Simulate Tolerance Range with Real Parts

[thehovie]

I am performing a gage R&R test for hydraulic hose fittings after being crimped on hoses. I have chosen 10 fittings and marked a location with a blue dot on each one. The upper spec is 21.459 and the lower spec is 20.959 but all of the measurements are coming in between 21.27 to 21.3. According to an article titled "Gage R&R (Repeatability and Reproducibility)" on qualitymag.com (not enough posts to insert a link), you should find parts that represent the full range of tolerance. We are producing these so consistently that I cannot find anything near the top or bottom end of the tolerance.

Should I simulate the upper end of the variation by applying a thin layer of some kind of durable and/or hard tape to the fitting? For the lower end I could maybe hammer a fitting down a bit .

 

[Jim Wynne]

What you want is the full range of the process, not the tolerance.

 

[Bev D]

This is a deep and complex subject. If you can post your data we can give further advice. I also strongly recommend MIner’s blogs and my paper on measurement error in the resources section. I will let you poke around as there is a wealth of knowledge there and I’ve almost always accidentally found really useful things when I wander around looking for something else.

 

[Jim Wynne]

The article is here, and the author contradicts himself by saying,

[You will need] Five to 10 parts (number each part) that span the distance between the upper and lower spec limits. The parts should represent the actual or expected range of process variation.

 

[thehovie]

What you want is the full range of the process, not the tolerance.

So in my Gage R&R spreadsheet which performs all the calculations, should I use the part tolerance for the upper spec and lower spec, or should I use the upper and lower measurements from our samples (which would represent the range of the process)?

 

[Bev D]

Well it is quality magazine and most articles in recent history have been puff pieces, usually by people trying to sell you something from their company. Caveat emptor

 

[Jim Wynne]

So in my Gage R&R spreadsheet which performs all the calculations, should I use the part tolerance for the upper spec and lower spec, or should I use the upper and lower measurements from our samples (which would represent the range of the process)?

Use the actual tolerance. I also highly recommend Miner's MSA blog articles, and @Bev D's article, which is here: Free - MSA (elsmar.com)

 

[Miner]

If you use the measurement device for inspection, you would use the % Tolerance metric to assess whether it is adequate. Part selection has NO impact on the results of % Tolerance. There is no benefit to making special parts at the extremes of the tolerance. Your calibration program probably checks the calibration throughout the measuring range of the device anyway.

If you use the measurement device for SPC, you would use %Study Variation (OR ndc). In this case the part selection is extremely important and MUST reflect the actual process variation (not the tolerance range).