CNC Process Validation Cpk Ppk & In-Process Inspection

[Validation2011]

Hi all,

New to the forum!

Hopefully you can help.

Currently validating a CNC process. Have competed prelimary capability studies based on 3 batches of 50 parts. Have got good cpk and ppk vales for the process (all above 2.57).

We are hoping to start validation of the process next week. All protocols are compete.

Problem is that we have in process inspection sheets (1 in 30 parts), however once the process is up and running we hope to run unmanned for up to 3 hours (100 parts).

I presume we need to validate based on an inspection frequency of 100 parts ? But we are currently not sure that we will be able to run for this period of time unmanned (based on 50 parts it looks good).

My question, Can we validate the process and ignore the in process inspection data for the moment? And when we do run the 3 batches of 250 parts for validation and prove capability over X number of part finalise our in process inspection sheets at this point instead?

Basically, will finalising our in process inspection sheets after validation as opposed to before validation void our validation effort?

Hope you can help!!
Many thanks
Validation2011

 

[bobdoering]

Re: Process Validation cpk ppk & in process inspection

Are you doing CNC machining? What are the tolerances? Is it turning or milling? I suggest reading this link before going too far down the wrong road.

It is possible to tell how long a CNC machining process can go without adjustment, and therefore pick a sane measurement frequency - but it requires performing a correct study. I am afraid you did a traditional study, and you will not get to the right process control doing that if you are doing precision machining.

 

[Validation2011]

Thanks for that.

Can you advise if it is possible to do x hi/lo-r charts in minitab.

 

[bobdoering]

Thanks for that.

Can you advise if it is possible to do x hi/lo-r charts in minitab.

I do not believe they have caught up with that technology, yet. There are some blank Excel charts in the Cove.

 

[Validation2011]

Thanks forthe help

 

[bobdoering]

Please feel free to ask for any help with your implementation!

 

[erinpristel]

Trying to implement X hi/ Low R charting for our small machine shop. (4 machinists). We are currently AS9100C Certified and I need to know if this is going to be acceptable for an audit in place of traditional SPC - of which we all know is of little value. But thinking to myself .........AS9100 only specifies that we use statistically sound practices if I am correct? Please let me know if I am wrong. Also another issue; determining the sample frequency as to when to measure and record to collect enough data for it to be reliable. How do I do this using the data I collect until I can get a grip on the process using the charts?

 

[bobdoering]

Trying to implement X hi/ Low R charting for our small machine shop. (4 machinists). We are currently AS9100C Certified and I need to know if this is going to be acceptable for an audit in place of traditional SPC - of which we all know is of little value. But thinking to myself .........AS9100 only specifies that we use statistically sound practices if I am correct?

I would hope that would be the case. Otherwise, the AS9100C certification would not be of much value - forcing one to use statistically invalid process control.

Also another issue; determining the sample frequency as to when to measure and record to collect enough data for it to be reliable. How do I do this using the data I collect until I can get a grip on the process using the charts?

Here is the key. You should have at least 7 data point between adjustments. There should be no adjustment until you hit a control limit (75% of the tolerance). The only thing that would trump that is surface finish (if your finish exceeds the surface finish callout before you hit the control limit, then you need to further tighten your control limits). The other trump is the range chart - if you see a sudden jump in the range chart, then your tool is too worn and is approaching failure. That tool may not last until you hit the X hi/lo control limits - in which case they will need tightened.