Choose Language Hide Translation Bar
Highlighted
tenzin
Level I

DOE - Fractional Factorial

Hi, I am running a DOE for coating thickness and uniformity. I have 5 factors and am running a fractional design to reduce the number of trials to 16. My responses are all the same metric, which is thickness. But they are in different fixed locations (Z1,Z2..etc). In the end of this test i want to determine which factors have the largest effect on uniformity, determine any interactions and most importantly determine which factors give me the best uniformity (smallest variation between all responses). So my question is: does this model make sense for what i am looking for? 

 

- Edit: Or should i create only 1 response as the average thickness of 5 locations and another response as the standard deviation and set that to minimize? 

 

 

DOE.PNG

4 REPLIES 4
Highlighted
txnelson
Super User

Re: DOE - Fractional Factorial

This works if you are treating each location as a separate DOE. If you want to determine any differences between the locations, and any interactions location has with your other factors, you will need to make location one of the factors.
Jim
Highlighted
tenzin
Level I

Re: DOE - Fractional Factorial

Ok i figured so, but i wanted to get around that because there are 5 locations. Which would mean 5 levels and that would make it much more complex. Also, i am coating all 5 locations in the same run. If i make that a factor i wouldn't be able to do that. Any suggestions? 

Highlighted
txnelson
Super User

Re: DOE - Fractional Factorial

If I am interpreting your model correctly, by choosing to coat all 5 locations at the same time would confound the location factor with time. If you have engineering knowledge that time is not a factor, then I would feel comfortable with coating all locations at the same time.
Jim
Highlighted
tenzin
Level I

Re: DOE - Fractional Factorial

These are performed in a reactor, so these locations are not too far away. They have to be coated in the same run to determine the uniformity. I am analyzing real-time coating using interference patterns for one of the locations. Appreciate your help so far.
Article Labels

    There are no labels assigned to this post.