Next: Parametrization of the shape
Up: Constrained Optimization
Previous: Numerical Results for the
Contents
The METHOD project
My work on optimization techniques for high computing load,
continuous function without derivatives available is financed by
the LTR European project METHOD (METHOD stands for Achievement Of
Maximum Efficiency For Process Centrifugal Compressors THrough New
Techniques Of Design). The goal of this project is to optimize the
shape of the blades inside a Centrifugal Compressor (see the
figure 11.1 for an illustration of the blades
of the compressor).
Figure 11.1:
Illustration of the blades of the
compressor

The shape of the blades is described by 31 parameters. The
objective function is computed in the following way:
 Based on the 31 parameters, generate a 3D grid.
 Use this grid to simulate the flow of the gas inside the
turbine.
 Wait for stationary conditions (often compute during 1
hour).
 Extract from the simulation the outlet pressure, the outlet
velocy, the energy transmit to the gas at stationary conditions.
 Aggregate all these indices in one general overall number
representing the quality of the turbine. (That's this number we
are optimizing.)
We simply have an objective function which takes as input a vector
of 31 parameters and gives as output the associated quality. We
want to maximize this function. This objective function has been
implemented by the Energetics Department ``Sergio Stecco'' of the
''Università degli Studi di Firenze'' (DEF). For more information,
see [PMM$^+$03].
Subsections
Next: Parametrization of the shape
Up: Constrained Optimization
Previous: Numerical Results for the
Contents
Frank Vanden Berghen
20040419