14. MSA – Type II Gage R&R
MSA’s type II study is applied when measurement system will be influenced by both appraiser variance (different operator) and equipment variance. It also will consider the part to part variance within the system. In general, the minimum sample size shall be at least 30 (10 non-destructive different samples measured for at least three times, each time is measured by different operator) to have close to the normal distribution.
For the example demonstrated below, 6 different operators measured 25 different samples for at least 3 times. Therefore 450 data points (25 x 6 x 3) were collected to analyze measurement system’s gage R&R, and the results running from Minitab can provide us the following explanation.
First, let’s start with how to interpret both the X-Bar and R chart from the analysis.
- R-chart helps to indicate part’s max and minimum (usually is max – min), so if the range is greater than the UCL, then it indicates the operation does not have a standardized method to measure for parts. And vice versa.
- X-Bar chart indicates and compares part to part variance and repeatability while the control limit is calculated based on the measurement’s quantity and average. And by looking at our example, since some data points were falling outside of control limit. It indicates the part to part variance is greater than the gage variance itself.
X Bar – R charts for the integration between different operator and the repetitive measures.
The interaction chart and measurement comparison chart on the left hand side is to explain the measuring pattern between the operators or the parts..
- For interaction charts, if all operators have similar trend. Then the operation is similar to measure all parts.
- If one of the operator has an uniformly high measurement, then the operator’s technique could seems suspect and need to be re-evaluated.
- If trend is not parallel or crossed, then the operator’s measuring capability is determined by the particular parts.
- For measurement comparison chart, it helps to evaluate measurement distribution and operator variances. By this case, all 6 operators measure similarly (if the distribution is different then the measurement could be different between operators).
Measure Data Distribution and Interaction Chart.
If measurement system is used for process improvement, use %StudyVar for better estimation for measurement accuracy.
If measurement system to used for evaluating the parts’ relation with respect to specification, then %Tolerance is a more appropriate indicator.
By looking at the analysis result given on right hand side, both measurement repeatability and reproducibility are less than 20% (total gage R&R is 10.40, Repeatability is 8.81 and Reproducibility is 5.52), this indicates the consistency and stability of measurement system is in good condition under spec tolerance. But potential improvement is still needed based on the variance tolerance contributed within the system.
The measurement system’s number of distinct category indicates it can detect 1 group (ndc = 1), but this is reasonable because this gage is used for evaluating one particular part.
MINITAB Data Summaries