Prediction of PM Emissions During Transient Operation of Marine Diesel Engines Using Artificial Neural Networks

EasyChair Preprint 4199

Abstract

Many internal combustion engine emission limits are already prescribed for land transport. Stricter regulations are also expected for international shipping in the future. The International Maritime Organisation (IMO), a subdivision of the UN, has been negotiating for years on direct regulation of particulate emissions from ships. Therefore, in addition to exhaust aftertreatment systems, internal engine and operational measures are also of interest. This article focuses on an operational measure. A new type of assistance software is presented, which shows the nautical ship officer the environmentally relevant consequences of his actions already during manoeuvre planning and later during its execution. The well-known "black flag" on the funnel of a ship is usually the result of transient engine operation. A corresponding assistance software is dependent on a model of transient engine operation including the resulting particle emissions. Over decades, the reaction kinetics for the formation and oxidation of soot have been investigated. Such physical models are to be preferred, provided they meet the quality criteria and the calculation time requirements. At present, however, no model exists which describes the emissions of particulate matters (PM) in transient operation and can make predictions for several minutes within a few milliseconds. An alternative to physical modelling is data-based modelling. The shorter computing time is a major advantage here. On the other hand, there is a great need for training (measurement) data. Two different approaches of Artificial Neural Networks (ANN) were investigated for their applicability to the special case of PM emission prediction under transient engine operating conditions. The advantages and disadvantages of both approaches are discussed in the paper. The results were examined for their applicability in the Maritime Simulation Centre in Warnemünde (MSCW).

Keyphrases: Artificial Neural Networks, Particulate matters, dynamic non-linear process simulation, prediction, ship emissions

@booklet{EasyChair:4199,
  author    = {Michèle Schaub and Michael Baldauf and Egon Hassel},
  title     = {Prediction of PM Emissions During Transient Operation of Marine Diesel Engines Using Artificial Neural Networks},
  howpublished = {EasyChair Preprint 4199},
  year      = {EasyChair, 2020}}