DOE setup for a complex system with an adjustable number of components

Question

Hello everyone, happy Friday. I would love your input on this problem posed to me by a customer.

The goal is to optimize a shredding machine. Material goes through a number of cutters in sequence. The main factors are:

-Cutter Type: 2 level categorical (Knife and Needle)

-Cutter Speed: continuous

-Cutter Depth: continuous

Here's the complication: The machine can have between 1 and 8 cutters in sequence, and each individual cutter can vary on all the factors. So for instance, one setup might have 3 cutters: Knife, then knife, then needle. Another setup might have 4 cutters, say needle then knife then needle then needle. There is a strong suspicion that it is not only the number of cutters that will be important for achieving the desired outcome, but also the specific sequence of operations. Knife before needle, or needle before knife, etc.

I did run into this paper from US Discovery 2019: The Design and Analysis of Experiments With “Order” Factors. It has some useful ideas, but is not directly applicable. The paper deals with order-of-addition effects for mixture components.

Has anybody run into this kind of situation before?

Thanks in advance,

Jordan

Bill_Worley · Answer

Hey @Jordan_Hiller ,

Maybe a dumb question, but I'll ask it anyway.

When you say "optimize the shredding machine" how is that optimization being measured?  Is it the size of the shredded bits that are the end result of the shredding process?  Or, is the speed of the overall shredding process measured and the size and scope of the shredded bits really do not matter?  Is it a combination of both speed and size which will likely require some sort of tradeoff?

You mention below the influence of setup which could include potential factors like cutter order and/or number of cutters.

If you add those two factors and a dummy factor you could run a DSD and although I have no experience with them this might be an opportunity to try a GO-SSD.

Best,

Bill