Extending the lifetime of modern gas turbines is becoming increasingly challenging. A proper method for exploiting further improvement potential is to utilize automatic design optimization techniques, where for example, optimization may help to prevent excitations of eigenmodes of compressor airfoils thus, decreasing service costs, by reducing the risk of high cycle fatigue. However, in some cases the avoidance of any excitation may be not possible or too expensive due to the loss of aerodynamic efficiency. Since different eigenmodes do not share the same risk of failure, this is also not always necessary. Numerical optimization may account for this by penalizing the various eigenmodes differently which, however, requires that a proper mode-recognition technique be robust against design and mesh changes during the optimization process. This paper will introduce such a technique based on self-organizing maps (SOM). It will be used to project the suction-side surface of different blade geometries and mesh types onto a standard map in order to make deformations comparable for classification techniques. This enables to apply e.g. the mode-assurance criteria (MAC) in order to find correlations between the actual and already classified reference modes. The method will be demonstrated on two rotor blades.

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