Dear Guys

Recently i have faced a situation where my customer was asking the GR&R value to be calculated with refer to part tolerance instead of process variation. Base on the hand out page 116 (MSA 3rd edition), the practice can be done to determine if the measurement system is acceptable for it’s intend application. Further more a statement say “ will not equal 100%. I would like to understand more as what is the intention behind of doing this,

Base on the practical calculation
A method - By refer to Total variation, GR&R = 20.05%
B method - By refer to part tolerance, GR&R = 3.13%
The B method seems to give the minimum value compare to A method.

Apart from that the book call for the tolerance value to be divided by six, example -/+ 1.00mm = 2mm/6

Appreciate you can give some clue as why there is a different :mad:

Periannan

In my case, I use two worksheets...

If the process tolerance is known, I use a worksheet which computes for the Total Variation (TV) as tolerance/6. This TV value will now be used to calculate the %GRR = 100(GRR/TV)...percent GRR to Tolerance.

If the tolerance is not known or not applicable, I use a worksheet which computes for the TV which is equal to SQRT[(GRR*GRR)+(PV*PV)]...percent GRR to Total Variation.

Either of the two approaches is acceptable.

Dear Daniel,

Yes both approach is acceptable and why there is different value seen? (related back to my earliest post). understand from one of my friend, part tolerance is used to calculate GR&R value to evaluate the equipment is able to measure its intend part and by using the process variation, it's calculate the measurement system effectiveness. is it true. i'm still blur

Periannan

Dear Daniel,

Yes both approach is acceptable and why there is different value seen? (related back to my earliest post). understand from one of my friend, part tolerance is used to calculate GR&R value to evaluate the equipment is able to measure its intend part and by using the process variation, it's calculate the measurement system effectiveness. is it true. i'm still blur

Periannan

Attached is the sample of it

Hi Periannan,

The computation for %GRR using Tolerance and Total Process Variation is the same. But the resulting %GRR value will be different mainly because of the difference in Total Variation (TV) calculation which is the denominator in both equation (%GRR=100(GRR/TV).
When it comes to the definition, I am not in authority to say whether it is true or not...but if we will go back to the definition or Gage R&R in the MSA manual - it is an estimate of the combined variation of repeatability and reproducibility in a measurement system. But Gage R&R is different from %GRR. If my understanding is correct on %GRR, if you get 10% GRR result, then you are allowing 10 marginal parts out of 100 parts inspected to be called good when it is actually reject or the parts are good but the measurement system calls it reject. So the lower the %GRR value the better.

I attached two worksheets using different methods. You can check it out and give me feedback.

See the following thread for a detailed explanation.

http://elsmar.com/Forums/showthread.php?t=18189

Dear Miner,

it's really help, thank you. just want to check with you, By using the part tolerance, is it the acceaptable figure will be the same like GR&R less 30% (ideal cases)

it seems the GR&R value in spec but part tolerance figure is more than 30%. it mean something wrong with equipment, the part or the processs.

periannan

By using the part tolerance, is it the acceaptable figure will be the same like GR&R less 30% (ideal cases)

it seems the GR&R value in spec but part tolerance figure is more than 30%. it mean something wrong with equipment, the part or the processs.

periannan

The acceptance criteria is the same for both measures.

Assuming all calculations are correct, an acceptable %GRR with an unacceptable P/T Ratio could only occur if the process variation was not capable to the tolerance, or if the variation of the samples did not match the variation of the process (that is, the sample variation was wider than the actual process variation).