Site icon 4TU.ResearchData

When to stop and when to skip: Public transport modelling during Covid-19

University of Twente Professor, Kostas Gkiotsalitis, shares his software models for modifying the public transport demand during the Covid-19 pandemic.

Kostas Gkiotsalitis is an Assistant Professor within the Transport Engineering and Management group at the University of Twente. His research focuses on public transport modelling and traffic demand prediction. 


After studying a Master’s degree in Transport and Sustainable Development at Imperial College in London, Kostas joined the Intelligent Transport Systems group as a Research Scientist at NEC Laboratories Europe in Heidelberg, Germany.

During his time at NEC, Kostas was part of three multinational Seventh Framework Programme (FP7) projects and a NEC Smart Public Transport project that funded the development of ‘REFLEX’, a decision support tool for public transport planning in Asian cities.

The road to reliable transport

“REFLEX is a software model that collects real-time data about the local traffic and provides bus drivers with instructions on how to drive according to the demand for public transport, the ongoing traffic congestion, and the location of other buses in the area,” explains Kostas. 

The tool was implemented in Singapore, Hong Kong and India to provide a reliable bus service for passengers. The tool interface can tell a bus driver when to stop and wait at a control point; when to skip a stop and advance along the route; when to schedule a bus to leave the station; and, when to suspend the bus service.

Kostas believes in the need for computational models to maintain the regularity of the bus service and reduce the waiting times for passengers.

“Buses are a critical part of public transport in cities. There are millions of bus rides each day, thousands of vehicles in operation, and often the service is high-frequency with buses running along lines every two to three minutes,” say Kostas.  

He explains that traffic congestion causes many problems such as ‘bus bunching’ and overcrowding. 

“Bus bunching occurs if buses don’t maintain even headways between consecutive trips, resulting in more than one bus arriving at the same stop at the same time. The first bus is overcrowded with passengers which delays the bus; the following buses are emptier, lighter and, therefore, advance ahead of schedule.”

In several cities, transport authorities provide monetary incentives (i.e. monthly bonuses) to bus operators that deliver a regular and reliable bus service, as well as penalties for those who leave passengers waiting longer than the estimated waiting time.

The coordination of public transport is a complex business and computational modelling task that has become somewhat essential for a smooth-running operation. 

Public transport modelling during Covid-19

Kostas explains that public transport modelling has been particularly important during the Covid-19 pandemic in order to minimise the spread of the virus and mitigate the economic loss incurred by transport companies. 

“Crowded public transport is considered one of the virus transmission factors. Since the pandemic hit many transport companies have closed stations, cancelled certain lines, limited their service hours, altered the frequency and routes of transport, and imposed vehicle capacity limits, leaving a significant number of unserved passengers.”

He continues, “The reduced demand for public transport during the pandemic, in conjunction with the financial penalties, has jeopardized the future of many transport businesses and many are concerned for their economic losses.”

Kostas received a fund from The Netherlands Organisation for Health Research and Development (ZonMw) to develop a model for modifying public transport service patterns to account for the imposed Covid-19 capacity

His Python-based model, that’s based on bus line 9 connecting the University of Twente with Hengelo and Enschede, aims to improve crowding levels inside vehicles and reduce the passenger wait time. 

Using OV-Chipkaart data from Dutch public transport service provider, Keolis, the model estimates the number of passengers waiting for a bus and the frequency of buses travelling along the line so that adjustments can be made to ensure that buses are not overcrowded. 

“In the morning, when the demand for public transport is higher, transport companies can schedule more buses to leave the terminal. At times when the demand is lower, companies can suspend buses so that they do not lose money,” he remarks. 

Sentiments about sharing software 

Kostas uses the version control software management system, GitHub, for managing his software code during active research projects and values the opportunity of being able to make it available for reuse in the 4TU.ResearchData repository. 

“I’ve published the software code that underlies my peer-reviewed publication as well as documentation about the model so that others can understand and reuse it,” says Kostas.

“I’ve worked for commercial enterprises in the past where it wasn’t possible to share my code. Now I’m working for the University of Twente, I realise how important these models are for wider society and I want others to be able to build upon my work and improve it.”

He also explains that having a stable URL for his research outputs is important to him. “I like that my data and code are assigned with a DOI and can be preserved long-term in a repository that has a reputable identity within the Netherlands.”

Planning ahead

Kostas’s next project involves working on a Joint Programming Initiatives collaboration with six universities in the Netherlands, Austria, Italy, Turkey, Belgium and Germany to study the impact of mobility hubs on public transport. For example, this could be a train station that integrates trains, buses, shared cars, shared bicycles/e-scooters and other modes of transport. 

Kostas is currently working with University of Twente data steward, Simone Fricke, to coordinate the data management of this international European consortium, and to plan for data publication at the end of the project.  

We thank Kostas for sharing his research story with us and for making his research software publicly available online. 

Please consult this guide for more information about connecting your GitHub and 4TU.ResearchData accounts or contact us for assistance.


Written by Connie Clare (4TU.ResearchData)

Cover image illustrated by Connie Clare (4TU.ResearchData)
Featured image by travelphotographer from Pixabay 

Skip to toolbar