Abstract
Freight transport is essential to the function of a city. Yet in urban areas, goods are mostly carried by road vehicles powered by fossil fuels, with road freight contributing to between 20–30% of transport-related CO2 emissions. Quantifying and examining the fuel use of freight vehicles at a high spatiotemporal resolution is essential to understanding the underlying patterns and inform future interventions towards fuel-efficient freight. With the widespread use of GPS for tracking vehicle fleets and advancements in driver activity surveys, instantaneous fuel consumption/emission models can be used to calculate fuel use at high resolution, using inputs such as speed and acceleration derived from GPS data. But while a wide variety of models exist, studies comparing fuel use estimates from different models are lacking, especially when applied with the constraint of limited calibration. This work (i) evaluates the accuracy of fuel use estimates from four fuel consumption/emission models using GPS data as inputs, and (ii) examines changes in performance when these models are supplemented with partial on-board diagnostics (OBD) data and payload information from a driver activity survey. Four models were chosen for the study: COPERT 4, CMEM, SIDRA TRIP, and MOVES. Results show that the chosen models predict a wide range of fuel use when tested on a standardised drive cycle, especially for heavier vehicles in the sample. These models were then applied to real-world GPS traces of ten diesel commercial road vehicles in Singapore using the method of space-time path segments (STPS) and road grade obtained from a digital elevation model (DEM). When compared to actual fuel use calculated from OBD data, the model SIDRA TRIP performed the best, especially when compared with other models at original settings. When supplemented with idle fuel rates from OBD data and payload information, model performance converged and some models improved consistently while other exhibited the opposite behaviour. Finally, the best model (SIDRA TRIP) from (i) and (ii) is applied to GPS traces from 712 heavy-duty vehicles in Singapore. With the estimated instantaneous fuel use for each vehicle, the CO2 emissions of freight transport in Singapore are examined together with its distribution across vehicles, industries, space and time. On average, the resultant CO2 emission factors were comparable but consistently higher than existing default emission factors, although large variations exist within the data set. Spatiotemporally, CO2 emissions closely followed driver working patterns and were more concentrated in the west of Singapore. With this work, various stakeholders – such as planners, policymakers, analysts and operators – can monitor vehicle fuel use at a microscopic level and estimate the environmental impacts of transportation on a larger scale, so that well-informed decisions can be made towards sustainable transportation in our cities.