Real-world driving cycles and energy consumption informed by large-sized vehicle trajectory data

Abstract

The inadequate representativeness of driving cycles used by regulatory lab tests is one significant factor leading to the large discrepancy between real-world fuel consumption and type-approval levels. Onboard measurement devices have been used in previous researches to collect vehicle activity data but the amount of data is sometimes limited. With second-by-second GPS trajectory data of 459 private passenger cars, covering over 17,000 sampling days, this study used big-data mining techniques to study the variations in real-world driving cycles. A Markov chain method was developed to generate typical driving cycles that have representative features of real-world driving. As a case study, two typical cycles, Off-peak cycle, and Peak cycle are constructed from six sub-cycles representing different road types and travel periods. The travel dynamics indicated the New European Driving Cycle (NEDC) would be too mild to represent real-world driving in China. The simulation results of vehicle fuel consumption showed that different driving cycles could lead to different lab-to-road gaps when comparing with NEDC type-approval levels. For example, the fuel consumption (median value) of Off-peak cycle and Peak cycle were higher than the NEDC type-approval level by 29.3% and 37.5%, respectively. This study highlights the importance of addressing real-world features in improving future fuel economy regulations. The practical approach to generate representative driving cycles ensured by massive travel profiles can be employed reliably for fuel consumption and exhaust emission assessment in the future.