Hard to change factors in DOE (Split Plot Designs) - Injection Molded Parts

[davis007]

Sample size

THis defect seems to take some time before it starts to appear. If we start up a machine that has been cleaned and polished we see no defects for ~4 hours. Then somtimes we begin to see the defects. And they tend to get worse over the first 24 hours until we reach some kind of steady state.

So my plan for each run is:

1. Clean machine and start up.
2. After four hours take a 1000 part sample ~45 min run time.
3. After 8, 12, 16, 20 and 24 hours take a 1000 part sample.
4. Stop machine and set up for next test.

We already 100% sort these parts because of this defect so I will get 6 1000 part tests at each condition. I can ether treat each as a seperate response, or lump all 6000 smaples together to "hopefully" increase power.

The last issue is that a batch of material is only enogh to run for 4 days. So I will need to block the tests over four batches.

THanks again to everyone for the help I will keep you postedon how this turns out. Should be done with the testing in about 30 days.

 

[Miner]

Another response variable that you can track and analyze is the time before the first defect appears.

 

[supreecha]

DOE: Handling Hard-to-Change Factors with Split-Plot Designs in Minitab
Consider the case of a Low Preessure Die Casting company that wants to use design of experiments (DOE) to simultaneously study three factors to optimize the formula for its Al wheel. Two of these factors are easy to change: the Die Coat Thickness and the Cycle Time of Die Casting in the mix. But another factor, Mold & Al. temperature, aren’t so easy to change. Changing this factor is difficult because the enormous Mold & Al Temp. the company uses take several hours to attain a stable temperature. An experiment that includes a hard-to-change factor, such as the AL Wheel’s Mold & Al. temperature, calls for a special type of DOE called a split-plot design.
Creating a split-plot experiment in Minitab 16 is easy—just choose the 2-level split-plot option under Stat > DOE > Factorial > Create Factorial Design to create a design with up to 3 hard-to-change factors.

 

[Statistical Steven]

I tend to use blocking if Split Plot is not feasible.

On the analysis side of things, use the Binary Logistic regression with a binary response like defect.

 

[supreecha]

DOE: Handling Hard-to-Change Factors with Split-Plot Designs in Minitab
Split-plot designs are experimental designs that include at least one hard-to-change factor that is difficult to completely randomize due to time or cost constraints. By making the creation of split-plot experiment designs simple, Minitab 16 makes the benefits of this powerful statistical technique accessible to everyone.
Consider the case of a baking company that wants to use design of experiments (DOE) to simultaneously study three factors to optimize the formula for its chocolate cake. Two of these factors are easy to change: the amount of flour and the amount of sugar in the mix. But another factor, oven temperature, isn?t so easy to change. Changing this factor is difficult because the enormous ovens the company uses take several hours to attain a stable temperature. An experiment that includes a hard-to-change factor, such as the bakery?s oven temperature, calls for a special type of DOE called a split-plot design.
Creating a split-plot experiment in Minitab 16 is easy?just choose the 2-level split-plot option under Stat > DOE > Factorial > Create Factorial Design to create a design with up to 3 hard-to-change factors.

 

[supreecha]

Hard to change factors in DOE (Split Plot Designs) :Tensile Test