GR&R for Gage Building

A study of the entire measuring system should identify the individual components that contribute to a lack of repeatability or reproducibility.

Gage repeatability and reproducibility, or GR&R, is a tool used to assess the reliability of gaging results. In a GR&R study, a few gage operators each measure a small number of parts several times. The results are compiled and, after some mildly confusing arithmetic, reduced to a single number that indicates the total expected spread of measurements for a single part, for all trials, by all operators. That number is presented as a percentage; a GR&R of 30 percent means that all the results fall within a range equal to 30 percent of the allowable part tolerance. (This is slightly simplified, but close enough for our discussion.)

GR&R doesn’t measure the gage in isolation; it measures the entire gaging process, with all of its influences and variables. If you want to achieve a GR&R of 10 percent, then you’ll have to be able to read the results to a considerably higher degree of resolution than 10 percent of the required tolerance. For example, a typical engine component might require 10 percent GR&R on a part with a total tolerance of 0.001 inch (±0.0005 inch). That means all the measurements for a given workpiece should fall within a range of 0.0001 inch.

By knowing this target, some basic gage design criteria become clear. For example, achieving that 0.0001-inch tolerance requires a gage that has a resolution better than 0.0001 inch and a repeat performance better than 0.0001 inch. If it does not, the gage itself will be taking up too much of the GR&R allotment. The flip of a 0.0001-inch digit on the display could mean the difference between passing or failing the GR&R requirements.

Remember the acronym SWIPE? There are five major factors that influence gaging results: standard, workpiece, instrument, personnel and environment. Each of these introduces a certain amount of variation to a measurement. Is the standard (the master) absolutely accurate? How about the workpiece’s geometry? If it’s out of round, it will generate different results every time you put it on the gage. The gage operators will introduce a certain amount of observational error, plus variability due to differences in gaging practice. Are you paying attention to the environmental factors that can influence a measurement: temperature, dirt, vibration, etc.? And, of course, there’s the instrument—the gage itself—which could have stiction, wobbles, a misaligned holding fixture or even, just possibly, a wrongly specified dial indicator.

As a gage manufacturer, we have a real responsibility for two out of the five components of the measuring system—the gage itself and, if we are supplying it, the master. As part of our process, we try to determine the performance of the gaging process by leaving out the workpiece and testing the rest of the components—SIPE. Since we are building and putting master (standard) and the gage (instrument) together, we have a pretty good chance of controlling the people and the environment. Thus, we use a different GR&R guideline. By properly taking into account the required resolution and the repeat performance of the gage, and by controlling the people and the environment, our preliminary testing target for a gage is 4 percent GR&R.

Using the single master and three operators, we know what the gage is reliably capable of doing. If 4 percent is achieved, the remainder (that 10 percent or 30 percent) can be allotted to errors influenced by the workpiece. From experience, though, we have seen the workpiece make or break GR&R testing. If the part is not round enough for the gaging requirements, or if the surface finish is such that a standard contact point picks up anomalies, then there is a pretty good chance the results of the gage will not meet expectations. When the part is too much of an influence, you have to find ways to take that influence out to the process. This could mean marking the part for location both axially and radially, and measuring it at the same point for every test by every operator. 

The moral of the story is: If a GR&R study produces failing results, don’t shoot the gage. You can’t expect it to correct errors from other sources. This also extends beyond the confines of GR&R. Any time you’re assessing a gaging program or trying to determine your gage requirements, remember that the instrument is only one-fifth of the equation.