Ship Propulsion Plant Performance Assessment Using An Artificial Neural Network

Nowadays, more than in the past, the attention towards the environmental impact of shipping has gained great interest. In particular, several international rules were issued to set new standards in terms of ship energy efficiency and emissions. Most of the actual worldwide fleets are not compliant with the new standards, and it is unthinkable that ship-owners will replace the existing ships with new buildings in a short time. According to this, the retrofit of either the propulsion plant or auxiliary system is the good compromise choice. The first task that the designer has to face is the evaluation of the actual propulsion plant performance to detect where to act. On the view of this, the authors present two different approaches to identify the performance of an existing ship propulsion plant equipped with a four-stroke diesel engine and a controllable pitch propeller. The first approach is the standard approach, relying on the static performance assessment of the required power and fuel consumption, starting from the design data of the hull and machinery, not always available several years past ship fabrication. The second approach is based on the application of an artificial neural network, trained using the results of sea trials. Ship speed, shaft revolution speed, pitch angle, engine torque and fuel consumption have been recorded, then part of the data have been used as a training set for the artificial neural network, and the remaining as a validation set to compare the two approaches. The main idea is to evaluate the best strategy, in term of developing time and accuracy, to obtain the global, even if static, evaluation of the propulsion plant performance, with the final aim to have a handy tool to be used to assess potential energy saving solutions. Eventually, a comparison between the two methodologies and sea trials is shown and critically discussed.