gemseo-multi-fidelity

Overview

A preliminary version of the GEMSEO® plugin for Multi-fidelity MDO.

This plugin originates from works performed at Airbus, in the frame of Romain Olivanti's CIFRE PhD funded by Airbus:

• Olivanti, R., Multi-fidelity aerodynamic shape optimization considering flexible adjoint at multiple operating conditions. PhD thesis manuscript. ISAE, Toulouse. 2021.
• Olivanti, R., Gallard, F., Brézillon, J. and Gourdain, N., Comparison of generic multi-fidelity approaches for bound-constrained nonlinear optimization applied to adjoint-based CFD applications. AIAA Aviation 2019 Forum. 2019.

It was later improved and ported by IRT Saint Exupéry, in the frame of the MDA/MDO project funded by Investments for the Future Programme (French acronym: PIA) and the European Union's Clean Aviation project ODE4HERA.

The goal of Multi-fidelity is to speed up optimization processes using degraded levels of fidelity (low fidelity), in addition to high-fidelity levels that are generally more accurate but as well more costly.

The plugin eases the creation of a main Multi-fidelity scenario, relying on several GEMSEO MDO scenarios based on different levels of fidelity.

This main scenario is executed as a usual GEMSEO scenario, however, during the process the driver is able to switch from a level of fidelity to another, based on criteria defined by the chosen Multi-fidelity formulation.

The level of fidelity associated with a scenario is defined from a variety of parameters (related to disciplines, MDO formulations, ...) that can be consulted in the following list of opportunities.

The Multi-fidelity scenario can be created and executed with simple commands, specifying pre-defined sub-scenarios and the formulation settings:

mf_scenario = MFMDOScenario(scenarios, formulation_name="Refinement", name="mf_process")
mf_scenario.execute(algo_settings_model=MFRefine_Settings(n_levels=2))

Installation

Install the latest version with pip install gemseo-multi-fidelity.

See pip for more information.

Bugs and questions

Please use the gitlab issue tracker to submit bugs or questions.

You can also contact the authors using our Discourse channel.

Contributing

See the contributing section of GEMSEO.

Contributors

• Romain Olivanti (Airbus)
• François Gallard (IRT Saint Exupéry)
• Yann David (IRT Saint Exupéry)
• Sylvain Béchet (IRT Saint Exupéry)
• Jean-Christophe Giret (IRT Saint Exupéry)
• Antoine Dechaume (IRT Saint Exupéry)

The funding comes from ANR FR2030 Investments for the Future Programme, in the frame of the MDA/MDO project, and European Union Clean Aviation project ODE4HERA, co-funded by the European Union under GA no. 101140510

The authors wish to acknowledge the PIA framework (SGPI, ANR) and the industrial members of the project Airbus, Airbus Defence and Space, Capgemini, Cenaero, CERFACS, Expleo, Liebherr, for their support, financial funding and own knowledge.