Modelling of air quality in and around railway station environments

Abstract

Air pollution is a major health concern, contributing to an estimated 1,300 premature deaths annually in Ireland. Most Irish rail services still use diesel traction, and over 97% of the network is not electrified. Busy, semi-enclosed termini can therefore become local hotspots of particulate matter and nitrogen oxides. This thesis examines air quality in and around two major diesel railway stations in Dublin: Heuston and Connolly. It addresses the lack of detailed emission tools and dispersion models for rail hubs, and the limited evidence on how train idling and operations affect pollution exposure for passengers, staff, and nearby residents. A systematic literature review of air quality modelling at transportation hubs is first conducted to identify research gaps and to guide the choice of statistical, emission and dispersion methods used in the case studies. Field campaigns then measure PM₂.₅, PM₁₀ and NO₂ on platforms and concourses, alongside background air quality and local meteorology. Timetables and on-train recorder data are used to derive hourly train movements and cumulative idling time in different platform zones. Random Forest models show that urban background concentrations set the main baseline, but that idling in the enclosed platform regions is the strongest operational factor driving station increments. A detailed idling analysis and Bayesian modelling link the distribution of idling time to timetable constraints and driver behaviour. The results show that small changes in layover times or shutdown practices could reduce unnecessary idling. To convert train operational activities into emissions, a physics-based tool is developed that combines fleet information, notch-based emission factors and movement data to produce time- and space-resolved NOx and PM₂.₅ fields for the main diesel fleets. These inventories reveal that station-area emissions form only a small share of national rail totals but are highly concentrated near platforms, depots and approach tracks, with Class 201 locomotives dominating per-train emissions. The same tool is also applied to quantify the emission reduction achieved when Class 201 locomotives operate on Hydrotreated Vegetable Oil instead of diesel. Finally, an ADMS dispersion model uses the rail and road emission inventories to map pollutant concentrations around Connolly station and to test separate rail and road contributions through scenario runs. Results indicate that station-related increments can be comparable in magnitude to nearby road-traffic increments at the most exposed receptors along the station-road corridor, but are highly localised and strongly dependent on wind direction and distance from rail stations. Together, the thesis provides an integrated framework that links monitoring, statistical modelling, emissions calculation, and dispersion modelling, and it offers practical measures to reduce air pollution from diesel train operations in semi-enclosed urban stations.