BookmarkSubscribeSubscribe to RSS Feed

Blocking in DoE

Highlighted
Roman-V

Community Trekker

Joined:

Jul 13, 2017

Hello,

 

I designed my experiments using custom design platform which gave me 15 runs in total (minimum was used).

There are 4 factors:

2 level categorical

3 level categorical (two of them)

2 level numerical countinuous

 

The experiment itself consist of mixing matrix components step (ratio is fixed, quality of 2 of 3 matrix components is controlled by 2 factors) followed by addition of other additives (another 2 factors).

I realized that random combination of grades of matrix components results in only 6 unique matrices.

Can I mix matrices in one go and use each of them for every individual run? Or will it irrevocably spoil a whole randomness of my design?

If I will do experiments during 2 or more days, should I also use block for this split?

 

1 REPLY
Byron_JMP

Staff

Joined:

Apr 26, 2012

there is something fishy with the design you're describing.

 

I think you said you are doing this

	Add Factor( Categorical, {"L1", "L2"}, "X1", 0 ),
	Add Factor( Categorical, {"L1", "L2", "L3"}, "X2", 0 ),
	Add Factor( Categorical, {"L1", "L2", "L3"}, "X3", 0 ),
	Add Factor( Continuous, -1, 1, "X4", 0 )

and I think the default runs would be something higher than 15

and also if you're days aren't identical, it might be smart to block by day. (i.e. 9 runs/day)

JMP Systems Engineer, Pharm and BioPharm Sciences