To meet the ACARE 2020 objectives, new green innovative A/C configurations were investigated. Following preparation of preliminary layout configuration for each green aircraft, optimisation needed to be performed so as to help reaching improvements in the design solution, but also assess key factors and trades and point out innovative design options. In this framework, the conjunction of mature high-fidelity CFD analysis, HPC facilities, and the most advanced optimisation algorithms was mandatory so as to maintain the computational effort within feasible limits. In this respect, the present proposal answered topic JTI-CS-2009-1-GRA-05-004, dedicated to the development of the high-performance tools that will allow the efficient aerodynamic design of new configurations.
An adequate and general answer to optimisation based on computationally expensive analysis lies in the exploitation of surrogate models in lieu of the expensive analysis results, i.e. Surrogate-Based Optimization. However, the performance of such methods was known to be largely dependent on the following key elements:
- the underlying optimization algorithm(s),
- the surrogate model(s),
- the training,
- the surrogate model(s) management scheme.
The present topic focused on the second element to be handled by Proper Orthogonal Decomposition (POD). Cenaero possessed the expertise and critical technologies to tackle all these aspects in a coherent way to take a leap forward and make POD techniques truly efficient in an industrial setting. Cenaero proposed to implement and demonstrate on a GRA representative case an online (i.e. dynamic) non-intrusive (Constrained) POD to perform the space reduction of the high-fidelity aerodynamic model, tackling the challenge of snapshot selection through innovative capture/recapture Design of Experiments and based on the most efficient generic interpolators for the reconstruction in the low-dimensional space.