Set constraint of two factors in Custom DOE

[#1]

I have 5 continuous factors. The sum of the values of 2 factors always has to be 100. Should I only enter 1 of these 2 factors in the design and constrain it in the range of 0-100? Or is there a way to add both factors and set a constraint for their sum to be 100 in Custom DOE?

[#2]

First welcome to the community.  I can't tell by your description, but are trying to design a mixture or formulationIf so, you might want to use the mixture design platform.  DOE>Classical>Mixture Design.  You can constrain any of the 5 factors.

https://www.jmp.com/support/help/en/17.0/?os=mac&source=application#page/jmp/mixture-designs.shtml#

[#3]

Not all my 5 factors are in one formulation/mixture. Only those 2 factors (A & B) are and the values of these 2 factors need always to be added up to 100. If I set the levels of A factor to be 40 and 60, my B has to be 60 when A=40 and be 40 when B=60. The other 3 factors do not have such constrain. Can this be done in Custom Design?

[#4]

Yes, when you add the 2 mixture factors (you get the options after you enter the number of factors Add N Factors and pull down the button labeled Add Factor), designate them (A & B) as mixture factors.  The others can be added as continuous.  You still need to decide what model you want to investigate (e.g., main effects, interactions, mixture terms, other higher order terms).

Obviously, you will not be able to run an orthogonal design with the mixture components.  Analysis of mixture factors is typically done with  contour plots.  If you start at the link I posted, you should be able to create the design you want with the appropriate levels.

[#5]

Hi @TC3303,

It's difficult to help you without having more information and context about all the factors in your study.
One other suggestion possible (but it may depend on what are your other factors) would be to add a continuous numerical factor "ratio A/B" in the design (instead of one factor for each components A and B), with the other continuous factors, and always keep the total amount of A+B at 100% in the experiments.

This way you'll have 4 factors in total in your design : 3 other continuous factors + 1 continuous factor "ratio A/B".
By specifying quadratic term for this ratio factor, you'll need three levels for this factor, and this may be easy to handle (something like 70/30, 50/50 and 30/70 for example).

This approach would be perhaps interesting to consider in an explorative approach, to better understand the significance of factors (and impact for example of ratio A/B). 

I hope this other suggestion may help you,

[#6]

I agree with @Victor_G . If you only have 2 ingredient amount factors that must sum to a fixed total, then they are really just one factor: the ratio of A to B.

So it is really quite simple because you can only control the proportion of A; the proportion of B is completely determined from that. (Alternatively you could think about proportion of B being the factor that you are controlling and proportion A is determined from that - it makes no difference).

If you have 3 or more ingredient amount factors that must some to a fixed total then you would need to define these as mixture factors in Custom Design.

[#7]

Just be aware that JMP expects the levels of mixture components to be proportions, not percentages. So enter 0.4 instead of 40, and you should be good. JMP will make sure that the proportions add to 1.