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; } } }

DOE on consecutive proesses

Danijel_V

Hello,

I wonder if DOE can also be applied on concsecutive processes. For example, the first step is granulation (you take the factor "amount of H2O"), the second step is spheronization (factors "speed" and "time") and the third step is compression (factor "main pressure"). Then your are looking at responses on the tablet (gained by the compression), for example response 1,2 and 3 and on response on spheronized particels (gained after spheronization).

I don't see any reason why not to use DOE on consecutive process, but of course you have many interaction effects, which are likely very big, so you have to take into account when selecting a proper design (I would chouse a design with resolution V).

What do you think, is there any problem or other things that I should be concerned about?

Thank you in advance,

Danijel

Accepted answers

louv

Hello,

Absolutely DOE can be applied to consecutive processes. If the number of experiments and testing cost are not a problem then a classical design with resolution V would certainly be a reasonable approach. You might also consider utilizing a custom d-optimal design especially if you have any constraints etc. as well. If you suspect some curvature in your response then Definitive Screening Designs are a very attractive and efficient approach. In any event JMP allows you to generate a variety of designs for your problem and also has the compare designs capabilty so you can assess the various candidate designs.

All comments

louv

Hello,

Danijel_V

Thank you.

Mark_Bailey

Since these processes are run serially, you should consider defining the granulation factors as very hard to change, the spheronization factors as hard to change, and the compression factors as easy to change (default setting) in Custom Design.

The serial nature means that you are unable to fully randomize the experiment. If you design it as if it were fully randomize it and analyze it as if it were fully randomized, but run it serially, then you will increase the type I error rate for the granulation factors and spheronization factors and increase the type II error rate for the compression factors.