An automated surrogate model generation framework for rapid aeroelastic structural sizing optimizations in conceptual aircraft design

The optimisation of aircraft performance is a crucial aspect of improving fuel efficiency and thus reducing the environmental impact of air transport. A potential improvement for the early design process of an aircraft in terms of time efficiency and accuracy, is the involvement of more accurate, computationally based data instead of statistically based handbook methods. To overcome huge calculation times for higher fidelity data acquisition, surrogate models, which are mostly regression-based models fed with results from higher order models, are a promising solution. In this paper, a process for the automated generation of surrogate models for aeroelastic structural sizing optimizations, using an open-source Python-library is investigated. After methodology development and implementation work, different surrogate algorithms are applied to a test case. The tested surrogate models for aeroelastic structural sizing executes within seconds, compared to hours for the original calculation method. This enormous performance gain enables the calculation of derivatives for an optimization-algorithm and thereby further increases the performance of a surrogate-based shape optimization.