Increasing rail transport throughput while avoiding incentives to compromise social distancing: agent-based quantification leading to guidelines

Lead Research Organisation: University of Sheffield
Department Name: Mechanical Engineering


Public transport is crucial to economic activity, functioning cities and access to work, but presents many pinch-points (doors, confined areas of queuing, ticket gates) where social distancing is easily compromised. These points determine people flow rates, creating conflicting priorities in enabling functioning transport while maintaining social distancing safety.

The proposed research will build on previous agent-based modelling of passengers at the railway platform-train interface conducted using massively parallel Graphics Processing Unit (GPU) simulations for parameter exploration and sensitivity analysis. Our current RateSetter model has informed rail sector policy and stakeholders through collaboration with Railway Safety and Standards Board (RSSB). Additional factors to be explored include: (i) Incentives such as imminent train departure to compromise social distancing. (ii) Limitations on personal situational awareness in complex confined space pedestrian flows. (iii) Differing personal assertiveness and its impact on confined space flow dynamics.

Modelling will focus on optimisation of passenger flow to avoid incentivising compromised social distancing, providing guidelines on effective timetabling and COVID safe station operation. This is expected to be very important in a semi-lockdown situation as large numbers of rail passengers are likely to be in the later cohorts to receive any vaccination yet will want to begin travelling again. To convert the findings to actionable insights for policy and practice validated predictions of passenger flow times for train boarding and alighting under a range of conditions will be transferred to RSSB for input to network level rail system modelling. This will reveal the network wide implications of behavioural change and management of passenger flow at individual stations. RSSB will facilitate data access, knowledge exchange and dissemination within the rail industry.

The work will increase confidence in rail use and enable higher passenger volumes with lower risk of compromised social distancing through: (i) Algorithms representing human movement in confined spaces subject to incentives to compromise social distancing. (ii) A validated model to rapidly test and optimise new ways of operating transport to aid national recovery. (iii) Guidelines on quantification of intervention effectiveness in limiting proximity and cumulative proximity (potential viral load) for passengers and staff. (iv) Input of validated passenger flow time predictions to rail industry network wide modelling to reveal impacts of station management policies.


10 25 50
publication icon
Hayes S (2022) Validation of Agent-Based Passenger Movement Modeling for Railway Stations Subject to Social Distancing During the COVID-19 Pandemic in Transportation Research Record: Journal of the Transportation Research Board

Description Our research focused on safe use of railway transport during the covid-19 pandemic, in particular the impact social distancing has on the boarding and alighting of trains. The rail industry typically wants to minimise this time so that network capacity is maximised and trains keep on the move for more of the time. However, they also aim to operate the system safely and needed to understand whether more time should be allowed in stations as people returned to transport after lockdowns were lifted. If more time is needed this would require creation of new timetables to enable the network to remain coordinated, avoiding unpredictable running and delays/congestion.

Predictions were made using models of pedestrian movement tailored to the constrained environment of the train and platform within a railway station. These were validated against CCTV showing real-world behaviour of passengers. Social distancing was implemented in the models so that passengers would aim to maintain distance from one another, but not from surrounding doors/walls etc, i.e. a more sophisticated representation than just modelling very large people. Steering behaviours and the ability to judge when to board an already crowded train were implemented.

Initial predictions showed dramatic extension of passenger movement times if large (2m) social distances were maintained. These extended flow times were less severe with 1m social distancing. CCTV observation of real boarding and alighting behaviour showed a tendency for people to greatly compromise on distancing at the train door location, and this moderated some of the flow time extensions as the behaviour was integrated into the modelling predictions. The outcomes were integrated into a regional rail network model developed through Rail Safety and Standards Board to provide train operators and Network Rail with insight to how the network would behave in a range of possible scenarios as passengers returned to rail transport at the lifting of lockdown, enabling planning for different potential outcomes.

The final stages of the work extended the model to look at groups travelling together (no social distancing between members of a group), and the impact of luggage on passenger flow times. As the pandemic evolved and vaccination began to make social distancing a less important defence against covid-19 there was engagement from Network Rail Station's Capacity Planning team. They identified that the tools developed have good potential to assist in timetable development after the pandemic, addressing an area of pedestrian flow modelling not covered by their existing modelling.
Exploitation Route The pedestrian modelling undertaken is particular to confined and complex geometries. We focused on the interface between a railway train and the platform, but there may be other applications in which the underlying algorithms could be applied (e.g. seated sports stadia, boarding/alighting bus transport). Currently our focus is to apply the work to post-pandemic railway system modelling.
Sectors Transport

Description Data from our UKRI research is of greatest relevance to the rail sector, and the specific impact during 2021 has been reported in the section "Influence on Policy, Practice, Patients & the Public". In 2022 we are using the techniques we developed in work with Network Rail on post-pandemic station capacity assessment. We are using Peckham Rye station as a case study to develop the integration between our outputs and Network Rail's existing modelling system. Peckham Rye is an example of a highly constrained station in which pedestrian movement capacity is limited, but where straightforward solutions such as building addition entrances/exits are not possible. This is typical of stations on viaducts or with island platforms between rail lines on which there is no additional land available for extension to passenger areas. Impact from our research is through modelling changes in behaviour due to environment changes (e.g. location of station furniture). These environmental changes have the potential to reduce crowding at much lower cost than changing the underlying infrastructure. We are exploring the effects using our models and discussing how a trial may be conducted if the outcomes are promising. Decisions on this are expected in June/July 2022.
First Year Of Impact 2022
Sector Transport
Impact Types Societal


Description RateSetter+: the impact of social distancing on network performance
Geographic Reach National 
Policy Influence Type Participation in a guidance/advisory committee
Impact The work was integrated into rail industry planning for recovery from the 2021 lockdown phase of the pandemic. Industry document "RateSetter+: the impact of social distancing on network performance": Providing insights to help the rail industry determine the impact of social distancing on network performance while passenger demand grows back to normal levels. Project Reference: COF-G22-01.
Description Return to Work: Passenger flow modelling at Peckham Rye station
Amount £36,903 (GBP)
Organisation Higher Education Innovation Funding (HEIF) 
Sector Public
Country United Kingdom
Start 01/2022 
End 07/2022
Description Network Rail 
Organisation Network Rail Ltd
Country United Kingdom 
Sector Private 
PI Contribution We have supplied Network Rail with our findings on an ongoing basis throughout the project. We have had a follow-up project with them in 2022 modelling a specific UK station. This focused on post-pandemic people flow modelling to evaluate potential improvements to the location.
Collaborator Contribution Network Rail's "Stations Capacity Planning Team" have been part of our steering group and subsequently in 2022 continued developing the research implementation with us. They have provided guidance on areas in which they are already well informed about passenger flows and risks, and those areas in which they welcome the additional insight we can provide with our research. Network Rail has also provided access to CCTV of stations (Birmingham New Street, London Bridge, Leeds) to enable calibration and verification of our modelling work.
Impact Network Rail, University of Sheffield, Peckham Rye Ratesetter Modelling Report, July 2022. Passenger movement modelling focused, crossing engineering and computing disciplines.
Start Year 2021
Description Rail Safety and Standards Board (RSSB) 
Organisation Rail Safety and Standards Board
Country United Kingdom 
Sector Public 
PI Contribution Data and algorithms developed in our UKRI project have been supporting rail industry planning for recovery of passenger numbers. This has been via transfer of data and algorithms, online meetings with Rail Safety and Standards Board, online meetings with Network Rail's Stations Capacities team, and with RSSB's independent contractor Risk Solutions who has been contracted by RSSB to prepare a regional scale model of the rail network to understand it's operation when station flows are affected by social distancing. The network level models in which our data has been used are reported in "Ratesetter+ The effects of social distancing at the PTI on service performance (COF-G22-01)", J. Hyde and C. Rees, Rail Safety and Standards Board, London, 2021,
Collaborator Contribution RSSB have coordinated steering input from the rail industry, and facilitated dissemination of the UKRI project findings. They also greatly assisted in securing access to CCTV footage from rail stations around the UK. This was required in the UKRI work for calibration of our modelling work showing passenger movement from a range of railway stations.
Impact The rail network level models in which our data has been used are reported in "Ratesetter+ The effects of social distancing at the PTI on service performance (COF-G22-01)", J. Hyde and C. Rees, Rail Safety and Standards Board, London, 2021,
Start Year 2021
Description Siemens Mobility 
Organisation Siemens AG
Department Siemens Mobility
Country Global 
Sector Private 
PI Contribution People flow modelling developed in Sheffield for station environments is supporting work with Siemens for a station on the UK mainline railway network.
Collaborator Contribution For research on passenger flow in the rail sector to reach implementation in equipment, signage, and route-finding for passengers requires working with a supplier of this equipment. Siemens is a University of Sheffield partner and we are collaborating to see how the Sheffield modelling can contribute to improved flows at a test location on the mainline network where Siemens is already planning to undertake some work.
Impact The work is ongoing with preparation of bids and contracts. It has not reached output stage yet.
Start Year 2022
Title RateSetter SteerSuite implementation 
Description This is a bespoke development from the open source Steer Suite modelling tool for agent based pedestrian simulation. The primary contributions are (i) Pedestrian navigation algorithms specific to the platform-train interface for passenger railways. (ii) Addition of social distancing as a behaviour followed by agents in the railway environment. (iii) Calibrated real-world behaviour on compromise in social distancing when subject to incentives to board or alight public rail passenger transport. (iv) Extension to consider behaviour of groups, and the influence of luggage on movement. 
Type Of Technology Software 
Year Produced 2021 
Open Source License? Yes  
Impact Data have been generated using this code on boarding and alighting times for trains during periods of social distancing. This has been supplied to rail network operators though the Rail Safety and Standards Board to support planning network operation during the COVID-19 pandemic. 
Description Dissemination meetings with rail operators, Network Rail and RSSB 
Form Of Engagement Activity A formal working group, expert panel or dialogue
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Industry/Business
Results and Impact Presentations and debate on findings with UK railway operators, Network Rail (UK rail infrastructure manager), and Rail Safety and Standards Board. Some of these meetings were with specific companies, and some organised via the existing RSSB Platform Train Interface Working Group.

These meetings disseminated our findings to the people in train operating companies who manage railway platforms around the UK to support their safe operation during the pandemic and use of social distancing. The work triggered questions and discussion about methods of managing to avoid crowding. There was particular interest during periods of lockdown as planning was undertaken to maintain safe operation when lockdown finished and passenger numbers began to rise.
Year(s) Of Engagement Activity 2021
Description Does social distancing impact network performance? 
Form Of Engagement Activity A magazine, newsletter or online publication
Part Of Official Scheme? No
Geographic Reach National
Primary Audience Public/other audiences
Results and Impact Public information article created by Rail Safety and Standard Board describing our research and its integration with rail industry planning for pandemic recovery.
Year(s) Of Engagement Activity 2021