Your Measurement Systems are Better than You Think!

The usual standards for judging whether a measurement system is adequate are arbitrary. Donald Wheeler has determined new criteria based on objective standards. These new criteria show that many systems formerly considered inadequate are just fine. A Gage Repeatability and Reproducibility (GR&R) study measures the noise in a measurement system. %GR&R is a common measure of this noise. It is the ratio of the measurement noise (as a standard deviation) to the total variation (as a standard deviation) expressed as a percentage. Total variation includes the measurement noise and the actual part-to-part variation you want to measure. Historically, %GR&R has been compared to the numbers 10 and 30 -- less than 10 %GR&R is a good system, between 10 and 30 %GR&R is a questionalble system, and greater than 30 %GR&R is a poor system. Unfortunately, these numbers are arbitrary. Donald Wheeler has determined a new set of numbers to replace 10 and 30 for comparison with %GR&R. These new numbers are based on objective criteria. These criteria are, the probability that you will detect an out-of-control point in your process, and the impact of noise on your measurements through "attenuation," or how much the noise will obscure what you are actually measuring. The table below provides a summary of the objective guidelines for judging the quality of a measurement system using %GR&R. Quality %GR&R Good < 45% Adequate < 71% Maybe < 89% Poor > 89% You can read the full details in Donald Wheeler's article, "Good Data, Bad Data, and Process Behavior Charts." You can learn Measurement Systems Analysis, including Gage R&R studies, in workshops from Objective Design of Experiments. You will learn practical skills for assessing your measurement systems that you can apply immediately to your work.