ABSTRACT. Smart card data offers an in-depth understanding of travel behaviour of public transport users. An efficient way to analyze public transport users is to group them into different clusters of similar behaviour. However, this clustering process should take into account space and time, because both of these dimensions characterize daily trips. Depending on the outcome, we might wish to give more importance to space or to time, or we might wish to balance the two. In this study, we present a spatiotemporal clustering tool that permits modulation regarding to the importance of space versus time. We then test this tool with different values for the space-time balance parameter to evaluate the influence of this parameter on the results. The method has been applied to 769,614 smart card transactions of the Réseau de transport de la Capitale (Quebec City, Canada). Results show that the influence of space and time can indeed be controlled, and that the types of clusters obtained vary whether one or both of the dimensions are considered.

ABSTRACT. Travel locations do not exist individually and independently. Movements between two travel locations are highly reciprocated over a day, but not within a day. This presentation explains an innovative framework to analyse diverse mobility by conceptualising locations as individual entities in a network and performing dyadic analysis to study each pair of locations for their dyadic intra-day and intra-week “interactions” and “relations”. If a location can have friends, can it be friendlier to one than another? We will demonstrate the application of the framework on London’s rail demand data.

ABSTRACT. We collect tracking data from individuals who agreed to provide this information via the OneBusAway application. The data is collected with Google’s “Android Activity Transition Application Programming Interface” and provides very detailed information as to the location of the person and the type of movement. To construct complete tours, however, significant data cleaning is required. We discuss these steps and show how the cleaned data can be used to create meaningful trip and tour records including information as to when a person starts to walk to a bus stop, which bus is taken and what the final destination is. Potential applications of the cleaned data are discussed.

ABSTRACT. Inadequate public transportation (PT) infrastructure in urban areas enforces many PT users to own private cars thereby substantially increase already overloaded traffic flows. To improve the situation, the PT structure should be adapted to the real PT users’ needs. The last may be achieved only if their perceptions are well known. This study extends the previous research conducted in New-Zealand by exploring the perceptions of the Israeli PT users. Our results show that distinct social-economic characteristics and different transportation systems result in quite diverse preferences and thus, may lead to setting distinct transportation policies.

ABSTRACT. Feeder services are public transit services that transport people from a low demand, typically suburban, area to a high demand area, such as a transportation hub or a city. Here, passengers continue their journey using traditional forms of public transport. On the one hand, on-demand feeder services have been a topic of discussion in a number of recent studies, since these services can serve the demand efficiently. On the other hand, traditional feeder services with predetermined routes and timetables provide predictability and easier cost control. In this paper, a demand-responsive feeder service is considered, which combines positive characteristics of both traditional services as well as on-demand-only services. This feeder service has mandatory bus stops which are always serviced, as well as optional bus stops which are only serviced when there is demand for transportation nearby. Experimental results on 14 benchmark instances illustrate that the LNS algorithm obtains solutions with an average gap of 1% or less compared to the optimal solution, within 1s of runtime. Larger instances can also be solved, typically in less than 60s. The results also show that the demand-responsive feeder service generally outperforms a traditional service in terms of service quality, often by more than 60%.

ABSTRACT. Public transport is about meeting the travel needs in a timely and cost-effective manner. Lee et al. (2021) proposed a zonal-based flexible bus system (ZBFBS) that divides the service area into geographical zones and categorizes the ride requests by the zonal origin-destination (OD) to provide customized door-to-door transit service to riders. They considered price elasticity, stochasticity of demand volume, as well as stochastic location, to improve the authenticity of the model while planning. This study proposes a two-phase framework to optimize the zone sizes and demarcations. The first phase optimizes the zone demarcations to cover the whole service area without overlapping, and the second phase is a two-stage stochastic problem to optimize the profit of the operator of ZBFBS. The proposed two-phase framework is applied to a scenario created from real ridehailing data of the city centre of Chengdu. By optimizing the square zoning design, the profit increases by 11.1% compared with the base scenario that divides the service area into 3x3 equal zones; the hexagonal zoning design increases the profit by 11.3% compared with the base scenario. Together with our previous research, this framework can be applied to cities with adequate data to determine the flexible bus zonal division, routes, and schedule to provide door-to-door flexible transit services to passengers.

ABSTRACT. We propose Hyper-pool, an analytical, offline, utility-driven ride-pooling algorithm to aggregate individual trip requests into attractive shared rides of high-occupancy. We depart from our ride-pooling ExMAS algorithm where single rides are pooled into attractive door-to-door rides and propose two novel demand-side algorithms for further aggregating individual demand towards more compact pooling. First, we generate stop-to-stop rides, with a single pick up and drop off points optimal for all the travellers. Second, we bundle such rides again, resulting with hyper-pooled rides compact enough to resemble public transport operations. We propose a bottom-up framework where the pooling degree of identified rides is gradually increased, thereby ensuring attractiveness at subsequent aggregation levels. Our Hyper-pool method outputs the set of attractive pooled rides per service variant for a given travel demand. The algorithms are publicly available and reproducible. It is applicable for real-size demand datasets and opens new opportunities for exploiting the limits of ride-pooling potential. In our Amsterdam case-study we managed to pool over 220 travellers into 40 hyper-pooled rides of average occupancy 5.8 pax/veh.

ABSTRACT. Between a conventional public bus system and a complete on-demand system, a range of demand responsive options exists which can be designed based on information about the actual demand. In this study, we focus on the morning peak hours where the passenger flows towards a city center are typically much larger than the flows in the opposite direction. We introduce a system where short-cut trips, away from the city center, are allowed for a single line. Based on the expected demand, it is decided, whether a bus should visit all the stops ahead or skip some of them to take a shorter way in the return trip so that it increase the frequency of the service towards the city center. We present a Mixed Integer Quadratic Program to mathematically model this problem. Due to its complexity, only small-sized problems can be solved optimally. Therefore, we also design a metaheuristic algorithm based on Variable Neighborhood Search that finds high-quality solutions within reasonable time for realistic instances. The results show that the demand responsive system can improve the total passenger travel time with up to 10\% compared to the conventional system.

ABSTRACT. From a passengers' perspective there are two types of timetables: periodic and non-periodic ones. Periodic timetables are repeated, e.g., every hour and provide a reliable and easily memorable service. Aperiodic timetables schedule the vehicle trips irregularly over the day. In this work we compare periodic and non-periodic timetables analytically. Since periodicity is an additional constraint, an aperiodic timetable can be better adapted to the demand. We quantify these effects on the simple example of one single line in which it is possible to compute the best periodic and aperiodic timetables. Our results allow conclusions also for more general transport networks.

ABSTRACT. We present a new mathematical formulation for tactical railway timetabling that aims at minimizing total passenger perceived travel time. This new formulation, based on the POT model by Polinder et al. (2022) relaxes the generally adopted assumption that line frequency is given as input and considers instead a minimum and maximum frequency. We expect experimental results to improve on current state-of-the-art methods where a trade-off between fewer trains running can lead to a lower average perceived travel time, coming from better scheduling of most utilised lines.

ABSTRACT. We develop an iterative approach to jointly create line plans and timetables for railways. Compared to existing iterative approaches that ban a whole conflicting line plan, our method is able to identify the smallest set of incompatible services accurately. Besides, by efficiently exploiting the available railway infrastructure, our method accounts for all the possible routing options of trains, which are usually neglected in integrated railway planning to preserve tractability. Using data from a railway company in Switzerland, we show that our approach is practical for solving real-life instances and significantly faster than existing benchmarks. Our insights on the necessity of banning conflicts and considering infrastructure inform decision making in railway planning

ABSTRACT. We develop an optimization model for the generation of a transit service timetable. Based on the desired headway and/or frequency preferences for individual routes and demand corridors, we generate an optimal timetable from scratch. The optimization minimizes the costs of operating the service under different constraints. In addition, using real-time data we maximize the on-time performance (OTP) measure of the service.

ABSTRACT. In this study, we develop a generic framework to quantify the robustness benefits of new rescheduling infrastructure for urban rail networks. The proposed framework identifies the rescheduling infrastructure where these benefits are maximised. These benefits can be included in appraisal studies or cost-benefit analyses, to support policy-makers in prioritising infrastructure investments.

ABSTRACT. During railways operations, unplanned events might occur which can result in traffic being heavily impacted. The paper proposes a passenger-centered resilience assessment for a disruption scenario which consists of multiple disruptions. It combines train traffic operations, passenger flows and network restoration. To evaluate resilience, an optimization based approach has been developed. Additionally, this approach develops mitigation plans for the best infrastructure restoration and traffic recovery and it captures the time-dependent transport network performance during disruptions. The performance of the proposed approach has been demonstrated on a Dutch railway network. This optimization-based approach shall enable decision makers to quantify accurately impacts of multiple disruptions by considering the created inconveniences to passengers in the railway operation due to these disruptions.

ABSTRACT. We present a topological approach to assess recoverability of public transport networks. The approach is based on previous works dealing with optical networks and adapted to the context of public transport. Two failure and three recovery strategies are defined and evaluated for the metro networks of Berlin and Paris using two performance metrics. Preliminary results suggest that simple greedy recovery strategies are able to quickly bounce back from the loss of performance inflicted during the failure process and that the proposed methodology is a promising approach to assess recoverability in public transport networks.

ABSTRACT. This article proposes to understand how the occupation of public transport vehicles and the proper use of face masks by passengers influences people's travel decisions, using data from a stated preference survey carried out in Santiago, Chile, in 2020. Through three modal choice models, we propose that the penalty for overcrowding increases with a non-linear (potential) trend, both with vehicle occupation and with the percentage of passengers not wearing face masks in buses and trains. Moreover, women and people with health conditions are more sensitive to passengers not using masks.

ABSTRACT. Extended abstract attached.

ABSTRACT. Across the world the Covid 19 pandemic has necessitated consideration of a fundamental trade-off: how to achieve sufficient containment of disease transmission while allowing for opportunities afforded by mobility. In this paper, we address this trade-off in the context of urban transit networks. We present a flexible agent-based simulation framework to maximise opportunities for safe mobility and economic interactions on urban transit networks while minimising the potential for Covid-19 contagion. We model a range of operational interventions with varied demand levels and social distancing constraints including: alterations to train headways, dwell times, signalling schemes, and train paths. We undertake a case study of the Victoria Line on the London Underground. Based on a multidimensional performance criterion accounting for operational costs and the societal economic benefits of travel, we determine the optimal headway for each operational scenario. We show that average performance gains of 352%, 302%, 141%, and 120%, are achieved when comparing the optimal with the least optimal value of headway for the base moving block, moving block with longer minimum dwell times, fixed block, and skip-stop schemes, respectively. Furthermore, we find that the more advanced moving block signalling system provides an average performance gain of 106% compared to the older fixed block signalling system, thus supporting the case for upgrading signalling technologies. Our results provide clear evidence-based support for interventions to manage transit demand and capacity during the Covid-19 pandemic.

ABSTRACT. The last decade has seen a considerable move forward regarding integrated vehicle and crew scheduling in various realms (airline industry, public transport). With the continuous improvement of information and communication technology as well as general solvers it has become possible to formulate more and more rich versions of these problems. In public transport, issues like rostering, delay propagation or days-off patterns have become part of these integrated problems. In this paper we aim to revisit an earlier formulation incorporating days-off patterns and investigate whether solvability with standard solvers has now become possible and to which extent the incorporation of other aspects can make the problem setting more rich and still keep the possible solvability in mind. This includes especially issues like delay propagation where in public transport delay propagation usually refers to secondary delays following a (primary) disturbance. Numerical results are provided to underline the envisaged advances.

ABSTRACT. This paper considers the railway crew scheduling problem taking Indian Railways as a test case. The paper proposes an algorithm to calculate a lower bound for the total crew requirement to operate a set of scheduled passenger train services. The algorithm is tested on seven sets of passenger train schedules and the lower bounds are compared with the optimal crew requirement obtained using an exact method. The lower bounds calculated using the proposed algorithm are found to be optimal in six out of the seven test cases. The proposed algorithm ensures a tight and feasible lower bound for the total crew requirement in quick time.

ABSTRACT. Zero emission policies in urban centers are forcing many transit agencies to convert their fleets to electric buses. The shorter driving range and the need for special charging infrastructure, among other things, must be taken into account when planning and scheduling electric buses. This talk presents the Multi-Depot Electric Vehicle Scheduling Problem with Stochastic Travel Time and Energy Consumption (S-MDEVSP). Vehicles are allowed to be partially recharged and a non-linear charging function is considered. Our model takes advantage of the full information on the current state of charge that is available in operation by allowing planned charge time to be extended when energy consumption deviations are observed. We formulate the S-MDEVSP as an integer program and a column-generation-based heuristic is provided to solve real-life instances.

ABSTRACT. We address the personalised crew rostering problem in a context where each employee has his own skills, seniority and preferences, where all work has to be assigned and optimisation goals include maximising preference satisfaction and fairness. We propose an approach that combines operational research with auction theory for addressing scenarios where employees may prefer different things and seniority may not be strict, which have not yet been addressed. We evaluate the approach with real world data.

ABSTRACT. We address the optimisation of rolling stock rotations with maintenance constraints as a profit maximisation problem with fixed costs. We propose an approximate solution method that combines in a new way existing techniques. We report real revenue growths and cost savings obtained by VIA Rail Canada while using this solution method in practice, as well as optimiality gaps for small problem instances.

ABSTRACT. The emerging modular autonomous vehicle technology allows vehicles to connect and disconnect with each other in motion so that MAVs can join into a platoon en-route for less energy consumption and to reposition on-board passengers and freight between vehicles. Prior routing models adapted the pickup and delivery problem with transfers to include synchronized transfers made en-route. However, the platooning capability was not considered. A mixed integer program is proposed for the pickup and delivery problem with synchronized transfers and vehicle platooning. A two-phase heuristic algorithm is proposed to construct routes and iteratively improve the solution quality. Computational tests are conducted on a series of randomly constructed instances and compared to benchmark heuristics and a prior heuristic developed for the model without platooning.

ABSTRACT. This paper studies the optimal planning and the operational design of public transit services with modular-vehicles. In such service operation, the vehicles are comprised of smaller modular units/pods, which can be assembled/dissembled in some specially designed transit stations. The planning decisions are to determine the locations of the specially constructed stations enabling station-wise assembling/dissembling of modular units. Meanwhile, the capacity of these stations, i.e., the maximum number of modular units that can be accommodated, will also be determined. At the operational level, the optimal vehicle formation at these stations will also be decided to consider the multi-period passenger demand. Aiming to minimize the total cost of the operator and passengers, we formulate this optimization problem into a mixed integer nonlinear program (MINLP). To solve the problem, we first transform the MINLP into a mixed-integer linear program (MILP) using various linearization reformulation techniques. The reformulated MILP is then solved by using the existing solution methods. A case study based on a proposed Singapore dynamic autonomous road transit (DART) line is conducted to illustrate the validity of the proposed model and solution method. The results demonstrate its service performance in significantly reducing the operating cost and passengers' travel time costs in the modular-vehicle transit system.

ABSTRACT. Public Transport electrification holds great potential in promoting sustainability and decarbonisation in the mobility domain. However, great care must be taken in ensuring that the shift towards electric vehicles is accompanied by appropriate decision support systems, in order to ensure that the transition does not introduce negative externalities for transit operators, instead enabling them to exploit novel technology to its fullest. In this work we focus on developing a solution scheme for the Mixed-Fleet Multi-Terminal Electric Vehicle Scheduling Problem which exhibits appropriate scalability and can therefore be applied to urban-sized instances.

ABSTRACT. In recent years there has been an increasing desire to develop alternative solutions to traditional energy sources. Since transportation is a significant fossil-fuel consumer, the development of electric vehicles, especially buses, has the potential to reduce fossil-fuel use and thus provide a better living environment. In this study, an enhanced optimal allocation model for designing the allocation of wireless bus charging stations is introduced and demonstrated in the shuttle route of Bar-Ilan University.