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; } } }

Optimising Concentration of Two Factors

miggledee

Hi all,

I am trying to develop a screening platform to find the best mixture of two factors. I have chemical A and B and want to run experiments with a variety of concentrations of just A, just B, and then a variety of experiments with A and B mixed together in different ratios. I have previously only performed two-factor DoE and am confused about where to start with this experiment. I initially through I would be running a screening experiment but now I think I might need to do some kind of optimisation.

Does anyone have any advice about what types of experiment I could look at to get an example? Is there a specific name for this scenario? I have been searching for examples of this DoE and keep just getting descriptions of two-level DoE rather than mixtures.

Thank you very much for your help!

Accepted answers

Ben_BarrIngh

Sounds like you want to look into a mixture design DOE, you can easily do a mixture design that measure the ratios of A/B and their effect as just A and B.

Here's an overview of the platform: https://www.jmp.com/support/help/en/18.0/index.shtml#page/jmp/mixture-designs.shtml

And a great video guide on the tool:https://community.jmp.com/t5/Mastering-JMP/Designing-Mixture-Experiments-Part-1/ta-p/546086

All comments

Ben_BarrIngh

Sounds like you want to look into a mixture design DOE, you can easily do a mixture design that measure the ratios of A/B and their effect as just A and B.

Here's an overview of the platform: https://www.jmp.com/support/help/en/18.0/index.shtml#page/jmp/mixture-designs.shtml

And a great video guide on the tool:https://community.jmp.com/t5/Mastering-JMP/Designing-Mixture-Experiments-Part-1/ta-p/546086

miggledee

This was very helpful! Thank you very much!

Victor_G

Hi @miggledee,

I have a different opinion on your topic.
If you have only two factors and you are interested in the impact of the different ratios on the response, then you only have one factor to consider in your study, which is the factor ratio A/B, as A and B are not independant.

You can then define several levels depending on the precision needed, for example 0, 0.25, 0.5, 0.75 and 1, and add replicates to have robust analysis and interpretation of your results. Make sure to randomize the order in which you realize the tests (even if you only have one factor), to make sure the ratio levels may not be correlated to external nuisance factor (temperature, humidity, day of the week, ...).

Here is a possible design you can create with Custom Design platform :

If you have more than two factors, then the documentation provided by @Ben_BarrIngh about mixture designs can help you.

Hope this answer may help you,