We should start with GRR or Bias first?

[hazwan2283]

Hi all, my name is Hazwan. Kindly would like to ask a question here .

Question 1: In what order should we start to do a variable MSA study?. To elaborate the first question, do we need to start MSA with GRR first followed by Bias, then linearity and followed lastly by stability study?.Since AIAG MSA book suggesting to check %EV first prior doing Bias, we need to start with GRR first then correct?.

Question 2: in AIAG book, it is stated that there are few items need to be checked when we evaluate MSA results such as the sensitivity of the measurement system and the measurement system needs to be stable. Does this mean we need to do stability study first before doing GRR?. I'm super confused at the moment.

 

[Jen Kirley]

My experience with GR&R tools is that bias, repeatability and reproducibility are all calculated by the same data inputs.

 

[Bev D]

If the study is conducted appropriately it can be. It depends on which bias you are concerned with.

In any case, repeatability must be established first. The manual doesn’t specify that stability has any sequence other than it must be established at some time before use. (Unless I missed something). But even stability is tough to establish without a solid repeatability unless you use a crap-ton of parts to ‘see thru’ any repeatability variation.

 

[hazwan2283]

If the study is conducted appropriately it can be. It depends on which bias you are concerned with.

In any case, repeatability must be established first. The manual doesn’t specify that stability has any sequence other than it must be established at some time before use. (Unless I missed something). But even stability is tough to establish without a solid repeatability unless you use a crap-ton of parts to ‘see thru’ any repeatability variation.

So i believe @Bev D what you are saying is yes we need to start with GRR to get the repeatability then we can proceed with bias etc..thank you Bev

 

[Miner]

I would recommend doing a bias and linearity study first as part of the calibration process. This would also be a good opportunity to assess the resolution (sensitivity) of the measurement device. I have found that in most cases you can perform a risk assessment on the need for a stability study. In years past, complex electronic equipment was subject to warmup and drifting over time. With today's technology, not so much. Mechanical measurements are typically only subject to thermal expansion and vibration. Thermal expansion is only a risk with precision tolerances. I have personally only encountered one occasion where a stability study was appropriate. A GRR study is usually the last study in sequence.