topic Re: Augmenting a Plackett Burman design to a Full Factorial Design in Discussions

Augmenting a Plackett Burman design to a Full Factorial Design

SaraA — Thu, 11 Apr 2024 16:04:00 GMT

After performing a Plackett Burman design with nine continuous factors at two levels with 3 replicates for each run, I would like to perform a second iteration by augmenting my initial PB design to a full factorial design. When using the augment design platform, JMP asks how many runs are desired (including the runs that were already performed in the first DOE). However, I cannot find a separate box to include the number of replicates for these additional runs. Moreover, when I add the additional runs and make a new table, I see that some of the new runs are replicates (sometimes I find 5 runs that are identical). Can someone explain to me what is happening here?

Thank you,

Sara

Re: Augmenting a Plackett Burman design to a Full Factorial Design

Mark_Bailey — Thu, 11 Apr 2024 16:31:00 GMT

You are correct. You are unable to request replicates of the new runs. On the other hand, you can generate the design with new, augmented runs, and then use simple copy and paste in the data table to make the replicates you want.

Re: Augmenting a Plackett Burman design to a Full Factorial Design

SaraA — Thu, 11 Apr 2024 18:19:36 GMT

But how come I do find identical runs when augmenting the design even though I did not request replicates of the new runs?

Re: Augmenting a Plackett Burman design to a Full Factorial Design

Mark_Bailey — Thu, 11 Apr 2024 19:41:59 GMT

You started with a P-B design, which is generated from Hadamard matrices. Augmentation uses a different method, the same as Custom Design. It uses many random starts and searches for the optimal set of runs based on a criterion. The optimal design might include replicates instead of all unique runs.