Nox emissions from passenger cars under real driving conditions
- Autores: Zamir A. Mera Rosero
- Directores de la Tesis: José María López Martínez (dir. tes.), Natalia Elizabeth Fonseca González (codir. tes.)
- Lectura: En la Universidad Politécnica de Madrid ( España ) en 2020
- Idioma: español
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
Pollution from motor vehicles affects air quality in all global cities. In real-world driving, most Euro 5 and early Euro 6 diesel passenger cars exceed the nitrogen oxides (NOₓ) emission limits of the type approval procedure. In addition, the prediction of instantaneous NOₓ emissions is difficult because of its high nonlinearity and dependence of aftertreatment action, which decreases the certainty of NOₓ estimations. These aspects restrict the applicability of policy strategies for the mitigation of pollution.
The objective of this works is to provide a better understanding of motor vehicle NOₓ emissions and better quantification criteria for microscopic emission models, in particular, from diesel Euro 6 passenger cars under real-world driving conditions. Therefore, in this study, real-world NOₓ emissions and operational data from Euro 6b passenger cars were examined. The vehicles had different powertrains and aftertreatment systems including gasoline vehicles with a three-way catalyst (TWC), namely, one gasoline direct injection (GDI) and one hybrid electric vehicle (HEV); four diesel vehicles with only exhaust gas recirculation (EGR), lean-burn NOₓ trap (LNT) and selective catalytic reduction (SCR). Descriptive statistical analysis was performed to determine the relationships between NOₓ emissions and the vehicle, engine, and aftertreatment operation conditions. Then, a physical, a data-driven, and a hybrid microscopic model were compared to understand the influence of the NOₓ emissions peaks on the prediction accuracy of microscopic models.
High instantaneous NOₓ emissions (NOₓ peaks) were clustered using a new developed method called high emission sets (HES). The results obtained using HES, indicate that the NOₓ peaks account for a large proportion of the total emissions produced in a short amount of time. The NOₓ emissions are reflected as a high nonlinear function dependent on the operating conditions of the powertrain, fuel injection, and particularly, of the NOₓ control system. From the sample of tested vehicles, the gasoline hybrid electric and diesel SCR-equipped vehicles showed real-world NOₓ emissions under Euro 6 limits. It was demonstrated that the efficiency of TWC and SCR systems—both continuous catalytic converters—depend on operating temperatures, and in the case of TWCs, on air-fuel ratios. Therefore, during the first five minutes of cold start, the HEV and SCR-diesel emission levels reached similar values to those observed for the other NOₓ control technologies. In addition, low loads in urban driving or downhill in rural areas cooled down the SCR catalyst, thereby decreasing efficiency. These aspects suggest the need for active heating for SCR and TWC (in hybrid applications) aftertreatments. High NOₓ emissions were measured from the GDI because of the recurrent leaner air-fuel ratios, where the TWC catalyst decreased the deNOₓ efficiency. Regarding microscopic models, in the case of selective catalytic reduction (SCR) systems, the results suggest the need for modelling NOₓ emissions in two blocks: one for engine-out outputs and another for tailpipe predictions. The low occurrence of the NOₓ peaks affects the accuracy of the predicted emission factors. “Low” NOₓ emissions affect instantaneous prediction metrics, such as R2 and RMSE. The use of vehicle activity parameters becomes useful inputs to model physics-based, and machine learning models. A hybrid approach that combines the physics-based model with a machine learning model for the aftertreatment system is a good alternative as enables the use of the advantages of the physics and data-driven models for reducing the amount of dimensioning and characterisation of the aftertreatment layout.
