cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Choose Language Hide Translation Bar
bbenny7
Level II

Run Order with JMP DOE

Hi,

 

I have to set-up a DOE (Response Surface) where I have a machine as blocking factor.

 

Since it is very time consuming to change the temperature of the machine, I want to have the run order of the DOE with temperature going from low to high.

 

Right now, when I design the DOE, JMP gives me high, low and then intermediate.

1 ACCEPTED SOLUTION

Accepted Solutions

Re: Run Order with JMP DOE

I understand how inconvenient it is to change temperature in this case. At the same time, obtaining the correct estimates from the analysis is necessary. The blocking factor Machine helps account for an external noise source in the response. The imposed order you want violates the assumption of independence in the regression analysis. You can design and analyze your experiment correctly if you make the continuous factor temperature hard to change. Click Easy in the Changes column of the factor list for Temperature and select Hard. This designation tells JMP that the experimental error for setting a temperature was sampled once and shared by all the runs in that 'whole plot.'

You might still have to endure some inconvenience, but your results will be usable.

View solution in original post

1 REPLY 1

Re: Run Order with JMP DOE

I understand how inconvenient it is to change temperature in this case. At the same time, obtaining the correct estimates from the analysis is necessary. The blocking factor Machine helps account for an external noise source in the response. The imposed order you want violates the assumption of independence in the regression analysis. You can design and analyze your experiment correctly if you make the continuous factor temperature hard to change. Click Easy in the Changes column of the factor list for Temperature and select Hard. This designation tells JMP that the experimental error for setting a temperature was sampled once and shared by all the runs in that 'whole plot.'

You might still have to endure some inconvenience, but your results will be usable.