Assessing on-road emissions from urban buses in different traffic congestion scenarios by integrating real-world driving, traffic, and emissions data

In recent years, the integration of traffic simulators and emission models has become the most preferred option for evaluating vehicle emissions in different traffic states. However, the definition of a 'traffic condition' is often subjec-tive, as driving patterns can vary significantly with the spatial domain of study. Alternatively, the implementation of 'Cooperative Intelligent Transport Systems' has led to a growing variety of devices being installed, both on the road and in public transport vehicles for monitoring traffic-flow conditions and vehicle speeds in cities. This study purposed an original approach for integrating real-world emissions (as an micro-emission model), real-world driving profiles, and city traffic sensor data to assess the effects of traffic congestion at the route level on emissions from urban buses in Madrid (Spain). The definition of the traffic scenarios was based on a K-means clustering analysis by linking station-ary (from city sensors) and dynamic (from bus driving profiles) congestion indicators. In parallel, a micro-emissions model based on vehicle-specific power (VSP) methodology was used to model second-by-second CO2 and NOx emis-sions from individual trips of the diesel and compressed natural gas (CNG) buses. Finally, the clustering and modelled emissions data were combined. A comparison of the free flow and the severe congestion scenarios showed that the average speed of the route de-creased by approximately 50 %, and the number of stops per kilometre increased by a multiple of 1.5; furthermore, the CO2 and NOx emissions from buses increased by approximately 50 % and 85 %, respectively. The diesel bus showed a lower sensitivity to variations in the congestion level at the route level, although the low-NOx emissions from the CNG buses were evident for all traffic scenarios. The results of this study, based on extensive real-world data, can be used to develop high-resolution vehicle emissions inventories.