i recently got a job as a quality engineer in a machine shop that fabricates metal parts using CNCs, depending on the customer specs. I am tasked in designing a quality management system from total scratch, for an upcoming ISO 9001certification..
I wonder if you could help me out in the inspection process. since i always go for simplicity, i had chosen FINAL INSPECTION to be the last process prior to shipment. of course there are in-process inspections (done informally by operators, checking dimension after milling,etc.)..
> how can i design an inspection procedure (FINAL) that will not be a hassle to process itself (since some parts have complex forms such that 100% is entirely time-consuming) as well as provide enough confidence in the inspection result...
>> what steps will i take, incase out of sampling i am able to detect a non-conforming product. do i need to inspect all the finish goods?
Thank you.

>> what steps will i take, in case out of sampling i am able to detect a non-conforming product. do i need to inspect all the finish goods?

Sampling is sampling, and you can not eliminate sampling error. But, final inspection can really only be sampling designed to pick out the higher probability problems. 100% inspection is a phenomenal increase in overhead that is difficult to justify - both in labor and equipment costs. The best way to ensure good product is proper control in the process. The information in the process is far more valuable - especially for machining. Predictive decisions come from sequential data - and that is lost in final inspection.

My suggestion is to look through Statistical process control for precision machining (http://elsmar.com/Forums/blog.php?b=79) to get a good grasp on the process itself.

i> how can i design an inspection procedure (FINAL) that will not be a hassle to process itself (since some parts have complex forms such that 100% is entirely time-consuming) as well as provide enough confidence in the inspection result....

100% inspection is often very wasteful. I would look at implementing a sampling inspection program at final. My preference is always a C=O plan.

>> what steps will i take, incase out of sampling i am able to detect a non-conforming product. do i need to inspect all the finish goods?
Thank you.

A C=0 plan requires you to reject the lot and sort out the defects.

Hope this helps a bit.

i recently got a job as a quality engineer in a machine shop that fabricates metal parts using CNCs, depending on the customer specs. I am tasked in designing a quality management system from total scratch, for an upcoming ISO 9001certification..
I wonder if you could help me out in the inspection process. since i always go for simplicity, i had chosen FINAL INSPECTION to be the last process prior to shipment. of course there are in-process inspections (done informally by operators, checking dimension after milling,etc.)..
> how can i design an inspection procedure (FINAL) that will not be a hassle to process itself (since some parts have complex forms such that 100% is entirely time-consuming) as well as provide enough confidence in the inspection result...
>> what steps will i take, in case out of sampling i am able to detect a non-conforming product. do i need to inspect all the finish goods?
Thank you.


I am curious how do you define "final inspection"? I believe that will drive the extent of inspection needed.

Do you have a First Article Process to qualify the process?

Do you have First Piece Inspection program in place?

Do you have a in process inspection program in place?

Answers to this questions may lead other to provide more help!

Good Luck going forward!

With regards to "hassle", I am not sure how you can design an effective inspection system that does not appear to be a hassle (burden) to the work force. IMHO, the goal should be a cost effective program that provides the assurance the customer requirements are met nothing more, noting less.

> how can i design an inspection procedure (FINAL) that will not be a hassle to process itself (since some parts have complex forms such that 100% is entirely time-consuming) as well as provide enough confidence in the inspection result...


I work in the same environment you do and we have recently revised our QMS as part of our process to become ISO certified. We have three levels of inspection at the Quality Manager level: 1. First Article Inspection 2. In-Process Inspection 3. Final Inspection. After the FA (first artcle) everything is inspected on a lot sample basis. We will sometimes require the operator to document measurements through out the production run. These are reviewed by the QM.

>> what steps will i take, incase out of sampling i am able to detect a non-conforming product. do i need to inspect all the finish goods?
Thank you.

We use ANSI/ASQC Z1.4 for lot sampling.

With regards to "hassle", I am not sure how you can design an effective inspection system that does not appear to be a hassle (burden) to the work force. IMHO, the goal should be a cost effective program that provides the assurance the customer requirements are met nothing more, nothing less.

Good point - that is a good difference between inspection and control. Inspection tells you if you made good parts - not very interesting, feels like a hassle. Control gives feedback that the process needs adjustment (when done correctly, before any bad parts are made.) To a good operator, that should be valuable information of ongoing interest.

Some good answers so far.

From the description given in the first post, we are dealing with a contract machining environment. Almost by definition, the use of CNC machining and turning centers describes a precision machining environment.

In a precision machining environment, it is not a good practice for in-process inspection to be informal, rather, it should be part of a well-thought out Control Plan for each different product, taking into consideration such things as critical characteristic dictated by the customer and machining "checkpoints" before a final dimension.

Following are links to posts about some practices I deployed in a precision machining environment - perhaps something similar will work for your organization.

Wes Bucey on an efficient shop - empowerment (This is a single post (#18) in a longer thread - the url leads directly to the post - it is associated with a follow-up in post #20)
http://elsmar.com/Forums/showpost.php?p=105469&postcount=18 (http://elsmar.com/Forums/showpost.php?p=105469&postcount=18)
Wes Bucey on quoting and empowerment (This is a single post (#20) in a longer thread - the url leads directly to the post)
http://elsmar.com/Forums/showpost.php?p=105566&postcount=20 (http://elsmar.com/Forums/showpost.php?p=105566&postcount=20)

In process and final inspection:
Re: Inspection Dimensional Check Sheets - Over 500 part numbers (http://elsmar.com/Forums/showthread.php?p=274032&highlight=redundant+first+article#post274032)

Who inspects?
I, too, ran a shop where primary responsibility... (http://elsmar.com/Forums/showthread.php?p=150031&highlight=redundant+first+article#post150031)

Control Charts
In my high tech machining business (1990 -2000),... (http://elsmar.com/Forums/showthread.php?p=97017&highlight=redundant+first+article#post97017)

What makes sense?
In point of fact, I have seen several operations... (http://elsmar.com/Forums/showthread.php?p=85248&highlight=redundant+first+article#post85248)

When I worked in a machine shop, our CNC machinists checked the tight tolerance features of their parts frequently based on tool wear, and variation introduced by the environment: temperature and humidity. We could never say we achieved process control, so sampling per ANSI would not work, and if too many parts went by without a check he was apt to lose some.

So, as Wes said a customized inspectionplan could be needed to address the likelihood of certain characteristics going out of tolerance and why.

When I worked in a machine shop, our CNC machinists checked the tight tolerance features of their parts frequently based on tool wear, and variation introduced by the environment: temperature and humidity. We could never say we achieved process control, so sampling per ANSI would not work, and if too many parts went by without a check he was apt to lose some.

That is often the case with traditional attempts at CNC controls, but process control in CNC is very possible when done correctly. The real problem in machining is the sorts after a broken tool. Often final inspection sampling will pick up the results of ineffective sorts for that condition - because the sorts are just that bad.

As was mentioned in one of the threads, sampling is god for finding very bad conditions. For very good conditions, sampling will not pick up the rare problem. And, it falls into "inspecting in quality" approach. That is why control and capability are the goal.

:agree1: That's a good follow up comment Bob, I do not think I could have said it so well.

In our case the recurring problem was a complete lack of environmental control. I left that company some time ago - they might have air conditioned the space by now, as I proposed.

In our case the recurring problem was a complete lack of environmental control. I left that company some time ago - they might have air conditioned the space by now, as I proposed.

That is very bad when machining aluminum! :frust: Also have to track machine warmup more carefully for some materials and machines than the ongoing checks after warmup. It is a special cause!

That is very bad when machining aluminum! :frust: Also have to track machine warmup more carefully for some materials and machines than the ongoing checks after warmup. It is a special cause!We were machining steel for the most part; so the issue from temperature/humidity variation was usually size, but chatter on sealing surfaces did crop up too, and was being frequently inspected for.

Be warned Sativ, all of these types of issues need to be considered when designing an inspection plan.

In a precision machining environment, it is not a good practice for in-process inspection to be informal, rather, it should be part of a well-thought out Control Plan for each different product, taking into consideration such things as critical characteristic dictated by the customer and machining "checkpoints" before a final dimension.
[/FONT]

True. :agree1: The best control is to at least SPC the tightest tolerance for each finishing tool. SPC some of those 'machining checkpoint' interim dimensions (e.g. roughing for a tight finishing operation or blanking for threads) can also be very beneficial in minimizing adjustments downstream. The criticality of the dimension really has little to do with the control - although they tend to be the tightest tolerance characteristics.

Control Charts
In my high tech machining business (1990 -2000),...
Good ideas for the time! The only exception now would be to go back to paper charts and use X hi/lo-R charts. Not likely available on the software SPC systems. Also, no need to calculate cpk. Capability in a properly controlled precision machining process is a constant.

We were machining steel for the most part; so the issue from temperature/humidity variation was usually size, but chatter on sealing surfaces did crop up too, and was being frequently inspected for.


Good point - surface finish can trump a lot of dimensional controls.

I am curious how do you define "final inspection"? I believe that will drive the extent of inspection needed.

Do you have a First Article Process to qualify the process?

Do you have First Piece Inspection program in place?

Do you have a in process inspection program in place?



>> Can I have a First Article Inspection only if required by customer? can i state it in my manufacturing procedure?

>> For the In-process inspection, currently is done depending on the operator's discretion.

Thanks for spending some time on this guys, i really appreciate it!

Btw, my previous job was a process engineer (look for ways to improve a product), and not focused on quality assurance inspection design...

>> Can I have a First Article Inspection only if required by customer? can i state it in my manufacturing procedure?

How do you qualify new parts if you don't do a first article inspection?

For the In-process inspection, currently is done depending on the operator's discretion.


How qualified are those operators to make that judgement? I would try to standardize the inprocess inspection. You may want to start by interviewing the operators to see what they use for their "discretion". Then find a way to apply that to all processes.

I agree with Carol, but I want to add that my insights are drawn from my experience in a job shop. That means:

1) Jobs and their specifications will vary, so their inspection needs will too.
2) Equipment was not the same - machines had unique quirks that also affected inspection design.
3) The uncontrolled heat and humidity added further uncertainty to the CNC process.

None of our inspections were set up exactly the same among all the types of parts we ran. Furthermore, even the same type of part might have different inspection needs if, as I said earlier, variation is being introduced via man, material, machine and/or environment.

The machinists were familiar enough with their produts and machines that they knew they needed to inspect at given intervals; they could sometimes stretch out the intervals, and sometimes had to shorten the intervals. Be very cautious, at least at first, to understand how well your machinists understand and apply these principles.

All of that is moot if the processes are in control and all the machinist does is put the part in, push a button and remove the part.

There - have I made that complicated and confusing enough? :bonk:

Thank you for sharing me some opinions.
>> What I mean regarding the FA inspection is the documentation (to be submitted to the customer) necessary, even if they dont ask?
>> My procedure in standardizing the INPROCESS inspection, will be the following:
i. inspection for the first part produced by a specific machine/tool replacement
ii. inspection for the last part produced by a specific machine or before tool
replacement
iii. inspection every "X" times(still to determine "X" depending on machine and
quantity to be produced)
iv. inspection on the previous part of which non-conformity was detected.

with all these, is it necessary to introduce another FORM for operator to jot down the reading results, or a simple checkbox will do in the router (to make things more simplier) to confirm that they have done ALL the inspection themselves (including the disposition for non-conforming parts)? ---- NOTE: THIS IS FOR INPROCESS INSPECTION ONLY

Thanks.

Thank you for sharing me some opinions.
>> What I mean regarding the FA inspection is the documentation (to be submitted to the customer) necessary, even if they dont ask?
Yep. If the customer wants it, just give him a photocopy (or, as I used to do, give each customer access to our network (password protected) to view the data in real time - saves forests in avoiding multiple bales of paper.) The point is you really want to ensure supplier and customer agree on product so supplier doesn't create a pile of scrap the customer will not accept.
>> My procedure in standardizing the INPROCESS inspection, will be the following:
i. inspection for the first part produced by a specific machine/tool replacement
ii. inspection for the last part produced by a specific machine or before tool
replacement
iii. inspection every "X" times(still to determine "X" depending on machine and
quantity to be produced)
iv. inspection on the previous part of which non-conformity was detected.

with all these, is it necessary to introduce another FORM for operator to jot down the reading results, or a simple checkbox will do in the router (to make things more simplier) to confirm that they have done ALL the inspection themselves (including the disposition for non-conforming parts)? ---- NOTE: THIS IS FOR INPROCESS INSPECTION ONLY

Thanks.You are ignoring the concept of in process inspection which requires taking a number of consecutive samples to detect any trends in variation. All this is predicated, of course, upon having large numbers of product in production versus small lots. Checkboxes are for attributes - if you want to check variation, you need metrics - actual dimensions, not just pass/fail.

Do you have operator empowerment to stop the process if actual nonconformance is detected (to implement corrective/preventive action)?FWIW:
In our high tech contract machining business (all custom work for other manufacturers), each part had a Control Plan.

In addition to all the other stuff in a Control Plan (agreed to in advance by each customer), we included an Inspection Process specific to the individual part. At first we did hand measurement and recording - later, automated measuring and direct input of readings to a computer.
Among other things, the Inspection Process included:


an engineering drawing of the part with each dimension to be inspected tagged with a number (in the order in which the dimension was to be inspected.)
A part-specific inspection check list with each dimension numbered in the order it was to be inspected, listing the target dimension with tolerance (column 1), together with the type of instrument (column 2), and the actual dimension measured (column 3), and a 4th column for a redundant check of the same sample by another inspector.
The measured sample was tagged and traveled with the rest of the documentation in the order traveler (the document package which traveled with each step in processing to final shipment.)

We specifically designed each part-specific inspection process for the most efficient use of instruments and operator time. The numbering directed a consistent measuring process on each sample, designed to identify most likely N/C (from the FMEA) first, before wasting time on subsequent measures. Often different parts of the same part had the same dimension, but had been produced in a separate operation and thus had to be measured separately and DISTINCTLY (to identify which process went with which dimension as an aid in problem solving in the event of an N/C.)

Different inspection sheets were created for in-process SPC charting and for both First Article and Final Inspection. Above all, we wanted to eliminate opportunities for error and to make the inspection processes as efficient and capable of being replicated as possible - either by our own in-house folks doing redundant inspections or by customers doing receiving inspection.

The identity of the type of instrument used on each dimension was important. We had a chart to identify each instrument used in an inspection of a sample by serial number, thus the instrument itself could be recalled if there was a suspicion a faulty instrument out of calibration was the cause of any N/C reading.

Anybody and everybody in our organization was empowered to halt manufacture if a N/C were suspected, calling in as much help as necessary to affirm or deny the suspicion before the manufacture was restarted.

Elsewhere in the Cove, over the years, I have described how our operation evolved from traditional manufacture and inspection departments to the point where manufacturing folk performed the routine inspections and quality department folk acted as designers of each part's inspection process in collaboration with each customer, then would act as trainers for the manufacturing folk to perform those inspections, and act as arbiters if a redundant inspection was required.

Everything was slanted to make every part of the manufacturing process flow as smoothly and efficiently as possible, removing bottlenecks and choke points along the way, making liberal use of mistake proofing concepts on every aspect of our operation.

If it's important, we could probably find links to those posts.

You are ignoring the concept of in process inspection which requires taking a number of consecutive samples to detect any trends in variation. All this is predicated, of course, upon having large numbers of product in production versus small lots.
I apologize if I havent truly describe to you our operation. Our customers are engaged in automation, therefore, orders arrive by many design types (unique designs or modification of previous designs, but generally speaking, different designs) of which we are to produce 1~20 (average of 4) units per design. I would like to have the challenge of large quantities of the same type being produced to get an accurate trending (maybe later on other customers, when we will be certified).

Do you have operator empowerment to stop the process if actual nonconformance is detected (to implement corrective/preventive action)?
Yes. With that, they would either re-work the part or discard it. Either way, immediate adjustments and tweaking is done upon detection. I just dont know the details on the tweakings, and etc., but i suggest to them to inspect the dimension of the finished part after every repairs and adjustments..

Thanks!!!

I have a hard time justifying a lot of time spent on a final inspection process. It is very expensive to do 100% inspection or even sampling. I have yet to see anybody who can inspect quality into the parts they are looking at. It is the control of your process that you are looking for. If you need final inspection for everything you do then your processes are out of control. Until your processes are in control you are going to waste a lot of money doing inspection after the fact. Ideally we make the parts right the first time with the operators who are runing the machines being both responsible for them as well as empowered to do the job properly.
NO ONE CAN INSPECT QUALITY INTO A PART!!!!!!!!!