Review and evaluation of emission models and vehicle simulation tools

One of the main concerns regarding the road transport sector is the fact that it constitutes one of the main sources of air pollution, especially in urban areas, since the combustion of hydrocarbon fuels in vehicles produces several pollutants. The most common approach for the assessment of traffic-related emission factors is the exhaust gas measurement of vehicles on chassis dynamometers over various driving cycles. A rather favourable approach in order to reduce the number of experimental procedures and thereby the cost of such tests is the development and calibration of vehicle simulation tools and emission models which could be used for the accurate evaluation and quantification of vehicle emissions without the necessity of expensive experimental campaigns. Today, there are several tools for the estimation of traffic-related emissions. Such tools are essential in any European or global policy dealing with emission projections, air pollution or climate change issues. This study presents a description of the current emission models (COPERT, EMFAC, etc) and vehicle simulation tools (ADVISOR, AUTONOMIE, PHEM etc). The review of existing models and methods provides evidence that there is a large variety of available tools to calculate traffic-related emissions and to develop road transport emission inventories, however, new trends and policies must also be fully incorporated in the existing tools. In addition, in order to use emission models and vehicle simulation tools in the proper way, detailed and precise measurements of vehicle operation are required, otherwise any potential benefits may be lost. This is likely to be rather difficult since such information is relatively expensive or difficult to collect. For example, certain input data may not be available, such as vehicle loading and gear-shift behaviour. The last point raises an important consideration regarding model complexity. More complex models have the potential to provide more accurate predictions as they take into account more variables. However, they also require more detailed input data which may not be readily available to the model user.