DETC2011-47288 A REVIEW AND EVALUATION OF EXISTING ADAPTIVE SAMPLING CRITERIA AND METHODS FOR THE CREATION OF NURBS-BASED METAMODELS

Abstract Metamodels are widely used during the design process in place of expensive simulation models when computationally efficiency is imperative. Non-Uniform Rational B-Splines (NURBs) based metamodels have proven to be an effective modeling technique. Although metamodels can reduce the computation expense of design analysis and optimization, the sampling methods used to create the metamodels also play an imperative role in the computational efficiency and overall model quality. The research presented in this paper reviews various adaptive sampling criteria and methods and evaluates the different methods to determine the best sampling technique for the development of NURBs-based metamodels