Estimation of ship flue gas emissions in dynamic operational conditions with ANN

Abstract

While ship emissions are incomparably less in comparison to land units, the threat of the world's population and certain regions are in a serious danger due to 70% of these emissions occur in areas up to 400 km from the coast and on certain maritime routes. The formation of ship emissions strongly depends on the dynamically varying travel conditions and it is important to identify the emissions in order to guide the rule-makers to take the necessary measures correctly. In this study, an emission estimation modelling based on dynamic variables, such as voyage duration, engine revolutions per minute, speed, displacement, weather condition, sea conditions and average draught, has been realized by using artificial neural networks (ANN). The difference between the results of ANN and traditional estimation methods was found to be 1.57%. Then, in order to determine the optimum route, the ANN model was implemented for June and January in the Northern and Southern routes of the Atlantic and Pacific Oceans and it was concluded that harder sea and weather conditions produced more emissions. Finally, fuel consumptions, fuel costs and social costs of the emissions for different routes were calculated.