Posts tagged modelling

PhD project: Robust train trajectory optimization

In cooperation with the Dutch Railways (NS), Alex Cunillera works in this PhD research on robust train trajectory optimization. Even two trains of the same model running on the same line show significant differences in their dynamics. This might be due to different passenger loads, weather, fault history, driving style of the train driver, etc. Moreover, there are uncertainties in the track data that may also have a strong influence on the train operation. This research focuses on determining the uncertainties and stochastics of these variations and developing methods to compute robust train trajectories that optimize the energy consumption and minimize delays in the presence of the mentioned variations.

Project contributions (ongoing):

Papers:

Train motion model calibration: research agenda and practical recommendations (ITSC 2022) 

Real-time train motion parameter estimation using an Unscented Kalman Filter (Transportation Research Part C) 

Train trajectory optimization under parametric uncertainty and roubst maximum principle analysis (COIA 2022)

Presentations: 

Real-time train motion parameter estimation using an Unscented Kalman Filter (RailBeijing 2021)

Train motion model calibration: research agenda and practical recommendations (ITSC 2022)

European Transport Conference 2022, Milano

The European Transport Conference (ETC) is taking place this week, September 7-9 2022 in Milano, Italy.

The following Smart PT Lab contributions will be presented:

Change in train travelling behaviour during and after Covid-19 due to anxiety (Presentation and research report)

G.B. Hafsteinsdottir, R. van der Knaap, N. van Oort, M. de Bruyn, M. van Hagen.

Shared micromobility and public transport integration. A mode choice study using stated

preference data (Presentation and research report)

A. Montes Rojas, N. Geržinic, W. Veeneman, N. van Oort, S. Hoogendoorn,.

Understanding the whole station choice concept by cyclists (Presentation and research report)

A Barneveld, R Huisman, N. van Oort.

The full program can be found here:

https://aetransport.org/etc

CO2 Barometer

In this PhD project by Marko Kapetanović, an integrated model for dynamic monitoring and prediction of CO2 emissions of regional railway services is developed, following a life-cycle approach. The project is performed in close cooperation with Arriva, the largest regional railway undertaking in the Netherlands. Possibilities and measures to improve the energy efficiency of railway operation and to reduce the total emissions on the network are identified and assessed, including alternative propulsion systems such as hybrid, plug-in hybrid, fuel cell-electric and battery electric, together with a range of energy carriers. Analyzed fuels and energy carriers include LNG, first and second generation biofuels, hydrogen and electricity, with examined various alternative production pathways. Check the main output of this project below.

Short video explaining the project and main results

Korte video over het project en resultaten (in Dutch)

Papers

Optimal network electrification plan for operation of battery electric multiple unit regional trains (TRISTAN XI 2022)

Analysis of hydrogen powered propulsion system alternatives for diesel electric
regional trains (Journal of Rail Transport Planning & Management 2022)

Reducing fuel consumption and related emissions through optimal sizing of energy storage systems for diesel-electric trains (Applied Energy 2021)

Analysis of Hybrid and Plug-In Hybrid Alternative Propulsion Systems for Regional Diesel-Electric Multiple Unit Trains (Energies 2021)

Sustainability of Railway Passenger Services: A Review of Aspects, Issues, Contributions and Challenges of Life Cycle Emissions ( RailNorrköping 2019)

Presentations

Life Cycle Assessment of Alternative Traction Options for Non Electrified Regional
Railway Lines
(World Congress on Railway Research (WCRR) 2022)

Optimal network electrification plan for operation of battery electric multiple unit regional trains (TRISTAN XI 2022)

Improving Sustainability of Regional Railway Services in the Northern Netherlands (RailTech 2022)

Subjective Beliefs regarding Waiting Times in Public Transport Networks in the Netherlands, Greece, and Portugal

Waiting times in public transport networks (PTNs) are inherently uncertain for travellers and, similar to other service industries, such uncertainty is likely to be a major cause for anxiety and frustration (Maister, 1985). While real-time information regarding waiting times is an important development in mitigating such negative feelings, they do not completely remove uncertainty. Even when information is provided, travellers process it on the basis of their individual attitudes, habits, experiences, and contemporary contextual variables. Yet, previous studies on behavioural responses to travel time unreliability have either (unrealistically) assumed that travellers know the objective travel time distributions or have studied behaviour within the artificial context of travel simulators. Quantifying travellers’ attitudes and perceptions — subjective beliefs — regarding waiting times may be critical for assessment of travel satisfaction and subsequently choice behaviour.

In this research, a stated preference experiment is used to quantify travellers’ attitudes and perceptions — subjective beliefs — regarding waiting times in public transport networks in three European countries. Results and potential policy implications are presented at the European Transport Conference (ETC).

Find the ETC poster of Sanmay Shelat HERE

Fietsen naar de tramhalte: simultane modellering van voortransport- en haltekeuze

Wereldwijd wordt er gestuurd op een toename van duurzame vervoerkeuzes voor een betere leefbaarheid en bereikbaarheid. Vooral in de steden waar de samenleving groeit en de dichtheden groter worden is een verandering in kijk op de mobiliteit noodzakelijk om de burgers tevreden te stellen. De integratie van fiets en openbaar vervoer (OV) kan hier aan bijdragen. Wanneer de fiets wordt gebruikt als voortransportmiddel wordt het invloedsgebied van het OV vergroot ten opzichte van lopen waarmee het een beter alternatief wordt voor niet-duurzame vervoermiddelen. Om de combinatie fiets en OV te vergroten zullen effectieve klantgerichte maatregelen genomen moeten worden. Hiervoor is meer inzicht nodig is de factoren die een rol spelen bij de keuzes voor voortransportmiddel en halte. Hier is tot op heden nog weinig over bekend op het stedelijk niveau. Door de keuzes in één onderzoek te combineren wordt de afweging duidelijk tussen het voortransportmiddel en de OV-reis, en kunnen de effecten op het invloedsgebied van het OV bepaald worden. Dit is gedaan op basis van data van HTM-tramreizigers in Den Haag middels een simultaan discreet keuzemodel van voortransportmiddel en halte keuze. Resultaten geven aan dat reizigers in het algemeen liever lopen dan fietsen naar de tramhalte. Daartegenover staat dat de afstand naar de tramhalte lopend 2,1 keer zwaarder weegt dan als men fietst. Dat betekent dat bij een langere afstand fietsen aantrekkelijker wordt dan wandelen. Frequente fietsers zijn meer geneigd om ook naar de tramhalte te fietsen, terwijl frequente tramreizigers juist minder vaak fietsen naar de tram. De aanwezigheid van fietsparkeervoorzieningen vergroot het invloedsgebied van een tramhalte, maar de grootste impact op het invloedsgebied van fietsers is de OV-reistijd. Verbeteringen aan het OV-systeem, zoals minder haltes en/of hogere frequenties kunnen dan ook zorgen voor een groter geaccepteerde fietsafstand (fietskeuze) tot de halte. Op basis van deze resultaten lijkt het mogelijk de fiets-OV combinatie ook op stedelijk niveau te stimuleren. Hierdoor kan duurzame mobiliteit op stedelijk niveau betere concurrentie bieden aan de auto, wat lijdt tot een aantrekkelijkere en beter leefbare stad.

Bekijk hier de presentatie en paper van Danique Ton et al.: Presentatie en Paper

OV en (deel)fiets: vriend of vijand? Inzichten in gebruik en reizigersvoorkeuren

In beleid en onderzoek is steeds meer aandacht voor duurzame vervoermiddelen, zoals de fiets en het openbaar vervoer (OV). Integratie van fiets én openbaar vervoer kan de voordelen van beide systemen combineren: De fiets zorgt voor fijnmazige ontsluiting van herkomsten en bestemmingen, is duurzaam en bevordert een gezonde leefstijl. De kwaliteit van het OV neemt de laatste jaren toe, onder andere door de introductie van hoogwaardig OV (HOV): snelle, frequente en betrouwbare bus- tram- en metrolijnen met een hoog comfortniveau. De halteafstanden van deze systemen zijn, net als bij het spoor, relatief hoog, waardoor de fiets een belangrijke rol kan spelen in de gebiedsontsluiting. Echter, op kortere afstanden zijn de fiets en het OV, naast een nuttige combinatie, ook elkaars concurrenten.

Om inzicht te krijgen in de aanvullende dan wel concurrerende rol van de fiets en OV, is onderzoek nodig over hoe de reiziger zich nu en in de toekomst beweegt. Dit inzicht helpt om een optimaal integraal fiets+OV systeem te ontwerpen en gebruik van dit systeem te stimuleren en te faciliteren. Dit paper laat de resultaten zien van vier recente TU Delft onderzoeken op dit gebied.

Resultaten van een literatuuronderzoek naar de first- en last-mile laat zien welke factoren belangrijk zijn voor modaliteitskeuze, waaruit bijvoorbeeld blijkt dat mannen die bekend zijn met de omgeving vooral gebruik maken van de fiets. Onderzoek in Den Haag laat het bereik van de tramhalte zien voor de fiets. Fietsers zijn bereid tot 3 km te fietsen om bij een tramhalte in de stad te komen. Ongeveer 50% van de gebruikers fietst verder dan de dichtstbijzijnde halte als deze halte minder overstappen, betere parkeervoorziening en meer reisopties biedt. Voor het natransport is de deelfiets een relatief nieuwe optie. Onderzoek naar Mobike in Delft (dockless bikes) laat zien dat ca.19% van de deelfietsritten gebruikt wordt om van en naar het station te komen. Met name het gebruik van Mobike voor ritten naar station Delft Zuid, met beperkte andere mogelijkheden, valt op. Ook voor andere deelfietssystemen in Delft, zoals OV-fiets en Swapfiets is onderzoek gedaan naar het gebruik. Door de beschikbaarheid van deze systemen geeft 9-16% van de gebruikers aan meer gebruik van de trein te maken, tegenover 34-60% minder van de bus. Ook lopen wordt vervangen door deze nieuwe modaliteiten in 35-42% van de gevallen.

Bekijk de presentatie en paper hier: Presentatie en Paper

Forecasting bus ridership with trip planner usage data: a machine learning application

Currently, public transport gives much attention to environmental impact, costs and traveler satisfaction. Good short-term demand forecasting models can help improve these performance indicators. It can help prevent denied boarding and overcrowding in busses by detecting insufficient capacity beforehand. It could be used to operate more economically by decreasing the frequency or the size of the bus if there is overcapacity. Moreover, it could help operators plan their busses during incidental occasions like big public events where little information is known. Finally, it could be used to reliably inform the travelers on the current crowdedness.
This study investigates the usefulness of a new data source; the usage data of a trip planner. In the Netherlands there are multiple trip planners available for users to help find the most optimal (multimodal) journeys. These trip planners require a date, a time and an origin and destination, which they use to construct multiple alternative journeys from which the user can choose. For this study the data of 9292 was used, being the major trip planner in the Netherlands including all public transport modes.
We developed a model for forecasting the number of people boarding and a model for forecasting the number of people alighting at a certain stop. These forecasts are defined at the vehicle-stop level. By summing the number of people boarding and subtracting the number of people alighting along the trip the forecasted number of passengers after a stop is calculated.

We compare five different machine learning models: multiple linear regression, decision tree, random forests, neural networks and support vector regression with a radial basis kernel. We compare these models with two simple rules: 1 predict the same number as last week, and 2 predict the historic average as number. The models are implemented in the Scikit-Learn library of Python. The data is stored in a PostgresSQL database.
The trip planner datasets and smart card dataset are merged and preprocessed. The resulted dataset is rather sparse; a lot of stops have zero passengers boarding or alighting or requests suggesting to do so. Therefore we investigated if subsampling is needed. From the datasets useful data is selected and features are constructed. The features are standardized. Different number of features are tested, these features are selected based on recursive elimination using a simple random forests model. Finally, the hyperparameters of the models are tuned and the optimal configurations are stored. The scores are validated by using cross validation.

Find more details in the following contributions by Jop van Roosmalen: Transit Data workshop presentation and MSc thesis

Combining Speed Adjustment and Holding Control for Regularity based Transit Operations

Vehicle bunching often occurs in high-frequency transit systems leading to deterioration of service reliability. It is thus necessary to control vehicles during operations. Holding control is a common solution for this situation, but it may result in longer vehicle running times. Speed adjustments can contribute to more regular operations while preventing prolonged trip times. This paper proposes a control strategy, which combines these two strategies to maintain the regularity of transit operations. The findings based on simulation study for trunk bus services in the Netherlands suggest that combining the two strategies implies both the positive and negative attributes of each control.

Find the MT-ITS presentation and paper of Aishah Imram HERE and HERE

Walking and bicycle catchment areas of tram stops: factors and insights

Pollution and congestion are important issues in urban mobility. These can potentially be solved by multimodal transport, such as the bicycle-transit combination, which
benefits from the flexible aspect of the bicycle and the wider spatial range of public transport. In addition, the bicycle can increase the catchment areas of public transport stops. Most transit operators consider a fixed 400m buffer catchment area. Currently, not much is known about what influences the size of catchment areas, especially for the bicycle as a feeder mode.

Bicycles allow for reaching a further stop in order to avoid a transfer, but it is not clear whether travelers actually do this.This paper aims to fill this knowledge gap by assessing which factors affect feeder distance and feeder mode choice. Data are collected by an on-board transit revealed preference survey among tram travelers in The Hague, The Netherlands. Both regression models and a qualitative analysis are performed to identify the factors that influence feeder distance and feeder mode choice. Results show that the median walking feeder distance is 380m, and the median cycling feeder distance is 1025m. The tram stop density and chosen feeder mode are most important in feeder distance. For feeder mode choice, the following factors are found to be influential: tram stop density, availability of a bicycle, and frequency of cycling of the tram passenger. Furthermore, the motives of respondents for choosing a stop further away are mostly related to the quality of the transit service and comfort matters, of which avoiding a transfer is named most often. In contrast, the motives for cycling relate mostly to travel time reduction and the built environment. Three important barriers for the bicycle-tram combination have been discovered: unavailability of a bicycle, insufficient and unsafe bicycle parking places. Infrequent users of the bicycle-tram combination are more inclined to travel further to a stop that suits them better.

Find the MT-ITS paper and presentation of Lotte Rijsman HERE and HERE

Impacts of charging methods and mechanisms of zero emission buses on costs and level of service

To limit global warming and strive for more liveable and sustainable cities, innovative zero-emission (ZE) buses are on the rise all around the world. Different alternative vehicle propulsion methods have been introduced during the last decades. However, for now, only trolley, battery and fuel-cell electric vehicles can be classified as (on the pipe) ZE-buses.

This research focuses on battery electric buses, since they are most cost-efficient and – therefore – the most promising option for the (near) future. An important limitation of battery electric buses is however the limited range of operations due to capacity restrictions of batteries. Batteries should be (re)charged before, during and/or after daily operations.

Different charging methods, including different charging power systems are available to charge battery electric vehicles. As far as known to the authors, scientific literature focusing on the operations and charging scheduling of electric buses is scarce. In this study, a comparison of different applied charging methods for electric buses is obtained.
A ZE-bus station simulation method is developed to assess charging methods and charging regulations with regards to their impacts on a variety of costs and level of service indicators. This simulation-based method is multi applicable, since it is particularly based on general automated vehicle location (AVL) data. To demonstrate our model, a case study at Schiphol (Amsterdam Airport) is performed.
This research concludes that the shift to ZE-bus transit is involved with higher costs and passenger disturbances. Investment costs of ZE-buses increase substantially: Most electric vehicles are around 60 to 80 percent more expensive than conventional diesel engine vehicles and additional charging infrastructure investments are required. Benefits of electric operations, including vehicle propulsion cost savings up to 70 percent, are not able to compensate these high investments.
The charging method choice contain trade-offs between level of service and (vehicle and charging infrastructure) investment costs. (Slow) depot charging offers opportunities for operations on short distance lines. However, additional vehicles are required in order to replace a vehicle which should be recharged. In this respect, conventional timetables could be complied and the level of service remains unchanged.

To prevent fleet overcapacity, vehicles should be recharged fast (with high charging power) along the line. Slight charging related delays could occur, especially when the number of charging systems is not sufficient, and/or the charging times are relatively long. Bus end stations and terminals are suitable as fast charging locations, since charging time could be combined with buffer time there.

Finally, dynamic/in-motion charging offers opportunities to prevent charging related delays completely due to combined charging and operation time. This charging method is still in its infancy stage yet, so focus is more on (innovative) static charging methods now.

Find the MT-ITS presentation wih Max Wiercx and Raymond Huisman: HERE

© 2011 TU Delft