I have a question that may require a bit of DOE theoretical background. I am trying to optimize a formulation. Because of the industry that I am in, it is typical to express the entire formulation as a percentage of some base ingredient. For example, if the base ingredient is always 100 and other components are expressed as proportions of the the base ingredient, the formulation {x = 0.1, y = 0.2, z = 0.3} means 100 parts base ingredient, 10 parts x, 20 parts y, and 30 parts z. Inside JMP, I am only setting x, y, and z as factors (and not the base ingredient) to lower the number of experiments I need to run, and the base ingredient is always 100 in my experiments. In addition, for cost considerations, I am setting the linear constraint Ax + By + Cz <= L, where A, B, C, and L are the costs and limits. My question is, if I ideally want to fix x+y+z to be a particular fraction (and hence x+y+z divided by the total formulation mass would always be fixed in all my experiments), would I still need to set this x+y+z linear constraint, or would the Ax + By + Cz <= L constraint suffice, because after doing the experiments, obtaining the results, and inputting them inside JMP would lead me to the optimal formulation anyway?
Thanks.
