Airfoil design and optimization using a parallel genetic algorithm

Abstract

In recent years aircraft liners have been blamed for incurring a substantial share of global warming. A report by the Intergovernmental Panel on Climate Change (IPCC), a group of experts affiliated with the UN warned that the share of global warming caused by air traffic could increase from 3.5% in 1992 to 17% in 2050. Airlines can therefore make their contribution to reducing global warming by buying more efficient aircraft and operating them efficiently. Aerodynamic shape design and optimization can contribute substantially in this regard (e.g. through drag minimization). Moreover, considering the number of miles travelled by major airlines each year, a small increase in aerodynamic efficiency could prove to be highly beneficial in terms of finances. The aerodynamic design of the two-dimensional cross-sections of a wing, known as airfoils, is a huge challenge for aerodynamic designers. Due to the number of parameters involved in wing and airfoil design it is almost impossible to find an optimal solution using the traditional "trial and error" method. Nowadays computational fluid dynamics (CFD) has developed to such extent that it can accurately tackle a wide range of flow problems. Over the time a variety of different optimization techniques have been employed in this field. Although it is difficult to state the overall superiority of one optimization technique over another, one of the most popular and successfully applied techniques is the Genetic Algorithm (GA). Apart their ability to contend with multiple objectives, their robustness, generality, efficiency and natural openness to parallelization, make the GA a popular choice. In this study, an asynchronous "island" Parallel Genetic Algorithm (PGA) is coupled with a CFD code called XFOIL. The main aim is to create a distributed application which will aid aerodynamic designers to design and optimize airfoils that comply with the defined objectives (e.g. drag-to-lift ratio minimization, lift-constrained drag minimization, constant lift drag minimization, etc.) and satisfy the defined constraints (e.g. maximum moment, minimum lift, thickness constraints etc.) and flow conditions.