Evaluation methodology: Process control vs Function control

[[Ri...]]

Process control vs Function control

As I go through the cookbook and read more and more regarding what choices to make regarding various evaluation methods the old machinist in me can't help but be somewhat confused as to the justifications, or lack thereof provided for the methodologies shown. Let's pretend we are machinists,and as such we are making a functional assembly. (A simple shaft and a block with a mating bore.)

Logic tells me that we should employ "Function Checks" as our methodology:

"Minimum Circumscribed" for our shaft
"Maximum Inscribed" for the mating bore.

This will yield us real world numbers that translate back to reality. In this way if calypso states that the pins and mating bores are both within tolerance, those the pins will fit in the mating bores back here in reality.

--

With process Evaluation however, I am unable to come up with any rational justification for its use. (At least with my limited understanding of it.)

How does "Process Evaluation" help that same machinist make consistent, FUNCTIONAL parts when it's not really a reflection of what he/she is holding in their hand at the end of the run?

If the evaluation method artificially alters the interpretation of your data in ways that won't necessarily translate back to the parts actual size/shape, how does that help to provide real world numbers one can apply as offsets to achieve and maintain nominals through a monitored production run?

[I]Basically, wouldn't you ALWAYS want to measure and monitor features in accordance with their final useage for functionally?![/I]

I'm genuinely confused as to why anyone would ever use these "Process Methods" as explained in the cookbook.

If anyone help me to understand the other side of this coin or can point me to some literature that will clarify this it would be GREATLY appreciated!

Thanks all

[[No...]]

Uhm, I never payed much attention to the cookbook, mostly because it's not made for plastic parts, but also because I personally don't like its "recipe" structure.
So I don't really know what "process" methods you're referring to. I basically agree with you that functional checks should have priority. But sometimes these methods are not suitable for (statistical) process control. Maximum inscribed and minimum circumscribed for example are extremely susceptible for outliers unless you perform heavy filtering and outlier elimination, and you don't like that kind of results in statistical evaluations.