A new reduced model of scavenging to optimize cylinder design

Abstract

This paper presents the development of a new scavenging model to optimize cylinder design. The developed model explicitly integrates some of the cylinder’s design and environmental variables to describe flows during the scavenging process. Then, based on the reduced model, an optimization phase was carried out in order to determine the optimal values of cylinder variables. From computational fluid dynamics (CFD) models to optimal cylinder design, all method steps used are described in this paper. Adaptable to any type of engine, here it is applied to the particular two-stroke diesel engine with ports only. In order to fully understand fluid flow, the methodology integrates a number of CFD calculations with different cylinder configurations to provide data. The CFD results are used as neural network outputs during the training phase, whereas the cylinder variables plus the crankshaft angle are the inputs. The trained network provides the analytical reduced model for gas composition transiting through ports, which characterize the scavenging process. Thanks to these, genetic algorithms are run to define the most suitable values of cylinder variables in order to improve scavenging. The entire process for establishing the reduced model and the optimal design of the chamber is presented in this paper.