I want to create a DOE that will include replicates of the same level and that the model has interactions as well as quadratic for 4 factors out of 5. The reason why is that each run, you can obtain 3 independent results. I just want to know if with this approach the number of run will be n/3 becuase I will obtain 3 results in 1 run.
Are you just wanting to replicate the whole design 2 additional times? From Custom Design, the easiest thing to do is just concatenate the outputted table to itself 2 times (Tables > Concatenate).
Are they really independent if they come from the same run? That suggests 0 run-to-run variability at the same treatment combinations. I would never assume that unless that has been supported with prior data. If they really are independent, the concatenated table is all you need. Otherwise, if you want to treat as repeat measures, you need to include a random block for each set of 3. An easy way to do this would be to give each run a unique ID before concatenating, then sort the concatenated runsheet by the ID. When you model the results, add the block ID as a random block to the model.
The are somewhat independent in a sence that the solution added to the each run come from 3 independent preparation. Concatenate does work however it does not minimize the number of runs event though I created more results.
You're going to have give more explanation. Can you upload an example of what you are trying to accomplish? There just isn't enough detail here for me to be very helpful to you.
x1, X2, x3
(each run will provide 3 independent results)
I will like to use RSM.
I hope this helps
Are x1, x2, and x3 supposed to be your 3 indepedent results or do they measure a different type of response?
What is an acceptable number of runs for you? For 4 factors, you would have an intercept, 4 main effects, 6 2-factor interactions, and 4 quadratic terms, totalling 15 model degrees of freedom. You would generally need about 24 runs to get decent power with a 1:1 signal to noise ratio. If I understand you correctly, you measure 3 variables each (assuming 24 runs, 72 measurements per response). Is that not acceptable?