In this study we present models for the parametric optimization of a centrifugal fan impeller using kriging-simulated annealing (SA) meta-algorithm. First, a kriging model is constructed using a limited number of CFD simulations for the centrifugal fan impeller to be optimized. The inlet and outlet blade angles are chosen to optimize the impeller. A dataset consisting of 22 different blade angles are determined by Latin Hypercube Sampling (LHS). After validation of the kriging model, it is used in conjunction with the simulated annealing and thus a meta-algorithm is developed for the solution of global optimization problem for the impeller optimization. Within the desired range of parameters, it is shown that this meta-algorithm provides a robust, reliable and fast optimization method. The procedures can be used to many problems in engineering. In this study a centrifugal fan impeller is successfully optimized using this procedure.

This content is only available via PDF.
You do not currently have access to this content.