Space-filling designs for mixtures

Abstract Uniform experimental designs (Space Filling Designs) are now widely used with independent variables, particularly in numerical simulation. A lot of construction methods have been developed to uniformly cover the domain of variables, but they remain rarely applied for the study of mixtures. In this article, we propose methods to build space-filling designs for mixtures that can be used to modelize complex phenomena in formulation. Various algorithms like Kennard and Stone, WSP, Strauss and Dmax algorithms, to construct these designs are detailed and compared with respect to uniformity criteria.