Simulated Annealing (SA), a stochastic optimization method, is applied to aerodynamic
shape design in which at least one CFD solve is required for each evaluation of the
objective function. A simple, short algorithm, SA is used as an outer loop and calls the CFD
solvers. It is found that objective functions which involve CFD in shape design, have small-
scale roughness due to discretization errors and incomplete convergence. SA is more robust
than the gradient-based methods, in the sense that the roughness creates difficulties for …
n/a