| In the present work a new software system for multi-objective flow optimization is described. The main features are flow simulation (CFD), bio-inspired optimization algorithms and artificial neural networks. The approach is based solely on free-of-charge OpenSource software to offer a cheap and adaptable alternative to commercial software packages. Several optimization examples of labyrinth seal geometries as used in gas turbines are presented to demonstrate the capability of the new software framework. A straight-through geometry is optimized by both CFD and ANN varying five parameters. As a more complex test case an optimization of a stepped seal is carried out with seven variable parameters. To find a geometry which shows low leakage both in an unworn (new) and worn state, a multi-objective approach is selected. The results presented reveal the high potential of the optimization framework. Its OpenSource character makes it perfectly suited for wide-spread application in many fileds of interest. |
