GRR based on part tolerance or process variation.

T

tongxiaozhi

Dear quality experts:

Can anybody kindly clarify the following content from MSA Fourth edition. Also, we have a molded plastic part with only 1 cavity. The dimesion between the parts are very close, so the process variation is also close to 0, which makes it impossible for GRR to be less than 10%, how to deal with this kind of situation? Many thanks in advance.


For Product Control situations where the measurement result and
decision criteria determine, “conformance or nonconformance to
the feature specification” (i.e., 100% inspection or sampling),
samples (or standards) must be selected, but need not cover the
entire process range. The assessment of the measurement system
is based on the feature tolerance (i.e., %GRR to TOLERANCE).
For Process Control situations where the measurement result and
decision criteria determine, “process stability, direction and
compliance with the natural process variation” (i.e., SPC, process
monitoring, capability, and process improvement), the availability of
samples over the entire operating range becomes very important.
An independent estimate of process variation (process capability
study) is recommended when assessing the adequacy of the
measurement system for process control (i.e., %GRR to process
variation).
 

GRP

Involved In Discussions
In order to calculate the process index (Ppk/Cpk) you use the results of a measurement system. This means the index is affected both by the real variation between parts and the variation of the measurement process. Your actual index will always be better than the calculated because your measurement system variation will never be 0.

If you had a process with Ppk=Cpk=2.0 and you had not performed a GRR, you could argue that both your process and measurement system are good, but you will not know if the actual process index is 10 and this low variation is masked by the measurement variation.

In practical terms, if you need a GRR for submission to your client in order to determine that your measurement system is suitable for a certain characteristic, then use the GRR as a % of tolerance.

Additionally, if you feed your GRR parts with very little variation you will fail your ndc. Omitting for the moment the many sound rebukes of the MSA by Dr Wheeler, just inject a few parts with lower holding pressure, change the tool temperature, increase and reduce cooling time and get parts with variation spanning the entire range of tolerance.

In cases when I needed the GRR and for practical reasons I could not inject parts with different parameters I have gone as far as deforming parts (with engineering judgement) to obtain the required variation for the feature.
 
T

tongxiaozhi

In order to calculate the process index (Ppk/Cpk) you use the results of a measurement system. This means the index is affected both by the real variation between parts and the variation of the measurement process. Your actual index will always be better than the calculated because your measurement system variation will never be 0.

If you had a process with Ppk=Cpk=2.0 and you had not performed a GRR, you could argue that both your process and measurement system are good, but you will not know if the actual process index is 10 and this low variation is masked by the measurement variation.

In practical terms, if you need a GRR for submission to your client in order to determine that your measurement system is suitable for a certain characteristic, then use the GRR as a % of tolerance.

Additionally, if you feed your GRR parts with very little variation you will fail your ndc. Omitting for the moment the many sound rebukes of the MSA by Dr Wheeler, just inject a few parts with lower holding pressure, change the tool temperature, increase and reduce cooling time and get parts with variation spanning the entire range of tolerance.

In cases when I needed the GRR and for practical reasons I could not inject parts with different parameters I have gone as far as deforming parts (with engineering judgement) to obtain the required variation for the feature.


Many thanks for your professtional answers. You also have good knowledge of injection process. I fully agree with you, more and more I feel that GRR is like Emperor's new clothes.
 
T

tongxiaozhi

To be honest, for injeciton mould, we had to manipulate the data in order to make GRR and ndc compliant. The dimension of molded parts are quite close, even between cavities. I don't see much value in this tool.
 

GRP

Involved In Discussions
To be honest, for injeciton mould, we had to manipulate the data in order to make GRR and ndc compliant. The dimension of molded parts are quite close, even between cavities. I don't see much value in this tool.
For me the GRR is a necessary tool. There are some nice threads to read in the Cove about honest GRR and also Miner´s blog on MSA
 
T

tongxiaozhi

For me the GRR is a necessary tool. There are some nice threads to read in the Cove about honest GRR and also Miner´s blog on MSA

Thanks for your answer. My knowledge of GRR is quite limited although I have read the mannual of 4th edition several times. Probably I jumped to conclusions that it is not useful. I will try to find more informaiton. I would appreciate it if you could share more of knowedge with me.
 

GRP

Involved In Discussions
The reason why I say GRR is necessary can be supported with a quick example: once I had a medium-sized plastic part that required a measurement fixture. I had to prove capability to the client or measure 100%. The measurements were really off. So what was the problem, the part/process or the measurements? Long story short, the measurements. I had to chase the fixture supplier throughout 12 weeks of redesigns and tweaks until the measurements were OK. This definition could not have been made without a bias study and a GRR.

The GRR is just one of all the tools described in the MSA. You will find many useful threads in the Cove by searching for GRR, and also here https://elsmar.com/elsmarqualityforum/resources/miners-msa-blog-series.1/

Try to use software like Minitab and be wary of templates found around the net.

Stick to the ANOVA method in preference of X-bar and range.
 

Dave Cx

Involved In Discussions
Thanks for your answer. My knowledge of GRR is quite limited although I have read the mannual of 4th edition several times. Probably I jumped to conclusions that it is not useful. I will try to find more informaiton. I would appreciate it if you could share more of knowedge with me.
If you are not getting an acceptable NDC then trying using components with a greater variation, but this isn't so easy with mouldings. What I would suggest is keeping mouldings from the different tool tuning loops which you know have variation to get 'range'. Its only when you see the actual results of repeatedly measuring the same component that you can appreciate the MSA
 
Top Bottom