| Autor: |
Stefano Piola, Pio Astrua, Andrea Silingardi, Federico Bonzani |
| Rok vydání: |
2012 |
| Předmět: |
|
| Zdroj: |
Volume 8: Turbomachinery, Parts A, B, and C. |
| Popis: |
This paper presents a flexible and effective optimization approach to design an axial compressor transonic blade for heavy duty gas turbines. The design goals are to improve design efficiency, choke margin and off-design performance while maintaining mass flow in design point as well as structural integrity. The new blade has to provide a wide operating range and to satisfy tight geometrical constraints. A database of aero-mechanical calculation results is obtained for three operating conditions. A number of 3D flow simulations are performed using a CFD solver with endwall boundary layer simplified model (thin layer) to reduce computational costs. The optimization process adopts a set of artificial neural networks (ANN) trained for each operating condition and a random walking search algorithm to determine the multi-objective Pareto Front. ANN enables speed up of the optimization process and allows high flexibility in choosing criteria for optimum member selection. Random walking algorithm gives a fast and effective method to predict the multi-dimensional Pareto Front. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
