Reproducibility on Different Machines of the same Configuration

W

wroosmalen

I performed a GRR on 5 lines of the same configuration. Intercepted 10 products on 1 line and measured them twice on each line with a measurement system of the same type. So reproducibility describes the variance added by the measurement systems on different lines and not by different operators. The measurement is fully automatic and with no operator influence.

I think it only makes sense to look at the "to process variation" value to see if the data can be used for SPC over the 5 lines. Does the "to tolerance" value makes sense having the reproducibility of different machines? I can use the data to perform a type-3 study resulting in 0 reproducibility and only looking at the repeatability. I think this value would make more sense to make conclusions about the ability to separate right and wrong products.
 

Miner

Forum Moderator
Leader
Admin
I performed a GRR on 5 lines of the same configuration. Intercepted 10 products on 1 line and measured them twice on each line with a measurement system of the same type. So reproducibility describes the variance added by the measurement systems on different lines and not by different operators. The measurement is fully automatic and with no operator influence.

So far, so good. This is a perfectly sensible approach.

I think it only makes sense to look at the "to process variation" value to see if the data can be used for SPC over the 5 lines.

It depends. If you only use the measurement systems for SPC, yes. On a side note: Be careful about the rational subgrouping when dealing with multiple process streams.

Does the "to tolerance" value makes sense having the reproducibility of different machines? I can use the data to perform a type-3 study resulting in 0 reproducibility and only looking at the repeatability. I think this value would make more sense to make conclusions about the ability to separate right and wrong products.

Again, it depends what you want to learn. If you include the different machines in the study, the reproducibility will tell you the overall ability of all machines to separate right and wrong products. If you leave this out, you will know the individual ability of a single machine to separate right and wrong products.
 
W

wroosmalen

Thanks for your answer.

If you include the different machines in the study, the reproducibility will tell you the overall ability of all machines to separate right and wrong products. If you leave this out, you will know the individual ability of a single machine to separate right and wrong products.
What does the "overall ability of all machines to separate right and wrong products" adds to the analysis? Isn't it sufficient to know the individual ability of GRR/T to separate right and wrong products. A product produced on a line will always be measured by the measurement system on the same line.
 

Miner

Forum Moderator
Leader
Admin
It depends on your particular situation. Others may have different needs. I am giving you the different options. It is your decision on which to use.
 
W

wroosmalen

It depends on your particular situation. Others may have different needs. I am giving you the different options. It is your decision on which to use.

Can you give me a situation in which the GRR to tolerance over all lines would be useful to use? I find it hard to imagine a situation where this value would be useful.
 
Top Bottom