Are we getting vehicle emissions estimation right?

Abstract

Simulation is a valuable prediction tool to evaluate the impact of climate change interventions prior to disruptive implementations that may yield unintended and unwanted consequences. Advances in agent-based simulation allow us to estimate the dynamic emissions produced by vehicles at a fine resolution, with each vehicle modelled individually. The scalability of these models allow decision-makers to evaluate large-scale scenarios. But how good are these models? In this paper, the authors apply a state-of-the-art emissions model in the Multi Agent Transport Simulation (MATSim) framework to simulate individual vehicle emissions. Simulation results are compared with real driving emissions tests, using a portable emissions measurement device on a light and heavy vehicle. Quantifying the gap between macro-level simulation and actual emissions, the paper contributes by showing that pollutants are significantly underestimated, especially for heavy vehicles. This confirms prior micro-scale studies and magnifies the detrimental impact of using uncorrected aggregated models to inform environmental policy affecting transport. • Quantifying the gap in emission estimates on large-scale traffic simulations. • City-scale activity-based transport model coupled with vehicle-specific emissions. • Variation in real driving emissions is much larger than in randomised simulations. • Macro-level models underestimate heavy vehicle emissions more than light vehicles.