axial points are not tken into account for designs?

Thommy7571

Hello,

I wanted to add this question to my last post, but there is no button to edit it.

Id I create adesign, do I not need into account that in a further optimisation design axial points might be necessary? 

If I use the real maximum values for the points |-1| and +1 snd later I want to realise a rotable CCD with a  (coded) axial point at |3.36| how will JMP handle this? Why does it not take this into account in the frame of the creation of the design?

 

Sincerely

 

Thommy7571

 

 

 

Victor_G

Hi @Thommy7571,

 

Did you check my responses to your previous post : extend Definitive Screening Designs to  Face Centered central composite designs (FCCCD) ?
Since we were discussing about factorial, axial and centre points, you might have already some answers there.

 

Concerning your questions :

Id I create adesign, do I not need into account that in a further optimisation design axial points might be necessary? 

If you want to create an optimization design (Response Surface Model design), you can use the platforms Custom Designs or Response Surface Designs.

Note that classical designs from Response Surface Design, like CCD, will use axial points with possible different distance values (depending if you want circumscribed (>1), face-centered (=1) or inscribed (<1) Central Composite design), whereas optimal RSM designs from Custom Design will only create axial points with distance = 1 (face-centered type).

If you already have a design you would like to augment to a RSM design, you can use the platform Augment Designs and see which points have been added to your existing design.

 

If I use the real maximum values for the points |-1| and +1 snd later I want to realise a rotable CCD with a  (coded) axial point at |3.36| how will JMP handle this? Why does it not take this into account in the frame of the creation of the design?

Concerning the value you would like to have for your rotable design, where does it come from ?

In a response from @Mark_Bailey to a previous post about CCD, you have the details behind the values of specific CCD designs, depending on the number of factors and/or runs : Re: Central Composite Design - Orthogonal Axial Values - JMP User Community

Does that mean your design has around 128 runs ? If less than that, I would recommend following the calculations provided by JMP in the platform, where the alpha distance value is calculated depending on your design specifications. Example here with a CCD for 4 factors :

If you still would like to change the axial value, you can edit your datatable after having created the design.
Note that on a practical side, I don't see the point of having axial points so far away from the centre of your experimental space (where the optimum is supposed to be, or at least close from it). The main objective of this optimization design is to be close from the optimum point to have a precise and useful model at the proximity of this optimum.

 

Hope this answer will help you,