|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.