Robust Optimization of Thermal Aspects of Friction Stir Welding Using Manifold Mapping Techniques

The aim of this paper is to optimize a friction stir welding process taking robustness into account. The optimization problems are formulated with the goal of obtaining desired mean responses while reducing the variance of the response. We restrict ourselves to a thermal model of the process and use the manifold mapping technique to solve the optimization problems using a fast analytical coarse and an expensive accurate fine model. The statistics of the response are calculated using Taylor expansions and are compared to Monte Carlo simulations. The results show that the use of manifold mapping reduces the number of fine model evaluations required and that the Taylor expansion approach gives good results when compared to Monte Carlo simulations.