TASCOPT - A Shape Optimization Plateform for Turbomachinery Applications - Archive ouverte HAL Access content directly
Reports (Research Report) Year : 1999

TASCOPT - A Shape Optimization Plateform for Turbomachinery Applications


We present TASOPT, a new plateform for shape optimization in rotating frames for in particular applications in turbomachinery. The theoretical background comes from the idea develloped at INRIA for aerodynamical shape optimization. TASOPT adopts the unstructured CAD-free framework for shape and mesh deformati- ons where the interface with CAD is realized at the final stage of the optimization. This framework is extended to 2D and 3D multi-block structured grids. Automatic differentiation through Odyssée is used to access discrete sensitivities in reverse mode. In addition, the incomplete gradient approach for functionals based on boundary integrals enables for a drastic reduction of the computational cost. The plateform uses various flow solvers depending on the physic of the problem ranging from incompressible to high speed flows. Various minimization algorithm based on linear and quadratic models for the functional are presented. These ingredients are illustrated through a first cascade configuration blade shape optimization. In addition, we present an exemple of the applicatio- n of the plateform in industrial environment for VALEO-TM in the design of 3D cooling engine devices. In particular, we show how the plateform has been interfaced with the commercial flow solver TASCFLOW. This extension requires the developpement of specific interface between TASCOPT, TASCFLOW and CAD tools used by VALEO.
Fichier principal
Vignette du fichier
RR-3803.pdf (1.59 Mo) Télécharger le fichier

Dates and versions

inria-00072856 , version 1 (24-05-2006)


  • HAL Id : inria-00072856 , version 1


Mugurel Stanciu, Bijan Mohammadi. TASCOPT - A Shape Optimization Plateform for Turbomachinery Applications. [Research Report] RR-3803, INRIA. 1999. ⟨inria-00072856⟩
94 View
155 Download


Gmail Facebook Twitter LinkedIn More