Disallowed combinations with If statements

[#1]

Hi,

I am challenged by the following scenario: I have two factors (Volume and Solution), for which there are some specific combinations requirements:

If Volume = 100: Solution = 1002MH2SO4
If Volume = 50: Solution = 2002MH2SO4 or 2005NHCL
If Volume = 75: Solution = all options

(see the attached .jmp table)

If I want to play with Disallowed Combinations Filter, I can't really add Volume more than once. And I can't define constraints with If statements (perhaps I don't know the syntax). Do you have any ideas how to put those constraints into disallowed/allowed combinations?

I also tried with a very large design, then manually removed the disallowed combinations, and used this as a Covariate candidate set, but this does not work for me since there are also other factors which are hard to change, and with Covariate candidate set you can't separately choose which factors are hard to change.

[#2]

Sorry, I just solved this using the following link:

Solved: Disallowed Combinations not working - JMP User Community

[#3]

Unfortunately it does not work when other factors are included (with some hard to change factors), and also those factors have some linear constraints...

[#4]

Hi @aatw,

If you're using Disallowed Combinations filter in the Custom Design platform, you can define appropriate constraints that match your requirements :

Or with JSL : 

Volume >= 90 & (Solution == "1005NHCL" | Solution == "2002MH2SO4" | Solution == "2005NHCL") | Volume <= 60 & (Solution == "1002MH2SO4" | Solution == "1005NHCL");

And you'll obtain such design respecting your constraints :

Regarding your comments, what are your other problems/requirements ?
Maybe this post could be helpful DOE with two type of constraints, as @Laura_Lancaster show how to deal with disallowed combinations and linear constraints by setting the linear constraint as a disallowed combination. I also showed how to use the Candidate set approach to create a candidate runs dataset on which to generate your DoE. Using covariates enable to have factors "Hard to change" so using one or mixing the two approaches could perhaps help you create your design ?

Hope this answer may help you,