Efficient Gradient-Based Shape Optimization Methodology Using InviscidlViscous CFD

In developing the automated methodologies for simulation-based optimal shape designs, their accuracy, efficiency andpracticality are the defining factors to their success. To thatend, four recent improvements to the building blocks of such amethodology, intended for more practical designoptimization, have been reported. First, in addition to apolynomial-based parameterization, a partial differentialequation (PDE) based parameterization was shown to be apractical tool for a number of reasons. Second, an alternativehas been incorporated to one of the tedious phases ofdeveloping such a methodology, namely, the automaticdifferentiation of the computer code for the flow analysis inorder to generate the sensitivities. Third, by extending themethodology for the thin-layer Navier-Stokes (TLNS) basedflow simulations, the more accurate flow physics was madeavailable. However, the computer storage requirement for ashape optimization of a practical configuration with the high-fidelity simulations (TLNS and dense-grid based simulations),required substantial computational resources. Therefore, thefinal improvement reported herein responded to this point byincluding the alternating-direction-implicit (ADI) basedsystem solver as an alternative to the preconditionedbiconjugate gradient (PbCG) and other direct solvers.