Are internally observable vehicle data good predictors of vehicle emissions?

• Autor(a) principal: Fernandes, P.
• Data de Publicação: 2019
• Outros Autores: Macedo, E., Bahmankhah, B., Tomas, R. F., Bandeira, J. M., Coelho, Margarida C.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/27143
• topic: Portable emissions measurement system; On-road emissions; Diesel; Internally observable variables; Vehicle specific power

Scientific research has demonstrated that on-road exhaust emissions in diesel passenger vehicles (DPV) exceeds the official laboratory-test values. Increasing concern about the quantification of magnitude for these differences has meant an increasing use of Portable Emissions Monitoring System (PEMS), but the direct use of Internally Observable Variables (IOVs) can be useful to predict emissions. The motivation for this paper is to develop an empirical approach that integrates second-by-second vehicle activity and emission rates for DPV. The objectives of this research are two-fold: (1) to assess the effect of variation in acceleration-based parameters, vehicle specific power (VSP) and IOVs on carbon dioxide (CO2) and nitrogen oxides (NOx) emission rates; and (2) to examine the correlation between IOV-based predictors of engine load and VSP. Field measurements were collected from four DPV (two small, one medium and one multi-purpose) in urban, rural and highway routes using PEMS, Global Positioning System (GPS) receivers and On-board Diagnostic (OBD) scan tool, to measure real-world exhaust emissions and engine activity data. Results suggest the relative positive acceleration (RPA) and mean positive acceleration (MPA) allowed a good differentiation with respect to route trips. IOVs models based on the product of manifold absolute pressure (MAP) and engine revolutions per minute (RPM), and VSP showed to be good predictors of emission rates. Although the CO2 correlation was found to be good (R2 > 0.8), the models for NOx showed mixed results since some vehicles showed a reasonable correlation (R2 ~ 0.7) while others resulted in worst model predictions (R2 < 0.6). IOVs models have potential to be integrated into vehicle engine units and connected vehicles, for instance, to provide real-time information on emissions rates, but other parameters regarding the thermal management on after treatment system must be included in NOx prediction. This would allow for a better understanding of true physics behind NOx emissions in DPV.