Surroopt: A generic surrogate-based optimization code for aerodynamic and multidisciplinary design

Surrogate-based optimization (SBO) represents a type of optimization algorithm which makes use of surrogate models to approximate to the expensive objective and constraint functions, driving the adding and evaluation of new sample points towards the optimum. SBO has been shown to be very effective for engineering design problems where expensive numerical analysis such as computational fluid dynamics (CFD) is often employed. Despite the increasing popularity of SBO, it is seldom used as a generic optimization algorithm, due to its insufficient convergence properties, the difficulties associated with the so-called "curse of dimensionality", as well as the incomplete functionalities of being a generic optimization algorithm. During the past decade, a number of researchers have continuously made effort to the development of SBO, towards an efficient global optimization algorithm which can solve arbitrary optimization problems with smooth, continuous design space. This paper reviews the recent progress in development of SBO in our research group, highlighting the development of a generic optimization code, "SurroOpt", and its recent applications to aerodynamic and multidisciplinary design optimizations.