nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Cannot perform experimental runs generated in DOE

Meier_G

I am a begginner with JMP, so I just started designing my experiment following the general guidelines in the DOE guide. I am working with polymers, and I am trying to figure out what are the best conditions and features of the polymers for my experiment.

After I click Make Table I obtained a set of experimental runs.

The problem is that in my case I cannot perfom all the experiments as in some cases I do not have the corresponding polymer.

How can overcome this? Is there a way to tell the software to pick between a selection of polymer that I have and not use others?

I'm only starting, thank you for your comments and suggestions.

Accepted answers

cwillden

If I understand this correctly, you will probably want to design this experiment using Factors A and B as covariates. What follows is based on an assumption that each polymer is characterized by 2 metrics, which are your Factors and A and B, correct?

1. Start by creating a data table ever polymer you have available in terms of Factors A and B (see screenshot for example). Make sure they are continuous columns, not categorical.

2. Open Custom Design. Click Add Factor > Covariate. Select your 2 columns of your factors.

3. Press Continue and specify your model. I'm guessing you'll want to do an RSM since you only have 2 factors.

4. Choose the number of runs for your experiment and press Make Design.

The resulting design will only consist of the combinations of A and B you have on hand.

Some other things to consider: If you want to force certain polymers into the design, you can highlight them in the covariates table you make in Step 1, and then check "Enforce use of selected covariate rows" right before the Number of Runs box and Make Design button. You can also choose an optimality criterion from the red-arrow menu, as well as number of random starts or design search time.

All comments

cwillden

The answer is almost certainly yes, but it’s not clear to me how you are getting your polymers in the experiment. Are they combinations of factors? Mixtures?
Can you post the run sheet you get so we can see exactly what you mean?
JMP has several ways to use only experimental runs that you have available such as linear constraints, disallowed combinations, mixture designs (special type of linear constraints), and using enumerated covariates. One or more of these should get you what you need, but we need to know more about how the polymers are being generated to give you more specific advice.

Meier_G

Thanks @cwillden.

So to make it clearer, I have two factors (A and that I consider for each polymer. I have about 40 polymers.

1. Theoriticaly these factors can be considered continuous, but in my case I have a limited number of polymer thus a limited values for these factors. When I obtain the runs, I have to be able to perform the experiments and producing other polymers (with different factors values) is not possible for me for the moment, so I decided to use categorical factors. (enclosed)

2. Second thing is that when I get the runs (run sheet enclosed) I end up with polymer with a combination of factors A and B that I don't have. For example, Run 1 would be to perform an experiment with a polymer with A=0.38 and B=68, which is not possible for me because the polymers that I have are with A=0.38, B=58 or 110, and another one with A=0.16 or 0.20, B=68.

If it's possible to use linear constraints, disallowed combination, or other methods could you advice me the best approach in that case and tell me how to do this effectively.

Many thanks !

Screenshot 2018-03-12 14.03.09.png Screenshot 2018-03-12 14.02.23.png

cwillden

1. Start by creating a data table ever polymer you have available in terms of Factors A and B (see screenshot for example). Make sure they are continuous columns, not categorical.

2. Open Custom Design. Click Add Factor > Covariate. Select your 2 columns of your factors.

3. Press Continue and specify your model. I'm guessing you'll want to do an RSM since you only have 2 factors.

4. Choose the number of runs for your experiment and press Make Design.

The resulting design will only consist of the combinations of A and B you have on hand.

Meier_G

Perfect ! thank you @cwillden for the step by step explanation and the advices !