Trends in Intelligent System Technologies for Airfoil Design: - Computer Algebra Techniques for Gradient Computation through the Continuous Adjoint Approach - Use of Approximate Fitness Evaluators in Evolutionary Optimization; Conference paper

摘要

Aerodynamic shape design is a challenging application for modem computational fluid dynamic. This is, in-deed, a very productive field, both in term of scientific publications and developed applications. Furthermore it is possible to find excellent analysis/design software released in the public domain. However, industry, in particular automotive and aerospace, but also emerging fields like wind and sea extracted energy, continuously needs increased product performance and competitiveness. This requires increasingly refined, advanced and complex computational models capable of describing flow behavior with higher and higher fidelity. Thus the design tools available to scientist and engineers are challenged by the growing complexity of simulation models. The increasing price to performance ratio of nowadays computers is a strong allied of designers in this though race, but often it is not enough. Objectives of this lecture is to explore and analyze some of the technologies that, together with the increasingly available computational power, may help to exploit the tremendous potential of high fidelity flow field models to aerodynamic shape design challenges. For the sake of simplicity we will concentrate on aerofoil design. Two aspects will be considered: the computer algebra assisted development of code related to gradient computation through adjoint techniques and the use of fitness approximation techniques to improve the performance of search procedures. The concern of this first lecture will be more methodological, while some applications will be illustrated in the next one. The outline of the first part of this lecture is as follows: some short remarks on adjoint problems will introduce the framework in which the Computer Algebra techniques will be mainly used here.