Specifying Number of Replicate Runs in DOE Custom Design Platform

[#1]

Hello,

I was hoping someone could please help confirm a detail related to specifying the number of replicate runs within the Custom Design platform for creating a DOE. As a novice JMP user, I was originally thinking that when you enter a number for the "Number of Replicate Runs" field, you're specifying how many different runs you want to have replicates for. As an example, if you type "1" into that field, you're allowing JMP to select only 1 run that will have replicates. However, after trying this several times, this doesn't seem to be true.

I *think* "Number of Replicate Runs" may actually be instructing JMP to use this user-specified number for any number of runs within the design space (this is determined by JMP via its design algorithm). For example, if you type "1" in this field, you're allowing JMP to use 1 replicate (or "duplicates") for any number of runs depending on your design inputs (i.e. mainly how many factors you're studying and if you're interested in interaction and/or quadratic effects). If you were to type "2", you're essentially allowing JMP to make any number of triplicate runs (aka 2 replicates) that obviously fit within the desired design space and its algorithm. Could this be confirmed?

Thank you! (Using JMP 17.0.0) 

[#2]

Does the explanation in the documentation help?

[#3]

Hi Mark,

Thanks for providing the additional resource. I included an excerpt from that resource below (highlight by me) that indicates my initial thought stated in my original post was correct after all. For example, it seems to say that even if you specify "1" or "5" for number of replicate runs, JMP may actually provide more based on it's algorithm.    

Thanks for replying to my post and the many posts by others that you've kindly responded to in the past!

-wjm2024

[#4]

Hi @wjm2024,

From previous answers done on similar topic, here is a response that may help you:

Replicates vs. Replicate runs : You may face different words depending on the DoE platform used :

• For Classical Design (like Central Composite Design), you can specify "replicates", which is the number of times all experiments will be realized independently in the DoE,
• For Custom (optimal) Design, you can specify "replicate runs", which is the independent repetition of one or several experiments in the design.

If you specify 1 replicate run, then JMP will replicate the run with the highest prediction variance. If you specify several replicate runs, JMP will select one or several runs to be replicated, based on the reduction of prediction variance over your experimental space. So depending on your design balance and where the highest prediction variance is, specify 5 replicate runs could result in a lot of different situations : replication of the same run 5 times, or 2 runs replicated 2 and 3 times, or 3 runs replicated 1 and 2 times, ...

Hope this answer will help you,