Adaptive Shape Parameterization for Aerodynamic Design

This report concerns research performed in fulfillment of a 2.5-year NASA Seedling Fund grant to develop an adaptive shape parameterization approach for aerodynamic optimization of discrete geometries. The overarching motivations for this work were the potential to radically reduce manual setup time and achieve faster and more robust design improvement, especially for problems where many design variables are required. The primary objective was to develop a fully-functional prototype system that automatically and adaptively parameterizes discrete geometries for aerodynamic shape optimization. In support of this, we also matured the discrete geometry engine that underlies the work. Various applications are presented that demonstrate broad support and uptake for the discrete geometry tools developed in this work. This report outlines our theoretical approach to adaptive parameterization, provides detailed, practical implementation notes, and contains several verification studies and design examples. These studies demonstrate substantial wall-clock time computational savings, smoother shapes, and superior final designs compared to a standard static-parameterization approach. They also demonstrate that the adaptive system can autonomously discover the parameters necessary to solve an optimization problem. As a whole, this work is an important step towards greater automation in solving the unfamiliar aerodynamic shape design problems of the future.