Air flow measurement system MSA (Measurement System Analysis)

Hello everybody,

that's the first time I write a topic, but I have read many topics and I think you have a lot of Knowledge about this thema.

So, my doubt is about a %GRR between 10% and 30%.
We have measured 10 parts with 3 operators, 2 times each one. Our tolerance is +/- 12.2% of nominal value.
We have tested two different part number, each one with a different nominal value of air flow.
Part number #1: %GRR=13% (nominal value approx. 20kg/h)
Part number #2: %GRR=16% (nominal value approx. 30kg/h)

The Cp and Cpk of the measurement system is lower then 1.33 for both nominal values, when the measurement are made with standards.

How we can make the %GRR better, I mean <10%?
How we can make a separeted analysis about the operator influence, the variation of the machine influence and the variation of our product?

Charle said:

Hello everybody,

that's the first time I write a topic, but I have read many topics and I think you have a lot of Knowledge about this thema.

So, my doubt is about a %GRR between 10% and 30%.
We have measured 10 parts with 3 operators, 2 times each one. Our tolerance is +/- 12.2% of nominal value.
We have tested two different part number, each one with a different nominal value of air flow.
Part number #1: %GRR=13% (nominal value approx. 20kg/h)
Part number #2: %GRR=16% (nominal value approx. 30kg/h)

The Cp and Cpk of the measurement system is lower then 1.33 for both nominal values, when the measurement are made with standards.

How we can make the %GRR better, I mean <10%?
How we can make a separeted analysis about the operator influence, the variation of the machine influence and the variation of our product?

Click to expand...

Welcome to the Cove, Charle

The weekends tend to be a little slow here, so I'll look for this on Monday and bump it up if necessary. In the meantime, you might want to provide a little more information. A GR&R result of 16% isn't necessarily undesireable; in fact, depending on the application, it might be perfectly acceptable. Why not tell us a little more about the product, how you're measuring it, and why you feel you need to improve the measurement process. How are you doing the calculations? GR&R software should be breaking the results down into operator error, device error and part variation, as well as Number of Distinct Categories (NDC). Give us some more information and we'll be able to give you a better answer.

Attaching the raw data would also help.

Data

Hello,

I'll try to explain a little bit about the product and the measurement method.
Our product have a throttle that we movement to change the air flow rate. We
proceed the measurement in two throttle positions (1. Completely close (approx.
2,5L/h); 2. Default idle position (approx. 30L/h)). In this measurement the
difference of pressure is controlled in 600 mbar and a sensyflow sensor measure
the air flow. The values of sensor are converted with the actual temperature,
humidity and pressure.
We had measure two standards (2,123Kg/h and 30,490kg/h) to calculate the Cp and
Cpk of the device. We found always values lower then 1,33.
We must to improve the measurement system because there is a rule about it in
our company. So we measure 10 parts 2 times each one with 3 operators for two
part numbers and we find the following values:

#1 part number
Position 1. GRR%=13,8 ndc=3 (nominal value<3,0)
Position 2. GRR%=16,3 ndc=4 (nominal value=28+/-3,4)

Operator A: Daniel Operator B: Sérgio Operator C: Eliéser
Measure 1 Measure 2 Measure 1 Measure 2 Measure 1 Measure 2
1 1,9 1,960 1,950 1,970 2,020 1,960
2 2,01 1,940 2,030 1,910 1,960 1,970
3 1,81 1,760 1,750 1,820 1,780 1,840
4 1,99 1,820 1,950 1,900 1,840 1,920
5 2,01 2,000 2,090 2,090 2,120 1,980
6 2,22 2,270 2,120 2,080 2,180 2,110
7 2,03 1,890 2,090 1,980 1,950 1,900
8 2,17 2,140 2,260 2,220 2,190 2,250
9 2,18 2,280 2,300 2,320 2,000 2,000
10 2,38 2,410 2,400 2,320 2,350 2,010

Operator A: Daniel Operator B: Jeferson Operator C: Eliéser
Measure 1 Measure 2 Measure 1 Measure 2 Measure 1 Measure 2
1 28,86 29,930 29,720 29,730 29,940 29,900
2 28,93 28,760 28,570 28,800 28,930 28,970
3 28,79 28,350 28,810 28,840 28,800 28,800
4 28,63 28,430 28,790 28,650 28,430 28,710
5 28,44 28,390 28,510 28,640 28,640 28,410
6 29,85 29,600 29,750 29,740 29,800 29,600
7 29,14 28,940 29,200 29,050 29,010 28,570
8 28,12 27,890 28,210 28,060 28,140 27,930
9 28,98 29,060 29,050 28,870 28,700 28,750
10 28,46 28,550 28,760 28,310 28,530 28,110


#2 part number
Position 1. GRR%=15,1 ndc=1 (nominal value<2,5)
Position 2. GRR%=13,8 ndc=3 (nominal value=20+/-2,4)

Operator A: Daniel Operator B: Jeferson Operator C: Eliéser
Measure 1 Measure 2 Measure 1 Measure 2 Measure 1 Measure 2
1 1,89 1,890 1,83 1,860 1,890 1,810
2 1,83 1,890 1,91 1,950 1,940 1,820
3 2,07 2,020 2,03 2,020 1,970 1,860
4 2,04 2,000 2,15 2,020 2,000 1,990
5 1,94 1,820 1,9 1,960 1,830 1,820
6 1,92 1,910 2,01 1,910 1,940 1,890
7 1,84 1,980 1,9 2,020 1,850 2,030
8 1,83 1,970 2,02 1,980 1,910 1,990
9 2,00 2,030 1,96 2,010 1,900 1,960
10 2,02 2,020 2,03 2,060 2,070 2,010


Operator A: Daniel Operator B: Jeferson Operator C: Eliéser
Measure 1 Measure 2 Measure 1 Measure 2 Measure 1 Measure 2
1 20,23 20,210 20,130 19,880 20,140 19,950
2 19,79 19,830 19,740 19,760 19,840 19,590
3 19,43 19,480 19,470 19,310 19,290 19,400
4 19,95 19,790 19,940 19,770 19,650 19,570
5 19,84 19,620 19,830 19,890 19,720 19,810
6 20,44 20,150 20,340 20,350 20,020 20,010
7 19,99 19,840 19,990 20,050 19,880 19,930
8 20,16 20,010 20,080 19,980 19,960 20,080
9 20,25 20,050 20,190 20,210 20,010 20,180
10 19,4 19,400 19,450 19,400 19,380 19,330


I didn't use a software to calculate the %GRR values. I create my own Excel file
for this.

Thank you in advance for your attention.

Please answer the following questions:
Is this gage used for product inspection to tolerance or for process control (e.g., SPC)?
What do you mean by the Cp and Cpk of the device? These indices normally represent the process, not the gage.
Is the < 10% requirement from an internal source?

I evaluated your first two sets of data and confirmed your P/T ratio and ndc results. The gage definitely has adequate resolution. Both studies showed an out-of-control point on the Range chart. This has slightly inflated your percentages, but is not a major source of your problem. However, you should still investigate and correct the issue.

The major source of measurement variation is coming from the equipment (i.e., Repeatability), not operator (Reproducibility).

This could be from the following sources: variation within-part (form, finish, taper, etc.), within-gage (wear, fixture, maintenance, etc.), within-standard (class, wear, quality), within-method (setup, clamping, zeroing, etc.), within-operator (technique, feel, etc.), within-environment (fluctuations in temperature,humidity, etc.), and stability, etc.

Check the basics such as the cleanliness of the parts and gage, state of repair, etc.

Since your measurements are converted based on actual temperature, humidity and pressure, I recommend that you perform a Stability study as well as make sure that the equipment is adequately protected from short-term fluctuations in these environmental conditions.

Also check the effect of part movement within the gage. If the parts are cylindrical, does rotation affect the result (see within-part variation). If the parts are not cylindrical, does movement within the fixture affect the result (see with-gage variation and within-method variation).