Publications

Aerodynamic shape optimization has become a well-established process, with designers routinely performing wing and full aircraft optimizations with hundreds of geometric design variables. However, with increased geometric design freedom comes increased optimization difficulty. These optimizations tend to converge very slowly, often taking many hundreds of design iterations. In addition, designers have to manually obtain suitable design variable scaling through trial and error in order to have a well-behaved optimization problem, which is a tedious and time-consuming task. In this work, we propose a sensitivity-based geometric parameterization approach that, while keeping the same optimization problem, maps the design space onto one which is better suited for gradient-based optimization. At the same time, we can automatically determine appropriate design variable scaling such that the new optimization problem can be solved more rapidly. We demonstrate the approach on aerodynamic optimizations, and show improved convergence behaviour compared to the traditional approach.