Experimenters often wonder which design they should choose. Below is a summary of each mixture design.
Mixture designs are used when the response changes as a function of the relative proportions of the components. All components must be entered in the same units of measure and each run must sum to the same total.
“Simplex” designs require the difference between low and high to be the same for all components. If the components do not meet this requirement then a more suitable custom design should be used instead. Simplex designs include an automatic augment by default.
Lattice designs are recommended when the simplex requirements are met. When augmented, (default) these design provide good coverage of the formulation space providing excellent precision for a low number of runs.
Centroid designs are more space filling than lattice designs, but lack the ability to model higher order blending effects.
Optimal or Screening designs are the most common designs used for new formulation work.
Screening designs are constructed to estimate linear effects (gradients) to determine which ingredients will be included in future experiments. Component ranges determine if simplex or non-simplex design interface is provided.
Optimal (custom) designs work with unequal component ranges, multi-component constraints, blocking, custom models, and specific augments. Run settings are chosen algorithmically to provide the best estimates for the chosen model.