The repeatability and reproducibility study, known as R&R, answers a specific question: how much of the variation you see in your data comes from the measurement system and how much from the parts. If the system contributes too much, it is not fit to control that process, however good the instrument may be in calibration.
How it is designed
- Parts: selected to represent the real variation of the process, including values close to the tolerance limits.
- Operators: those who actually measure in production take part.
- Repeats: each operator measures each part several times.
- Randomization: the measurement order is shuffled and the parts are identified in such a way that the operator does not remember previous results.
- Conditions: the usual method, instrument and work environment are used.
How the data are analyzed
There are two common approaches. The average and range method is simple and allows manual calculation. Analysis of variance, ANOVA, is more complete because it also estimates the interaction between operator and part. Both break the variation down into repeatability, which is that of the instrument with a single operator, and reproducibility, which is that due to changing operators.
How to interpret the result
The result is compared with the total variation of the study or with the tolerance, depending on the purpose. A low percentage indicates that the system distinguishes well between parts; a high one, that the system hides real differences. Acceptance criteria are set by each organization or applicable guide, and there is no universal value.
What usually goes wrong
- Choosing nearly identical parts, which makes the system look terrible.
- Letting the operator know which part is being measured.
- Taking the repeats back to back, without reinstalling the part.
For destructive measurements, or where measurements cannot be repeated on the same part, specific designs are required.