DoE where choice of factor changes with the levels of another factor

[#1]

Hello - This was briefly answered in another question, but it was suggested that I put it as another question.

I am also using two different heat treatments (microwave and thermal), which normally would be fine as a categorical variable, but there is also a switch of a second factor associated with the different levels of the "heat treatment type" factor: Microwave makes more sense to control the "power input" as an additional factor, while thermal makes more sense to use the "temperature" set point as a factor. My two options as I see it are to do a few different DoE's, one for microwave, the other for thermal; or I could bite the bullet and use the microwave temperature set point, even though the temperature will not be constant during the run, due to exothermic/endothermic reactions while the solid material is nuked. I assume JMP doesn't handle this kind of situation and wondered if anyone had suggestions.

[#2]

I hope I understand your situation...I would suggest that "power input" is nested in "heat treatment type".  And yes, JMP can handle nested factors.  You will lose the interaction between the factors, bu should learn about main effects.

[#3]

Hi @statman Thanks for the feedback! How would you tell JMP to nest the factors? I'm not seeing a button for that on the design window.

[#4]

I am not sure that having power input as a factor is the right thing here. Power input is not something that you can set for thermal heating, is it? 

If the temperature set point is the thing that you can control (even if that is not the actual temperature), then it makes sense to have that as the factor.

I would imagine that the set point and actual temperature will be closer in the case of thermal, but they would still not be exactly the same.

Ultimately, would it be useful to have a model of how your response(s) change as a function of set temperature?

You should make sure to add the interaction of Heating_Type*Set_Temperature as you could expect the effect of set temperature to be different for the 2 heating types.

[#5]

How do you think you change the temperature?  Somehow you have to change the energy input.  How, of course, depends on whether the thermal heat treat is electronic or gas driven, but in either case I think it makes sense from an engineering perspective.

[#6]

Nesting is not so much needed for deigning the experiment.  You run the experiment with the appropriate levels.  Two for heat treat type (A: categorical) and then for Power Level (B: continuous), use 2 different set points of power for Microwave and two different temperature settings for Thermal.  The key to nesting is in the model:  Y = A + B[A]