ABSTRACT. Japan has a long history of managing natural hazards due to its geography and geology. Simultaneously, it is known as a country with a railway system that is highly punctual, reliable, and safe. This research uses an exploratory approach to review climate change adaptation and disaster risk reduction practices in the Greater Tokyo Area. Unstructured interviews with various stakeholders and a literature review discuss how the climate is changing in Japan, what strategies are used amongst railway organisations to manage meteorological hazards, and how railway organisation acknowledges climate change adaptation alongside disaster risk reduction. Results indicate that disaster risk reduction practices still dominate within the railway sector however, more efforts of climate change adaptation set forth by the government may shift practices.

ABSTRACT. How to promote public transport to the level before COVID is what government in all levels attempt to achieve. In Taiwan, regional level monthly public transport pass, TPASS, introduced in July 2023. This study aims to explore travel pattern change before and after TPASS introduction with big data (i.e., smart card data). There are around 30% users transfer to TPASS with monthly increase rate of 11.4%. Further, the current analysis provides spatial analysis for TAPSS usage. The outcomes can allow policymakers to understand how to revise or improve current TPASS ticket to attract target group.

ABSTRACT. Sustainable development is critical in the transport field, particularly in the form of promoting green vehicle (e.g., electric vehicle, EV) development. This study focuses on the behaviours of operators on EV type identification through meta-analysis with a logistic regression. Four vehicles (i.e., trucks, vans, buses, and cars) are identified for 52 samples collected from 43 literature. In particular, heavy policy incentives have been implemented via vehicle purchase subsidies for the public transport sector. Key findings reveal that fuel and infrastructure costs positively influence the selection of electric buses (E-bus). The results provide valuable insights for fleet operators about EV type recognition and promote E-bus adoption in terms of other vehicles.

ABSTRACT. Bike-Sharing System is one of the most used transport modes in urban areas. This study examines the impact of urban greenery and shading of cycling routes using Street View imagery data from the smart card data of 367 YouBike stations in Taipei. Utilizing Geographically Weighted Regression and the transformed gravity model by using Markov Chain Monte Carlo, the findings show that street-view elements exert varying influences under different urban structures, urban greenness in suburban areas, and shading levels in urban areas may enhance the cycling environment for bike-sharing usage. These insights can guide in designing green transport to encourage cycling.

ABSTRACT. Accurate energy consumption evaluation of zero-emission buses (electric and hydrogen buses) benefits operational optimization and decision-making. This study collected data from 710 trips of double-deck electric buses (DDEBs) and double-deck hydrogen buses (DDHBs) in Hong Kong. Features were extracted across six aspects: route, bus status, environment, traffic, driver, and driving behavior. Four distinct machine learning models were employed to predict energy consumption. The results show prediction errors of 5.745% for DDEBs and 7.173% for DDHBs. Ambient temperature and congestion index exhibited a significant positive correlation with energy consumption for both bus types. Additionally, the initial state of charge (SoC) of the supplementary battery onboard DDHBs was found to have a notable impact. Ensuring sufficient battery charge before each trip can reduce energy consumption in DDHBs.

ABSTRACT. This paper addresses the challenge of jointly optimizing transit service frequency and fare in urban public transportation systems with elastic demand. We propose a mathematical model that maximizes social welfare while accounting for the strategic behavior of passengers within the network. It also incorporates constraints related to operational resources, passenger flow, and vehicle capacity. A branch-and-cut algorithm with outer approximation-based cuts is proposed to solve the problem. Numerical results are presented for a small case study. The presented model will help transit agencies maximize their profit without compromising transit user satisfaction.

ABSTRACT. Electric moped scooter sharing (EMSS) has emerged as a key solution to the first-last mile problem in public transportation, experiencing substantial growth recently. However, passengers encounter risks due to inherent uncertainties during their travels, primarily related to potential service unavailability, such as a lack of nearby scooters or available parking spaces. Many prior studies have overlooked these uncertainties, leading to overly optimistic evaluations of shared mobility's potential, and research specifically focusing on EMSS remains limited. This study investigates the impact of uncertainty on the intention to use EMSS. A hybrid choice model (HCM) is utilized to integrate latent variables within a discrete choice framework, enabling the simultaneous consideration of both tangible and intangible attributes from the passenger's perspective. Using survey data from Taiwan's metropolitan areas, we propose a mixed logit model for variable analysis alongside a behavioral reasoning theory model to assess latent factors. Our findings indicate that the "probability of finding a parking space" significantly influences mode choice utility. Additionally, perceived risks, including scarcity and functional risks, negatively impact EMSS usage. The managerial implications derived from this study are expected to enhance the understanding of influential factors for system operators and city governments regarding EMSS utilization.

ABSTRACT. This study explores the social effect of a new type of public transport in Japan known as Green Slow Mobility (GSM). Using structural equation modeling, we examine how the introduction of GSM influences the development of social capital. Results show that a positive impression of GSM, interest in GSM activities, and GSM usage significantly contribute to social capital building. The findings highlight GSM's symbolic value and operational influence, offering valuable insights for regional transport policy fostering social inclusion.

ABSTRACT. This study investigates the effects of a 2023 policy banning privately-owned e-scooters on public transport in Barcelona. Prior to the ban, e-scooters served as critical connectors to transit, particularly for intermunicipal commuters with limited mobility alternatives. A two-wave longitudinal survey (n=311 pre-ban, n=111 post-ban) revealed that 73% of trips were intermodal. The ban disrupted these routines, disproportionately affecting women and lower-income users who lacked viable substitutes. Many users adapted by walking more or relying solely on public transport, but these strategies often increased travel time and reduced satisfaction. Some shifted to full e-scooter or car use, experiencing improved travel times but disengaging from public transport. The most significant declines were emotional: stress and frustration increased, and overall satisfaction dropped, especially among users with strong ties to e-scooter commuting. Anticipated responses to the ban frequently diverged from actual behavior, highlighting unpredictable adaptation dynamics. The findings underscore the risks of restrictive micromobility policies, which can exacerbate existing inequalities and undermine intermodal transport goals. Policymakers should consider flexible alternatives, such as partial access windows or designated transfer hubs, to maintain equitable connectivity. This case emphasizes the need for inclusive, evidence-based regulation that preserves both safety and mobility integration.

ABSTRACT. Bus drivers are among the occupational groups most exposed to air pollution due to long driving hours. This study aims to assess the exposure risk for drivers of non-Air Conditioning (AC) buses in Delhi using continuous mobile monitoring. To comprehend the objectives, PM2.5 concentrations were measured for 30 bus drivers in 15 buses (morning and evening shifts) using low-cost air quality devices over eight months. The results demonstrate substantial temporal and spatial heterogeneity in PM2.5 concentrations for bus drivers during both shifts. The cumulative inhalation dose is found to be very high for all bus drivers, indicating a higher risk of exposure. The findings indicate that bus drivers face a very high exposure risk while driving, necessitating immediate attention to mitigate it. Various policies aimed at lowering exposure risk values are formulated, and proposed policies can offer drivers practical guidelines for reducing health risks associated with PM2.5 concentrations.

ABSTRACT. Rapid motorization and urbanization have significantly increased air pollution exposure of urban residents. The present study explores integration of real-time air pollution data into choice modeling which helps in identifying pollution levels that prompt changes in travel behavior. Data collection is conducted through a face-to-face questionnaire survey using the TSaaS portal, and a total of 471 responses are retained after data preprocessing. The survey collects data on trip and socio-demographic characteristics, and users' mode preferences for their current mode and public transit alternatives, considering real-time exposure and travel time information. The study examines the impact of travel time, cost, waiting time, and pollution exposure on mode choice between private and public transit for Delhi city. A nested logit model is developed, grouping buses and metro under the public transit nest. The findings indicate that increased travel time and degrading air quality negatively affect mode choice. These insights can assist policymakers and transport planners in enhancing public transit and encouraging adoption of sustainable travel modes.

ABSTRACT. Free-floating shared electric mopeds are gaining popularity as flexible, low-carbon mobility solutions in urban areas, emphasizing the need to understand user preferences. This study examines how attributes such as battery level, walking distance, and price (unlock and per-minute ride fee) influence user preferences for shared e-moped usage. Using survey data from the Netherlands, we employ a Multinomial Logit model to estimate utility and a Mixed Logit model to capture preference heterogeneity. Results underscore the significant impact of battery range on user choices, identifying an indifference threshold and offering practical insights for user-based relocation strategies and fleet management.

ABSTRACT. To combat climate change and air pollution, electrifying urban bus fleets is crucial for reducing vehicular emissions. Past studies often overlook long-term phased transitions as well as the network effect. This study proposes an optimization model addressing urban bus network fleet transition and charging facility installation over multiple periods. A solution method called DRL-HetGNN is proposed, leveraging deep reinforcement learning and heterogeneous graph neural networks to solve large-scale problems efficiently. Focusing on Hong Kong's bus system, this study examines scenarios involving future price fluctuations and policy support mechanisms, aiming to guide policymakers in achieving a sustainable, zero-emission public transportation system.

ABSTRACT. As urban areas worldwide increasingly prioritize environmental sustainability and energy efficiency, battery electric bus (BEB) systems have emerged as a pivotal alternative to traditional diesel bus systems. This paper addresses the pressing need for integrated replacement planning and scheduling models for BEB systems. An integrated replacement planning and scheduling model for multi-terminal BEB systems is proposed and converted to an integer linear programming (ILP) which can be solved by commercial solvers. The proposed model simultaneously determines the number of electric buses, the number of charging stations, and develops optimized dispatch and charging schedules. The model takes deadheading, maximum number of chargers, charging time threshold, and time-of-use electricity prices into consideration as well. A branch and price framework and a heuristic algorithm are proposed to solve large-scale problems. Through a case study using data from Hong Kong’s bus company, the proposed model demonstrates significant cost savings over traditional diesel bus systems, even after accounting for higher vehicle purchase cost and charging infrastructure costs. Several sensitivity analyses provide practical insights into the replacement planning of BEB systems.

ABSTRACT. Urban metro systems produce extensive operational data, often analysed through benchmarking to identify best practices and prioritise improvements. This process informs strategies and policies by comparing key performance indicators (KPIs) across systems. However, traditional benchmarking has limitations, particularly for top performers who gain little actionable insight and for systems focusing solely on emulating successful operators without accounting for unique contexts. While normalisation techniques attempt to address such disparities, they often fail to capture heterogeneity in system characteristics. This study advances benchmarking by using counterfactuals to create a ‘synthetic’ unit, offering personalised benchmarks based on each metro system’s specific attributes.

ABSTRACT. New land-use planning configurations can have wide-ranging crowding effects on the public transportation system, given the ongoing increase in urban agglomerations worldwide. In this study, we propose a method for quantifying the network-wide crowding implications of new developments accounting for their socioeconomic and planning characteristics. Size and proximity to a high-capacity connection are highly influential factors in determining crowding implications’ extent and geographical spread. Interestingly, the income level can have a twofold effect on crowding contributions (increase or decrease). The proposed method can serve as a tool for the ex-post quantification of the crowding impacts using automated data sources.

ABSTRACT. Public transit inequalities persist globally, yet traditional assessment methods remain infeasible in data-poor environments. This study develops a framework leveraging mobile phone data to simultaneously evaluate supply and demand-side transit inequalities. Analyzing six global cities, we find three distinct equity tiers, with demand-side disparities consistently exceeding supply-side inequalities. High-density cities achieve better equity through gradual transit evolution, while rapidly growing cities show stark access disparities despite similar infrastructure levels. These findings suggest transit agencies should prioritize alignment between infrastructure distribution and actual usage patterns, particularly in rapidly urbanizing regions.

ABSTRACT. The ageing population presents a global demographic challenge, as the elderly mobility declines rapidly due to their unstable income and limited physical and cognitive abilities. Autonomous demand-responsive transport emerges as a potential solution to cater to their specific transport needs. In order to comprehend their acceptance and adoption of autonomous demand-responsive transport, a stated preference survey was conducted, interviewing 232 elderly individuals aged 60 or above. The design of choice experiments was pivoted with respect to revealed trips from the surveyed elderly to simulate realistic choice scenarios. Using the collected data, logistic regression models were developed to identify contributory factors influencing their mode choice selections between their currently chosen alternative and the new transport mode. The model results indicate that travel fare, walk time, on-street wait time, in-vehicle travel time, seat availability, and the provision of on-board staff are the significant attributes. An unobserved heterogeneity among the elderly is identified in the provision of on-board staff. About 35% of them hold a negative perception of this arrangement, and preferred a fully autonomous service, while others prefer having an on-board supervisor for safety concerns. The study highlights that the elderly are hesitant to shift to autonomous demand-responsive transport due to their concerns about autonomous driving and digital proficiency.

ABSTRACT. Initial waiting time at stations contributes to passengers‘ generalized travel time. Timetable awareness is its most influential factor. This study reports from a survey in Zurich, Switzerland about timetable awareness for different headways, age groups and times of day. We link the surveyed share of timetable awareness with the estimated share of randomly arriving passengers, based on passenger arrival distributions from the literature. In our case study, estimated shares of randomly arriving passengers differ from the share of timetable aware passengers. We discuss the possibility that timetable independent passengers do not arrive perfectly randomly but also time their arrivals at least sparsely.

ABSTRACT. This study examines the decision-making process behind planning outings, specifically focusing on the timing and origin of travel choices. It also investigates the impact of the COVID-19 pandemic on the number of route searches and assesses the recovery of bus routes post-pandemic. Furthermore, the study analyzes the correlation between population distribution, facility locations, and the characteristics of bus routes to determine which routes have successfully recovered from the effects of COVID-19. The results indicate that many routes have recovered or are expected to recover. Route searches for these routes may be related to business and leisure travel.

ABSTRACT. The rise in air traffic demand necessitates sustainable and efficient airport connectivity. This study analyses the modal choices of passengers and workers at Santiago de Chile’s airport, focusing on the impact of the new bus service (line 555) launched in 2023. Based on a revealed preferences survey, the proposed Mixed Logit model revealed key factors influencing choices, including cost, income, luggage, and travel purpose. Results highlight challenges with line 555, such as its indirect connection to terminals and low user awareness. These findings underline the need for improved connectivity to enhance airport accessibility and support sustainable urban development.

ABSTRACT. Accessibility is a key factor influencing railway demand, and improving access trip quality could increase ridership. However, there is no practical way to accurately determine where and how people travel to stations. As a result, access trips are often assumed to be uniformly distributed with identical speeds and travel patterns. Variables such as access time and speed are also applied as fixed values in mode choice models, leading to errors in railway demand forecasts and potential failure in preliminary feasibility studies.

To address this, we propose an algorithm that defines access trips and analyzes regional travel patterns using empirical mobility data instead of survey-based data. We calculated the rate of change in traffic volume from mobility data to define access boundaries and applied the DBSCAN clustering algorithm to group similar access trips.

A case study at Osong Station in South Korea was conducted to validate the algorithm. The results showed that while some areas had well-distributed public transport services, others had coverage concentrated in limited zones. In one region, travel times exceeded 70 minutes, yet Osong Station was still used, suggesting a need for more stop routes. These findings highlight the effectiveness of the proposed algorithm in identifying access trip characteristics from real-world mobility data.

ABSTRACT. Mobility as a Service (MaaS) offers a promising solution to urban transportation challenges by encouraging public transport use. In Taiwan, the Public Transport Pass (TPASS) serves as a simplified MaaS model. This study analyzes data from 498 non-users through latent profile analysis, identifying four user profiles: flexible multimodal commuters, economy-oriented commuters, private transport enthusiasts, and sustainable transport supporters. Findings indicate that psychological factors, such as perceived economic benefits and habitual behavior, significantly influence TPASS adoption, highlighting the need for targeted strategies to promote its uptake and foster sustainable travel behaviors.

ABSTRACT. This paper examines the determinants of commuting modal choice by focusing on aerial cable car (ACC) users and transport-related social exclusion (TRSE) factors. The case study is set in Manizales (Colombia), a medium-sized city known for pioneering ACC as a transit mode. Manizales is characterised by steep topography and housing development in hilly terrain, making it a unique context for studying ACC’s impact on mobility. Data was collected through a questionnaire survey conducted in two phases, and responses were analysed using discrete choice models and exploratory factor analysis. We estimated an MNL model with time and cost parameters, which will be the initial step in employing a hybrid choice model, including TRSE latent variables, to better understand modal choice behaviours.

ABSTRACT. This study examines the influence of physical facilities and station surroundings on public transport attractiveness. Using an online survey (N = 4,801) on train stations within Denmark, we identified three factors: "Cleanliness & Maintenance," "Crowding & Environment," and "Safety & Urban Life." Newer transport modes, like light rail and metro, scored higher in cleanliness but were more isolated with less lighting and urban life. Conversely, S-train stations in Copenhagen’s suburbs scored low across all factors, highlighting significant challenges. These findings align with previous studies suggesting that aesthetically pleasing and well-maintained stations positively influence passengers’ perceptions.

ABSTRACT. Introduction of a new transit system influences the travel choice of commuters. This paper assesses changes in intercity travel mode share in the case of the proposed regional rapid transit system (RRTS), National Capital Region (NCR), India. A stated choice experiment of Meerut – Delhi commuters analyzed using the random parameter logit (RPL) model. The findings reveal RRTS as the preferred alternative for intercity travel. Increase in RRTS mode share was significantly linked to improvement in comfort during travel. It is also observed that the qualitative attributes like comfort and safety have a significant impact on mode choice of commuters. Further, the sensitivity analysis performed considering five different scenarios also indicated a major shift in the travel choice of commuters with change in comfort during travel. These results suggest that transportation planners should prioritize enhancing qualitative aspects to attract commuters and boost adoption of new transportation modes.

ABSTRACT. This study aims to identify who rides rural microtransit and how changes in travel behaviour due to microtransit vary between rural and non-rural riders through a case study evaluation of the Yolobus BeeLine, based in Yolo County, California. Through the conducting of ride-along interviews and surveys with riders of the BeeLine, we find that rural microtransit provides vulnerable populations, such as women, those without drivers’ licenses, and those with physical or similar limitations, means to complete trips that they would not take otherwise. Both rural and non-rural riders decreased use of other modes of travel, especially taxi/ride-hailing services, because of the microtransit service. However, rural riders tended to use the microtransit service more frequently. While riders indicated that the availability and waiting times for the service could be unreliable, most riders still viewed the BeeLine positively, indicating that microtransit services like the BeeLine provide a valuable mobility option to rural transit riders.

ABSTRACT. Stemming from the Extended Unified Theory of Acceptance and Use of Technology (UTAUT2), we develop a model to explain the effects of passenger’s service expectancy, innovative traits, safety concerns, entertainment, environmental perception, and facilitating factors on their acceptance of autonomous ferry. We analyze the survey data from 594 passengers using the ferry routes between Cijin Island and Kaohsiung City in Taiwan. Preliminary results indicate that passengers’ safety concern, facilitating factors, innovative traits, entertainment, and entertainment significantly influence their use intention of autonomous ferries. The influence of service expectancy on the use intention of autonomous ferries was supported in this study.

ABSTRACT. This study develops analytical tools aimed at enhancing public transport (PT) transfers to improve efficiency, comfort, and safety. Building on Hadas and Ranjitkar (2012), and on Hadas and Ceder (2010), it emphasizes the significance of transfer quality over route selection for users. Recognizing the anticipated increase in transfers driven by services similar to the London Tube, this research identifies critical transfer attributes that users find intolerable and proposes optimal locations for transfer points requiring the most significant improvements. Utilizing optimization models, the study allows passengers to select routes tailored to their preferences while minimizing transfer-related costs. A toy example illustrates the application of network-flow models to detect bottlenecks and optimize route selection, fostering a user-centric PT experience.

ABSTRACT. This study examines customers’ longitudinal MaaS bundle choices during Semester 1, 2023, using data from the University of Queensland Mobility as a Service trial. Analysing 709 student panels, a latent class choice model identifies two customer classes. Specifically, Class 1 shows strong inertial effects for 30-day public transport-only (PT30) bundles, while Class 2 exhibits similar tendencies for 7-day public transport-only (PT7) bundles. Furthermore, both classes also demonstrate the existence of some variety-seeking behaviours in multi-modal bundle choices. These findings reveal heterogeneous roles of individuals’ past choices in shaping their bundle preferences, offering valuable insights for MaaS operations.

ABSTRACT. This study investigates regional disparities in public transport satisfaction by comparing the Seoul metropolitan area and provincial regions in South Korea. Drawing on data from a large-scale national satisfaction survey (n = 82,000), we employ a Structural Equation Modeling framework using the Multiple Indicators and Multiple Causes (SEM-MIMIC) approach to identify key service attributes affecting satisfaction. To guide service enhancement strategies, we further apply Importance-Performance Analysis (IPA), mapping perceived performance against its influence on satisfaction. Results reveal region-specific determinants: in the Seoul metropolitan area, congestion reduction and accessibility are the most influential, while in provincial areas, service frequency (headways) and fare affordability are more critical. The IPA results reinforce these findings, highlighting congestion and user environment as top improvement priorities in metropolitan areas, as well as accessibility and comfort in provincial regions. These findings emphasize the necessity of region-tailored transport policies that account for spatial differences in service needs and user expectations. By addressing these regional disparities, public transport providers can more effectively promote modal shift, enhance user satisfaction, and support sustainable urban mobility in high-density and low-density contexts.

ABSTRACT. To support the investigation of the mobility behavioural pattern by different traveller groups, we propose a framework that segments the activities from public transport (PT) smartcards by their regularity and frequency. It uses the activity start time, duration and location to assess the spatiotemporal regularity. Utilising data from London, we demonstrate that this framework works for infrequent and frequent users and activities, and that regularity can be aggregated to card and location levels. The insights can assist operators and planners to optimise staffing, service planning and marketing, and inform the analysis of home-working, night working patterns and night-time activity locations.

ABSTRACT. Express bus services, which use faster routes and make fewer stops than regular bus services, are a potential solution to reduce travel time. This study aims to solve the regular bus network design problem, the express service design problem, and the frequency setting problem simultaneously. A hybrid genetic algorithm is modified to solve this integrated problem. A waypoint mechanism is proposed to facilitate the design of both regular and express routes. The numerical result shows that the solution obtained by designing the regular routes and express services simultaneously is better than the sequential approach that designs them separately and sequentially.

ABSTRACT. This study develops a robust multimodal network design integrating metro and bus services, addressing correlated stochastic disruptions. Using a vine copula technique, we model dependencies among disrupted metro links for realistic correlation representation. A two-stage correlation-aware stochastic program is formulated to optimize bus routing and frequencies, minimizing total costs while accounting for unmet demand. By designing the entire network to withstand metro disruptions, the proposed approach demonstrates superior performance, reducing demand losses by 99% compared to benchmark that neglects correlation. This framework enhances the resilience of public transport systems under uncertainty, ensuring efficient and reliable service during disruptions.

ABSTRACT. Ridesharing could offer a solution to urban mobility challenges by enhancing affordability, convenience, and sustainability. This study explores a system designed for transportation hubs with parking facilities, integrating personal vehicles with carpooling options and shuttle services as dynamic transit solutions to serve inbound and outbound ride requests. To capture diverse interests and priorities within urban transportation, we apply column generation to minimize four objectives: user costs, operator costs, emissions, and vehicle underutilization as a system-wide objective. Our evaluation of the proposed system across various scenarios demonstrates its potential to address the complexities of large-scale transportation systems.

ABSTRACT. The Rolling Stock Rotation Problem (RSRP) is to find a cost optimal set of cycles, i.e., rotations in order to cover given timetabled trips by rail vehicles. In this paper, we consider a dedicated variant of the RSRP arising at our partner DB Fernverkehr AG, which is the largest passenger railway operator in Europe. There, rail vehicles need to be composed in order to form vehicle compositions and also need to be maintained within given maintenance intervals. While the first requirement is known to be efficiently manageable by a hypergraph approach in our application, the maintenance constraints appear harder as they have a similar computational flavor as, e.g., constraints in vehicle routing applications.

We contribute the utilization of a heuristic concept, called regional search. Its idea is to mime an exact iterative improvement procedure for a hard problem's relaxation, in order to derive an efficient large neighborhood search procedure. Here, we construct promising neighborhoods for the RSRP from alternating cycles found via linear programming arguments by the primal version of the famous Hungarian method. We show that the procedure is able to produce near-optimal as well as optimal solutions for industrial scenarios by a proof-of-concept computational study.

ABSTRACT. This study addresses the Train Platforming and Passenger Management Problem (TPPMP), integrating train platforming and passenger flow management problem. A mixed integer linear programming (MILP) model is proposed to optimise the passenger flow within stations. A case study at London Euston Station shows that reassigning trains to flexible platforms can significantly reduce passenger delays and improve station flow. By incorporating passenger flow into platforming decisions, it enhances the efficiency and safety of station operations.

ABSTRACT. The Train Unit Scheduling Problem assigns train unit configurations to trips, ensuring capacity meets passenger demand. UK train operating companies (TOCs) set a mandatory minimum demand level and a higher desirable level, derived from various sources. Limited resources force TOCs to prioritize capacity allocation, addressing only a subset of trips' higher demand. This is complicated by an "empty-carriage dilemma", where traditional manual methods and fixed thresholds often fall short of optimal solutions. This paper proposes a Demand Assignment with Dempster-Shafer Theory (DADS) method to fuse demand data and define two-level capacity. Testing on real-world data demonstrates superior performance compared to existing methods.

ABSTRACT. Train Rescheduling consists of retiming, reordering and rerouting trains in real-time when unexpected delays or disruptions cause the infeasibility of the planned schedule. In this work, we extend the well-known RECIFE-MILP model, a Mixed Integer Linear Programming formulation introduced in "Pellegrini, P., Marlière, G., & Rodriguez, J. 2014. Optimal train routing and scheduling for managing traffic perturbations in complex junctions. Transportation Research Part B: Methodological, 59, 58–80", to include a new piecewise linear objective function and new constraints accounting for capacity and travel time restrictions, that were considered in the time-space graph-based heuristic algorithm proposed in "Bettinelli, A., Santini, A., & Vigo, D. 2017. A real-time conflict solution algorithm for the train rescheduling problem. Transportation Research Part B: Methodological, 106, 237–265". Computational results on realistic instances show that RECIFE-MILP is successfully extended to the new setting.

ABSTRACT. The Jakarta-Bandung High-Speed Rail (HSR) represents Indonesia's first high-speed rail, significantly reducing travel time between Jakarta and Bandung by over 50%. However, transportation options from the HSR terminus to Bandung city centre remain limited and time-consuming, creating a critical last-mile connectivity gap. This presents an opportunity for innovative transit solutions like Urban Air Mobility (UAM) to complete the journey. This study investigates user preferences and attitudes toward UAM as a feeder mode for the Jakarta-Bandung high speed rail station to Bandung city center. Using a stated choice experiment and PLS-SEM analysis, key determinants of UAM adoption were identified, including travel cost, travel time, attitude, performance expectancy, perceived safety, and social influence. While feeder trains and ride-hailing services remain preferred, UAM shows potential if cost and time efficiencies improve. Findings highlight the importance of leveraging social norms and enhancing perceptions of convenience and value, providing actionable insights for policymakers in advancing sustainable urban mobility solutions.

ABSTRACT. Integrating bike-sharing programs with public transport enhances accessibility and car-independent mobility, yet a comprehensive societal cost-benefit analysis of this integration remains scarce. This study addresses this gap by conducting an ex-durante analysis of the OV-fiets program in the Netherlands, a station-based round-trip bike-sharing system designed to improve last-mile connectivity for train commuters. The analysis reveals that, on average, with a balanced view of costs and benefits, the net present value (NPV) of the OV-fiets scheme is positive, with a benefit-cost ratio (BCR) of 1.5. Primary benefits include enhanced accessibility, reduced road congestion, and improved health outcomes.

ABSTRACT. Abstract: This study examines shared mobility preferences focusing on two distinct settings: a train station, where shared mobility serves as an access and egress mode for train commuters, and an inner city, where local mobility hubs provide shared modes as main travel options for residents travelling within and beyond the city centre. The city of Delft in the Netherlands is used as a case study. Findings reveal that younger users prefer cost-effective shared bicycles and e-scooters as access/egress modes for short distances, while older adults and families favour shared bicycles and cars as main modes for longer trips. Insights support tailored mobility policies to reduce car dependency and enhance accessibility.

ABSTRACT. Access/egress travel to train stations continues to pose a significant barrier to increasing the number of train travellers. Shared micromobility (SMM), including bicycles, e-bikes, steps and mopeds, is often cited as a prominent solution, especially for the activity-end of the trip chain. Using a stated preference survey, we analyse activity-end mode-choice preferences for SMM, walking and public transport (PT) among the Dutch population. By means of a latent class choice model, we uncover three user groups with respect to activity-end mode choice behaviour, namely the Multimodal sharing enthusiast (58%), Sharing hesitant cyclists (16%) and Sharing-averse PT users (27%). The three classes differ in their modal preferences, travel time valuation, as well as in their willingness and ability to use SMM. All three exhibit a high preference to walk for short egress distances, reaffirming the need for transit-oriented development policies. For longer egress distances, PT should be the primary focus at stations in high-density areas with high demand, where high frequencies and dense networks are justified, while stations in lower demand areas are better served by SMM. Providing multiple SMM options would result mainly in competition for the same travellers.

ABSTRACT. Hydrogen buses (H-buses) and battery electric buses (E-buses) are viable zero-emission bus (ZEB) options for achieving carbon neutrality. In Hong Kong, where bus companies are profit-driven and do not rely on subsidies, economic considerations are crucial. The choice between ZEB types depends not only on costs in the vehicle-usage stage such as purchase, fuel, and maintenance, but also on upstream energy supply chain costs, including fuel sources and facility development. The choice of ZEB can, in turn, influence market conditions, affecting costs through economies of scale and infrastructure decisions over the long term. The H-bus market, in particular, may significantly impact upstream energy facility decisions, serving as a potential catalyst for a hydrogen economy in Hong Kong. To address these complexities and long-term interactions, this paper uses a system dynamics (SD) approach for a lifecycle economic assessment from a fuel-cycle perspective. It incorporates three feedback loops—H-bus facility, E-bus facility, and the ZEB market—enhancing traditional lifecycle cost (LCC) assessments by integrating dynamic feedback mechanisms. The findings aim to inform LCC assessments of Hong Kong's ZEB market and offer insights for developing tailored LCC frameworks and modeling feedback decisions in broader contexts.

ABSTRACT. Electric buses play a critical role in advancing urban green transportation. However, a transition period remains before the full conversion to electric buses is achieved. During this period, it is essential to address the scheduling challenges associated with operating a mixed bus fleet of electric and diesel buses. This study considers the scheduling problem for mixed bus fleets using a time-space network modeling approach. To evaluate the proposed model, test instances of varying sizes are generated based on real-world data. The findings provide practical insights and serve as a reference for bus companies managing mixed fleets in their scheduling operations.

ABSTRACT. Bike-sharing has emerged as a crucial first- and last-mile solution in urban mobility, particularly around subway stations where multimodal connectivity is essential. Accurate demand forecasting and intelligent dispatching of shared bikes in these areas present significant challenges due to complex spatial-temporal dependencies. To address this issue, this study proposes a collaborative prediction model—CoGLSTM—for bike-sharing demand, considering its spatiotemporal dependence on subway travel demand. Specifically, metro passenger flow, bike-sharing trip orders, and built environment data are fused to analyze usage patterns of bike-sharing around metro stations. A Graph Convolutional Network (GCN) is employed to model spatial heterogeneity among stations, while a Long Short-Term Memory (LSTM) network captures temporal evolution. Experimental results based on real-world data from multiple bike-sharing platforms in Shanghai show that the proposed CoGLSTM model substantially outperforms baseline approaches in short-term demand prediction. The model offers robust support for dynamic shared bike redistribution and operational optimization in urban transportation systems.

ABSTRACT. In metropolitan areas, low vehicle occupancy rates exacerbate traffic congestion, fuel consumption, and greenhouse gas emissions. To address these challenges, this study proposes a peer-to-peer (P2P) ridesharing model without predetermined drivers. The model enhances matching efficiency through a set-packing formulation and column generation (CG) with a clique-based method to form multi-passenger groups, aiming for maximum cost savings. It incorporates constraints including maximum waiting time and detour distance to improve user experience. Importantly, the model integrates the value of time (VoT) to reflect participants' dissatisfaction with extra travel time. Validated with Chicago taxi data, the model shows a network effect: higher participation improves matching rates, and increases cost and distance savings. Sensitivity analysis reveals that VoT has a more significant impact in low-fare scenarios, especially with low distance rates. For example, under a $0 base fare and $0.10/km distance rate, increasing VoT from $0/minute to $1/minute reduces cost savings by 20% and lowers average occupancy from 2 to 1.5 passengers per vehicle kilometer. These results emphasize the necessity of incorporating VoT in ridesharing models for practicality and user satisfaction. The proposed model is a scalable and policy-relevant solution to increase vehicle occupancy, reduce travel distances, and promote sustainable urban transportation.

ABSTRACT. Shared micro-mobility services have attracted attention worldwide as a smart, sustainable, affordable, and active transportation option. This study focuses on assessing the service quality of bicycle-sharing and e-bike-sharing systems in the National Capital Territory (NCT) of Delhi, India, employing the Relative to an Identified Distribution Integral Transformation (RIDIT) method to analyze and compare user satisfaction. Users of bicycle-sharing systems value these services for being cost-effective, promoting physical well-being, and benefitting the environment, whereas those who opt for shared e-bikes emphasize the convenience of e-bikes and the ability to improve first or last-mile connectivity. The evaluation pinpoints specific strengths and weaknesses across both shared micro-mobility modes, offering crucial recommendations for operators and policymakers to meet user needs better, address service gaps, and encourage frequent usage as well as adoption. The findings underline the role of shared micro-mobility in advancing sustainable urban mobility, especially within the context of developing economies.

ABSTRACT. E-scooters represent a personal micro-mobility solution that enhances sustainable transportation and improves urban mobility. They have the potential to complement public transit and promote mobility equity within cities. This study examines the extent to which e-scooters facilitate connectivity with the subway system in Seoul. Specifically, we identify mobility zones for e-scooter trips that start or end within 100 meters of a subway station and propose a methodology for comparing first-mile and last-mile impacts. Ultimately, this research aims to establish a framework for integrating various transport modes and advancing the concept of Mobility as a Service (MaaS) in the future.

ABSTRACT. This study proposes a method for determining the optimal location and capacity of pick-up and drop-off (PUDO) sections by utilizing Flex Zones as shared autonomous vehicle (SAV) service points. Using taxi Digital Tachograph (DTG) data, high-demand travel patterns and frequently used road segments are identified, and candidate PUDO sites are selected based on clustering of taxi activity. Key spatial variables such as road classification, proximity to public transit, and surrounding facility density are extracted using machine learning models to inform location suitability. A Mixed-Integer Linear Programming (MILP) model is developed to optimize the spatial distribution of Flex Zones. The objective function minimizes total travel distance for passengers and improves service accessibility while satisfying constraints on zone capacity, assignment logic, and maximum number of active PUDO sites. Time-of-day variations in demand are incorporated to ensure operational feasibility under dynamic conditions. The model recommends strategically placing Flex Zones along arterial and sub-arterial roads with strong transit connectivity and commercial accessibility, while dynamically adjusting the number of vehicles served per site to match demand and spatial constraints.

ABSTRACT. This study examines the impact of differentiating electric and gasoline vehicles in order allocations to promote ride-sourcing fleet electrification and aims to provide practical guidance for ride-sourcing platforms. Specifically, this study develops an aggregate and static model of a ride-sourcing platform with electric vehicles that experience longer per-trip service times due to additional charging downtime but incur lower per-trip operation costs than gasoline vehicles. The platform uses differentiated order allocation to promote fleet electrification. Our findings suggest that, unlike monetary incentives that require direct financial investment, differentiated order allocation offers a flexible and cost-effective approach to electrifying the ride-sourcing fleet, potentially improving social welfare without adversely affecting the platform’s profitability.

ABSTRACT. This study develops a semi-dynamic link-based transit assignment model to identify congested railway sections in the Tokyo Metropolitan Area (TMA). While path-based models could address the complexities of through services, multiple fare levels, and systems (flat or distance-based) in the TMA, they require explicit path enumeration, which is infeasible for more extensive networks. Using a link-based structure that avoids path enumeration and incorporating time dynamic effects through a semi-dynamic assignment framework, we successfully predict congestion rates of railway sections across time intervals within the Yamanote Line, though the model requires further calibration.

ABSTRACT. In this paper we present a new algorithm to solve the all-to-all passenger assignment problem,assuming all passengers take the shortest path over a transit network. Compared to previous studies we gain a significant scaling advantage, that depends on the number of transfers in the network. This method is furthermore capable of solving the passenger assignment problem with individual demand requests, and a line plan that includes frequencies and time-tabled busses.

ABSTRACT. This study explores the use of eigenvalue analysis in multilayer networks representing multimodal transportation systems composed of layers of varying sizes. This study demonstrates how eigenvalue analysis can reveal critical structural characteristics of multimodal transportation by using multilayered test networks. Specifically, the largest eigenvalue of the Adjacency matrix (LEV) is used to identify key high-accessibility hubs commonly shared across multiple transportation modes. In contrast, the second smallest eigenvalue of the Laplacian matrix (2nd SEV) is shown to reflect fluctuations in network connectivity, providing a measure of topological equality across the entire system. The results underscore the importance of evaluating integrated transport networks as a whole, rather than analyzing each mode in isolation. The method proposed in this study is to incorporate new transportation modes as additional layers and to determine their optimal placement within the existing network to enhance overall equality. This framework extends the applicability of the eigenvalue analysis and supports the development of equitable, efficient, and sustainable urban mobility systems.

ABSTRACT. Existing public transportation and ridesharing are similar services in that they transport non-drivers. Therefore, the level of service of public transportation may affect the diffusion of ridesharing. This study aims to determine the effect of the level of service of public transportation on the diffusion of ridesharing by comparing assignment results for cases with different levels of service of public transportation using a logit-based stochastic ridesharing user equilibrium model (RUE model). In the proposed model, travelers have three options: as a Rideshare Driver (RD), who travels in his/her own vehicle to his/her destination while allowing for shared seating; a Rider (R) who rides with an RD; and a Public Transportation User (PT), who takes public transportation to his/her destinations. The proposed model explicitly considers the en-route transfer-free condition of R/s by generating a set of paths from a path of the RD. As a result of applying the model to a hypothetical network, we confirmed that only the ratio of RD increases while the ratio of R remains unchanged if the level of service of public transportation declines. The result suggests that the efficiency of ridesharing deteriorates as the level of service of PT decreases.

ABSTRACT. Dwell time makes up a significant portion of total transit trip time and is important for transit agencies and users alike. This study examines the relationship between bus dwell time, the probability of stopping, numerous dwell time determinants, and the level of compliance with holding policies during operations. Different grouping methods are employed to group stops, then dwell time is estimated using the probability of stopping and historical transit data. The accuracy of estimated total trip dwell times are assessed and then improved by accounting for holding policies at timepoints and the probability of adhering to the policy.

ABSTRACT. This study explores the real-time assignment decisions for extraboard transit operators, who are responsible for covering open work resulting from unexpected events such as driver absenteeism. The problem is formulated as a Markov decision process to capture its stochastic nature. Due to the problem's very large state space, an approximate policy is proposed and solved using a backward dynamic programming algorithm. The proposed policy produces high-quality solutions for a test case based on real-world operations.

ABSTRACT. Evaluation and prioritization of public transport infrastructure remain pivotal challenges for urban planners and transport agencies. Traditional Level of Service (LoS) metrics, while useful, often overlook the compounded impacts of delays on passengers, particularly on heavily used transport corridors. This paper introduces the "Men-Hours M-H Factor", a novel evaluation metric that integrates vehicle operational data, travel time deviations, and passenger occupancy to quantify the potential cumulative burden of delays on passenger time. Using Manhattan's and Kaohsiung's public bus network as a case study, the methodology uses extensive data, including GPS-based travel times and hourly passenger counts, to recalibrate LoS metrics and identify high-priority inter-stop segments for intervention. The results reveal significant man-hour savings potential in select inter-stop sections and demonstrate how Man-Hour (M-H) factor shifts prioritization to heavily utilized routes, offering a more equitable and actionable framework for decision-making. By incorporating passenger-centric metrics, this study provides a scalable, data-driven approach to the evaluation and planning of transport infrastructure, with broad implications for sustainable and equitable urban mobility systems.

ABSTRACT. Predicting individual passengers' next trips in metro networks is critical yet challenging due to complex spatiotemporal contexts and long-term dependencies. We propose a spatiotemporal context-aware transformer-based multi-task learning framework that integrates station and time embeddings, a transformer encoder for capturing trip patterns, and a multi-task module for origin-destination prediction with auxiliary time slot estimation. Validated on real-world data, the model outperforms baselines, with ablation studies highlighting key components' contributions. Analysis of contextual factors reveals the influence of passenger regularity, station, and temporal attributes. This work offers a solution for mobility prediction, enhancing transit management and service personalization.

ABSTRACT. Rapid e-commerce growth is straining urban streets and carbon budgets, turning the last-mile into a costly source of congestion and emissions. Crowdshipping—where regular train commuters are paid to carry small parcels between station parcel lockers—offers a low-carbon alternative that repurposes existing passenger capacity rather than adding new vehicles. Guided by the Theory of Planned Behavior, this study combines a survey of 405 Australian rail users with nineteen expert interviews to identify the factors that will determine commercial success. Binary logistic regression shows that environmental concern, confidence in reliable earnings, and peer endorsement are the strongest predictors of a commuter’s willingness to participate, physical comfort and routine train frequency also double or triple uptake. Interviews corroborate these findings and stress three operational imperatives: strategically placed lockers that do not disrupt passenger flow, simple yet transparent pay-and-insurance schemes, and referral programs that convert early adopters into ambassadors. Demographic variables proved largely irrelevant once these design features were addressed, indicating that a single, well-structured offer can attract a broad user base. The paper concludes with actionable recommendations for policymakers, transit agencies, and logistics providers seeking to invest in public-transport-based crowdshipping

ABSTRACT. We propose a generalized discrete choice modeling framework that unifies and extends previous AI-integrated approaches. Our model features three components, including (1) feed-forward neural networks to identify heterogeneous latent classes through nonlinear combinations of demographic variables helpful to derive policy/equity implications and resolving endogeneity and (2) class-specific embedding matrices to reduce the dimensionality of high-cardinality categorical variables and to provide rich behavioral insights, and (3) a conventional linear utility formulation to preserve econometric interpretability. Results highlight the importance of appropriately assigning input variables to specific roles and demonstrate that the integration outperforms existing models.

ABSTRACT. Public transport enhances urban mobility, making accurate bus arrival predictions essential for Advanced Traveller Information Systems that promote public transport use. This study develops a unified multi-citycity-agnostic, multi-city model for bus arrival time prediction using data from seven Chilean cities. Leveraging GTFS-RT feeds, historical commercial speeds, and real-time headway information, we compare the performance of XGBoost, CatBoost, and LightGBM. CatBoost consistently outperformsoutperform all other approaches, including city-specific models and considering complex scenarios where data from the target city is excluded during training. We further demonstrate that this unified approach can be extended to city-agnostic predictions, where reliable estimates are provided for cities not included in the training data. Using SHAP analysis, we identify key predictive features and their contributions across different urban contexts. These findings demonstrate the feasibility of scalable, unified prediction models across diverse urban environments, with significant implications for transit authorities managing multi-city systems with limited resources. the other models, especially in complex scenarios where data from the target city is excluded during training. These findings demonstrate the feasibility of scalable, generalizable ETA prediction models across diverse urban environments.

ABSTRACT. Smart card data in public transportation provides a continuous and detailed source of information on passenger movements. Although typically limited to entry validations, this data can be enhanced using destination estimation algorithms to reconstruct origin-destination trips. These reconstructed trajectories are essential for calculating key performance indicators such as passenger-kilometers, occupancy rates and schedule adherence supporting both daily operations and long-term planning. While existing algorithms are generally effective at estimating destinations for most trips, some cases remain difficult to resolve particularly isolated or incomplete transactions. To address this an improved estimation model is proposed. It combines criteria based on stop sequence analysis, the travel history of the individual smart card and a final criterion leveraging the collective historical data of all cards in the system. This last approach uses a weighted random sampling method to select the most probable alighting points. The combination improves both the coverage and coherence of the estimations. To ensure reliability the algorithm’s outputs are compared with passenger counts at tram stations. Although validation revealed some errors and left uncertainty about accuracy this comparison significantly reduces the risk of inaccuracies compared to relying solely on smart card data.

ABSTRACT. During boarding and alighting in urban mass transit, some train doors become overcrowded while others are underutilized. This imbalance increases dwell time as trains wait for the last passengers to board or alight. We propose a MINLP which may be nonconvex to minimize boarding and alighting times—a function of the number of boarding and alighting passengers—by optimally allocating passengers to doors. To capture the process of passengers’ choosing doors, we developed a choice model with financial incentives to guide passengers to specific doors. Optimal discounts are determined within the choice model and integrated into the optimization framework to adjust passengers per door.

ABSTRACT. This study improves university transit system ridership forecasting using Direct Demand Models (DDMs). Key objectives include improving prediction precision and exploring factors affecting passenger counts. Models were divided into undergraduate and graduate categories, as well as all-day and peak-period models. Fixed effects regression addressed route-level variables, while log transformations handled heteroscedasticity and allowed elasticity interpretations. Findings show delays, weather, enrolment patterns, new housing, academic quarters, and day of the week significantly impact ridership. This research translates the findings into a practical tool designed for transit agencies operating in university settings to support better-informed service planning decisions, enhance operational efficiency, and improve the overall transit experience for riders.

ABSTRACT. This study assesses spatial spillover effects due to public transportation network changes on ridership using a spatial Difference-in-Differences (DiD) method. Traditional DiD approaches neglect spatial spillovers and dependencies; hence, this study integrates a spatial dimension with distance-based spatial weights to capture both direct and indirect effects. Data from Nicosia, Cyprus (2018–2023), comprised of bus routes, frequency, socio-demographics, and land use, the analysis reveals mixed effects for urban areas and positive effects for rural areas. The findings highlight statistically significant causal effects, emphasizing the importance of spatial considerations in policy evaluation for public transportation improvements.

ABSTRACT. Traffic volume indicates potential for mode shifts towards public transportation. Estimating traffic volume for rural roads is challenging because most transport models are developed for urban areas. This study investigates a multiple linear regression model for traffic volume prediction in the Norwegian County Vestland. The best regression model utilizes the road width, the census unit population, the municipal household density, the number of registered vehicles, and a dummy for build-up places and forests to explain 0.64 of the variation in traffic. The analysis reveals a need for more rural measurements including vehicle lengths and small-scale socioeconomic data for accurate estimations.

ABSTRACT. Modern cities are implementing various public transportation policies to address issues like traffic congestion and improve urban mobility. Autonomous buses are expected to initially run in exclusive median bus lanes, making accurate bus speed prediction increasingly important. In this study, a model for predicting bus speed in EMBLs was developed by comprehensively considering Bus Management System and smart card data, along with factors that can affect autonomous driving constraints. An ensemble machine learning approach was applied to enhance prediction performance. The findings provide accurate speed predictions and support stable and efficient operations during the initial adoption of autonomous buses.

ABSTRACT. Nowadays, urban travellers choose from or jointly utilize multiple transportation modes to complete their trips. Prediction of travel demand in this multi-modal context is crucial yet challenging due to the dynamic and complex interactions between modal usage. In this work, we propose a Dual-hierarchical Dynamic Graph Neural Network (DHDG) for simultaneous prediction of arrival and departure demand for multiple transportation modes. Specifically, DHDG introduces a copula-based module to generate subgraphs at two hierarchies: OD-related intra-mode demand, and local inter-mode demand, and a fluctuation detection module update them when triggered. Preliminary experimental results show that our proposed method achieves comparable high accuracy in predicting long-term multi-modal transportation demands.

ABSTRACT. On-demand transit (ODT) service, provided by transit agencies, have grown rapidly in the last few years in different countries and cities around the globe. The popularity of ODT resulted in multiple challenges, a common challenge is frequent ride cancellations and no shows. Unfortunately, very little is known about ODT cancellations because of the complex data sharing agreements of trip records, due to the sensitivity of such datasets. This study examines a rich dataset of individual ODT trip records across the Regional Municipality of Durham in Canada for 18 months between January 2022 and June 2023. The trip records feature both, completed trips and cancelled trips. This paper presents models that can predict cancellations based on trip and rider characteristics using choice modelling. These models would provide transit agencies with influencing factors that increase cancellations and therefore help formulate better cancellation policies. A mixed-effect logistic regression model is used, with varying random effects of the service zone. This allows us to capture unobserved effects based on individual ODT service zones. The results showed that increased walking time at pickup and dropoff, increase schedule deviation, and early booking significantly increase trip cancellations.