A Simplex lattice is usually good for these sorts of mixture screening studies. If you have three or less components then a Simplex Centroid is possibly a good choice.
To answer your question better, some more info would be helpful. How many components are in the formulation? Do all of the components (As a percentage weight of the formulation) add up to 1? Are there any constraints (Max or minimum value of amount) on any of the components?
I have found both the The National Institute of Standards and Technology (NIST) and Penn State hosts of excellent Design of Experiments information.
Not sure what software you have, but you may want to consider looking into using Minitab and R. R has BIG plus in that it's FREE. I have attached a paper that describes using the package 'mixexp' to create and evaluate constrained mixture designs in R. See below for package info and well documented user manual.