Optimization of a segment of the main mirror for the “OWL” Telescope

Summary of the Case History
Building on the success of its 8-m Very Large Telescope (VLT) and the coming to maturity of controlled optical systems, ESO is undertaking the design of a giant, next generation optical and near-infrared telescope, dubbed OWL for the eponymous bird's keen night vision, and for OverWhelmingly Large. With a diameter of 100 meter, OWL will combine unrivalled light gathering power with the ability to resolve details down to a milli-arc second.
This work is focused on the crucial issue to optimize the support configuration and the geometry of the mirror modules used for the OWL main and secondary optical surfaces.
The goal is to design a lightweight structure (minimizing the areal density) keeping the deformation of the mirror surface at least under a fixed maximum limit.

In the frame of the project one crucial issue is the optimization of the support configuration and of the geometry of the mirror modules that are used for the OWL main and secondary optical surfaces. The goal is to design a lightweight structure (minimising the areal density), while keeping the deformation of the mirror surface within design limits.
Such target was hit through a modeFRONTIER based procedure, controlling a parameterised model which was built in the ANSYS environment. The starting configuration was a relatively good design which was obtained by running the standard ANSYS optimiser with default settings. Working on that with modeFRONTIER an amazing theoretical 31% weight saving was obtained, which, compared to the total mirror size, leads to a global weight saving of 167 tons. And this without any loss of optical performances and quality, as well as complying with all other design requirements and constraints!

The coupling between ANSYS and modeFRONTIER proved to be the right choice and, moreover, straightforward and very efficient. After some preliminary tests to find out which approach and algorithm among the variety offered by modeFRONTIER should better suite the application, the choice was the new set of multi-objective optimization algorithms based on the game theory. A total of only 107 models were evaluated to reach the result.
All in all, the different tests carried out, as well as the comparison with the ‘standard’ ANSYS optimization and the nature of the algorithms that were employed, leads to conclude that the solution is likely the ‘global optimum’ within the current bounded input variables domain space.


For more informations:
info@enginsoft.it

Images from:
http://www.eso.org/projects/owl