Calculating part variation from historic data for GRR study

[joonasan]

We made a GRR study for measuring gear outside diameter. We had too little information about GRR so the study failed. The part variation was bad, only 13,4%. That means we got very bad results for %GRR (GRR/TV).

I understand now that with better PV we will get more acceptable results. We could continue the study with more samples but I was thinking if we could use our history measurement database to calculate part variation, instead of using the GRR study's PV? I'm a little confused how and if this can be done.

However %GRR based on the tolerance(0,063 mm) is still too bad, so we need to do something in any case.

 

[Miner]

Using historical Process Variation is the best approach as you have discovered.

If you use Minitab, there is an option to enter the Historical standard deviation under the Options tab. This will add a % Process metric that you use instead of the % Study Variation metric.

If you do not use Minitab, you can use my Excel based MSA file found in my blog. If you enter your historical process variation in cell T59 of the GRR Report tab, it will calculate the % Process Variation metric, which is what you are seeking.