Statistical Analysis of Parameter Variations using the Taguchi Method
A. Wolf, D. Henes, S. Bogdanski, Th. Lutz and E. Krämer, BMWI-Verbundvorhaben MUNA, to be published, 2009
Abstract:
This project is part of the MUNA-project. It deals with the investigations of uncertainties of a CFD-Simulation caused by variation of numerical parameters, inflow conditions and geometrical parameters. To evaluate the influence of different parameters on the results of an experiment or simulation usually a full parameter analysis must be performed. A full parameter analysis requires a high number of simulations, rising with the number of parameters, which are investigated. In this study the Taguchi method, which is based on DoE (Design of Experiments) methods and well known for the optimization of production cycles, is applied. It reduces the number of required simulations dramatically and therefore also the costs. In this article the basics of the Taguchi Method are explained and a summary of the
ANOVA-analysis, is given. The ANOVA analysis is used for the analysis of the results obtained by the Taguchi method. It delivers the relative influences of each investigated parameter and also the influence of interaction among different parameters can be obtained. The effect of the parameter variations on the CFD result is shown for two- and three dimensional simulations. Numerical, geometrical and also inflow conditions are investigated.
