Comparing an energy-based ship emissions model with AIS and on-board emissions testing

On-board emission testing data for two ocean-going vessels is used to assess the performance of a new Australian ship emissions model, and to also assess the impact of local currents on emission predictions. Prediction performance is only marginally affected by AIS post-processing method and inclusion of local current information. Model performance was assessed for three different aspects, fuel-based emission factors (g/g CO2), engine work-based emission factors (g/kWh) and distance-based emission factors (g/km). Analysis of fuel-based and engine-work based emission factors suggest good performance and small to reasonable mean prediction errors for CO2 (±10%), PM10 (±15%) and SO2 (±20%). For NOx and CO, on-board emissions testing suggest that model emission factors are biased high and low with mean prediction errors +60–70% and −60%, respectively. The results for distance-based emission factors were not considered to be meaningful due to spatial and temporal inaccuracies in linking on-board testing with the AIS data that could not be resolved. Given the importance of AIS data as input to fuel and emissions modelling, it is recommended that the spatial and temporal accuracy of AIS data is investigated and confirmed in future studies. Moreover, the differences found in this study between model predictions and on-board measurements highlight a few limitations in application of generic fleet-based models. © 2022 The Authors