To add a bit to @Mark_Bailey 's advice, sometimes people use more classic DOE families of designs because, well that's what they were taught in their DOE 1 class. Me, case in point. Back in the early 1980's when I took my DOE 1 class, methods such as the Box Behnken and Central Composite designs were kind of the defacto RSM design types. And as Mark states, they each have their own unique properties and requirements. As statistical methods research AND more powerful desktop computing hardware/software became more prevalent, additional methods were introduced into the mainstream through first the literature and software, and now training and classroom type experiences. Optimal DOE methods is one very large case in point example. JMP houses all optimal DOE methods in the Custom Design platform. Often times using these methods can be much more efficient than the classic methods wrt to solving the practical problems at hand.
So what you may want to consider, if you haven't done so already, is create several designs in JMP (Box Behnken, one or more optimal designs, etc.) and then use JMP's Evaluate Design platform to compare and contrast the performance characteristics of each candidate design and then pick one. My choice is usually guided by the same basic principle I was introduced to back in my first DOE 1 class..."Pick the design that provides the required information for the least expenditure of resources."
Good luck! Hope this helps?
