Repeatable Expectations For GR&R

It seems axiomatic that the more precise we try to be, the less certain we become. In today's manufacturing world, nothing is flat, round, smooth or exactly the same length. Tenths have turned into thousandths and now millionths. While it seems easy to put a dimension on a part print, it's neither easy to make the part nor to design a gage to measure that tolerance with some certainty.

The process of determining the capability of a gage—GR&R—requires analyzing the complete measurement process involving that gage. This includes looking at each of the individual elements, which we remember as SWIPE: the standard, the workpiece, the instrument, the people and the environment. Each element can be a source of error and if you change any one element, you have to reanalyze the entire process again.

However, the heart of the measurement process is still the gage. There are many design characteristics that need to be incorporated to help the gage meet its requirements. The tighter the tolerance, the more consideration needs to be given as to how the part is staged, how the contacts meet the part and how deviation is displayed. It is easy to sometimes put a very tight tolerance on a part; it's just as easy to require that a gage meet a 10 percent GR&R requirement on that very tight tolerance.

Because the gage is only part of the measuring process, it is allowed only a percentage of the total GR&R spread. In the worst case, the repeatability allowance for the gage, by itself, on the master, is 4 percent of the tolerance. So what does that mean with today's shrinking tolerances? The following chart might help. If we look at the tolerance spread and take 4 percent of it, small numbers begin to appear.

There are a few things to consider when looking at this list. Most importantly, it puts into perspective just how good the gage must be to achieve the 10 percent GR&R. Then we have to consider what the repeatability test should consist of to verify the requirement. What this spec emphasizes is how well the gage can repeat only on the master. This says nothing about the gage calibration, bias or linearity. It solely refers to the repeatability. Even so, we are really dealing with the whole measurement process, but trying to isolate a small portion for the gage.

Let's assume that the environment is stable and one operator is capable enough to put the master in the gage the same way every time. Usually, the test consists of one operator measuring the master 30 times. Therefore, the master should be appropriate for the test: it must be the right form and surface finish, and stabilized for the test. After 30 measurements—done in a series over a short time—the results are analyzed for total span. If the span falls within 4 percent of the tolerance, then there is a good chance that the gage repeatability will not be the determining factor in the 10 percent GR&R requirement.

Some of the 4 percent requirements are marked with a double asterisk (as shown on the chart) to note that in many gaging applications, this may not be attainable. We are referring to those gaging applications that would be performed in the typical shop environment. To even think about achieving these types of performances, you must consider the measurement as a millionth class gaging requirement. There are gages that can accomplish this—gage block comparators and precision length machines—but these are specialized machines designed to meet specific requirements. This is not to say that production gages cannot be designed to this level (they can and have been). However, they require automatic control, special environments and design criteria for millionth measurement, along with a lot of investment.

The point is that it's often a casual thought to put a GR&R requirement on the gaging request for quote. But before it's put on the paper, both the tolerance and the expectations need to be reviewed to see if the requirement is achievable.