High Repeatability? – Why it may not be a Gage Problem

What is repeatability?

This is the definition of Repeatability from Automotive Industry Action Group Measurement System Analysis Guide Book:
“Repeatability is the variation in measurements obtained with one measurement instrument when used several times by one appraiser while measuring the identical characteristic on the same part. This is the inherent variation or capability of the equipment itself.”
OK, so if I measure a single part 100 times using the same gage and do the calculations, I will have captured the variation that’s driven by two things:

  1. The consistency of my technique
  2. The gage’s inherent variation

– Right? – Well No – not necessarily !

The Effect of Within Part Variation

To illustrate why my repeatability measures may be corrupted, let’s look at a simple example. Let’s say I produce a 1″ diameter x 3″ long shaft. The diameter tolerance is +/- 0.002″, and I measure it using a 1/10th of a thou calibrated Micrometer. Now, I’ve done a gage r&r study, and have found a very high repeatability, which I don’t understand, so I take a part to a Laboratory, and they measure it using their super-expensive system and give me back this crappy, but informative, sketch:

Diameter with excessive within-part variation

Whoops! – What I thought was high gage repeatability was actually nothing to do with the gage – it was variation within the part itself!
If I measure the part at the small end and record the measurement, and then measure the part at the large end and record the measurement, I have introduced a source of variation that is nothing to do with the gage.

Dealing With High Within Part Variation

So, assuming that I don’t want to improve the process yet, I need to reduce the effect of variation within the part.
This is pretty easy to do in this case – I just have to make sure that all my measurements are taken in exactly the same location on the part.
Now, it could be argued that taking the measurements at the same location on the shaft is part of the operator technique. However, I would counter argue that this particular shaft has no features that allow me to recognize one end from the other, or to identify the across diameter measurement points. Furthermore, due to the taper, and out-of-roundness on these shafts, any decent operator would be taking multiple measurements across the whole part, to ensure the part is within spec.

An alternative approach would be to get hold of a “master” part. What I mean by master part is one that is geometrically consistent – i.e. insignificant cylindricity relative to the specifications.
Using the master I could have run a small repeatability study where all the variability was either down to inconsistent technique, or inherent gage variation

A Couple of Other Options

You could take a couple of parts and run them across a Co-ordinate Measuring Machine (or any other very accurate system) to look for within part variation, before you do the gage r&r study
Before the gage r&r study, you could take a production part, and measure it in exactly the same place, say 10 times, and then measure anywhere on the part 10- times. When you compare the results this will give you a feel for whether within part variation is a concern or not.

More Difficult Parts

This article was written using a pretty extreme, but simple, example to illustrate the point, but think about this:
What if we need to do gage r&r studies on complex systems like Electro-Hydraulic systems, or delicately balanced spring loaded systems. In these cases there is massive within part variation – apply the same load twice, and the system reacts differently both times.
So, in these cases, if you want to determine pure gage repeatability you have to have a standard that reacts consistently.

Tagged with,
This post was written by
Comments are closed.