, but with two additional interfaces. One is connected to a kind of AVL system, and the other is for driving IBL

signals. Although simple loop detectors are not sufficient for bus location, some AVL technologies, like GPS system, have been successfully used for bus location with accuracy around 5 m. The IBL lights would generally consist of LED arrays installed in a small box inserted in the pavement. LED arrays are extremely visible and reliable, because they have multiple small lights working in parallel. And joint consideration of

IBL signals and traffic light signals at intersections leads to lower time losses in bus operation, but these gains can be significantly improved if there is an integrated control of several intersections along the bus line, with bigger advantages obtained for bus movements, with less or similar delays imposed to other traffic flow. The intermittent bus lane strategy is more efficient in improving the bus flow than the ordinary two-lane traffic and maintaining the car flow at a higher level at the same time as the DBL, and the ordinary two-lane traffic suppresses the public transportation and is not advantageous at easing urban traffic congestion [27].

IBL was implemented in Lisbon, Portugal as the result of a protocol between the

University (where the concept was developed), the Municipality and the urban bus operator (CARRIS). During the six months of demonstration, starting from September,

2005, increase in bus average speed in all routes that use the target road link is 15 to 25% along with no significant impact in general traffic main attributes for instance, flow, vehicle speed and queues etc. In Melbourne, Australia, IBL was implemented under the

Dynamic Fairway project started in 2001 on Toorak Road and continues today. This 1.3- mile stretch is a two-lanes-each-way urban street that utilizes VMS signing and embedded LED flashing lights similar to Lisbon. The transit mode is a TRAM operating

in the center lanes as the restricted lane. Travel speed increases of 1 to 10 percent were reported [28].

Overall, the performance of the IBL influenced by the road configuration, transit mode, levels of congestion, and the newness of the technologies involved, as well as good driver compliance with transit lanes. Such as Longer lengths of IBL lane, buses will obtain more advantage than with shorter ones regarding bus time, but for shorter one it is possible to obtain the advantage activating the signals before or after the arrival of bus at entrance [27]. In general IBL systems has some limitations i.e., IBL applications do not work well in saturated traffic flow conditions at peak and it seems better suited for lower bus headways (three buses per hour) [29].

IBL variant is „„BLIP‟‟ whose elaboration if bus lane with intermittent priority which provides a compromise between dedicated bus lanes and buses operating in mixed traffic lanes. BLIP is similar to IBL, but it clears traffic out of the lane reserved for the bus and does not rely on TSP. Therefore, the BLIP concept is easier and less expensive to implement. With BLIP, other traffic can make use of the lane as normal. As a bus approaches, other vehicles are instructed to safely leave the lane (or are prevented from entering the lane), yielding right-of-way to the bus. Dynamic signage will communicate the status of the BLIP to other users of the roadway, potentially including overhead signalization, roadside signalization and in-pavement lights. The findings of this theoretical evaluation of BLIP are : 1) BLIP does not significantly reduce street capacity; however, it does increase average traffic density, including causing some delays to traffic, 2) delays are more than offset by the benefits that bus passengers receive, as long

as traffic demand does not exceed the maximum flow possible on non-special lanes, 3)

BLIP is not appropriate for roadways nearing or in excess of capacity, and 4) the main factors determining the time savings of BLIP are traffic saturation levels, bus frequency, bus travel time improvements from BLIP, and the ratio of bus-to-car occupancy flow [29,

30]. BLIMP concept that is modeled in Eugene, Oregon and unlike the IBL traffic is forced out of the lane when it is activated. This system causes decrease in the travel time

(60 sec / 14%) and improved reliability (17 sec / 28%) reduction in travel time standard deviation). These impacts are further enhanced with the minimal impact on intersection delay [31].

2.2.3. Review of Performance of Queue Jumpers

Federal Transit Administration (FTA) defines queue jump lane is a short stretch of bus lane combined with traffic signal priority which is often restricted for transit vehicles only. A queue jump lane is usually accompanied by a special queue jumper phase which allows vehicle in the queue jump lane to move ahead over other queued vehicles and can therefore merge into the regular travel earlier to the general traffic. This idea basically enables buses to by-pass waiting queues of traffic and to cut out in front by getting an early green signal. Knapp [32] found that right-turn lanes must be long enough to make the queue jumper lane more effective so that buses can bypass the entire queue in almost every instance. Head [33] mentioned some signal design considerations for queue jumper lanes without much detail.

Figure 2 shows the queue jumper lane at Charlotte, the largest urban area of North

Carolina for Independence Boulevard Bus way.

Figure 2 Queue jumpers at Charlotte, NC

Several studies have shown that TSP is ineffective during peak hours because very long queue gathered prior to intersection and buses are not able to bypass the long waiting queues and it is also difficult to justify the use of the exclusive bus lanes during peak hour as it covers one lane and cause fall in capacity. However queue jumper lanes has the likely to avoid these shortcomings as it consists of additional travel lane so bus can bypass the long queue and simultaneously it does not occupy the mixed traffic lane.

Hence queue jumper can reduce the bus travel time and can be a possible solution to assist in bus schedule adherence. Despite their likely advantages, queue jumper lanes have been the subject of a limited number of research studies so far.

Foremost importance of queue jumper lane is it can provide significant travel time savings for buses. Nowlin and Fitzpatrick [34] presented the findings from both field study and computer simulation (TexSIM) for a high volume intersection in Tuscon,

Arizona. The models included a far side open bus bay with and far side open bus bay without queue jumper. Reported travel time savings to buses over 180 m section ranged from 0 to 14 seconds with an average of 6.5 seconds. Their investigation showed that queue jumper lanes could provide significant time savings when traffic volume exceeded

250 vehicles per hour per lane (vphpl). However, when the through traffic volume exceeded 1,000 vphpl (i.e., near saturation), the benefit of queue jumper lanes began to decrease quickly.

Zhou and Gan [5], investigated impacts of queue jumpers with TSP for a VISSIM based network of three intersections and four bus stops. Authors evaluated the impacts of several traffic parameters as well as alternative TSP strategies on successful implementation of queue jumpers. Traffic parameters are TSP strategy, bus stop location, check-in detector location, bus headway, main-street through volume, and main-street right turn volume etc. Comparison was made for four scenarios: base scenario without queue jumpers, queue jumpers without TSP. queue jumpers with general TSP (green extension, red truncation, phase skip), queue jumpers with jumper TSP (green extension, red truncation, phase skip and, phase insertion). They reported that queue jumpers with jumper TSP effectively reduced bus delay by 8% to 34% better than other three scenarios. Authors also informed that the rate of the increase is marginal at low and medium demand levels (v/c <0.9) but increases quickly at high demand levels (v/c >0.9).

However, when the main street through v/c tends to 1, the standing queue will generate and block entry of buses to the queue jumper lanes. They presented that near side bus stops upstream of the check-in detectors for queue jumpers with jumper TSP can reduce bus delay by up to 14% when compared with far side bus stops. For far side bus stops, buses often cannot make use of the jumper phase to move through the queue jumper lanes completely because of dwell time variations and additional bus delays is generated when the bus yields to the mainline traffic. For nearside bus stops downstream of check-in detectors, the variations of bus dwell time may lead to missing or early activation of TSP strategies for buses. Detector location is an important issue to activate the TSP at right moment. The optimal detector location is about 500 feet away from the stop line for near side bus stop upstream of check-in detector and queue jumpers with jumper TSP. A shorter distance than the optimum location will not provide a sufficient time for signal phase transition, and distance longer than 500 feet, might cause that buses to miss the priority phase and would increase the delay [36]. Zhou and Gan [35] also reported that normal bus headways (>120 s) did not significantly affect the delay for buses and other vehicles. However, at high bus frequencies (bus headway <120 s), the delay for the entire intersection will increase because the high volume of buses will cause frequent TSP activate and simultaneously adverse impacts on cross street performance.

Lahon [37] investigated transit signal priority (TSP) and queue jumpers at six signalized intersections for a 2-mile long corridor for the Livermore Amador Valley

Transit Authority‟s Bus Rapid Transit (BRT) system in the City of Pleasanton, California using VISSIM. Model included TSP operation with right turn queue jumpers only and

TSP operation with both left turn and right turn queue jumpers. It was found that TSP and queue jumpers helped reduce bus travel time by 30 percent without adversely affecting automobile traffic in the corridor. Author recommended far side over near side bus stop so that buses can avoid nearside queues at intersection and can eliminate the needs to wait multiple signal cycle.

2.3. Overview on Calibration

Basic guidelines of calibration was given by Hellinga et a. [38], described a calibration process consisting of seven component steps: setting study goals and objectives, determination of requisite field data, choosing measures of effectiveness

(MOE) , establishing evaluation criteria, study network representation, modifying driver behavior parameters, and evaluation of model outputs. Park and Schneeberger [39] proposed a procedure which is a more detailed step-by-step approach, through linear regression model of calibration parameter. In the process of model calibration, model parameters are adjusted to obtain a (qualitative and quantitative) congruency between the model and field observed data. However this trial-and-error method of calibration is very time consuming. More systematic approaches include the gradient approach and Genetic

Algorithms (GA). These approaches accounts model calibration procedure as an optimization problem in which a combination of parameter values that best satisfies an objective function is searched. Kim et al. [40] applied GA to find suitable calibration parameters through Mean Absolute Error Ratio (MAER) using both CORSIM and

TRANSIMS, and reported the benefits obtained of using a GA for calibration. Cheu et al.

[41] applied a GA to find a suitable combination of FRESIM parameter values for a real

network. A similar effort was made by Lee et al. [42] to find suitable calibration parameter for PARAMICS which they reported reliable. However, most of these approaches used a few selected calibration parameters due to the complexity of the optimization surface because of the stochastic nature of the calibration parameters.

Zlatkovic et al. [2] presented a calibration procedure by comparing traffic counts for turning movements of intersections from real field and simulation network and this method was applied along 3500 S in West Valley City, Salt lake County, Utah and it showed co-efficient of determination more than 90%. This method is adopted for this research work.

2.4. Summary of the Literature Review

The first part of the literature review summarizes the previous studies on BRT, started with introduction of BRT, graze through communication technologies integrated with

BRT and later the successful implementation of BRT. Second part covers the overview of

TSP, several TSP strategies, benefits and disbenefits of TSP, sample example of successful TSP implementation etc. Third Part includes the overview of exclusive bus lanes i.e., dedicated bus lane, intermittent bus lane, its benefits and drawbacks etc. and also the previous studies on queue jumpers, what are the potentials of this strategy over other preferential treatments, what are the factors affecting the performance of queue jumpers etc. Fourth part summarizes several procedures to calibrate the models and the technique is adopted to calibrate the model of this research study.

3. MICROSIMULATION MODEL DEVELOPMENT

Proper model building is required to evaluate the impacts accurately. This chapter focuses the case study area and in depth discussion of model building procedure in microsimulation i.e., preliminary model building, signal setting, bus stop re positioning, formation of queue jumpers etc. Finally a summary of model building is mentioned.

3.1. Study Area

The first BRT line in the state of Utah was implemented along 3500 South in Salt

West Valley City, Salt Lake County, Utah. 3500 South is one of the major arterials which connect the fast growing western part of the county and Magna City with North – South highway and transit routes, such as I-15, I-215, Bangerter Highway, and light rail transit system (TRAX). It carries a significant amount of traffic, with Average Annual Daily

Traffic (AADT) between 33,000 and 51,000 vehicles per day.

Salt Lake County‟s transit agency, Utah Transit Authority (UTA), started a project called “MAX”, which refers to BRT implementations along the County are planed according to the 2030 Regional Transport Planning. Seven BRT projects are planned for the future. 3500 South corridor was chosen for the first BRT implementation, as it is one of the busiest routes for regular bus route RT 35. The 3500 South BRT line will run from

Magna to 3300 South TRAX station, covering 10 miles with 23 bus stops and would provide fast and reliable connection from Magna and West Valley to the TRAX line.

The old RT 35 continued to operate along with the BRT line, but less frequently with

30 to 60-minute headway but it would be off at Sunday. UTA bought ten new buses from

Belgian manufacturer, Van Hool and these buses are equipped with stainless steel frames and body panels, top mounted cooling systems, object detection systems, full low-floor boarding capabilities, center ADA boarding, wider isles, and more windows which had been assigned to the BRT line. Each bus has seats 60 passengers and boarding and alighting would be possible through any of the three doors and simultaneously. The buses had a new and unique paint scheme which will give identity to the new MAX system.

The BRT line will use fewer bus stops than RT 35. There will be twenty-three BRT stops in both directions along the line. For better accommodation and protection of the passengers each BRT stop will be sheltered and lighted. BRT stops will be equipped with passenger information displays.

To decrease bus stop dwell times and approve accessibility, Ticket vending machines were installed on BRT bus stops (the same machines already exist on TRAX stations).

Two types of ticket are available: one-way tickets ($2.00) or all-day passes ($5.00). After buying a ticket, passengers could board through any of the three doors and they would not need to show the driver their fare.

UTA‟s expectations were that the ridership will increase 18-20% after the BRT line implementation. After seven months preliminary survey was made which showed 33% increase in ridership. Figure 3 shows the whole route of the RT 35 and the BRT line with

BRT bus stop locations.

Figure 3 3500 South RT 35 and BRT route with BRT bus stop locations

3.2. Simulation Network

Test bed includes the demanding section of the new BRT corridor along 3500 South, from 2700 West to 5600 West Street, with a small detour from 2700 West to 2820 West, where the line makes a turn in order to service West Valley City‟s Valley Fair Mall. This busiest corridor is four miles long with thirteen signalized intersections along it shown in figure 4 shows the overall network consists of fourteen signalized intersection, as there is an additional signalized intersection in between West 3650 South and South 3200 West.

Figure 4 Study Areas along W 3500 S

3.3. Traffic Control

All the signalized intersections in test bed are part of a coordinated system except the intersection 3650 South and 2700 West which is a free-running intersection. Cycle lengths on intersections from 3450 West to 4000 West streets are 150 seconds except

Bangerter highway. For other intersections cycle length is 120 seconds except 3500

South - 4155 West and 3500 South - 5200 West intersection, where cycle length is 75 sec and 60 sec respectively. These data were obtained from the Utah Department of

Transportation (UDOT). Two UDOT‟s data sources were used: SYNCHRO files for PM peak period for this corridor, and I2 software which enables direct on-line connection to traffic controllers and downloading signal control information.

Traffic operations were modeled based on historical traffic data for the corridor

(traffic counts collected in recent years). Based on traffic counts, traffic was generated and distributed on the network using static assignment.

3.4. Transit Operations

The model has two transit lines: RT 35 and BRT line. RT 35 uses thirty-nine bus stops within the field of study, nineteen in eastbound and twenty in westbound direction.

Besides for the BRT line, there are ten bus stops, five in eastbound and five in westbound direction. Both lines are operated in mixed traffic condition lane, and they were simulated as that. All sort of transit operations, such as bus routes, locations of bus stops, time scheduling, bus ridership, passenger loadings on each bus stop and bus stop dwell times, for both transit lines and for both directions are considered here and simulated in such a way to get the exact real field condition. 31

Dwell time is a major for transit operation. Passenger boardings and alightings were used for simulating dwell times at bus stops. There are two ways to define dwell times in

VISSIM: Normal distribution and empirical distribution. A normal distribution is defined by the mean value and the standard deviation (in seconds). In VISSIM constant dwell time can be modeled by introducing standard deviation as zero and negative dwell time results from the normal distribution it is automatically cut to 0s. An empirical distribution is defined by providing a minimum and a maximum value and any number of intermediate points to build a graph of various shapes. Dwell time can be calculated based on passenger boardings, alightings and clearance time for buses [43].

This method is easier for determining the dwell time and incorporated with normal distribution in VISSIM. Boarding and alighting times depend on many factors, such as bus stop design, number of doors used for boarding or alighting and their width, bus floor height, payment process, etc. After reviewing data from FTA [44], boarding time was 4.0 seconds per passenger, alighting time 2.6 seconds per passenger and clearance time 10.0 seconds per bus stop was taken to build the model.

3.5. Preliminary Model building

For model building, VISSIM simulation software is used. VISSIM is a microscopic, time step and behavior based simulation model of urban traffic and public transit operations and it is easily programmable. The modeling process was started in VISSIM

Version 4.30, and continued in Version 5.30.

The simulation network includes the busiest section of the new BRT corridor in Salt

Lake city, Utah. The existing network was modeled, calibrated and validated based on 32

real data from the field, including network geometry, traffic and transit operations. For this project, VISSIM‟s features for transit operations modeling were very useful. It enabled the modeling of some basic parameters of transit, such as routes, transit stops, time scheduling, passengers movements (passenger arrivals at stops based on Poisson‟s distribution, passenger boarding for each stop, passenger alighting based on a user define alighting probability for each stop), stop sign, vehicle input, detector, signal head, and

TSP. The final output from this process was a validated and calibrated simulation model of the existing conditions for PM peak period (4 PM to 6 PM, with 15-minute build-up time).

3.6. Scenario Design

In order to see the advantages that this system provides, it needs to be compared to the existing system in multiple ways. In order to analyze all impacts of the new systems, three scenarios were compared:

 Scenario 1: No TSP with no queue jumpers (NTNQ), introduces line RT 35 and

its transit (both BRT and Bus) operations in mixed traffic lane. It does not

consider either queue jumpers or TSP.

 Scenario 2: No TSP with queue jumpers (NTQ), introduces queue jumpers along

with special queue jumper phase (8sec) but no TSP. BRT line and RT 35 is

shifted through queue jumpers. Queue jumper bays are developed for all

intersections except those where queue jumpers would not make any sense (i.e.,

left turns and unconventional intersections). Queue jumper phase is overlapped

with regular through moving phases. 33

 Scenario 3: TSP with no queue jumpers (TNQ), introduces TSP for all signalized

intersections along RT 35 and BRT line. Two types of TSP strategies are

implemented: green extension and red truncation, each of which has a maximum

time span of 10 sec. but no queue jumpers were implemented.

 Scenario 4: TSP with queue jumpers (TQ), introduces queue jumpers along with

special queue jumper phase and simultaneously with TSP. Implementation

scheme of TSP and queue jumpers are same as scenario 3 and scenario 2

respectively.

3.7. Queue Jumpers

FTA defines queue jump lane as a short stretch of bus lane combined with traffic signal priority. It enables buses to by-pass waiting queues of traffic and to cut out in front by getting an early green signal. A special queue jumper phase may be required.

Queue jump lanes are designed to facilitate straight-ahead movements through intersections or turning movements (left or right). Bus Rapid Transit Service Design

Guideline [45] mentioned five typical configurations of queue jumpers:

1) right-turn only lane with transit exemption

2) queue jump lane adjacent to right turn only lane

3) queue jump lane with advanced stop bar

4) curbside bus-only lane with transit exemption

5) curbside bus-only lane with “porkchop” island

Since queue jumper does not take a mixed-traffic lane and simultaneously it has a queue bypassing capabilities, it has the potential benefit of avoiding the shortcomings of both exclusive bus lanes and regular TSP. Figure 5 shows queue jumper in Ottawa.

Figure 5 Queue jumper lane (Ottawa) (Courtesy: Bus Rapid Transit service design guideline)

3.8. Queue Jumper Bus Bay

According to Transit Cooperative Research Program (TCRP) [46], bus bays can be considered for queue jumpers, particularly when right-turn only lane is used as queue jumper lane. These bus stops consist of a near-side, right-turn only lane and a far-side open bus bay. Queue jumpers bus bay provides benefit to buses, circumventing traffic congestion and private traffic operation, removing bus stoppage from the main traffic stream. Figure 6 shows a queue jumper bus bay.

Figure 6 Queue jumper bus bay

Figure 6 show that far-side bus bay consists of three parts: stopping area, acceleration lane, and exit taper and near-side right-turn only lane should be of minimum 240ft. In the existing network no change was made for entrance or exit queue jumper lane. The length of the stopping area is 50 feet for each standard 40-foot bus and 70 feet for each 60-feet articulated bus (Appendix A). 70 foot long stopping area was used as 60-foot BRT used the queue jumper. The length of acceleration lane and exit taper varies with respect to through speed (mph) and entering speed (mph) (Appendix A). Speed of the transit varies from 36.0 to 42.3 mph so the queue jumper lanes needed to be designed for 40mph through speed and 30 mph entering speed. However, due to space constraint queue jumper lanes were designed for 35 mph through speed and 25mph entering speed [46].

Queue jumper bus bays were for scenario TQ and NTQ and for those intersections where queue jumpers would not make any sense. Therefore, queue jumper lanes were added for nine signalized intersections excluding the intersection where BRT had to turn 36

left or right i.e., 3500S - 2820W, 3500S-2700W, 3650S-2700W etc. Moreover, far side bus bay and near side right-turn only lanes were added for only those intersection where existing condition didn‟t have those facilities. Table 1 enlists the existing scenario both for eastbound and westbound direction.

Table 1 List of intersections does not have entrance/ exit queue jumper lane

In table 1, “x” sign denotes the absence of the facility and “√” denotes the presence of the component. Three intersections has right-turn only lane in westbound and five intersections has right-turn only lane in eastbound direction but no intersection does not have far side bus bay in any direction. Moreover, three intersections have near side right- turn only lane both in westbound and eastbound direction.

Queue jumpers were developed in “TQ” and “NTQ” scenario following TCRP report. A sample of queue jumper bus bay is shown in figure 7 for 3500S-4000W intersection. Bottom part of figure 7 shows the intersection without queue jumper bus bay. Far side bus bay is highlighted by purple color. This intersection has one near side 37

bus stop (PT stop no.11) in westbound direction and one far side bus stop (PT stop no.25) in eastbound direction. Therefore, considering 35 mph through speed and 25mph entering speed the length of the far side bus bay should be 70 (stopping area length for articulated bus stop) + 250 (length of acceleration lane) + 170 (length of exit taper) = 490ft.

Figure 7 Intersection with or without queue jumpers

In some case the specification was not followed as there was not sufficient space to form the far side bus bay. Near side right-turn only lane was modeled for 3500S - Bangerter intersection. Length of the near side right-turn only lane in VISSIM model was 303.164 feet which is more than 240 feet compliant with the requirement. To facilitate queue jumper a special queue jumper phase was provided which was overlapped with regular through flow signal group 2 or 6 and the duration of special queue jumper phase was 8 seconds. Figure shows the screen shot of overlapped queue jumper phase.

3.9. Relocating Bus Stops

For TQ and NTQ scenario both BRT and RT 35 moved through queue jumpers.

Moving bus stoppages to queue jumpers provide advantage to general traffic as regular flow does not blocked by stopping buses. Bus stop should be placed to the restricted bus stop zone or no parking zone of street parking lane. Performance of queue jumper lanes is impacted by the location of bus stop and detectors [35, 36].

There are three main types of bus stop locations corresponding to the intersection: near side stops, far side stops and middle block stop. Far-side bus stop is located immediately after intersections, in the direction of bus travel. Near-side bus stops are located prior to intersections in the direction of bus travel. Mid-block bus stops are located at least 400 feet away from intersections. All three types of bus stops are shown in figure 8.

Figure 8 Different types of bus stops

Different agencies have different specifications for bus stop zone or no parking zone of street parking lane. PIMA County Department of Transportation‟s (PCDOT) “Transit

Guidelines for Roadway Design and Construction” [47] provides guidelines of the location and placement of bus stops within Pima County and the City of Tuscon

Metropolitan Area. According to PCDOT near side bus stop should be placed 30 to 100 feet upstream from beginning of curb return. If more than one bus uses a bus stop concurrently, 40 feet shall be added for each additional standard bus and 60 feet for each additional articulated bus. Moreover minimum 30 feet clearance should be provided between rear end of the bus to beginning of no parking zone. So the overall bus stop zone varies from 100 to 190 feet. Far-side bus stop should be placed 70 to 200 feet from the intersection curb return. A minimum 30 feet clearance should be provided between rear end of the bus to end of no parking zone. So the overall bus stop zone varies from 90 to

220 feet. Mid-block bus zone should be of 110 feet. If more than one bus uses a bus stop concurrently, 40 feet shall be added for each additional standard bus and 60 feet for each additional articulated bus. Figures of all types of bus stop zones according to PCDOT are shown in Appendix A. 40

OMNITRANS [48] have not provided any specific range of value for the placing of near-side bus stop from beginning of curb return. Additional 100 feet for low speed and low volume street and 120 feet for high speed and high volume streets should be added.

Far-side bus stop should be 100 feet and 120 feet from the end of the curb return for straight and turning movement respectively. A clearance of minimum 40ft for low speed and low volume street or 60ft for high speed and high volume street should be added from the front end of the bus to end of the no parking zone. Length of the middle-block bus stop should be minimum 140 feet and 180 feet for low speed and low volume streets and high speed and high volume streets respectively. Figures of all types of bus stop zones according to OMNITRANS are shown in Appendix B.

According to TCRP report 19 [46] the minimum length of the bus stop zones for near- side bus stops, far-side bus stop and, middle-block bus top is 100 feet, 90 feet and 150 feet respectively. (Appendix C)

In this study far-side bus stop was placed 70-200 feet from the end of the curb return and near-side bus stop was placed 30-100 feet from the beginning of curb return and the length of the bus stop was 60 feet considering 60 feet BRT. Bus stop position for far-side and near-side was remained same just they are moved to the queue jumper lane. But for few mid-block bus stops are moved to far-side bus stop zone and near-side bus stop zone regarding suitability otherwise it will affect the traffic flow impending to enter queue jumper lane. Table 2 enlists mid-blocks bus stops; need to be moved to the queue jumper lane.

Table 2 List of bus stoppage needs to be shift to queue jumper lane PT Stop No Intersection Previous type Present type 16 3500S -5600W Mid-block Far-side 17 3500S -5600W Mid-block Near-side 49 3500S - 4800W Mid-block Far-side 11 3500S - 4000W Mid-block Near-side 25 3500S - 4000W Mid-block Far-side 8 3500S - 3450W Mid-block Far-side 6 3500S - 3200W Mid-block Far-side

3.10. Transit Signal Priority Implementation

TSP Handbook defines Transit Signal Priority (TSP) [18] as cost-effective method to enhance regional mobility by improving transit travel times and reliability, providing vehicles a little extra green time or a little less red time at traffic signals to reduce the time they are slowed down by traffic signals. The objective of transit signal priority strategies is the reduction of delay for transit vehicles at signalized intersections. In recent years, a number of studies have attempted to evaluate TSP using either empirical, analytical, and/or simulation tools. Because of the costliness of its field implementation

TSP impact analysis relies greatly on simulation software packages such as VISSIM

(PTV 2003), CORSIM (FHWA, 2003), and PAMRAMICS (Quastone 2004) etc.

In this research work VISSIM was used to implement two TSP strategies: green extension and red truncation/ early green. TSP can be implemented using Ring Barrier

Controller (RBC) emulator. Figure shows a screenshot of Ring Barrier Controller (RBC) emulator. To implement TSP properly few terminology needs to be known: parent SG, priority mode, extend time, recovery minimum green, call transit SG, travel time, travel slack time, presence, check-in, and check-out detector etc.

Parent SGs are the signal groups that the overlap will be allowed to time with. When one parent signal group is timing and another parent signal group is next, the overlap will remain green (unless a negative vehicle or pedestrian signal group is next). When the last parent signal group terminates, the overlap will also terminate. For calling a transit SG call mode is used. A transit SG will not turn green unless it receives a call. The call modes are: recall, locked, non-locked and soft recall. Priority requests of transit SG varies with respect to estimated travel time. A transit priority requests with lower estimated travel times are served ahead of transit priority requests with higher estimated travel times. For higher priority transit movements, the Priority should be defined to a higher value than that of lower priority transit movements. Priority of a transit SG can be: none, early/ extend and extend only. For this project “Early/ Extend” option was used. How long the transit SG will get priority defined by “Extend limit” [43].

In this research work transit signal priority was provided for 12 intersections except

Bangerter and 3650 South – 3200 West and maximum duration for early green and red truncation was adopted 10 seconds using “Extend Limit” option from RBC. TSP was activated for only BRT and detector type “Presence” was used to call transit SG 301 –

308 [43].

3.11. Queue Jumper Phase Implementation

Queue jumpers are generally accompanied with special signal setting which allows transit to cut out the intersection and pass ahead of the other vehicles. The duration of the queue jumper phase was 8 seconds. This phase was overlapped with regular through phase in East – West direction. VISSIM was equipped in such a way that when the BRT

was sensed by detector it called overlapping phase. Detector was positioned on the queue jumper lane. When BRT was sensed by this detector it called overlapping phase.

3.12. Summary of Model Development

This chapter explains a methodology to build a model and designing scenarios to evaluate the impacts. Besides it provides guidelines of how to set the queue jumpers, special queue jumper phase, and bus stops zoning. The next will chapter presents results for the model calibration and validation.

4. CALIBRATION AND VALIDATION OF THE MODEL

This chapter includes the calibration and validation procedure of the VISSIM model to obtain a model close to the real field. The chapter will explain the driver behavior parameters need to be checked and what parameter need to be compared with the model output. Without calibration and validation model will not be able to achieve a significant level of confidence.

4.1. Why necessary to calibrate?

Calibration is a process where the modeler selects the correct model parameters to reproduce the observed traffic conditions as accurately as possible to the real field and this process is called "calibration". The calibration of microscopic simulation models has received widespread attention in the transportation and traffic engineering professions because of the use of simulation models in operations and planning applications.

This process is a trial and error effort that sets the parameter values within an acceptable tolerance of error. After obtaining satisfactory estimates of the parameters for all models, the models must be checked to assure that they perform the functions for which they are intended like traffic volumes on transit and roadways adequately and accurately. VISSIM uses car-following, lane-changing parameters for calibration. The driver behavior of this model is dependent on parameters whose values can be modified.

The software has a set of default values for these parameters which cannot represent all possible traffic conditions. The objective is to modify these parameter values such that the simulation software can best reproduce the driver behavior and replicate traffic performance of the study location.

4.2. Overview of VISSIM

The simulation model used in this research was VISSIM, version 5.30. VISSIM is a microscopic, time step, and behavior-based simulation model. The model was developed at the University of Karlsruhe, Germany, during the early 1970s and the commercial distribution of VISSIM was launched in 1993 by PTV Transworld AG. In the United

States, ITC Inc. distributes and supports the program. As a microscopic simulation it can simulate each individual entity (vehicle, pedestrian etc.) considering each relevant property. VISSIM uses a discrete time step based model that runs at fixed time intervals.

The time interval is user-definable and ranges between 0.1s and 1s. The heart of the

VISSIM is car following model of Wiedemann, a psycho-physical car following model for longitudinal vehicle movement and a rule-based algorithm for lateral movements.

Several important technical features about VISSIM are detailing network, vehicle and pedestrian assignment, control strategies and user interfaces. It has the capability of simulating traffic networks including freeways, arterials and pathway for pedestrians. An additional feature of VISSIM is that it can simulate transit and multimodal operations including pedestrians under constraints such as traffic composition, lane configuration, traffic signals, transit stops, etc. Therefore, VISSIM is the ideal tool to simulate different traffic scenarios before starting implementation. It thus allows finding a solution which

takes traffic and transportation quality, safety and cost into consideration. As VISSIM provides state-of-the-art presentation options, even 3D animations, besides of transportation professionals, more and more decision makers and local authorities are choosing VISSIM to convincingly show how effective a projected measure might be, regardless of whether a new road is going to be constructed or a new tram line is being planned. VISSIM therefore offers the unique opportunity to integrate citizens into the decision-making process [34].

4.2.1. VISSIM Car Following Parameters

VISSIM uses the psychophysical driver behavior model developed by Wiedemann

[43]. The assumption of Wiedemann model is that a driver can be in one of four driving modes

 Free driving: Free driving means no influence of preceding vehicles observable.

And in this mode the driver wants to reach and maintain a certain desired speed

which cannot be kept constant, but oscillates around the desired speed.

 Approaching: The process is actually adapting of driver‟s own speed to the lower

speed of a preceding vehicle. A driver applies a deceleration so that the speed

difference of the two vehicles turns to zero in the moment he reaches his desired

safety distance.

 Following: The driver follow the preceding car without any conscious change of

acceleration in speed and maintain a safe distance more or less constantly.

However due to imperfect throttle control and imperfect estimation the speed

difference oscillates around zero. 47

 Braking: Braking allows medium to high deceleration rates if the distance falls

below the desired safety distance. This can happen if the preceding car changes

speed abruptly, or if a third car changes lanes in front of the observed driver.

The concept behind the psycho-physical model used in VISSIM is that the drivers have a desired speed at which they would like to travel when they are not constrained by other vehicles or signals or any other factors. The driver starts to decelerate as he approaches a slow-moving vehicle or a traffic signal. Figure 9 depicts the car-following logic and shows the thresholds, characteristic distances and associated driving procedures for a vehicle-driver-unit. The horizontal axis represents the speed difference with positive values characterizing a closing process where the vertical axis represents the distance to the vehicle ahead.

Figure 9 Car following logic (Courtesy: VISSIM manual 5.3)

AX = desired distance for standing vehicles (front-to-front distance).

BX = speed dependent term in the desired minimum following distance

ABX = desired minimum following distance = AX + BX.

SDV = perception threshold of speed difference at long distances.

SDX = perception threshold of growing distance in following process.

CLDV = perceptual threshold for recognizing small speed differences at short, decreasing distances.

OPDV = perceptual threshold for recognizing small speed differences at short but increasing distances [43].

Distributions of the speeds and spacing thresholds characterize the variety of driver behavior characteristics. These distributions were obtained by multiple field measurements at the University of Karlsruhe, Germany. Wiedemann 1974, model mainly suitable for urban traffic, car following model has the following parameters:

1. Average standstill distance (ax) defines the average desired distance between

stopped cars. It has a fixed variation of ± 3.28 ft. Default average standstill

distance is 6.56 ft.

2. Additive part of desired safety distance (bx_add) and Multiplic. part of desired

safety distance (bx_mult) affect the computation of the safety distance. The

distance d between two vehicles is computed using this formula:

d = ax + bx

Where ax is the standstill distance

bx = (bx _ add + bx _mult * z) * v

Where v is the vehicle speed [m/s], z is a value of range [0, 1] which is normal

distributed around 0.5 with a standard deviation of 0.15. Default value for

Additive part of safety distance and Multiplic. part of safety distance is 2.00 and

3.00. [43]

4.2.2. Lane change

There are basically two kinds of lane changes in VISSIM: Necessary lane change (in order to reach the next connector of a route) and Free lane change (because of more room

/ higher speed). In case of a necessary lane change, the driving behavior parameters

contain the maximum acceptable deceleration for the vehicle and the trailing vehicle on the new lane but they count the distance to the emergency stop position of the next connector of the route. In case of a free lane change, VISSIM checks for the desired safety distance of the trailing vehicle on the new lane. This safety distance depends on its speed and the speed of the vehicle that wants to change to that lane. In both cases the first step is to find a suitable gap (time headway) in the destination flow which dependent on the speed both of the lane changer and the vehicle that “comes from behind”. In case of a necessary lane change it is also dependent on the deceleration values of the

“aggressiveness” [43].

The following parameters are considered for adjustment in order to closely match lane change values available from the field and simulation model:

 Waiting time before diffusion - defines the maximum amount of time a vehicle

can wait at the emergency stop position waiting for a gap to change lanes in order

to stay on its route. When this time is reached the vehicle is taken out of the

network (diffused). Default value is 60 seconds.

 Min. Headway (front/rear) - defines the minimum distance to the vehicle in front

that must be available for a lane change in standstill condition. Default value is

1.64ft.

 Safety distance reduction factor – takes effect for the safety distance of the

trailing vehicle in the new lane for the decision whether to change lanes or not,

the vehicle‟s own safety distance during a lane change and the distance to the

leading lane changing vehicle. Default value is 0.6 [43].

In this research, waiting time before diffusion is lowered from default 60 to 5 seconds for the link after intersection 3500 S – 4000 W. As there is an on-street bus stop after the intersection so there is possibility that it would obstacle the regular movement of the general traffic flow. In order to get rid of this waiting time before diffusion is lowered so that vehicle does not need to wait long time for a gap to change the lane. Figure shows the

4.2.3. Driving Behavior Parameters for the Model

The driver behavior in VISSIM was modeled through the car following and the lane change models. The driving behavior is linked to each link by its link type. For each vehicle class, a different driving behavior parameter set may be defined. By default, five parameter sets are predefined. These are shown in Figure 10 (numbers 1 to 5). Thus, the parameters described in this section apply equally to all vehicle types, but were adjusted for each link type. New links with modified driver behavior parameters was defined for the reproduction of existing traffic conditions. These link types are shown in Figure 10 and are numbered 6 to 9. Based on the type of link, driver behavior was modified using the position of the driver/vehicle in the network. No correlation between vehicle type and the driver behavior. VISSIM includes two car-following models – urban driver and freeway driver. Only the urban driver type was used.

Figure 10 Behavior parameter sets for drivers in VISSIM

The car-following mode of the freeway driver model includes 3 parameters: average standstill distance, additive part of desired safety distance, and Multiplic. Part of safety distance. For instance prior to intersection 3500 S – 5600 W at WB direction link behavior of approaching links are connected with driver behavior type “Modified

Following” newly created driver behavior for which waiting time before diffusion was reduced to 5 sec from 60 sec cause after that intersection there was on-street bus stop which obstacles the movement but the vehicle needed to move there fast.

4.3. Model Calibration

The basic existing network model had to be calibrated. Calibration was based on the traffic data collected in the field. Model calibration was performed based on traffic

movement counts for each signalized intersection in the network. Most of the traffic counts were collected in 2006, except for two intersections (3500 South and 3450 West in

2007, and 3650 South and 2700 West in 2008). VISSIM was programmed to collect the same data on all 13 signalized intersections. Calibration was performed by comparing data from the field counts to the data from the simulation.

Figure 11 shows this comparison of traffic movement count from simulation and real friend from 4pm-5pm. The R Square value was 0.948 which shows high correlation between the two data sets.

3000

y = 0.938x + 18.755 2500 R² = 0.948

From From the 2000

/h)

5 5 PM

– veh ( 1500

1000

Simulation Simulation Movements Movements 4 of

500 Traffic Traffic

0 0 500 1000 1500 2000 2500 3000 Traffic Movement Counts of 4 – 5 PM from Real Field (veh/h)

Figure 11 Model calibration results – traffic movement comparison Moreover, figure 12 shows this comparison of traffic movement count from simulation and real friend from 5pm-6pm. The value of coefficient of determination (R2) is 0.944 which shows high correlation between the two data sets.

3000

y = 0.938x + 2.552 2500

R² = 0.944 the the

2000

/h)

6 PM From 6 From PM

– veh

( 1500 of 5 5 of

1000 Simulation Simulation

500 Traffic Movements Movements Traffic

0 0 500 1000 1500 2000 2500 3000

Traffic Movement Counts of 5 – 6 PM from Real Field (veh/h)

Figure 12 Model calibration results – traffic movement comparison

4.4. Validation

Validation is a process where the modeller checks the overall model outputs for observed values of traffic performance, e.g. traffic flows, travel times, speeds, queues and delays. This uses data not used in the calibration process and it can be described as an independent check of the calibrated model.

4.4.1. Model Validation

The corridor along 3500 South, from 2700 West to 5600 West, was split up into twenty-two segments (eleven in EB and eleven in WB direction). These segments are parts of the corridor between each pair of signalized intersections. Travel times for each segment were measured in the field using GPS in PM peaks. Travel time measuring

points in VISSIM were set for the same segments. Travel times from the field were used to validate those from the model.

Figure 13 shows comparison of travel times after the validation was completed.

Depending on the random seed that is used in the simulation, R Square value for travel times varies between 92% and 98 %.

180.0

160.0

140.0 in 120.0

(sec) 100.0

80.0

60.0

Westbound Westbound 40.0

Average Travel Time Time Travel Average 20.0

0.0

Intersections

VISSIM GPS

Figure 13 Model validation results – travel time comparison

120.0

100.0 in

80.0 (sec) 60.0

40.0 Eastbound

20.0 Average Travel Time Time Travel Average

0.0

Intersections

VISSIM GPS

Figure 14 Model validation results – travel time comparison

4.4.2. Summary

For calibration turning movement were used as key parameter to compare with the real filed and for validation travel time for different travel time segments were compared with the real filed data. Driver behavior was changed for few links other defaults values were used. The Coefficient of determination for calibration and validation revealed that the model was good enough to show the characteristics of the real filed.

5. RESULTS AND DISCUSSION

The queue jumper lanes were created as described in the previous chapter. In order to see the advantages of this system, it was needed to be compared with the existing system in multiple ways. In order to analyze all impacts of the new systems, four scenarios were compared for different criteria. This chapter includes the details comparison and analysis to evaluate the impacts properly.

5.1. BRT Travel Times

The main goal of the project is to evaluate the performance of queue jumper lanes for

BRT operations. The most visible parameter affected by these implementations was transit travel time. Travel time data was obtained from VISSIM for ten random scenarios.

Table 3 shows comparisons of private vehicular travel time for four scenarios.

Table 3 : BRT travel times comparison Directions Intersections NTNQ TNQ NTQ TQ 2700W-2820W 166.6 146.4 165.6 147.1 2820W-3200W 98.5 45.8 83.1 54.0

3200W-3450W 44.0 47.2 36.4 42.1

3450W-3600W 31.8 35.1 37.0 30.3 3600W-Bangerter 86.6 87.3 77.4 77.3 Bangerter-4000W 81.8 107.5 83.3 99.3 4000W-4155W 31.7 29.3 28.0 23.8

WESTBOUND 4155W-4400W 42.7 41.4 39.3 34.3 4400W-4800W 70.4 59.8 62.3 57.8 4800W-5200W 85.5 80.8 85.6 82.4 5200W-5600W 180.4 153.1 170.3 146.0 5600W-5200W 63.4 50.2 55.5 51.8 5200W-4800W 88.9 59.1 82.7 53.0 4800W-4400W 82.0 78.7 83.7 79.0 4400W-4155W 34.8 35.5 32.5 33.2 4155W-4000W 48.0 35.7 44.3 37.1 4000W-Bangerter 105.6 108.1 88.3 95.2 Bangerter-3600W 49.1 37.8 53.6 39.5

EASTBOUND 3600W-3450W 56.9 58.8 58.7 58.3 3450W-3200W 97.2 77.8 57.7 53.5 3200W-2820W 56.3 78.9 54.3 61.8 2820W-2700W 323.9 261.2 227.7 221.2 Westbound 915.6 835.7 868.6 796.3 Total Eastbound 1005.0 887.3 841.4 793.0

Figure 15 and figure 16 shows difference in travel times for BRT along the segments for BRT. Through the segment 5200W-5600W and 2820W-2700W travel time is significantly higher than any other segment for all four scenarios.

Figure 15 Travel time comparison of BRT in EB

Figure 16 Travel time comparison of BRT in WB

5.2. Private Vehicular Travel Times

Besides of providing facility for improvement of BRT operation another vital thing is travel time for private vehicles so that it might not adversely impacted. In order to evaluate the impact of queue jumpers and TSP on private vehicular travel times along

3500 South from 2700 West to 5600 West a comparison between travel times for the four scenarios was made. Table 4 shows these comparisons for private traffic travel time.

Table 4 Private traffic travel times comparison Direction Intersections NTNQ TNQ NTQ TQ 2700W-2820W 16.3 16.9 16.3 17.1 2820W-3200W 63.9 64.7 68.1 67.5

3200W-3450W 48.0 48.0 48.0 48.0

3450W-3600W 44.2 43.2 48.8 46.7 3600W-Bangerter 46.1 46.6 47.2 50.3 Bangerter-4000W 46.8 45.6 50.1 49.1 4000W-4155W 31.8 30.9 33.7 33.6

WESTBOUND 4155W-4400W 41.2 41.3 41.5 41.4 4400W-4800W 71.6 70.2 71.2 70.7 4800W-5200W 53.9 53.6 53.3 53.4 5200W-5600W 191.6 183.4 173.0 189.8 5600W-5200W 60.0 58.7 56.7 55.9 5200W-4800W 82.0 76.1 77.4 73.7 4800W-4400W 53.5 51.8 54.5 51.7 4400W-4155W 36.7 37.0 36.7 36.7 4155W-4000W 38.5 37.4 38.7 37.0 4000W-Bangerter 60.7 62.4 60.9 61.1 Bangerter-3600W 42.9 43.0 62.2 61.0

EASTBOUND 3600W-3450W 23.5 23.2 32.8 32.3 3450W-3200W 92.3 84.0 89.1 84.4 3200W-2820W 64.3 79.1 65.6 71.4 2820W-2700W 68.3 74.2 64.0 69.0 2700W-5600w 651.7 640.0 647.2 660.7 Total 5600W-2700W 625.3 627.7 641.2 638.5 Figure 17 and figure 18 show these comparisons of eastbound and westbound private traffic travel time in each segment for four scenarios. 61

Figure 17 Travel time comparisons for private traffic in EB

Figure 18 Travel time comparisons for private traffic in WB 5.3. Bus Travel Times

Implementation of queue jumpers with TSP has large impact on existing RT 35. RT was not activated for TSP however route was modified along queue jumpers when queue 62

jumpers were implemented on the system. Alike BRT, RT 35 received benefit using queue jumpers to bypass the queue. It might get benefit from the TSP depending on arrival time. Table 5 shows RT 35 travel times comparison. This comparison was also made for the four scenarios of ten random seeds.

Table 5 Bus travel times comparison Intersections NTNQ TNQ NTQ TQ 2700W-2820W 169.4 168.3 168.6 168.5 2820W-3200W 102.5 101.9 86.8 84.4 3200W-3450W 104.2 109.9 97.2 102.0 3450W-3600W 53.0 48.7 54.5 59.4 3600W-Bangerter 49.7 45.6 47.0 52.2 Bangerter-4000W 56.1 64.2 54.5 56.0 4000W-4155W 35.4 36.2 31.1 31.6

4155W-4400W 47.3 42.7 41.3 40.4 WESTBOUND 4400W-4800W 82.2 89.6 70.9 76.5 4800W-5200W 52.8 52.7 53.6 52.2 5200W-5600W 189.9 169.9 142.8 158.3 5600W-5200W 62.3 58.1 55.1 52.1 5200W-4800W 87.6 79.7 83.8 67.5 4800W-4400W 72.6 64.2 70.5 70.8 4400W-4155W 37.1 36.1 36.6 37.4 4155W-4000W 84.6 71.5 89.7 79.2 4000W-Bangerter 66.2 79.1 50.7 72.0

Bangerter-3600W 39.8 41.8 49.4 47.7 3600W-3450W 38.6 38.3 42.1 44.1 3450W-3200W 112.4 101.5 78.6 79.1 3200W-2820W 90.6 107.8 88.8 97.4

2820W-2700W 282.4 280.3 203.7 213.2 EASTBOUND Westbound 945.1 929.9 850.7 878.3 Total Eastbound 976.3 960.4 851.1 862.8

Figure 19 and figure 20 show these comparisons of eastbound and westbound private traffic travel time in each segment for four scenarios

Figure 19 Travel time comparisons for bus in EB

Figure 20 Travel time comparisons for bus in WB

5.4. Overall Comparison of Travel Times

For the existing scenario NTNQ the travel time for BRT is 915.6 sec and 1005.0 sec in WB and EB direction respectively. However Minimum travel time for BRT is 796.3 and 793.0 sec in WB and EB direction respectively for TQ scenario. The reduction in travel time is between 13.0- 22.0 % which shows much improvement. The reasons behind of it are the implementation queue jumpers to bypass the queue, special queue jumper phase which duration is 8sec and of transit signal priority to provide early green or green extension for BRT. When TNQ is compared with NTNQ scenario travel time reduction is

1.60-1.63% in WB and EB direction. Parallel analysis shows that NTQ reduces the travel time in between 8-12% for BRT in both directions.

For NTNQ private traffic average travel time is 651.7 sec and 625.3 sec in WB and

EB direction respectively. Among four scenario private traffic travel time is highest in westbound direction for TQ and in eastbound direction for NTQ. The reason behind of this is special queue jumper phase and TSP. Travel time is increased 1.4% in TQ and

2.1% in NTQ.

For the existing scenario NTNQ the average travel time along for RT35 is 945.1 sec and 976.3 sec in WB and EB direction respectively. However Minimum travel time for

RT 35 is 850.7 sec and 851.1 sec in WB and EB direction respectively for NTQ scenario.

In NTQ scenario, the reduction in travel time is 9.0-13.0%. When TNQ is compared with

NTNQ travel time reduction is 1.60-1.63%. Parallel analysis shows that in TQ scenario reduction in travel time is in between 7.0-11.7%.

Figure 21 and figure 22 show these comparisons of travel time in WB and EB direction for private traffic, RT 35 and BRT for all fours scenarios.

1200.0 (sec) 1000.0

800.0 in Westbound Westbound in 600.0

400.0

200.0 Average Travel Time Time Travel Average 0.0 NTNQ NTQ TNQ TQ

Scenario Private Vehicle Bus BRT

Figure 21 Overall travel time comparisons in WB

1200.0 (sec) 1000.0

800.0 in Eastboundin

600.0

400.0

200.0 Average Travel Time Time Travel Average 0.0 NTNQ NTQ TNQ TQ

Scenario Private Vehicle Bus BRT

Figure 22 Comparison of travel time in EB

5.5. Network Performance Evaluation

Impacts and benefits of the different scenarios can be assessed on a network wide level. “Network Evaluation” tools in VISSIM provides output of several parameters i.e., average stopped delay, average speed, average delay, and average number of stops etc. aggregated for the whole simulation run and whole network.

5.5.1. Average Stopped Delay

For private vehicle average stopped delay increases to 9.64%, and 16 .26% respectively for NTQ and TQ phases compare to the existence scenario. But for bus and

BRT significant improved is found. Average stopped is reduced to 7.03% and 25.76% for bus in TNQ and TQ phases. For BRT total reduction is 50.40% to 57.70% for TNQ and

TQ scenario. Overall network average stopped delay for private vehicle, bus and BRT is

105.3s, 108.6s and 121.7s respectively. Figure 23 shows the comparison of average stopped delay for all four scenarios.

300.0

250.0

200.0

150.0

100.0

Per Vehicle (sec) Vehicle Per Average Stopped Delay Stopped Average 50.0

0.0 NTNQ NTQ TNQ TQ Scenario Private Vehicle RT 35 BRT

Figure 23 Comparison of network wide stopped delay 68

5.5.2. Average Speed

Comparison of average speed of private vehicle, bus and BRT is shown in figure 24.

Average speed for private vehicle, bus and BRT is 16.9 mph, 16.2 mph and 15.8 mph for

NTNQ phase. For private vehicle average speed decreases to 16.3 mph, 16.7 mph and

15.9 mph respectively for NTQ, TNQ and TQ phases. But for bus and BRT significant improved is obtained. Average speed is increased to 17.7 mph, 16.8 mph and 17.8 mph for bus in NTQ, TNQ and TQ. For BRT speed increases to 17.4 mph, 18.3 mph and 19.3 mph for NTQ, TNQ and TQ. Overall network average speed for NTNQ, NTQ, TNQ and

TQ scenario is 16.9 mph, 16.3 mph, 16.7 mph and 15.9 mph respectively.

25.0

20.0

15.0

10.0 Average Speed (mph) Speed Average 5.0

0.0 NTNQ NTQ TNQ TQ Scenario Private Vehicle Bus BRT

Figure 24 Comparison of network wide speed

5.5.3. Network Wide Comparison of Average Delay

Comparison of average delay of private vehicle for different scenarios is shown in figure 25. Average delay for private vehicle is 178.9 sec, 191.3 sec, 183.4 sec and 201.3 sec for NTNQ, NTQ, TNQ and TQ scenario. For TQ network wide average delay for private vehicle is increased 12.52%. However for TQ network wide average delay for

BRT was reduced about 33%.

650.0 600.0 550.0

500.0 ) 450.0 400.0 350.0 Delay Delay (sec 300.0 250.0

200.0 Average Average 150.0 100.0 50.0 0.0 NTNQ NTQ TNQ TQ Scenario Private Vehicle RT 35 BRT

Figure 25 Comparison of network wide delay

5.6. Average Cross Street Delay

When TSP and queue jumper lane is provided along the main corridor, some impacts on side streets traffic are expected. The TSP strategies (Green extension and red truncation) facilitate transit operations along the main corridor, which increase delays for the traffic on the side streets. Comparison is shown in figure 26. For 3500 S-5600 W,

3500 S-5200 W, 3500 S-4155 W, 3500 S-3450 delay varies in between 0-10%, for W

3500 S-4000 W delay varies between 10-20%, for 3500 S-4800 W, 3500 S-4400 W, 3500

S-3600 W delay varies between 20-30% and for 3500 S-3200 W delay is increased about

32%.

Figure 26 Comparison of cross-street delay

5.7. Impacts on Intersections on Major Corridor

Intersection performance was measured for percentage of stopped BRT vehicles at red light and BRT waiting time at red light at intersection. Comparison is shown in table

6 and in table 7:

Table 6 Comparison of waiting time and percentage of stops along mina corridor in WB Base QJ TSP QJ & TSP WB Stop (%) WT (s) Stop (%) WT (s) Stop (%) WT (s) Stop (%) WT (s) 3500 S-2700 W 89 494 100 313 67 204 78 223 3650 S-2700 W 0 0 33 57 0 0 33 52 3500 S-2820 W 100 424 100 340 89 230 89 181 3500 S-3200 W 100 386 100 272 0 0 33 80 3500 S-3450 W 22 51 11 2 33 32 33 62 3500 S-3600 W 0 0 22 50 56 71 44 92 3500 S-Bangerter Hwy 44 154 44 148 78 271 22 63 3500 S-4000 W 22 21 22 84 22 16 0 0 3500 S-4155 W 22 14 0 0 33 7 0 0 3500 S-4400 W 33 26 0 0 0 0 0 0 3500 S-4800 W 33 64 33 57 11 11 0 0 3500 S-5200 W 0 0 0 0 0 0 11 1 3500 S-5600 W 100 740 89 574 78 232 78 226 Average/Total 44 2374 43 1897 36 1074 32 980

Table 7 Comparison of waiting time and percentage of stops along mina corridor in EB Base QJ TSP QJ & TSP EB Stop (%) WT (s) Stop (%) WT (s) Stop (%) WT (s) Stop (%) WT (s) 3500 S-5600 W 100 382 100 426 0 0 100 131 3500 S-5200 W 86 25 0 0 0 0 14 1 3500 S-4800 W 100 174 100 214 57 46 0 0 3500 S-4400 W 0 0 0 0 0 0 0 0 3500 S-4155 W 0 0 0 0 0 0 14 4 3500 S-4000 W 43 143 43 80 43 57 57 100 3500 S-Bangerter Hwy 71 173 43 155 86 237 86 305 3500 S-3600 W 57 94 86 210 29 24 43 19 3500 S-3450 W 29 43 29 48 0 0 0 0 3500 S-3200 W 86 318 43 101 57 110 14 45 3500 S-2820 W 0 0 0 0 57 52 14 14 3650 S-2700 W 86 802 86 327 100 713 100 198 3500 S-2700 W 0 0 0 6 14 1 0 0 Average/Total 51 2154 41 1561 34 1240 34 817

5.8. BRT Time-Space Diagrams

VISSIM has the option to record BRT positions and link coordinates for every simulation step. These data were used to plot and compare BRT vehicle trajectories for the four scenarios. There were eleven BRT vehicles in the WB direction, and eight BRT vehicles in the EB direction that started and completed their trips during the evaluation interval. The example diagram for one randomly seeded simulation is given in Figure 27.

The diagram shows three consecutive westbound BRT vehicles for the four scenarios and their progression between 2700 S and 5600 S intersections.

Figure 27 Sample BRT time-space diagram

5.9. Summary of Overall Result Analysis

Overall result analysis revealed mixed upshot. TQ scenario had positive impacts along the main corridor as it reduced BRT and bus travel time and not much negative impacts on private traffic flow. However it has adverse impacts on overall private traffic as well as cross streets.

6. CONCLUSIONS

The first part of this chapter presents the key findings of this research study. The second part provides limitations of the research and suggests potential ideas for future research.

6.1. Conclusions

Based on the results presented in Chapter 5, major findings are mentioned in following:

 For TQ scenario average travel time for BRT was reduced about 13-22% however

the travel time for private traffic along W 3500 S increased only 1.3 – 2.1%.

 Implementation of only queue jumpers or only TSP didn‟t show significant

difference in reduction of travel time as reduction varies 5-16% for NTQ and 9-

12% for TNQ. However implementation of queue jumpers and TSP together

provided benefits almost twice than the individual implementation.

 For buses along RT 35 travel time was minimal for NTQ scenario and reduction

varies between 9 to 12% almost close to the reduction achieved in TQ scenario (7-

11%). Therefore buses didn‟t achieve much benefit from the implementation of

TSP.

 Analysis on the impacts on individual intersections along the main corridor

showed that for TQ scenario PT waiting time at red phase and no of stops were

minimal.

 When the overall network was considered for TQ average delay for private traffic

was increased up to 12.5% which was almost close to individual implementation

of queue jumpers or TSP. however average delay for BRT was reduced upto

33.5%.

 Adverse impact was observed on performance of cross-street of main corridors

specially where queue jumpers were implemented. The range varied in between 0

to 33%.

6.2. Limitations of the study and future Research Work

Result analysis showed that implementation of queue jumpers along with TSP can provide benefit to reduce travel time without causing adverse impacts on private traffic movements along main corridor. However implementation of queue jumpers is very costly. In this research work queue jumper bus bay was modeled for 9 intersections. This bus bay consists of near-side, right turn only lane and far side bus bay. However in real filed no intersections in both direction have far-side bus bay and only three intersections in WB direction and five intersections in EB direction has near-side right turn only lane.

Therefore massive amount to money will be required to construct or re designing of queue jumpers. Therefore future research work can emphasize on cost benefit analysis of queue jumpers along with TSP.

Another important issue is TSP. In this research work only two active TSP strategies was considered: green extension and red truncation. However several strategies reported that when phase insertion is included in TSP strategy it shows better performance.

Therefore future research work can focus on the implementation of other two TSP strategies.

APPENDIX A

STANDARD BUS STOP LOCATION ACCORDING TO PCDOT

APPENDIX B

STANDARD BUS STOP LOCATION ACCORDING TO OMNITRNAS

(A) Typical dimension of far side bus stop

(B) Typical dimension of near side bus stop

APPENDIX C

TYPICAL DIMENSION OF ON-STREET BUS STOP ACCORDING TO TCRP

BIBLIOGRAPHY

1. Schrank, D. and T. Lomax. 2005. The 2005 Urban Mobility Report. College Station,

TX: Texas Transportation Institute, Texas A&M University System.

2. Zlatkovic, M., A. Stevanovic and P.T. Martin and I. Tasic. 2012. Evaluation of

Transit Signal Priority Options for the Future 5600 W Bus Rapid Transit Line in

West Valley City, UT. Transportation Research Record, Journal of the

Transportation Research Board, in print.

3. Valley Transportation Authority. 2007. Bus Rapid Service Design Guideline.

4. Garrow, M. and R. Machemehl. 1997. Development and Evaluation of Transit Signal

Priority Strategies. Southwest Region University Transportation Center,

University of Texas, Austin, 1997.

5. Ngan, V., T. Sayed and A. Abdelfatah. 2004. Impacts of various parameters on transit

signal priority effectiveness. J. Public Transp., 7(3), 71–93

6. Zhou, G. W. and A. Gan. 2005. Performance of transit signal priority with queue

jumper lanes. Transportation Research Record 1925: 265-271.

7. Federal Transit Authority. Queue Jumper. < www.fta.dot.gov/12351_4359.html >

8. Thomas, E. 2001. Bus rapid transit. Presentation at the Institute of Transport

Engineers. Annual Meeting Chicago, August, 19-22.

9. Levinson, H. S., S. Zimmerman, J. Clinger, and S. Rutherford. 2002. Bus rapid

transit: An overview. Journal of Public Transportation 5 (2): 1–30.

10. APTA Standard Development Program. 2010. Implementation of BRT Intellinget

Transportation System.

11. Weinstock, A. and W. Hook. 2011. Recapturing Global Leadership in Bus Rapid

Transit - A Survey of Select U.S. Cities. Institute for Transportation and

Development Policy.

12. Callaghan, L. and W. Vincent. 2007. Preliminary evaluation of metro orange line bus

rapid transit project. Transportation Research Record: Journal of the

Transportation Research Board, No. 2034, pp.37–44, Transportation Research

Board of the National Academies, Washington.

13. Cain, A., G. Darido, M. R. Baltes, P. Rodriguez, and J.C. Barrios. 2006. Applicability

of Bogotá‟s TransMilenio BRT System to the United States. National Bus

Rapid Transit Institute (NBRTI), Center for Urban Transportation Research

(CUTR), University of South Florida, FL-26-7104-01, Tampa, FL.

14. Zlatkovic M., A.Z. Stevanovic, F. Cevallos and H. Johnson. 2010. 35M MAX: the

first bus rapid transit system in Salt Lake County. World Review of Intermodal

Transportation Research, Vol. 3 Nos. 1/2 (2010) 103-120.

15. Spencer, S., K. Kitasaka, R.A. McNally and H. Vahidi, H. 2003. 98 B-line bus rapid

transit evaluation study. Transport Canada‟s ITS Deployment and Integration

Program, pp.3–57, Vancouver, Canada.

16. Rodriguez, D. and C. Mojica. 2008. Land value impacts of Bus Rapid Transit: The

case of Bogotá‟s Transmilenio. Land Lines, April, pp.2-7, 24.

17. Evans, H., and G. SkileS. 1970. Improving Public Transit through Bus Preemption of

Traffic Signals. Traffic Quarterly, 24 (4), 531-543. 82

18. Smith, H., B. Hemily and G. F. M. Ivanovic. 2005. Transit Signal Priority (TSP): A

Planning and Implementation Handbook. U.S. Department of Transportation.

19. Lewis, V. 1996. Bus Priority Study. Tualatin Valley Highway, Tri-Met, Portland,

OR.

20. Kloos, W.C., A. R. Danaher and K. M. Hunter-Zaworski. 1994. Bus Priority at

Traffic Signals in Portland: The Powell Boulevard Pilot Project. ITE

Compendium of Technical Papers.

21. Atherley, R. 2000. Rainier Avenue S.TSP Field Evaluation. King County Department

of Transportation.

22. Rakha, H. and Y. Zhang. 2003. Sensitivity Analysis of Transit Signal Priority Impacts

on Operation of Isolated Signalized Intersections. Paper submitted to the 2003

Summer Computer Simulation Conference, Montreal

23. Daniel, J. 1997. Bus signal priority along congested corridors. Proceedings of the

ASCE conference on Traffic Congestion and Traffic Safety in the 21st Century.

24. Huffman, B., J. Paquet, J. Weesner, and M. Pitstick. 1998. Signal priority

demonstration in suburban Chicago. Proceedings of the APTA 1998 Bus

Operations, Technology and Management Conference.

25. Zhu, H. B. 2010. Numerical study of urban traffic flow with dedicated bus lane and

intermittent bus lane. Physica A, Volume 389, Issue 16, p. 3134-3139.

26. Viegas, J., J. Manuel, R. Roque, B. Lu, and J. Vieira (2007) “The Intermittent Bus

Lane System: Demonstration in Lisbon. 86th Transportation Research Board

Annual Meeting, Washington, D.C.

27. Viegas, J. and B. Lu. 2001. Widening the scope for bus priority with intermittent bus

lanes. Transportation Planning and Technology, vol. 24, no. 2, p. 87-110

28. Currie, G., and H. Lai. 2008. Intermittent and Dynamic Transit Lanes – The

Melbourne Experience. Paper 08-0626 TRB, 87th Transportation Research

Board Annual Meeting, Washington, D.C., 2008.

29. Eichler, M.E. 2005. Bus Lanes with Intermittent Priority: Assessment and Design.

Masters of City Planning Thesis, Department of City and Regional Planning,

University of California, Berkeley.

30. Carey, G., T. Bauer and K. Giese. 2009. Bus Lane with Intermittent Priority (BLIMP)

Concept Simulation Analysis. FTA-FL-26-7109.2009.8. FTA, U. S. Department

of Transportation.

31. Federal Transit Administration. Bus Lane with Intermittent Priority (BLIMP)

Concept Simulation Analysis. FTA –FL- 26-7109.2009.8.

32. Knapp, K. K. 1995. Safety-Related Characteristics of Arterial High-Occupancy

Vehicle (HOV) Roadway and Lane Treatments. Report SWUTC95/ 721944.

33. Head, K. L. 1998. Improved Traffic Signal Priority for Transit. Interim Report, TCRP

Project A-16. TRB, National Research Council, Washington, D.C.

34. Nowlin, L. and K. Fitzpatrick. 1997. Performance of Queue Jumper. Proc., Traffic

Congestion and Traffic Safety in the 21st Century: Challenges, Innovations, and

Opportunities. Chicago, Ill., ASCE, New York.

35. Zhou, G. W. and A. Gan. Performance of transit signal priority with queue jumper

lanes. In Transportation Research Record: Journal of the Transportation

Research Board, No 1925, Transportation Research Board of the National

Academics, Washington, D.C, 2005, pp. 265-271.

36. Zhou, G.W. and A. Gan. 2006. Optimal detector locations for transit signal priority

with queue jumper lanes. Transportation Research Record 1978: 123-129.

37. Lahon, D. 2011. Modeling Transit Signal Priority and Queue Jumpers for BRT. ITE

Journal, Vol.81, No. 12, P.20.

38. Duong D., and F. Saccomanno, B. Hellinga. 2010. Calibration of microscopic traffic

model for simulating safety performance. Compendium of papers of the 89th

Annual TRB Conference held Jan. 10 – 14, 2010 in Washington, DC.

39. Park, B. and J. D. Schneeberger. Microscopic Simulation Model Calibration and

Validation: A Case Study of VISSIM for a Coordinated Actuated Signal

System. In Transportation Research Record: Transportation Research Board.

40. Kim, K., and L.R. Rilett. 2001. Genetic-algorithm-based approach for calibrating

microscopic simulation models. 2001 IEEE Intelligent Transportation Systems

Conference Proceedings: 298-704.

41. Cheu, R.L., X. Jin, K.C. Ng, D. Srinivasan. 1998. Calibration of FRESIM for

Singapore expressway using genetic algorithm. ASCE Journal of Transportation

Engineering, 124(6): p. 526-535.

42. Lee, D. H., X. Yang and P. Chandrasekar. 2001. Parameter Calibration for

PARAMICS Using Genetic Algorithm. 80th Annual Meeting Preprint CD-

ROM, Transportation Research Board 01-2399.

43. VISSIM Version 5.3 Manual. PTV Planug Transport Verkehr AG, Innovative

Transportation Concepts, Inc., Jan. 2003. 85

44. Federal Transit Administration. Characteristics of Bus Rapid Transit for Decision

Making. 2004.

45. Wright, L. 2004. Bus Rapid Transit Planning Guide. Deutsche Gesellschaft für

Technische Zusammenarbeit (GTZ), Division 44 Environment and

Infrastructure Sector Project, Transport Policy Advice, Germany.

46. Transit Cooperative Research Program Report 19: Guidelines For The Location And

Design of Bus Stop. Transportation Research Board, National Academy Press,

Washington, D.C., 1996.

47. Pima County Department of Transportation. Transit Guidelines for Roadway Design

and Construction. 2009

48. Omnitrans. Bus Stop Design Guidelines. 2006.

49. Wiedemann, R. 1974. Simulation des Straßenverkehrsflußes. Technical Report,

Institute for Traffic Engineering, University of Karlsrühe (in German).