Design Optimization of Axial-flow Fan on Various Operation Conditions

In this paper, the Design of Experiment and metamodel-based multi-objective design optimization of the axial-flow fan is investigated. The Design of Experiments proceeded for analyze performances on the various operation conditions of the axial-flow fan. The 5-level full factorial design is selected for operation conditions and the 3-level near orthogonal array for design variables. The metamodels are constructed on proper pitch angles for each flow rate and then evaluated based on predicted R-squared. The RBFs are selected on low flow rates and the kriging is selected on high flow rates. For multi-objective design optimization, the mean values and the standard deviations for performances are selected as the objective functions and constraints. Optimization using NSGA-II is investigated and the Pareto front and compromised optimal solution which improves the objective functions with satisfying constraints are acquired. To confirm the accuracy of the optimal, the validation process and flow characteristics analysis are investigated. The errors of metamodel are under 0.4% for mean values and 4.2% for standard deviations and reasonable flow characteristics are acquired.