1

I would like to create an optimal experimatal design (DoE) for a machine which will be used to produce 20 results in batch mode to find the most important factors which do have an impact on certain quality attributes.

Usually, I would set blocking factors for those runs because they take several hours per run and span multiple days. Unfortunately, the results might deteriorate from the 6th run onwards due to wear within the maching. Unfortunatly, it is not possible to change the parts which do wear down. Therefore, the runs are not independend from each other because the result of the next run will depend on the result from the previous run.

I would like to find operational parameters that ensure consistent product quality for all of the runs. Ideally, the runs will be different only by the intercept I suppose and could be covered by the blocking factors. Otherwise, probably by the interaction of the blocking factors with process parameters. However, I am not sure if the OLS results would be correct if the runs are correlated with each other.

Are there any options to consider such an autocorrelation in the design?

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Moritz
  • 131
  • 4

1 Answers1

0

How many runs can you do before you have to change the parts that run down? Maybe define those runs as a block, and use a split-plot design, see Understanding the split plot.

If I have misunderstood, and you never change those parts, then maybe you just needs to model the correlation and take account of it in the analysis.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467