Multi-objective optimization of internal combustion engine by means of 1D fluid-dynamic models

The definition of an efficient optimization methodology for internal combustion engine design using 1D fluid dynamic simulation models is presented. This work aims at discussing the fundamental numerical and fluid dynamic aspects which can lead to the definition of a best practice technique, depending on the complexity of the problem to be dealt with, on the number of design parameters, objective variables and constrains. For these reasons, both single-and multi-objective problems will be addressed, where the former are still of relevant interest (i.e. optimization of engine performances), while the latter have a much wider range of applications and are often characterized by conflicting objectives. The Mesh Adaptive Direct Search (MADS) was chosen among the class of direct search methods and compared with the Genetic Algorithms to solve single-objective problems, and similarly two different algorithms were chosen and compared to solve multi-objective problems: the [epsilon]-constraint method and the NSGA-II (Non-Dominated Sorting Genetic Algorithm) A single cylinder spark ignition engine, used in a motorbike application, was chosen as test case, to allow reduced computational times, without any loss of generality of the results. The analysis evaluate the convergence and efficiency of each methodology for the different problems which are solved. The achieved goal is not the definition of an ever valid mathematical strategy, but here focus is given on the parallel application of a detailed fluid dynamic analysis and automated optimization techniques to suggest a best practice technique to be employed depending on the characteristic of the optimization problem to be solved.