Choosing In-Process Inspection Characteristics

Q

QAPLL

Hi everybody:

I just joined and this is my first post; please forgive me if I do it wrong. My question is: does every characteristic in the part print need to be checked in the regular in-process inspections? To elaborate: I work in a job shop that makes metal formed parts. I was always taught that you should be able to "control" your process with around five characteristics. If you need more than that, you don't really understand your process. These five (or so) characteristics are what you routinely check and control chart. The reasoning being that it is better to check the "critical" characteristics more often than to try and do full dimensional layouts every hour or so. We do a full layout at set up. Some where I work insist on checking every characteristic every inspection, but this means many hours of production occur between checks. I hope I gave enough information for you folks to help. Thanks for your advice
 
I do the same in a similar environment - What I do minimizes the overall time and ends up by checking almost the entire part -
I conduct a First Piece Inspection at every department station. It starts with raw material incoming (stainless steel sheets), then I do one at the shear, waterjet or laser to verify those processes, then bending, then welding, then assembly, then finishing, etc. When done, the part is almost completely inspected, and as a first piece, this means not every part is done this way. We do repeat this every X amount of parts, depending on the run size.
In the end I am using a sampling inspection for the batch. The full inspection (every dimension and characteristic) is done with a FAI.
 

inspector625

Involved In Discussions
In our company we do a FAI at the start of each department process, ex. Heading dept. Then during the run we do a visual inspection at 500 pcs. At 1000 pcs we do an in-process inspection and all characteristics are inspected and documented on an in-process inspection form.
 
M

msec0990

It's been a few years since I worked in a shop manufacturing "commercial" products, but the inspection process was fairly straightforward there. Some parts had as few as three critical dimensions and I can't remember anything having more than a dozen. We did first piece/article inspections and from there we performed periodic floor inspections to verify operators were doing their own hourly inspections. Not very proactive. We did material testing similar to what was described above (raw material, plating thickness, etc.) We also performed inspections at each operation, but they were constantly moving toward cellular operations. So the floor layout changed. A lot. I could comment more, but then I'm afraid this would start to sound like a rant. :notme:

Mark
 
Last edited by a moderator:
Q

QAPLL

HI:

Thanks for the reply. When your operators did their hourly rounds, did they check every characteristic on the print, or just several chosen ones? We can have prints with over 24 characteristics on them, and it takes a while to check all of them.
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
The correct answer: it depends. For example, if you are machining, it is meaningful to measure the one tightest characteristic per tool path. In other areas, you amy need to check them all, simply because your process can randomly fail at any point in the process.
 
M

msec0990

To clarify, in the case I mentioned the operators were only responsible for checks on products they were running. The inspectors were essentially “policing” them with their rounds, although the inspectors did also perform some in-process testing. This “policing” isn’t something I agree with, but this is getting a bit off topic.

In this case, each check was a full inspection of every characteristic. I believe this was necessary because the processes were not very well controlled and the machine capabilities were not even known. Unlike what the OP described, these checks could be completed within a few minutes, and even the in-process testing could be performed relatively quickly. They were using control charts for performance checks such as drill testing, so there were some good elements of their system. But they could have been much better.

To Bob Doering’s point, without knowing exactly what it is you’re running, there is no definitive answer. But I do feel that many companies could benefit from understanding their processes better.


I hope that this post is more to the point :lol:


Mark
 

bobdoering

Stop X-bar/R Madness!!
Trusted Information Resource
Two points to consider:

A comprehensive quality program uses Process Failure Mode Effects Analysis, or PFMEA. It answers the Why? of doing inspection for any characteristic. These are the bad things that will happen if you don't inspect that feature. If there is no "bad thing" then the need to measure it is low.

As far a frequency, here is the risk: if you find a dimension out, you should sort through all of the bad product to the last good inspection. If you inspect every 2 hours, that is a much lesser issue than ever four hours, and so on. Balance this simple risk.
 
Top Bottom