Design of Experiments (DOE)

[jtk921]

Any one here know 6 sigma/DOE? I want to do a 4 factor study, but only 2 variables can be used at a time. The 4 factors are catalysts for a reaction, but ideally we only want a combination of two for the final mix. This doesn't lend itself to factorial designs because all factors have to be used at varying levels in each run. Mixture designs don't seem appropriate, either, for the same reason. Is there any other way to do this statistically other than holding one variable constant and varying the other?

 

[m777182]

Divide your entire factorial design into blocks where you include only two type of catalyst in a single block and compare results among blocks.
m777182

 

[thermcool]

I dont know how detailed statistical analysis you want with your experiment. May be DOE Software named "JMP" can help you with this. It will need some learning time before you can get some outcomes but the effort are worth putting if you want precise statistical analysis.

 

[Failed]

Hi tipers,

Try asking in a more appropriate place

Statistics in engineering

Keep tiping.

 

[dogtop]

jtk921:

Visit

http://www.stat-ease.com/dx7descr.html.

The easiest software for DOE.

Other options: Mini-tab.

The book: Practical Experiment Designs : for Engineers and Scientists, 3rd Edition (Hardcover) by William J. Diamond is very useful.