A Data-driven Scalable MDO Problem To Compare MDO Formulations

Published in AIAA AVIATION 2021 FORUM, 2021

Optimizing a multidisciplinary system may not be straighforward because of the dimension of both coupling structure and design space, as well as the shape of both objective and constraint functions. Different formulations of the multidisciplinary design optimization (MDO) problem exist in order to deal with this issue. Nevertheless, selecting the most appropriate MDO formulation with respect to the problem dimension by running a set of such formulations is rarely possible because of the prohibitive computation time involved. To face this, a data-driven scalable methodology has recently been introduced in the litterature. In the MADELEINE project, we apply it for the very first time on two large-scale aerostructure wing design problem. For the Airbus XRF-1 use case, we compare a sequential and a all-at-once ad-hoc formulations while for the Dassault GBJ use case, we compare two formulations from the MDO litterature. For that, we use GEMSEO, a recent generic Python engine for MDO scenarios. First results are promising and areas for improvement are emerging.

Link to paper: https://arc.aiaa.org/doi/10.2514/6.2021-3053

Recommended citation:
De Lozzo, M., Gallard, F., Gazaix, A., Abu-Zurayk, M., Roge, G., Fougeron, G., & Ilic, C. (2021). A data-driven scalable MDO problem to compare MDO formulations. In AIAA AVIATION 2021 FORUM (p. 3053).

Recommended Bibtex entry:
@inproceedings{deLozzo2021data, title={A data-driven scalable MDO problem to compare MDO formulations}, author={De Lozzo, Matthias and Gallard, Francois and Gazaix, Anne and Abu-Zurayk, Mohammad and Roge, Gilbert and Fougeron, Gabriel and Ilic, Caslav}, booktitle={AIAA AVIATION 2021 FORUM}, pages={3053}, year={2021} }