ABSTRACT. Several cities around the world are contemplating the potential of an urban air mobility (UAM) system based on aerial vehicles with vertical take-off and landing capabilities, with or without pilots, to provide relief for congested urban streets. In this talk, we discuss various service design concepts for UAM, and present an agent-based simulation framework to evaluate various dynamic operational strategies for fleet assignment and routing. The framework is applied to evaluate potential market size and system performance in a large metropolitan area under different service design parameters. The talk also examines issues of airspace capacity, and the extent to which congestion might become a consideration in the design and operation of UAM systems. Taking separation requirements and UAM vehicle attributes into consideration, a microsimulation is developed to model UAM traffic flows with a decentralized conflict resolution scheme. The results show that like highway network traffic, UAM traffic can be characterized by a macroscopic system fundamental diagram with density and flow defined consistently with a 4D adaptation of Edie’s definitions. These findings also suggest the presence of a capacity of airspace that should be carefully managed by operators to achieve optimal system performance.

ABSTRACT. Urban Freight Transport is a continuously growing market mainly based on the use of vehicles with combustion engines, whose environmental impact has become unsustainable. Because of the technological improvement of electric vehicles and their growing economic feasibility, the introduction of electric fleets for urban freight distribution is now a considerable opportunity. Cities are rapidly adapting, in need of tools to properly guide and manage these changes, as the rise of electric vehicles must be encouraged by an appropriate infrastructural system, from charging stations to dedicated areas. What is proposed in this work is an aggregate approach to the freight system, transport demand and supply, to support the design of a distribution system based on electric vehicles by means of an accessibility indicator that takes into account the supply of facilities, vehicle performances, and freight demand patterns. A study case regarding the Metropolitan City of Rome is also presented to interpret and understand the potentialities of this approach.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Transport demand modeling is a crucial stage to be implemented for a variety of optimization techniques in the field of transportation management. The paper describes the model for simulations of demand for goods’ deliveries by cargo bicycles in the area with the restrictions of motorized traffic. The proposed model is based on the classical four-step approach with additional procedures for generations of demand as the flow of requests for cargo deliveries. The paper presents the case study of the demand generation in the Old Town district of Krakow city: the obtained simulation results are characterized by high adequacy. The simulation model, implemented in Python programming language, can be used as the key tool for solving optimization problems, such as the localization choice for loading points or substantiation of the optimal number of cargo bicycles used for deliveries in the region.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This paper presents a GIS-based approach to evaluate the feasibility of crowdshipping services in the context of university community. The case study analysed focuses on e-commerce deliveries and takes into account a campus with venues located in different zones in the city of Catania (Italy). The methodology is designed according to spatial considerations related to the proximity of delivery points and home addresses, students’ flows between origins and destinations and main mode of transport used. Results are useful to design the service in a well-established community, which could be considered more inclined to be involved.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The urban consolidation centre (UCC) is the most popular city logistics (CL) initiative for both, researchers and practitioners. Urban freight transport (UFT) models are often used to assess the impact of the UCC on urban freight flows. The number of UFT models are based on the origin-destination (O-D) matrices for freight distribution. However, the static nature of O-D matrices imposes the limitations to the O-D based models regarding the CL initiatives impact assessment. We propose a new way of aggregated urban freight flows modelling, whereby the O-D matrices are transformed depending on the number, location and the share of freight flows attracted by the UCCs introduction. To exemplify the idea, the model is applied in the case of Novi Sad, whereby three different UCC locations and seven distribution patterns were compared. The results might be of interest for the researchers interested in the CL initiatives impact assessment and practitioners involved in the UFT decision-making process.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Many benefits are expected from autonomous vehicles (AV), still, simulation models that assess AV services seems focused at the tactical level. This work investigates on how the performance of AV services has been assessed on the literature based on simulation models with an emphasis on socioeconomic aspects. Results show that technical and economic indicators are considered in all papers. In contrast, environmental and socioeconomic indicators are three times less studied and their level of treatment is also poorer. In addition, most of studies are based on agent-based models, which are more accurate to assess dynamic on-demand services. Future research should focus on refining socioeconomic KPIs and the environmental measure methodology. Agent-based models should be useful to conduct deeper analysis on a strategic level.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The multi-commodity network flow problem (MCNF) consists in routing a set of commodities through a capacitated network at minimum cost and is a relevant problem in applications that include transportation and telecommunication. In this paper, we study a generalization of the MCNF in which time limits are imposed as soft constraints on commodity travel time, and delays are permitted but penalized using a non-decreasing penalty function of the travel time. We propose an extension to both arc and path formulation of the MCNF to account for soft time constraints and, for the latter formulation, we outline a column generation procedure based on a resource-constrained shortest path variant. Our numerical experiments conducted using realistic liner shipping instances show that our column generation approach can solve large instances to optimality within a few seconds.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The paper deals with the problem of minimizing the reshuffling of containers in an inland intermodal terminal. The problem is tackled according to a hybrid approach that combines a preliminary selection of heuristics and a genetic algorithm. The heuristics are used to determine the initial population for the genetic algorithm, which aims to optimize the locations of the containers to store in the yard in order to minimize the operational costs. A simulation model computes the costs related to storage and pick-up operations in the yard bay. The proposed optimization method has been calibrated by selecting the optimal parameters of the genetic algorithm in a toy case and has been tested on a theoretical example of a realistic size. Results highlighted that the use of a suitable heuristic to generate the initial population outperforms the genetic algorithm, initialized with a random solution, by 20%.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. To improve efficiency and sustainability in synchromodal logistics, it becomes crucial to mitigate the disruptive effects caused by non-synchronized operations. In this work, we study the Synchronized Location-Transshipment Problem in which the decisions on the facilities to select are fundamental to ensure the synchronization of flows and the correct exploiting of just-in-time logistics and cross-docking operations. We propose a Mixed-Integer Linear Programming formulation for the problem and perform an economical analysis in order to address several realistic situations based on different sizes of the transshipment facilities and location types. Results of the tests allow us to derive managerial insights and to define future research lines.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The urban distribution of freight is a challenge that is gaining more and more momentum over the years because of a skyrocketing use of e-commerce. Even if market places could almost fully rely on TIC platforms in future years, physical networks for the distribution will be always needed. In this supply chain, the last mile represents on average 29% of the total delivery costs (Lopez, 2017). To make these operations more efficient and reduce their economic weight, new types of robotic technologies – mainly Ground Autonomous Delivery Devices (GADD) and drones – are emerging. Unfortunately, the research in this field is nowadays essentially oriented towards operations and focused on finding new algorithms to optimize the distance travelled by different combinations of these vehicles when they operate in the city. More strategic insight is needed to help cities define adequate frameworks. The main objective of the thesis is to gain knowledge about the economic potentialities of drones and GADDs as well as their future impacts on cities in the context of the urban last-mile freight distribution. Viable Business Models for these brand new logistics services will be defined, resulting in an improvement of business opportunities for carriers and more sustainable operations from the city perspective.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. In this work, we focus on the usage of drones (unmanned aerial vehicles, UAV) in the last-mile distribution and we introduce a new variant of the flying sidekick traveling salesman problem (FS-TSP). The FS-TSP is a drone assisted parcel delivery problem aimed at defining the distribution of a driver-operated truck assisted by an unmanned aerial vehicle. The truck and the UAV must depart and return to a single depot, either in tandem or independently. Each customer must be served once by either the truck or the UAV. The truck has an infinite capacity and acts as a mobile depot for the UAV, replenishing its batteries and providing the parcels to be delivered. The drone is launched from the truck to serve a customer and then it comes back to the truck to pick up a new parcel to deliver or to the depot. The launching node and rendezvous node of each drone sortie must be different. In this work, we consider the case where the drone can be launched and return in the same node. We propose an original formulation for this problem and develop a row-and-column generation approach to solve it. The proposed method has been experienced on test-bed and new instances. The obtained results show that our method can eectively solve to optimality instances with up to 20 customers and prove that this variant can reduce the overall delivery time, so improving the performance of the delivery process.

Download: (a) Extended Abstract

ABSTRACT. Air pollution remains as a key unresolved problem in many urban areas. Cities with such problem are gradually implementing Traffic Management Strategies (TMS) to reduce the total kilometers travelled by vehicles and subsequently decrease emissions. However, a prior evaluation of such TMS is needed if the target goals want to be achieved. In this sense, the combination of traffic simulation with emissions and air quality models can be of great use to assess the potential impacts of such policies. This paper presents an integrated modelling system tool for Barcelona that allows to estimate the changes induced by the implementation of TMS on traffic activity, emissions and air quality levels at a very spatial (street level) and temporal (hourly level) resolution. Some illustrative examples of the outputs that can be generated with the tool are presented.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. To promote active modes to counteract the increasing negative impacts of the transportation sector, this study develops an information system for cycling navigation based on 7 different bikeability indicators: travel time, human energy expenditure, effort distribution, infrastructure performance, safety, comfort and emission hotspots. Therefore, field data was collected in a selected cycling network map of Aveiro, Portugal, by a single cyclist during the weekdays’ afternoon peak hour period. A conventional aluminium bicycle equipped with a GPS data logger, a wireless heart rate recorder device, a video camera and a carbon monoxide detector was used. Using the defined methodologies as well as GPS Visualizer and ArcGIS, a total of 8 hours of video and approximately 100000 second by second data points were analysed and organized through a 460-link map. Several optimal solutions for different OD pairs were studied through multivariate analyses tools. Results show significant tradeoffs between the traced routes according to the chosen type of indicator, pointing the information system’ utility in providing useful information to cyclists and support management systems.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Road traffic accidents represent one of the leading causes of death across all age groups globally. Most of these accidents can be directly attributed to drivers’ failure to select correct driving speed. Thus, actions aimed to mitigate inappropriate driving performance, including speeding, are needed. This study, using a driving simulator, investigates the effects of different real-time coaching programs on driving performance, specifically on the occurrence of Elevated Gravitational-Force Events (EGFEs).

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This study examines the connection between an agent-based transport simulation and Nash flows over time. While the former is able to represent many details of traffic and model large-scale, real-world traffic situations with a co-evolutionary approach, the latter provides an environment for provable mathematical statements and results on exact user equilibria. The flow dynamics of both models are very similar with the main difference that the simulation is discrete in terms of vehicles and time while the flows over time model considers continuous flows and continuous time. This raises the question whether Nash flows over time are the limit of the convergence process when decreasing the vehicle and time step size in the simulation coherently. The experiments presented in this study indicate this strong connection which provides a justification for the analytical model and a theoretical foundation for the simulation.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Enhancing the sustainability of mobility systems is increasingly critical for cities and metropolitan areas. We must therefore understand mobility requirements much better and need to guarantee a constant monitoring and evaluation of the actions implemented to improve sustainability, through acceptable methodological tools. In such context, this research develops a decision support tool for the implementation of sustainable mobility policies in emerging cities. The approach is based on system dynamics to model the relationships between state and flow variables, organized in feedback loops. We propose a multi-layered model to analyze the cause-and-effect relationships within a system that integrates high-level policies, specific strategies and actions, and a sub-system to measure the impact of those policies and actions, in terms of sustainability. For this purpose, a specific set of indicators and dimensions was developed and validated in some case studies

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. User assignment models in transit networks are relevant for the planning of new lines but less relevant for management. Other types of models are used for management, which is very efficient for the simulation of supply but less efficient for the simulation of users, who nevertheless influence train traffic on urban railway lines. This paper proposes to use a combined train and passenger simulation model on a railway line to propose line management solutions in order to minimise the passenger travel time. An application of this method on line 13 of the Paris subway network, shows how the relationship between frequency and dwell-time impact the railway traffic. Precisely, the current frequency of the line associated with a maximum dwell time of the 40s minimise the travel time. These optimal time conditions correspond to significant congestion in the line's train traffic.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Selecting the best scenario of intervention to improve the functioning of complex systems represents a troublesome task both for public and private decision makers, since it requires the deployment of appropriate analysis and evaluation tools. In the case of intermodal transport systems, complexity is given by the execution of different types of activities using various resources, and by the presence of several actors operating in the same environment with diverse goals. In this paper, an integrated approach to assess design alternatives regarding rail port operations is proposed by combining business process modeling and multi-actor multi-criteria evaluation. In fact, a railway process has been graphically represented by means of the standardized modeling language called Business Process Modeling and Notation (BPMN) at different levels of detail, i.e. taking into account not only the actual transport operations but also the necessary documentary procedures to perform the freight transfer services. In addition to the identification of possible bottlenecks, the analysis of the considered railway process has enabled the determination of its most significant features. These parameters have been subsequently used as some of the criteria according to which the performances of the examined scenarios of intervention have been evaluated, adopting the Analytic Hierarchy Process (AHP) technique. Furthermore, the appraisal has been enhanced by explicitly including the key stakeholders involved in the railway process at hand. The developed methodology has been applied to the case study of the Port of Trieste, Italy, in order to investigate possibilities for an increase in railway capacity.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The Locomotive Scheduling Problem (LSP) plays a crucial role in the railway industry. The aim is to assign a fleet of locomotives to a set of scheduled trains such that the overall costs are minimized, while several assignment restrictions and operational requirements are met. As the rolling stock represents one of the main costs, this is a key challenge of a rail company. We formulate this optimization problem on a sparse weighted directed multi-graph. Based hereon we propose a Mixed-Integer Linear Program (MILP) that scales well and delivers optimal solutions for many LSP instances in freight transportation in reasonable time. However, in practical applications, maintenances for the rolling stock are of central importance. Thus, we adapt our model by introducing the required maintenance constraints, now speaking of an LSP with maintenances (LSPM). With increasing complexity our adapted MILP becomes intractable even for small instances, making an alternative solution approach necessary. In this talk we present an efficient two stage approach that yields high-quality locomotive assignments implementable in real-life. In the first stage, an optimal solution for the related LSP is determined by solving the MILP. In the second stage, a heuristic procedure is applied to iteratively resolve maintenance violations. We evaluate our approach using instances generated based on real-world data provided by Rail Cargo Austria, the largest rail company for freight transportation in Austria and one of the largest in Europe.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This work reports on the recent collaboration between Roma Tre University and SBB AG to develop and test advanced decision support tools for the computation of optimized train timetabling solutions at a tactical level. The methodology proposed is based on the alternative graph representation of the studied problem, in which the variables are related to train timing, sequencing, and routing decisions; the constraints are related to the satisfaction of time windows for the implementation of the proposed service intentions plus several additional operational constraints dealing with passengers satisfaction and limited infrastructure, rolling stock, and crew resources; the objective function is focused on computing the least possible degree of maximum deviation from the intended list of service intentions. Two solvers are compared: AGLIBRARY developed by Roma Tre University and CPLEX developed by IBM ILOG. The former solver is based on state-of-the-art exact and (meta)heuristic algorithms, while the latter is utilized to solve a classical big-M mixed-integer linear-programming formulation of the alternative graph model. The two solvers have been tested on local networks of the Swiss railways. For the given computational settings, the former solver outperforms the latter solver. AGLIBRARY can be applied to compute feasible microscopic timetables solutions with acceptable maximum deviations and a compatible computation time.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The last century urban planning strategies and mobility concepts based on cars led to city sprawl, increasing traffic congestion, and diminishing the importance of transit stations. These strategies ended up worsening people’s quality of life, increasing environmental degradation, and eventually stifling the economic competitiveness of cities. In recent years, public transportation has gained a new glow as a solution for car-based city problems. This new perspective has a significant impact on shaping urban structures with lasting effects and changing the way people live in cities. Transit hubs, in particular, have proved they can play an important role in addressing these problems, building more accessible, sustainable, and livable neighborhoods. These intermodal places embody different dimensions, not only by providing mobility and accessibility, but also by creating an identity, adding value, addressing social needs, and actively contributing to decarbonization – thus improving the overall quality of life in cities. Based on the literature review and considering international examples, this paper explores the implications of the transit hub’s different dimensions, setting a framework for decision-making.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Spatiotemporal traffic forecasting models become a popular tool of urban transport engineering. Validity of spatiotemporal model specifications and their generalisation abilities are the key aspects that intensively addressed in practice and methodological studies. This paper proposes a spatiotemporal cross-validation approach to estimating model performance, which extends classical temporal cross-validation techniques to a complex spatiotemporal structure of traffic flow relationships. The proposed approach allows estimating model generalisation abilities in the spatiotemporal dimension – ability to forecast traffic flows at unobserved nearby road segments. Additionally, the spatiotemporal cross-validation provides clues for stability of model performance in respect to minor modifications of the spatial structure. Advantages of the proposed spatiotemporal cross-validation approach are demonstrated on a large citywide traffic data set.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Each year vast budgets are spent to keep roads and bridges in functioning and the same time safe state. To achieve this, infrastructure databases have been widely used to obtain statistical deterioration models that assist decision makers in optimally allocating available budgets. Although infrastructure databases can provide plethora of information for the contained structure’s characteristics and structural condition, additional trusted data sources, can be used effectively to model environmental or other factors which could be altering deterioration rates. At the same time, introducing many variables of different types and following different distributions, increases the complexity of the problem hindering the use of commonly used statistical models. In this work Artificial Neural Networks and pattern recognition are used in the National Bridge Inventory (NBI) to capture structural deterioration. The study uses 60 variables in total in a dataset containing more than 600.000 structures for the conterminous US and discusses the results and precision of the methodology followed.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The paper investigates the capability of modeling the car following behavior by training swallow and deep recurrent neural networks to reproduce observed driving profiles, collected in several experiments with different pairs of GPS-equipped vehicles running in typical urban traffic conditions. The input variables are relative speed, spacing and vehicle speed. In the model, we assume that the reaction is not instantaneous. However, it may occur during a time interval of the order of a few tenth of seconds because of both the psychophysical driver’s reaction process and the mechanical activation of braking or dispensing the traction power to the wheels. Preliminary results confirm the reliability of this assumption and highlight as expected that the deep recurrent neural network outperforms the simpler feed-forward neural network.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The paper presents a data-driven framework and related field studies on the use of supervised machine learning and smartphone technology for the detection and georeferencing of speed bumps. The study proposes a low-cost and automated method to obtain up-to-date information about speed bumps, with the use of smartphones mounted on vehicles. The proposed methodology is based on readily available and accurate technologies, it can be utilized in crowd-sourced applications for pavement management systems (PMS) and geographical information system (GIS) implementations, and it has already been field-tested for the detection and classification of cracks, rutting, raveling, patches and potholes, exhibiting accuracy levels higher than 90%. The smartphone-based data collection and speed-bump detection algorithms discussed in this paper are complimented with robust regression analysis and Random Under Sampling (RUS) Boosted trees classification models. Ongoing work will further investigate the automated measurement of the geometric properties of the detected bumps and their compliance to regulatory requirements.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. While cellular network data allows to estimate travel patterns made with all modes, it is often important to also understand travel patterns with a specific travel mode. In this paper, we present methods to classify intercity trips, described by a sequence of antennas recorded by the cellular network, as either made by train or road travel mode. A nearest neighbour classification method, which compares how close the antenna sequence is to the fastest route options given the road network and public transport network, is tested. Further, supervised learning methods based on decision trees and Linear Discriminant Analysis methods are investigated. We present results for two pairs of cities in Sweden using both using a dataset of trajectories that have been split into trips and manually tagged with travel mode as well as for a large-scale dataset of identified trips.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. We develop and investigate reliable services for autonomous freight and passenger transport in urban areas. In particular, we consider the individual requirements of travelers for reliable arrival times. To ensure these in an environment of congested urban traffic networks, we investigate how much buffer we need while avoiding excessive idle times. Using the traffic simulation MATSim, we derive time-dependent travel times, integrate them into offline buffer planning, and evaluate the resulting tours.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Risk-free decisions are uncommon in business environments. Uncertainties may come from supply and demand sides and they should be taken into account when assessing decision making. Moreover some of those decisions are extremely remarkable in the success of a particular project as the case of facility location decisions in industrial or commercial firms. Consequently, Facility Location Problems (FLP) have been widely studied from many perspectives due to their undoubtable impact on the profitability and survival of location-sensitive projects. In addition, location decisions are particularly risky as they hold resources for a very long time-horizon and influence on forthcoming tactical and operational decisions (Daskin, 2014). Therefore, consideration of uncertainty within the strategic decision of facility location is of utmost interest for stakeholders. Moreover, energy sector is undertaking a revolution as a new economic and social context is being settled up. Firstly, the European Union (EU) is a major importer of energy resources, especially oil, thus EU countries are seeing oil dependences are threating their economic progress. Secondly, the increasing concern about environmental issues is focusing on transportation sector. Therefore, cleaner fuels and technologies are replacing traditional fossil-fuels vehicles as alternative-fuel vehicles are continuously increasing. In this sense, biofuels are a renewable alternative to fossil fuels in the EU's transportation sector, helping to reduce greenhouse gas emissions. Actually, 10% of transport fuel in EU is aimed to come from biofuels by 2020 (European Commission, 2017). In light of the above, EU countries are seeking to diversify their energy resources by producing alternative and renewable fuel such as biofuels. Therefore, construction of new biorefineries are spreading all around Europe as new and more efficient technologies for biorefining are being developed (Aresta, 2012). In this context, this paper proposes a methodology to locate a lignocellulosic biorefinery in northern Spain for the production of bioethanol (Luo et al. 2010). Biorefinery facilities may last for up to 20 years involving a huge investment, therefore, a careful analysis of its emplacement is required with the attention paid on robustness. Likewise, supply chain decisions must be also deeply analyzed when deciding infrastructure location, considering crop and biomass selection from the surrounding fields. Uncertainties in biorefinery location mainly come from two sources: the potential lack of biomass availability and the prices volatility. On the one hand, availability of resources for supplying a biorefinery is a critical issue because biomass is weather-dependent and heavily seasonal. In our work, it is considered a wide range of biomass with different physic-related attributes such as humidity and depreciation, what causes a model formulation with different scenarios which consider the most frequent uncertainties (prices changes, biomass availability, or use of biomass distribution fleets) . On the other hand, biomass prices are determinant for selecting a biomass type. To this respect, real current prices based on agricultural surveys are implemented. Official estimations of prices evolution are also considered within the biorefinery lifetime. Finally, warehousing policy is further analyzed as renting is allowed during the project duration. The problem has strategic decisions about the location of infrastructure, tactical decisions about location and timing of warehouse infrastructures, and operational decisions on the available network at the periods. The uncertainty in the strategic and tactical side is represented in a multistage scenario tree, while the uncertainty in the operational side is represented in two-stage scenario trees which are rooted with strategic and tactical nodes. Therefore, a robust multi-stage biorefinery location model is developed to cope with uncertainties based on biomass availability and prices. Thus, three kinds of decisions are taken into account: firstly, we consider the strategic decision of biorefinery plant location; secondly, the tactical decisions about intermediate warehouses having temporal location and crop/biomass selections; and, finally operational decisions regarding fleet management. Outstanding managerial implications are expected being the model able to provide a robust solution about biorefinery location based on biomass-related uncertainties. Furthermore, a detailed supply chain is also reported subject to the chosen robust optimal location.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The use of electric buses in public transportation has been of growing interest in recent years. In the context of vehicle scheduling one of the greatest challenges is the lower range of vehicles using battery technologies. Especially the process of charging during ongoing operations is a central issue. Since the recharging of the battery requires additional time, dwell times of current schedules may not guarantee for enough charging time. This work examines vehicle scheduling for electric buses while allowing small modifications of the underlying timetable. We propose a metaheuristic solution approach aiming at minimizing fleet and operating costs while taking timetable quality into account.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The environmental challenges and the initiatives for sustainable developments in urban areas are mainly focused on eco-friendly transportation systems. Therefore, we introduce a new green logistics solution for last-mile deliveries considering synchronization between e-vans and e-cargo bikes, developed as a Two-Echelon Electric Vehicle Routing Problem with Time Windows and Partial Recharging (2E-EVRPTW-PR). The first echelon represents an urban zone, and the second echelon represents a restricted traffic zone (i.e., historical center) in which e-vans in the first and e-cargo bikes in the second echelon are used for customers’ deliveries. The proposed 2E-EVRPTW-PR model aims to minimize the total costs in terms of travel costs, initial vehicles’ investment costs, drivers’ salary costs, and micro-depot cost. The effectiveness of the proposed solution has been demonstrated comparing two different cases, i.e., the EVRPTW-PR considering e-vans for the first case, and the 2E-EVRPTW-PR considering e-vans and e-cargo bikes for the second case. The comparison has been carried out on existing EVRPTW-PR instances for the first case, and on novel 2E-EVRPTW-PR instances for the second case, in which customers of initial EVRPTW-PR instances have been divided into two zones (urban and restricted traffic zones) by using Fuzzy C-mean clustering. Moreover, results encourage logistics companies to adopt zero-emission strategies for last-mile deliveries, especially in restricted traffic zones.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. In this paper we introduce the metro scheduling problem for a large event, e.g., a football match or a concert. We consider a metro line serving the venue of such an event. The ordinary metro timetable, which is typically based on constant headway between all stations, may lead to prolonged waiting when serving large events. To better handle such situations, we propose the Metro Scheduling for Events (MSE) problem to optimize the scheduling of a metro line serving a special event. In the MSE each train is scheduled individually, and trains are allowed to short-turn (i.e., the trains are allowed to reverse direction before reaching a terminal station). We develop a MILP formulation for the problem and an efficient iterated local search heuristic. We consider four objective functions representing different service measures. These are compared through computational experiments, which also demonstrate the effectiveness of the heuristic.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This paper presents a timetable-based approach to assess the capacity of a railway node, based on the microscopic simulation and saturation of the timetable. Saturation consists in scheduling additional train paths (saturating train paths) without introducing any traffic conflict, while respecting the required technical and operational constraints. The proposed approach is applied to analyze the potential effectiveness of the results obtained during a recent RFI project, which involves several infrastructure improvements of the rail freight node of Novara, Italy. The capacity of this railway node is evaluated by adding the maximum number of saturating train paths to the actual timetable. To highlight the bottleneck of the network, proper KPIs are measured on the saturated timetable and an UIC-based compression analysis is performed as well. For the latter analysis, we use SASTRE, an analysis environment for railway systems developed at Politecnico di Torino, which combines a MILP formulation for the timetable saturation problem with a saturation strategy layer. The saturation strategy considers some given priorities between the different network areas and the train types to be used during the saturation. The results reveal that using a microscopic model to simulate traffic flows on a complex railway node allows for a good accuracy of the timetable, but this can only be achieved at a high computational cost.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This work focuses on a particular node of the intermodal chain of transportation, i.e. the maritime port area that represents the linkbetween rail and sea transportation modes. Since this exchange node of the chain has not been addressed much yet, the aim ofthe present study is to provide a description of the process of rail operations in port area, to give an overview of the optimizationchallenges and to review the existing literature on it. The purpose of the paper is to attract the researchers attention on this particularintermodal node, where there is room for improvements.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. In this paper, we study the problem of discerning different reasons for which train delays occur. Given the total amount of delay a specific train builds up at a specific station, we discern the primary delays that would have occurred if there was no other train in the network, such as vehicle problems, from secondary delays which are knock-on delays. Our approach is to train an ML model that predicts the additional delay of a train, given a set of primary features such as weather and secondary features such as the delays of nearby other trains. Methods from explainable AI such as SHAP values help to classify to which amount the primary features and to which amount the secondary features contribute to a specific prediction of the model. We use this classification to discern the different reasons for the specific delay.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The purpose of this study is use data analytics methods for departure delays from shunting yards. Shunting yards are large areas where freight trains are made up and dispatched to the railway network. Shunting yards’ performance influences the reliability of rail freight transportation. In Sweden, late departures from the shunting yards has been one of the main causes of delays for freight trains. This has risen interest to analyze departure delays from shunting yards with systematic approaches. In this study, we aim to use data analytics to analyze departure delays from shunting yards. We use empirical data of freight trains departure from two main shunting yards in Sweden which are also suitable representatives for shunting yards in Europe. The data encompasses arrival and departure data for almost 250 000 trains over a seven-year period. The insight obtained by these analyses can help the infrastructure and yard manager improve punctuality of the rail freight services. It also can be used to develop rescheduling models for rail freight trains.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Travel time measurements are an important indicator for evaluating traffic flow conditions or the effectiveness of various traffic management and control measures and thus their accuracy is of great importance. This paper compares travel time measurements from four different data sources including Floating Vehicle Data (FVD), local detectors, CONSYST data and Bluetooth re-identification. The different data sources are evaluated depending on their penetration (for FVD and Bluetooth) and detection (for inductive loops) rates, the vehicle composition as well as their travel time values on different days. Travel time values are compared for days with and without an incident occurrence separately, as the traffic evolution significantly differs. Firstly, data source characteristics are identified based on the observation of travel time plots and differences and similarities are further assessed and estimated via statistical analysis and pairwise comparison. Finally, a statistical analysis and pairwise comparison are conducted to determine the travel time correlations between all data sources on different days and under various traffic conditions.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. This paper suggests and tests a method to the estimate penetration rates of floating car data at signalized intersections with no access to full-scale volume information (e.g. detection data) by means of data mining techniques and assuming the knowledge of maximum density (average vehicle length and gap). The trajectories are clustered to determine their associated cycle time and green start in order to aggregate them in a single referencial time-scale where the well-known Edie's definitions are applied to extract aggregated probe-scale macroscopic variables (flow, density, speed). A grid of space-time regions produces a scatter of points that are fitted to a fundamental diagram. The model-based congestion density is compared with the full-scale congestion density to produce the observation penetration rate which is reduced by the portion of unobserved time. The algorithm is tested with real data with two different fundamental diagram models and in simulation with different penetration rates and sampling rates. Results show better estimates and accuracy of +/- 10% is reached when the sampling rate is below 15s and the penetration rate is over 2%.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Currently, cities should respond the challenge of sustainable mobility. Travel time prediction is crucial for the improvement of traffic management systems, like route planners, which could mitigate the environmental, social and health problems related with traffic congestion. Moreover, the increasing penetration of connected car makes Floating Car Data a suitable data source to be exploited. This work presents a comparison of three methods (ARIMA, linear regression with varying parameters and neural networks) that use probe vehicle data to predict the travel time on freeways. This paper analyses through microscopic traffic simulation the performance of the selected algorithms and how the penetration rate of floating cars and other factors affect the prediction accuracy in a realistic model of the AP7 freeway (Spain). The obtained results show that the ARIMA and neural networks outperforms in all tested scenarios the linear regression proposal for all the experiments designed.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. In order to provide advanced models and tools for the prediction of congestion effects in urban areas, this paper investigates the problem of the network performance forecast by using traffic data acquired via Bluetooth devices. It applies statistical models within a methodology in which the above data are filtered, cleaned and fused for a better prediction of path travel times. The proposed approach has been applied to the city of Rome, for which the promising results of such data usage for online forecast are presented and discussed.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Traffic state estimation (TSE) is an important problem with significant applications in transportation management, traffic control and advanced traveler information systems that enables a wealth of applications related to the monitoring and control of intelligent transportation systems (ITS). TSE refers to the process of the inference of traffic state variables such as the flow, density, speed and any other equivalent variables, using partially observed traffic data. Connected and Automated Vehicles (CAVs) provide useful traffic data as they can acquire wide-range spatiotemporal information at relatively low cost. Assuming that a CAV can measure the spacing to its neighbour vehicles we propose a Bayesian framework to estimate traffic flow characteristics. We utilise numerical Monte Carlo approximation techniques to estimate the traffic density and build on existing methodology, by taking into account information provided by the CAVs, to derive a probabilistic estimation of traffic density. Our methodology is applied to traffic flow data from a specific road in Germany.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The locations and types of controllers employed in a transportation network directly impact the level of performance reachable by traffic control policies. In order to guarantee that the highest level of performance is reachable, fully exploiting the available network capacity, we explore methods to identify efficient locations while minimizing the number of controllers used. Previous work provided an exact method to identify pricing controller locations on networks. However, this method exhibited complications when applied to large networks containing bi-directional links. We therefore propose simple heuristic algorithms to solve this problem, while avoiding heavy computation, thus being able to determine a solution to the minimal controller location set problem on large networks. Based on the existing framework, we provide a large experimental setup, with diverse network sizes and configurations to compare the different heuristic methods performance, in order to develop an efficient algorithm.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. Urban population is increasing fast, particularly the elderly population. This is creating new challenges to public transport systems since these citizens may have sensory, cognitive and/or motor impairments that need to be addressed. This work explores the potential of a Demand Responsive Transport (DRT) system for the elderly population in urban environment. For this purpose, the Dial-A-Ride Problem (DARP) was implemented using a multivariable minimization approach. In this approach, an Assigning Request to Vehicles (ARV) algorithm is used order to obtain an initial solution, and then a Multi-Objective Tabu Search Algorithm (MOTSA) applied to optimise the initial solution. In this optimisation, the total travelled kilometres, the deadheading kilometres and the number of vehicles were minimised. The performance of the model was computed combining different parameters’ values of the number of requests, time boarding for each customer, the number of seats in each vehicle, vehicle’s speed, the total number of iterations, and candidate threshold number. The computational results found a strong positive correlation between the number of requests and the number of vehicles in the initial and in the optimised solution (rs =0.816, p<0.001, rs=0.823, p<0.001, respectively).

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. The negative impacts of urban freight transport have fostered the search for new distribution systems for inner city deliveries. One of the proposed solutions is the consideration of two-echelon distribution systems based on mobile depots, where loads arriving from the periphery of the city are directly transferred, at intermediate locations, from larger vehicles to smaller vehicles more suited to operate in the city centre. Four types of two-echelon distribution systems based on mobile depots can be identified, according to the degree of mobility of larger vehicles and to the accessibility to customers. In this paper, we propose a generic three-index arc-based mixed integer programming model for two-echelon vehicle routing problems, with synchronisation and multi-trips at the second echelon, for the particular case of mobile depots, where larger vehicles can only visit one transfer location and do not perform direct deliveries to customers. Additionally, we show how this generic model can be extended to address each of the other types of systems based on mobile depots, by changing the underlying graph and the base formulation. Furthermore, we propose symmetry breaking constraints and valid inequalities derived from the VRP literature, to tighten the formulations. The generic model, extensions and impact of valid inequalities are tested using adapted instances from the VRP literature. The results show the flexibility of the proposed generic model to address different two-echelon distribution systems with synchronisation and multi-trips at the second echelon, and the positive impact of adding the valid inequalities to strengthen the proposed formulations.

Download: (a) Extended Abstract (b) Presentation video

ABSTRACT. In this paper, we present a MIP based optimization approach to reduce timetable deviations in railway operations. We analyze historical train delay data and identify systematic delays that occur frequently. Next, we use an optimization model to adjust the scheduled running times of a given timetable to the actual running times from the past in order to avoid such systematic delays. The objective function of the optimization model is to minimize deviations between scheduled and actual (historical) train running times. In doing so, the original timetable is mostly preserved. Train routes remain unchanged and the loss of connections is allowed but penalized. We apply the model to real data from Deutsche Bahn and analyze trade offs between keeping connections and reducing deviations. The results show that only small adaptions, that are performed simultaneously in the whole network, are sufficient to improve the on-time performance of trains.

Download: (a) Extended Abstract

ABSTRACT. The Railway Crew Scheduling Problem (RCSP) consists of determining the most efficient combination of shifts, i.e., sequences of consecutive scheduled trains and tasks, over a given period of time. Each trip must be covered by exactly one shift while operational, legal and labor constraints are satisfied. Crew scheduling arises on an operational planning level where anonymous shifts are formed but not yet assigned to specific crew members. In our work we consider the RCSP for the Rail Cargo Austria (RCA), which is the largest railway company for freight transportation in Austria and one of the largest in Europe. We propose a multi-graph model and solve the RCSP using a matheuristic consisting of a breadth-first search heuristic and an Integer Linear Program. We aim to minimize the overall paid working time while incorporating all RCA specific constraints. These are requirements on depot capacity, line and traction knowledge, total working time, consecutive and total driving time, shift symmetry, night shift conditions, and break regulations. The latter belong to the most complex constraints. Although schedule efficiency and employee satisfaction are in general conflicting, a cost-efficient schedule will not be implemented if it does not reach acceptance of the crew. Since satisfaction among train drivers is really important to the RCA, we incorporate several constraints to achieve it. Besides that, the main goal of this work is to create robust crew schedules, which we encounter with robust train schedules, generated with special approaches from previous work. We evaluate our approach on real-world data provided by the RCA, based on the Austrian railway network, with focus on the impact of robust train schedules on railway crew schedules.

Download: (a) Extended Abstract (b) Presentation video