Paul joined IRT SystemX in January 2014 to write his thesis on a Bayesian approach to constrained multi-objective optimization. Here he revisits the doctorate he completed while working on the ROM project (Model Reduction and Multiphysics Optimization) supported by CentraleSupélec.
What was the subject of your thesis?
Throughout my doctorate, I was interested in the issue, initially raised by Safran, of numerical optimization of complex systems. The chains of calculations used to simulate complex systems are often quite resource intensive and it can take super-computers many hours to generate a result. Together with my supervisors, we perfected an algorithm which allows us to identify concepts relevant to performance and that comply with lower cost specifications. It is important to understand that it is not the simulation chains that are faster, but rather the way in which they are utilized by optimizing the selection of test geometries.
While writing my thesis, I had the opportunity to put my work to the test with concrete applications such as optimizing the geometry of a Safran turbo-machine fan blade, a powertrain for the Renault Twizzy and an air-conditioning system for the Airbus A320.
What did you gain from your doctorate?
SystemX gave me the opportunity to work with different manufacturers on a range of issues and to build a solid professional network – an essential part of my integration into working life. Thanks to my interactions with SystemX’s partners, I was able to apply the algorithm to a number of research projects, each more exciting than the last.
What is your fondest memory of your time at SystemX?
I couldn’t name just one; there were many. I would say probably the general atmosphere at IRT SystemX that provided me with a unique, flexible, serious, but most of all, friendly working environment.
What projects do you have in the pipeline?
Since April, I have been working at Safran Aircraft Engines as an optimization methods engineer. I am a part of the Methodologies and Tools Development department. My main duties are to offer support to the design office on their optimization issues and to lead R&T projects on multi-objective optimization and topology optimization. As such, I will no-doubt be asked to follow the progress of IRT SystemX’s TOP project (Topology Optimization Platform).
Find out more about Paul Feliot
Thesis subject: A Bayesian approach to constrained multi-objective optimization
Diplomas: ISAE-SUPAERO Engineer and Master of Fundamentals and Applied Mathematics from the Institut Paul Sabatier
R&D project: ROM – Model Reduction and Multiphysics Optimization
