Outdoor air pollution is considered the largest single environmental health risk and is estimated
to cause 4.2 million deaths every year. Despite the major vehicle emissions reduction achieved
over the past two decades, road transport remains a major source of air pollutants such as
nitrogen oxides (NOx), contributing 39% of the total EU-28 NOx emissions in 2017. The
regular exceedances of the annual mean concentration limit for NO2, particularly in urban areas,
have been largely attributed to the discrepancies between type approval limits and real-world
driving emissions as well as the tting of defeat devices on diesel vehicles. In order to design
e ective air quality mitigation strategies, it is therefore crucial to improve our understanding
of and ability to model real-world driving emissions. Based on the largest recorded Portable
Emissions Measurement System (PEMS) dataset, which comprised 287 Euro 5 and Euro 6
diesel and petrol vehicles, this PhD thesis aims to ll these gaps by providing new emissions
models based on an extensive dataset. It is demonstrated in this thesis that while physical parameters such as vehicle weight or
engine size did not show any correlation with real-world NOx emissions, external parameters,
particularly driving dynamicity, are directly correlated with real-world driving emissions. The
e ect of driving dynamicity on real-world emissions is shown to decrease with the successive
regulations (Euro standard), indicating a general improvement of aftertreatment systems. The
rst model presented is an aggregated emission model, while the second emission model is an
instantaneous emission model but both directly account for driving dynamicity, although the
way they do di er signi cantly. Both models are reliable and accurate, presenting relative error
in prediction smaller than 20%. Additionally, this PhD thesis also intends to gain insights
on the real-world impact of an air quality mitigation strategy on the emissions and local air
quality. Application of the developed models to the assessment of the impact of a tra c
intervention on air quality demonstrated that although the chosen mitigation strategy had
locally a measurable impact on emissions and air quality, this impact was small compared to
the variations in pollutant concentrations induced by the meteorological conditions. The ease
of use of both models, as well as their wide range of applicability make them ideal operational
tools for policy makers aiming to build emission inventories or evaluate emissions mitigation
strategies.