2020

Sewmini Panambara Arachchillage, Queensland University of Technology
Jonathan Bunker, Queensland University of Technology
Ashish Bhaskar, Queensland University of Technology
Marc Miska, Queensland University of Technology

The passenger specific area (PSA) on a Bus Rapid Transit (BRT) station platform is a key measure of operational performance. In the current literature, transit station platform operation is generally evaluated using PSA by considering passenger activities separately. The aim of this study is to combine platform characteristics and various passenger activities using Time-Space (TS) analysis, and to develop a methodology to determine the PSA and Level of Service (LOS) of stationary passengers, circulating passengers, and passengers overall, on a cell by cell basis and across the whole platform. For a study BRT platform, we evaluate LOS using PSA, finding that it varies from LOS A to LOS C between the platform cells, while we find LOS B across the whole platform. This highlights the importance of evaluating BRT station platform operation both on a cell by cell basis and across the platform. As platforms are multi-activity areas, this methodology is useful to evaluate BRT station platform operation by considering the various passenger activities in a combined fashion.

Konstantinos Gkiotsalitis, University of Twente

In this literature review, we systematically review studies on public transit control with a specific focus on bus operations. In our synthesis of the relevant literature, we consider three perspectives: (1) the mathematical models of the proposed methodologies; (2) their complexity; (3) their applicability in real-time operations; and (3) their advantages/disadvantages considering their practical implications. The reviewed dynamic control methods include bus holding, rescheduling, and stop-skipping. Dynamic control methods, that have attracted more attention in recent years due to the advancements in automation and data availability, try to alleviate the negative effects of service disruptions. However, dynamic transit control tends to improve the service regularity and the operational costs at the expense of certain groups of riders. Identified problems are the inconvenience of onboard passengers when a bus is held at a stop, the long waiting times of "stranded" passengers in the case of stop-skipping, and the "sliding" of the crew and vehicle schedules in the case of rescheduling. All those critical issues are discussed and future research directions are drawn considering the complexity of the mathematical models for dynamic control and their practical implications.

Beda Büchel, Swiss Federal Institute of Technology (ETH Zurich)
Francesco Corman, Eidgenossische Technische Hochschule Zurich

Understanding the variability of bus travel time is a key issue in the optimization of schedules, transit reliability, route choices analysis, and transit simulation. The statistical modeling of bus travel time data is of increasing importance given the increasing availability of data. In this paper, we introduce a novel approach to model day-to-day variability of running times of urban buses on a section level. First, the explanatory power of conventionally used distributions is examined based on likelihood and effect size. We show that mixture models are a powerful tool to increase fitting performance, but the applied components need to be justified. We propose a time varying mixture distributions of fixed components, by which we can ensure meaningful component distributions. Hence, with our modeling approach, we reduce the complexity of mixture models and increase the explanatory power and fit compared to conventional models.

Wen Xun Hu, University of Toronto
Amer Shalaby, University of Toronto
Baher Abdulhai, University of Toronto

Transit signal priority (TSP) is a broadly used traffic signal control strategy designed for reducing transit delays at signalized intersections. Like the case in Toronto, conventional TSP strategies applied to high-frequency lines are usually unconditional, grant priority to all detected transit vehicles. While effective in reducing signal delays, this strategy does not guarantee an improvement in transit reliability. Instead of focusing solely on alleviating delays, this paper proposes a dual-objective TSP algorithm that optimizes both transit delays and reliability for high-frequency transit lines using deep reinforcement learning. The core idea of this approach is applying a neural network function approximator to estimate the action-value function of conventional reinforcement learning. The proposed TSP algorithm was trained and tested in a microsimulation environment that modeled a transit route segment in the City of Toronto, Canada. The performance of the new TSP algorithm was compared with two baseline scenarios, one with a current TSP algorithm used in the field by the TTC and the City of Toronto and the other without TSP. The proposed algorithm demonstrated its advantages in reducing headway deviations, percentage of extreme headways, and travel time.

Zhuang Dai, Beihang University
Xiaolei Ma, Beihang University
Xiaoyue Cathy Liu, University of Utah

It is a common practice for transit lines with fluctuating passenger demands to use demand-driven bus scheduling to reduce passenger waiting time and avoid bus overcrowding. However, current literature on the demand-driven bus scheduling generally assumes fixed bus capacity and exclusively optimizes bus dispatch headways. With the advent of connected and autonomous vehicle technology and the introduction of autonomous minibus/shuttle, the joint design of bus capacity and dispatch headway holds promises to further improving the system efficiency while reducing operating and passenger costs. This paper formulates this problem as an integer nonlinear programming model for transit systems operating with mixed human-driven and autonomous buses. In such mixed operating environment, the model simultaneously considers: (1) dynamic capacity design of autonomous bus, i.e., autonomous buses with varying capacity can be obtained by assembling and/or dissembling multiple autonomous minibuses; and (2) trajectory control of autonomous bus, i.e., autonomous bus can dynamically adjust its running time as a function of its forward and backward headways. The objective of the model is a combined value of operating costs and passenger costs. The model can be solved using commercial optimization solvers. We show that the proposed model is effective in reducing passenger waiting time and total operating cost. Sensitivity analysis is further conducted to explore the impact of miscellaneous factors on optimal dispatching decisions, such as penetration rate of autonomous bus and passenger demand level.

Kinjal Bhattacharyya, Indian Institute of Technology Kharagpur
Bhargab Maitra, Indian Institute of Technology Kharagpur
Manfred Boltze, Technische Universitat Darmstadt

In mixed traffic streams, especially near the intersections, bus priority treatments are necessary to prevent buses from significant travel time losses. This paper aims to assist the practitioners and field engineers to identify the design parameters and potential impacts of implementation of a queue jump lane (QJL) with pre-signal based on several locational characteristics.  The impacts of nine different factors on the operations of QJL are evaluated using a validated traffic micro-simulation model. The levels of each factor are specified based on the urban transport operations in India. In order to assist the practitioners, simplistic travel time savings and maximum queue length prediction models are developed using multi-parameter regression analyses based on the factors that reveal significant impacts. Finally, a guidance is provided to practitioners in the form of an eight-step procedure to implement QJL in emerging countries with analogous operating conditions.

Michael Bundschuh, PTV
Klaus Noekel, PTV Group

The vehicle schedule optimization for transit vehicles needs to respect additional constraints, if the vehicles are battery powered. Battery capacity often does not suffice for a full operating day. Several strategies exist for re-charging during the day, differing in cost and charging time. This paper describes an extension of vehicle scheduling which optimizes line blocks under driving range and charging time constraints. An example demonstrates how this helps to determine the number of additionally required vehicles and evaluate different charging strategies. The proposed method generalizes to solve further extensions of the vehicle scheduling problem.

Zhiwei Chen, University of South Florida
Xiaopeng Li, University of South Florida

Timetable design with time-varying vehicle capacity is a relatively new topic in urban mass transit (UMT) studies. This paper studies a new problem of this kind where vehicles can change their capacity at any stations along the corridor. We formulate the investigated timetabling problem into a mixed integer linear programming model, taking into account general operational characteristics in transit corridors, which include the minimum headway requirement, vehicle movement dynamics, passenger boarding dynamics. Due to the consideration of time-varying capacity at each station, the solution space of the mathematical problem increases rapidly with the instance size. To solve the mathematical model more efficiently, a customized branch and bound (B&B) algorithm is proposed based on theoretical properties of the investigated problem. These theoretical properties reduce the solution space of the B&B algorithm to a narrow band that would have increase unboundedly without them. Numerical experiments demonstrate the applicability of the proposed discrete model, the computation performance of the customized B&B algorithm, and the effectiveness of this new timetabling paradigm.

Nick Chaloux, Access Planning
Ahmed El-Geneidy, McGill University
Ehab Diab, University of Saskatchewan

As many cities in North America, Montréal has been seeing shrinking bus ridership trends over the past few years. Nevertheless, most of the recent literature has focused on the broader causes for ridership decline at the metropolitan or city level, none have considered ridership at the route level. As service adjustments take place at the route level and are felt by riders at this level, our study explores the determinants of STM bus route ridership between 2012 and 2017 using two longitudinal multilevel mixed-effect regression models. Our findings suggest that increasing number of daily trips and increasing the average route speed are keys to bus ridership gains. In contrast, an increase in bus stop spacing decrease bus ridership, while controlling for the impact of a few important external variables related to built environment, residents’ socioeconomics, and gas prices. Our models also show that reducing service frequency along a parallel route will lead to an increase in ridership along the main route. This study can be of use to transit planners and policy-makers who are striving to increase bus ridership, by exploring the factors affecting ridership at the route level, where most of the policies are implemented and where riders actually feel them.

Yuqiong Wang, Beijing Jiaotong University
Shunping Jia, Beijing Jiaotong University
Ruibin Wei, Beijing Jiaotong University

Optimizing Schedule of Flexible Feeder Transit (OSFFT) to serve passengers to the transfer station with dial-a-ride service is typically significant and theoretically challenging for the real-world applications. This paper proposed a bi-level programming model that could be solved by programming software like MATLAB. The bi-level model aims to minimize travel time of passengers and difference between the actual transfer time and the expected one, and minimize travel time of feeder transit vehicles. The weight coefficient of passenger profit- Common.EditSubmissionSteps.Transform.EquationText used for the computation of the proposed model. Computational experiments are conducted in Beijing residential area. The results of real-world experiments show the model and solution approach could be applied in the real-world operations. Moreover, this paper regard the road travel speed as time-dependent to closer to reality, and the proposed bi-level model could output the detailed timetables of flexile feeder transit vehicles with rigorous computation and optimization.

Runxuan Zhou, Southeast University
Shanshan Liu, University of Illinois, Urbana-Champaign
Xiucheng Guo, Southeast University

Feeder bus can fill the existing service gap left by urban rail transit to satisfy feeder demand. Service gap needs to be identified, and the coordination of multilevel urban traffic modes needs to be strengthened. The first contribution in this paper is the construction of a nonlinear mixed-integer programming model for multiple circular feeder bus routes design problems attempting to minimize total traveling costs of passengers. The model takes into account the impacts of existing bus lines by defining and introducing a feeder demand coefficient into the model. The second contribution is that an algorithm based on tabu search is proposed to tackle this sizeable combinational optimization problem. Considering the complexity of the problem, the proposed algorithm decomposes it into three subproblems containing feeder area partition, candidate stops selection and routes design problem. The lower bound of TSP estimated by 1-tree and an optimization stopping criterion are introduced into the algorithm to improve the performance. The algorithm is verified of great performance by an application in a downtown area of Suzhou, China, in terms of the quality of optimal solutions and computing efficiency. Besides, a possibility of simplifying a multiple circular routes design problem to a single route design problem utilizing spatial clustering is provided.Feeder bus can fill the existing service gap left by urban rail transit to satisfy feeder demand. Service gap needs to be identified, and the coordination of multilevel urban traffic modes needs to be strengthened. The first contribution in this paper is the construction of a nonlinear mixed-integer programming model for multiple circular feeder bus routes design problems attempting to minimize total traveling costs of passengers. The model takes into account the impacts of existing bus lines by defining and introducing a feeder demand coefficient into the model. The second contribution is that an algorithm based on tabu search is proposed to tackle this sizeable combinational optimization problem. Considering the complexity of the problem, the proposed algorithm decomposes it into three subproblems containing feeder area partition, candidate stops selection and routes design problem. The lower bound of TSP estimated by 1-tree and an optimization stopping criterion are introduced into the algorithm to improve the performance. The algorithm is verified of great performance by an application in a downtown area of Suzhou, China, in terms of the quality of optimal solutions and computing efficiency. Besides, a possibility of simplifying a multiple circular routes design problem to a single route design problem utilizing spatial clustering is provided.

Eva Chioni, National Technical University of Athens (NTUA)
Christina Iliopoulou, National Technical University of Athens (NTUA)
Christina Milioti, University of West Attica
Konstantinos Kepaptsoglou, National Technical University of Athens (NTUA)

Efficient operation of bus networks is vital for urban areas. However, a series of factors such as uneven passenger loads and congestion hinder adherence to posted schedules, leading to poor reliability of public transport lines. Bus bunching has been identified as a significant reliability problem, with impacts both to users and operators. Bus bunching is typically treated as a route-level problem, while spatial patterns in explanatory factors are overlooked. In this context, this study exploits Automatic Vehicle Location data to investigate factors affecting bus bunching, while taking into account their spatial variability. For this purpose, a Mixed Geographically Weighted Regression Model is applied for 360 bus stops in Athens, Greece, using stop attributes and network characteristics as explanatory variables. Results underline the superiority of the proposed model to Ordinary Least Squares Regression and corroborate the presence of spatial variability in the factors affecting bus bunching. Indeed, the number of roadway lanes at the stop level is positively associated with bunching in heavy traffic segments, whereas a higher number of lanes is negatively linked to bunching in less congested regions. Furthermore, the number of bunching occurrences generally increases with the number of routes serving each stop, as well as with the distance from metro stations in the outer parts of the city. Such findings highlight the need to consider spatial structures in relevant models and can help improve their reliability and accuracy.

Linghui He, Tongji University
Dongyuan Yang, Tongji University
Jian Li, Tongji University

Exclusive bus lane has been widely regarded as an effective way to promote bus priority. In China, planners usually pay more attentions to the infrastructure and facilities, but few attentions to the passenger satisfaction of exclusive bus lanes. In this paper, a passenger satisfaction survey of exclusive bus lanes was conducted in Shanghai, China, to understand the level of passenger satisfaction and associated influential factors. The Structural Equation Model (SEM) was employed to investigate the mechanism of relationships between passenger satisfaction and influential factors. The results indicated that passengers were dissatisfied with the current service of bus lanes, however they would still prefer to take buses in the near future. It implied that either passengers had preference for public transit, or they could not bear the financial burden of private motorization. In addition, the model also showed that the variables of travel environment, facilities and convenience, rather than operational efficiency, had significant effects on passenger satisfaction, and what passengers most wanted to improve were “crowdedness in the bus during peak hours” and “driving stability and comfort”. The proposed model and results may benefit for the planning and operations of exclusive bus lanes in Shanghai and other similar cities around the world.

Xianghai Meng, Harbin Institute of Technology
Enze Zhang, Harbin Institute of Technology
Xinyu Liang, Harbin Institute of Technology

Feeder bus network is an important issue for transport capacity and influence scope of urban rail transit, even the whole public transport system. However, the conflicts between system trip cost and personal trip cost will occur, when the traditional hypothesis is used to solve the design problem of feeder bus network. This study provides an optimization method of feeder bus network based on belt-shaped passenger-attracting area, which is demonstrated with a case study of Harbin in China. This method considers more trip modes and focuses more on individual trip cost. The walking zone and feeder bus zone were defined based on the analysis of the rail transit influence scope and the trip characteristics of various trip modes. Further, the feeder bus zone was divided with the principle of minimize travel cost of passengers and the passenger-attraction zone with a hyperbolic boundary was established. Taking the passenger-attraction zone as constraint condition, an optimization method of bus feeder network was proposed, and the dynamic programming method of integer programming was used to solve the problem. This research provides a relatively simple approach to optimization feeder bus network and this method can be easily applied to other case studies around the world.

Nikhil Menon, University of South Florida
Pei-Sung Lin, University of South Florida
Rakesh Rangaswamy, University of South Florida
Achilleas Kourtellis, USF Center for Urban Transportation Research
Cong Chen, University of South Florida
Robert Bertini, Oregon State University

Emerging transportation technologies, such as automated vehicles (AVs) have created revolutionary possibilities in the way we might travel in future. Past research has highlighted the prospect of using low-speed automated shuttles as an alternative to personal vehicle modes in a variety of contexts. Many states, and cities have tested automated shuttles in controlled environments but not many trials have been conducted in mixed traffic. Like any other emerging technology, automated shuttles would have to overcome not just the technological challenges, but also the social barriers in order for successful penetration into the marketplace. The current study will reveal findings from an automated shuttle demonstration at a public university in Florida. A one-week demonstration was held in February 2019 with over 500 riders riding in the automated shuttle. Each rider provided feedback on their ride experience, and several opinions regarding the use of automated shuttles as a future transportation mode in low-speed environments. Results reveal an overwhelmingly positive response to using automated shuttles for various trips on campus with the potential for the shuttle to replace some/all trips by various modes. Results from this study provide a more nuanced insight into the future of automated shuttles as a mode of transportation. While automated technology continues to evolve temporally, studies like this could serve as an interesting and valuable baseline for measuring the change/technological progress as more studies are developed in the future.

Xiaoling Luo, Chongqing Jiaotong University
Liang Xia, Southwest Jiaotong University
Yangsheng Jiang, Southwest Jiaotong University
Wenbo Fan, Southwest Jiaotong University

Although transit stop location problem has been extensively studied, the two main categories of modeling methodologies, i.e., discrete models and Continuum Approximation (CA) ones, seem have little intersection. Both have strengths and weaknesses, respectively. This study intends to integrate them by taking the advantage of CA models’ parsimonious property and discrete models’ fine consideration of practical conditions. In doing so, we first employ the state-of-the-art CA models to yield the optimal design, which serves as the input to the next discrete model. Then, the stop location problem is formulated into a multivariable nonlinear minimization problem with a given number of stop location variables and location constraint. A heuristic solution algorithm is presented to find the optimal design that is ready for implementation. In numerical studies, the proposed model is applied to a variety of scenarios with respect to demand levels, spatial heterogeneity, and route length. The results demonstrate the consistent advantage of the proposed model in all scenarios as against its counterparts, i.e., two existing recipes that convert CA model-based solution into real design of stop locations. Lastly, a case study is presented using real data and practical constraints for the adjustment of a bus route in Chengdu (China). System cost saving of 27.96% is observed by before-and-after comparison.

Amirreza Nickkar, Morgan State University
Young-Jae Lee, Morgan State University
Mana Meskar, Sharif University of Technology

Demand responsive feeder transit can minimize passenger travel times and operating costs by optimizing routing based on real demand. One question about the demand responsive feeder transit operation is whether it can be optimized with door-to-door service or with temporary stops for picking up and delivering passengers. Obviously, door-to-door service eliminates passengers’ walking distances, but it increases passenger in-vehicle travel times and vehicle operating distances. On the other hand, demand responsive feeder transit with temporary stops, which designates temporary locations for picking up and dropping off passengers, minimizes bus operating distances as well as passenger in-vehicle times, although it increases passengers’ walking distances and times. This study develops an optimal routing algorithm for demand responsive feeder transit accommodating temporary feeder stops. The algorithm developed an optimal algorithm for clustering and grouping of passengers considering their physical locations and time windows; then it was integrated with the authors’ previously developed algorithm for the optimal flexible feeder bus routing as a mixed integer model with the objective of minimizing total costs, including both passenger travel times and  operating costs. The results of this study show that feeder networks with temporary stops always lower operating costs when compared to those with a door-to-door option. However, the results of the algorithm are highly sensitive to the location of passengers, and when passengers are located far from each other, feeder networks with a door-to-door service can be more efficient and reduce total passenger costs and total costs more than those with temporary stops.

Travis Glick, Portland State University
Miguel Figliozzi, Portland State University
Crumley Miles, TriMet

Dwell, the amount of time a bus services a bus stop, is a primary contributor to transit travel times and variability. One factor that is commonly excluded from dwell time models is the impact of overlapping routes and bus bunching. By utilizing a very large dataset and several bus interaction scenarios and variables, this research presents novel results that quantify the impacts of multiple overlapping buses and bunching at bus stops. Results utilizing linear and log-linear regressions are discussed and compared. Bus interactions have a measurable and significant impact on dwell durations. Dwells are larger as the number of buses that interact at a stop increases. The results indicate that overlapping routes creates more variability in dwell times and therefore may contribute significantly to bus bunching. In addition, there is a substantial difference regarding number of passengers boarding.  These results may be applied to monitor bus operations or to estimate the potential impact of route overlaps and bunching on dwell times.

MD Sultan Ali, Florida International University
Priyanka Alluri, Florida International University
Thobias Sando, University of North Florida

Transit is continuing to be a priority, as more agencies are looking for strategies to increase transit ridership. Transit Signal Priority (TSP) is a strategy that prioritizes the movement of transit vehicles through a signalized intersection to avoid transit delay and improve travel time reliability. The objective of this study was to quantify the mobility benefits of TSP. The analysis was based on a 10-mile study corridor in South Florida. Two high fidelity microscopic simulation VISSIM models, a Base and a TSP model, were developed. Note that the Base model was calibrated and validated to represent the field conditions. When TSP was deployed at the signalized intersections, the study corridor experienced up to 8% reduction in travel times for buses and all vehicles, and up to 13.3% reduction in average vehicle delay time for buses and all vehicles. To better quantify the mobility benefits of TSP strategy, this study developed Mobility Enhancement Factors (MEFs). An MEF is a multiplicative factor used to estimate the expected mobility level after implementing TSP at a specific site. An MEF < 1 implies that the TSP yields mobility benefits. The MEFs based on travel time for buses and all vehicles were estimated to be 0.91 and 0.96, respectively. The MEFs based on average vehicle delay time for both buses and all vehicles was estimated to be 0.87. The study results indicate that the TSP deployment along the study corridor has enhanced mobility for buses and all other vehicles.

Yuxiong Ji, Tongji University
Yuhan Ji, Tongji University
Yu Shen, Tongji University
Yuchuan Du, Tongji University

This study presents a trajectory data-driven approach for planning the zonal shuttle bus services, which is able to reflect the travelers’ path choice preferences, road network topology, and changes in traffic patterns. Taking into account the travel demand towards the airport with a many-to-one road network, a zonal airport shuttle bus service is designed. The zonal shuttle service picks up the passengers from a service zone then carry them directly to the airport. To determine the service zones, the developed approach first discovers the “meeting points” in the road network and identifies their associated trajectories. The effectiveness and efficiency of the developed approaches are demonstrated in a case study based on the taxi GPS data in Shanghai, China. The results show that the discovered zones are reasonable in real world and the developed approach is able to provide insights for adjusting current shuttle bus services and for planning the additional zonal airport shuttle bus lines.

Seyedmostafa Zahedi, Northeastern University
Haris Koutsopoulos, Northeastern University
Zhenliang Ma, Monash University

Dynamic interlining of buses is an operational strategy for routes that have a terminal station at a common hub. The strategy keeps a portion of the fleet unassigned to any specific route and allows them to be shared among the routes belonging to the hub (shared fleet). The shared fleet is then dispatched on an on-demand fashion to avoid delays and regulate service. In this paper, a simulation model is developed to examine the impacts of dynamic interlining on service reliability. Using a subnetwork of Boston area bus lines as a case study, the performance of the strategy as well as factors that affect its performance are investigated. Results show that dynamic interlining can improve service reliability and decrease the variations of departure headways at the hub. The Size of the shared fleet has the most dominant impact on the performance. Dispatching rules that govern the use of the shared fleet can also impact the performance.

Chen Guo, Chang'an University
Jianjun Wang, Chang'an University
Zhengyu Wang, Inner Mongolia University
Yueying Huo, Inner Mongolia University

This paper proposes a model of sparse and dense hybrid grid transit network for some square cities of downtown in the corner, which can improve spatial coverage and time coverage and avoid increasing agency cost. The network is aiming to design high performance transit systems with friendly structure and good accessibility for any O-D pair by high speed and low headways. The transit network is composed by two parts: dense grid in the downtown and sparse grid in the periphery. The model faces to square cities of downtown in the corner. The objective function derived based on the uniform and linear decrement demand function includes agency cost and user cost and the decision variables are downtown-to-city ratio, headway, stop spacing (line spacing) and headway ratio of downtown and periphery. This topologically is conducive to the expansion and development of the city with changeable headways. The paper validates the model advantages through the demand of San Francisco. The robustness of the model with decision variables and input parameters shows stable results and good feasibility for actual bus network design. Consistent with the actual situation, the concentrated of time and spatial distributed demands result in less system costs.