Gage R&R - how to conduct?

How do I conduct a Gage R&R study correctly? If I do 3 trials, use 3 appraisers, and 10 parts. Do I want the parts to vary or not? I thought about using gauge blocks rather than parts because I know that they are accurate and what the measurement is. If I have 10 parts do I have my appraisers measure the same 1 dimension on all 10 or measure 10 different dimensions on each part?

Comments

  • Matt SMatt S PQ Systems Employee

    Hi @jgergen,

    There are guidelines, but let me start with these three recommendations:

    1.       You do want the parts to vary. The expression “Bad parts make for good studies” applies here. You want to get a sample of parts that represents the true variability of the parts/process. So you might get one part from Monday morning, one from Monday afternoon, one from Tuesday morning, etc.

    2.       You want one appraiser to measure the 10 different parts. Then they will repeat this two additional times, but in a ‘blind’ study. I think of it like a blind taste test where the person doesn’t know what they are sampling. So you might label the 10 parts in a way that you know what they are measuring, but they don’t know which part.

    3.       Ideally this study will be done by the actual people who use the gages in as close as possible to the environment that they are measuring their production parts in. In other words, if they take measurements on the production floor, the R&R study should not be conducted in a laboratory type of environment.

     I hope this gets you started. Our expert, Eric Gasper, can provide additional details if needed.

  • Hi, Matt and jgergen. I agree with Matt. Not only do you want the parts being measured to vary, you also need to see variation when your appraisers report a series of measurements of the same thing with the same device. Method won't work without that.  Ideally you can randomize the parts and variation will naturally follow, if as I said, your measuring devices have enough resolution to generate data for your study. Speaking of gage blocks, use the very best resource as a standard that you can get your hands on.  This is because gage studies are great for learning about and distinguishing "between" and "within" variation, but they have a very tough time producing an unbiased assessment of the measuring system behind your numbers.  Check out Dr. Don Wheeler ("An Honest Gauge R&R Study") on this topic.

    Stephen
Sign In or Register to comment.