Screening DOE: A coming up headache story

Hi all,

i'm looking for the most important factors that critically influences the EPS molding parts welding.

My issue is the huge quantity of factors to take in consideration (20 more or less), some of that uncontrollable, that should need a very expensive design ( in term of time and money). moreover i can't neglect at least the second order interaction so i can't use III resolution methods.

Any suggest?

A Quick Bump!

Can someone help??

Thank you very much!!

Try a Resolution IV design. 2-way interactions are clear of main effects, but will be aliased with other 2-way interactions, so you will have to run additional experiments such as a fold-over to isolate the true interaction.

Another possibility, if your software supports it, is an optimal design.

Thanks for the advices,

yes, i was thinking about IV resolution, in order the clear the main effects, but i have to deal with a acceptable number of runs. actually now i'm trying to reduce the number of factors by means of a discussion with the prosess cheif. moreover at present i'm looking for the adequate number of replicates to do trough a repeatability test.

Have someone some suggest about the determination of number of replicate of the experiment?

Thanks a lot to everyone

If your factors are continuous, I would use 3 - 5 center points instead of replicates. These will provide a good estimate of experimental error, and with the number of factors in your design, you have a very good likelihood of having hidden replication as you remove non-significant factors from your model.