Comparison of Data Analysis Methods for European Real Driving Emissions Regulation 2017-01-0997

The gap between regulated emissions from vehicle certification procedures and real-world driving has become increasingly wider, particularly for nitrogen oxides (NO_x). Even though stricter emission regulations have been implemented, NO_x emissions are dependent on specific, short-duration driving events which are difficult to control, therefore high concentrations of these pollutants are still being measured in European cities. Under certification procedures, vehicle emissions compliance is evaluated through standards, recurring to driving cycles performed on chassis dynamometer under controlled laboratory conditions. Different countries use different standard cycles, with the US basing their certification cycle on FTP-75 and Europe using NEDC (Euro 5/6c)/WLTP (Euro 6d). However, the representativeness of standard driving cycles has been under discussion and, consequently, new updates on the light duty regulation from the European Commission include a Real Driving Emissions (RDE) Regulation, establishing that vehicle emissions must be measured on the road with a Portable Emission Measurement System (PEMS). However, RDE tests have shown weak points, namely regarding testing boundary conditions and data analysis methods. Moreover, reproducibility of tests is hardly achievable and both dynamic and environmental boundaries (such as ambient condition, route, etc.) are unique for a specific geographical location, which can be representative for one location but not for another test-site. On-road data analysis methodology is defined by the European Commission, with two main methods are being tested: the Moving Averaging Window (MAW); and the Power Binning (PB). As result, the scope of this study was to evaluate the RDE data analysis methods, including the Vehicle Specific Power methodology (used on MOVES model), applied to 1 Hz on-road data from 3 vehicle propulsion technologies (spark-ignition, compression-ignition and hybrid), collected in Lisbon, Portugal. This approach allowed analyzing and identifying the differences between the methods for each vehicle propulsion technology. Regarding total CO₂ emissions and NO_x emissions, the application of the methods indicates that the MAW provides an overall difference of around 7% for CO₂ and 10% for NO_x comparing with the PB method. The PB is conceptually similar to VSP, although an overall 10% for CO₂ and 19% for NO_x difference was found. Differences on the methods results are due to the use of different approaches and, consequently, lead to different results, which are more explained on this paper.