A218 L13 by Nastasia Nabilah

GR&R Study (Crossed)

nastasianabilah — 2023-02-16 08:25:58 UTC

A GR&R study is a methodology to determine the variation of measured data due to the measurement system and comparing it with the total variation.

The goal of finding out the variation is to keep the measurement variation < process variation

210103791 — 2023-02-16 11:42:51 UTC

Based on the data:
- the small P-value of 0.000 shows that the operators measured the nozzle diameter differently
- the nozzle*operator with the p-value of 0.707 is larger than the alpha value of 0.05, hence the interaction between the operator and the nozzle is insignificant

nastasianabilah — 2023-02-16 11:48:48 UTC

4. Diam by Nozzle
The measurements done by both operators are relatively similar, although operator 1 is slightly higher than operator 2. This means that the measurements done are consistent

5. Diam by Operator
Measurement done by operator 2 appears to be slightly lower than operator 1

6. Nozzle*Operator Interaction
All the nozzles indicated disagreement between the two operators. Parts 1, 3, 4, 6, 7 and 8 show the overlap of operator 1 being higher than operator 2, Part 9 where operator 2 is higher than operator 1 and Parts 2 and 5 where both the operators are on the same level. This may be because the measurement procedure does not account for within-part variance. For instance, the part thickness is inconsistent, and operators measure at several locations.

nastasianabilah — 2023-02-16 11:49:19 UTC

A joint probability is used when we do not have any prior knowledge of the acceptability of the part. As we do know the USL and LSL of the nozzle, which is 9012 +/- 7, a joint probability would not be useful.

nastasianabilah — 2023-02-16 11:49:31 UTC

Based on the data;

- Total Gage R&R is 0.80% which is <1%.
The measurement system is acceptable -> most variation caused by parts and not equipment or people

- Part-To-Part is 99.20% which is the most variability in the measurement -> captured around 100% of the part variation -> the system can distinguish between parts

nastasianabilah — 2023-02-16 11:54:27 UTC

Based on the data;

- The Study Variation is 8.97% which is <10
The measurement system is acceptable -> most variation caused by parts and not equipment or people

- The Tolerance is 4.73% which is <5
The measurement system is acceptable -> A value of 5 or more denotes an acceptable measurement system

- Number of distinct categories is 15 which is >5 based on AIAG

nastasianabilah — 2023-02-16 12:01:00 UTC

whereas a conditional probability is used when we have prior knowledge about the acceptability of the part. Hence, as we know the USL and LSL of the nozzle diameter, which is 9012 +/- 7, conditional probability is useful.

Gage Run chart

21021276_2 — 2023-02-16 12:07:05 UTC

Comments on part to part variation:
From this chart we can see there operator to operator variation are generally very small both operator 1 and 2 have little variation when measuring the same part.

Comments on repeatability of each operator :
For operator 1, when comparing the measurement taken the first and second time for each part, the second measurement taken is lower as compared to first measurement of the same part.

Operator 2 have generally less variation in the measurements taken for part 1,4 and 8 both measurements taken for the respective parts were the same.
For operator 2 the first measurement taken is smaller than the second measurement taken for the other parts 2,3,5,6,7,9.

R&R (ANOVA) components of variation

21021276_2 — 2023-02-16 13:14:13 UTC

This shows that the largest %contribution to variation is part to part variation which is variations due to the different part measured which is a good thing. We want most of the part-to part variation to be have the highest %contribution to variation and low Gage R&R contribution %contribution

R chart by Operator

21021276_2 — 2023-02-16 13:18:53 UTC

From here we can see that there is no variability of measurement of nozzle 1,2,3,4,5 for Operator 1 as compared to measurements of all nozzles taken by Operator 2 shows more variability as compared to Operator 1.

XBar chart by Operator

21021276_2 — 2023-02-16 13:24:01 UTC

The points in the x bar chart are statistically out of control which indicates that the variation between measured parts ( part-to-part variation ) across the two operators is greater than expected. This also supports why the %contribution of part-to-part is higher than the %contribution of Gage R&R.