Hello, all:

I need your expert opinions on something. Prior to performing a Gage R&R study, it’s usually wise to run a traditional control chart on the gage in order to establish basic stability, isn’t it? I’m talking about using a standard of some sort (not actual product), running 3 or more trials for each subgroup, sampling about once a week or so, and developing a traditional X-bar/R chart to make sure there are no special causes working on the measurement system. Some people I've spoken to think that Gage R&R itself establishes basic control of the process. My belief is that basic statistical control is a *prerequisite* to Gage R&R.

Any thoughts?

Talk to you soon,
Craig

Interesting theory, Craig. I never heard of anyone doing such. Every one of my experiences (in-house, supplier, customer, gage house) all involved straightforward R&R studies. Many did make some provision for testing under ambient conditions, rather than in temperature controlled lab.

I don't recall anything in the automotive industry requirements under the heading:
Measurement Systems Analysis (MSA) / Gage R&R
which might add fuel to your concept.
Perhaps Hershal or one of the other metrology specialists will weigh in with a definitive response.

I can't think of a good reason to do this as a matter of course. If there were a situation where the measurements were unusually critical and the gaging system questionable, maybe. In general, standard instruments (calipers, micrometers, indicators, etc.) that are in good condition and have been well maintained will be stable under controlled environmental conditions, in the hands of experienced operators.

Wes and JSW05,

This issue of statistical stability is covered in the MSA book I have (2nd edition, 1995), and it's mentioned as the first step in a true measurement system analysis. Whether it's practical or not is another matter. I suppose stability could be established simply through an ongoing calibration program with an analysis of results.

More thoughts?

Craig

Wes and JSW05,

This issue of statistical stability is covered in the MSA book I have (2nd edition, 1995), and it's mentioned as the first step in a true measurement system analysis. Whether it's practical or not is another matter. I suppose stability could be established simply through an ongoing calibration program with an analysis of results.

More thoughts?

Craig
Unfortunately (but perhaps predictably) the 3rd Edition of the MSA manual took a complex subject and made it even more difficult to comprehend. Here's an example, from page 39 of the 3rd Edition, on the subject of "Measurement Issues":
The long-standing tradition of reporting measurement error only as a percent of tolerance is inadequate for the challenges of the marketplace that emphasize strategic and continuous process improvement. As processes change and improve, a measurement system must be reevaluated for its intended purpose. It is essential for the organization...to understand the purpose of measurement and apply the appropriate evaluation.


As a practical matter, stability studies should always be performed on a production "master" part, and not on a generic reference standard. The feature to be measured is an integral part of the system. One size doesn't necessarily fit all in the case of stability.

It might be argued that evaluation of stability plays a role in selection of a measuring device and system, thus the idea that the stability study should be first. But sometimes, to paraphrase Thomas Huxley, a beautiful hypothesis is slain by an ugly fact, and the fact in this case is that there is generally no value added by confirmation of what is (or should be) already known.

Often, a statistical study is a tool that can be used to establish the calibration frequency. Its use as a prelude to a R&R is a bit unusual, but likely to give more predictable and solid results.....and to reduce the measurement uncertainty.....let us know how it goes.

Hershal

Hershal, Wes, and JSW05:

Thanks for all the information. It's interesting how confusing those MSA manuals were, and how few people seemed to follow them to the degree someone intended. I'm going to hold off on the X-bar / R charts, except for the super critical instruments.

Thanks again,
Craig

MSA manuals are confusing enough, but FWIW, here are my 2c:

Stability study should be performed as an ongoing exercise, not specifically in preparation of or as a prerequisite to a GRR study.

I believe it is important to know the status of stability chart when you are going to make any decision about the measurement system based on a GRR study. The frequency and time period of the stability study are important here. If the measurement system has large inherent variation (shown on stability control chart), it would be difficult to make any decision based on the GRR results. It is like calculating Cpk from an unstable process.

Stability has to do with predictability. And as Hershal says, it can also be used to establish calibration frequency.

Atul,

Great information! This is confirmation of what I believed. Thanks a bunch.

Craig