Search

• A two-step modal emission model for HEVs was developed using real-driving emissions. • The HEV can achieve over 50% CO 2 and CO emission reduction in on-road tests. • The model was applied to large real-driving datasets from Toronto and Beijing. • Emission reductions were lower for Toronto, which has more aggressive driving. • The study highlights the conditions where HEV can yield the most emission benefits. This study tested exhaust emissions and fuel consumption for a hybrid electric vehicle (HEV) in real-world conditions using a portable emissions measurement system (PEMS). A gradient boosting model was developed to predict the electric motor's operation and emissions using only vehicle kinematic data. The model was applied to estimate the potential emission reductions that would be achieved with HEVs compared to conventional vehicles, using two large real-driving activity datasets collected in Greater Toronto and Metropolitan Beijing. The emission reductions estimated for Toronto were 21.6%, 31.3%, and 53.0% for CO 2 , CO, and NO x , and 41.0%, 28.9%, and 68.5% for Beijing. We observed higher emission reductions for CO 2 and NO x under low power demand vehicle operations, which occur more frequently in Beijing, while more aggressive driving was noted in Toronto, leading to smaller estimated benefits of HEVs. Compared to previous studies, our explainable gradient boosting model improved prediction accuracy and robustness substantially by achieving an average Pearson correlation of 0.741 from cross-validation. This study goes beyond an analysis of HEV emissions from engine and motor operations by applying real-world driving data from two large metropolitan areas to the model. By doing so, a novel investigation of the traffic situations, roads, and driving behaviours that yield the highest emission benefits for HEVs was conducted. [ABSTRACT FROM AUTHOR]