Abstract
Diesel engine combustion chamber which reduces exhaust emission has been designed using CFD analysis and optimization techniques. In order to save computational time for design, the Kriging model, one of the response surface models, is adopted here. For a robust exploration, both the estimated function value of the model and its uncertainty are considered at the same time. In the present problem, the k-means method is used to limit the number of additional sample points to a reasonable level. Among the additional sample points, two combustion chamber shapes dominate the baseline configuration in terms of all objective functions. Compared with the previous optimization with the evolutionary algorithm, its computational time for design was cut by 95%. The results indicate that the present method is a practical approach for real-world applications.
