Recording Dimensional Data.. Why.. at times?

I have always wondered why certain aspects of quality.... especially when it pertains to dimensional inspection, has always been to document the actual dimensions... formally and sometimes informally to provide objectionable evidence even if there is no traceability of part and its dimension or no control over variables that can effect the actual dimension. I was wondering if you know of books/articles that deal with this and when its appropriate to document dimensions or not... We spend a huge amount of time and effort documenting dimensions when they are not even traceable back to the part once separated. I completely agree with documenting when doing statistics for trending... gage r&r.. capabilities.. etc. etc. but when this day and age of CNC capabilities are near perfect on proven programs/ tooling ,etc. and we are only running one and two parts (very low production) at anytime..... with no traceability requirements.... why would there be a need to record dimensional data.... ? and how would I explain this to management as to eliminate wasteful practices. We currently use stamps for acceptance/ rejection... so why would a work auth document with a stamp not satisfy objectionable evidence... It just seems there is a better way to perform metrology beyond writing down dimensions... would have been talked about by now. Considering all the variables with operators... equipment, environment, etc. etc. why would someone even attempt to use that data for a root cause investigation? With measurement uncertainty... it can be assumed the product was acceptable to a high degree of certainty but an actual number is not very helpful considering the low volume of production. It seems.... the days of manual machine created a need to document dimensions and the quality world/ environment has never moved away from that thinking even though technology has created many situations where it is not even needed.... just a CofC or stamp will do. look at CT scanning capabilities now days... don't even need to measure individual parts... can check an entire assembly to CAD with no need to actually document a dimension.

Just wondering if anyone... an article... etc has touch on this and gone into depth as to why certain situations need actual dimensions documented... what good they are... and the supporting requirements that make this all make sense. All dimensions inside of the budgeted tolerance should maintain fit, form and function. If a failure happens... it can not be a dimensional error... from inspection... maybe a design error in the dimension but if the product is in tolerance per print.. the root cause is in design... not in the actual measurement.

Thanks for any input....

Sorry for the all over the place..... there is just so much to this topic... and so many scenarios... hard to express the areas of concern without jumping around.

 

Even CNC capabilities need to be validated. While the software may execute perfectly, it may also perfectly execute a bad part if it was programmed in that way. I cannot tell you how many parts I have found to have a "mistake on the DXF". So at some point you can 'skip documenting inspection data' IF you are making parts in quantity, all the same off the same CNC program. At that point simple occasional SPC or sampling will suffice to verify everything is still on track. If there is no traceability of the part to the dimensions, this is essentially all that the data is used for. So the data you collect would be more to validate the process than the product.
When traceability is required, then the dimensional data does need to be documented, as in these products there is a concern of liability should something be defective. Even there is other attribute data needed, (function, fit, appearance), then records of this are needed as well, if for nothing else than to show the process was running as expected. The 'run data' of dimensional sampling would be necessary here as well (to prove process stability).
Much of this would be outside expectations also, the customer requirements, regulation, and QMS in place. All of these may mandate documented information of one sort or another.

 

This post really makes me stop and go "What???". So, you never have tooling off the CNC wear? Drill bits never dull? Maintenance needed on the tooling? Do your customers care about getting parts that fit/function the way they intend? What do you check when you uses your "acceptance/rejection" stamps? Do you just look at a part and say, "yes, there's the hole"? Dimensional data is not "objectionable" - it is the complete opposite. You don't have to record dimensional data for each part - we usually only do this at first article, but at a minimum you should have an attribute gage check. This is not a "old way of thinking - technology is infallible" argument. Tools wear, parts break - these are facts with any type of tool or method of manufacture. The only way to identify these changes over time is by recording the dimensional data. How do you know when the optimal time to do preventive maintenance on your tools is without documenting data over time?

Where is your evidence that this constitutes a "wasteful process"? Do your parts never get rejected by a customer? What is the argument behind "if a failure happens..it can not be a dimensional error..from inspection.. maybe a design error"? I know that I have asked a lot of questions, but I have seen way too much in tooling to place that much faith in technology over actual results that can be verified in measurement.

 

I think the OP is asking why (we) document the measurement data with no ability to backtrack the data to a specific part (serial number?) and how he/she can talk management into not having records created just to have records (since they don't do much good). I assume (evil word) that the OP records the dimension on a piece of paper/print but has no traceablility to what part was measured or when. If this is the case, I don't have a good answer for the OP since it depends on the company and the company culture.

As I was reading the OP's post, I was getting the impression that they were recording the dimensional data because they have 'always' recorded the dimensional data and he/she wants to find out how/if they can proceed in a different direction. If I mis-read the post, then I would ignore my above statement.

 

MMurphy... Exactly! We do dimensional check all parts.. but the documenting the data is not used in SPC... and since this is not volume production.. not much use for trending. The data is not traceable to the specific part.. Basically it is only used as objectionable evidence... and even the customer typically does not ask for dimensional data... only proof our process works and can catch deviation. Management/ engineers still cling to the notion that dimensions need to be recorded... just trying to explain/ prove how cost consuming this way of doing things are and there is better ways to provide objectionable evidence. As for physically checking dimensions.... yes we do... to a fault... do just critical but all. Engineering argues everything is critical and refuses to assign critical. I try to explain.. that yes... all features are critical but some just need to be verified they are there and not actually measured... like broke edges... non-critical chamfers.. large budgeted tolerances.. +- .010 and up... CNC are capable of producing this easily. Do you have to watch for wear of tooling... wrong setups.. operator error... yes but that is a combination of the machinist (quality starts there and not at the metrology lab) and quality sampling to catch deviation trends. Just would like a view point/ approach to explain this so that both management and engineers understand so a better way can be explored.

 

There is another view I believe that has not been investigated to length in today’s technology and abilities. Changing the way metrology in general is performed. Its seems all the information is already available to determine if dimensional inspection has occurred and if it was conforming or not without adding additional data input. Mainly data collection, in my view, is only adding to the cost of quality and not much value added to the overall objective to ensure conforming parts. Let’s go over what is already available... First.. the drawing/ CAD/ design characteristics are available to provide intent and acceptance/ rejection criteria. Inspection now knows what the requirements are and what constitutes what is not conforming. This has nothing to do with dimensionally inspecting product or not... The inspections should and will occur but is recording dimensions a necessity.. value added or just a time honored tradition? There is many that provide dimensional data with no measurement uncertainty statements... and probably, for the most part, they do not even know what their uncertainty is... from the operators... equipment or environment.... most do not perform this function because they deem it complicated... non-value added and very seldom requested by the customer (if they even understand the implications of measurement uncertainty). So now we are talking about data that could not possibly be used for reverse engineering or even an investigative root cause. Of course most measurements could never be repeated to precisely the same number over and over. From different operators, equipment, environment and just the order of operations can make the results subjective. Which is exactly why we have a "95% confidence" the product conforms... But this is another... long... discussion. Basically if the product conforms... all the information is already on the drawing. Any measurement that falls within the budgeted tolerance should perform as intended. If a failure happens... then the design must be reviewed... not the inspection. The inspection has already deemed the product was within the design budgeted tolerance. A Certificate of Conformance stating that the product conforms... has been dimensionally verified should suffice. If the product falls outside the budgeted tolerance... that gets documented as a non-conformance and becomes the means of trending, reporting and process capability. This is not the same as intentionally performing gage R&R.. or other SPC. Those can and should be performed but that method of acquisition should not be arbitrarily flowed down to day to day processes. Those activities are time consuming and for a purpose but does not need to be a part of the product acceptance during production. Secondly... Dimensional inspection is just a link in a chain to ensure product conformance. From the machinist/ quality verification and assembly.... dimensional acceptance is validated during this completed process. I cannot recall ever a failure... even catastrophic... that was attributed to a dimensional error from inspection... maybe dimensional from a design standpoint... but never because a dimension was determined to be conforming or not. This is not the same as accepting bad product or rejecting good. Talking about dimensionally acceptable to design characteristics which are already available on the drawing or documented as a non-conformance. Just like the shuttle explosion... I have heard too many say the o-ring... dimensionally.. was at fault... BS... the o-ring did exactly what it was designed to do. They hoped the o-ring would perform outside its design parameters. The o-ring was expected to fail outside the design temperatures and one engineer even tried to stop the launch for that very reason. The o-ring was the root cause of the explosion but not the failure. The failure was a result of taking the design past its intended range. I just think if you have acceptance/ rejection criteria already documented on a drawing/ CAD, etc... there is no reason to document an actual (subjective) if it falls within the acceptance criteria.. Just sign off the WAD.. or CofC... or even stamp the drawing. The non-conformance will be documented and really should be the only thing that needs to be documented. Creating dimensional data sheets is very time consuming/ expensive just to have objectionable evidence. A certificate of conformance should suffice... because if it does not... your basically saying I do not believe you. Which is ironic... since who is making sure the dimensions on a data sheet are legit. If one can falsify the CofC... then they are quite capable of the same on a data sheet.

 

I agree with Murphy entirely . And with you too, mcaffey. The question now appears to be - how to convince Management and the Engineering not to insist on old practice without any reasonable technological basis for this practice which is only adding to the cost. As for as the customer is concerned, if he is happy with your product, he would only like to see some objective evidence that you have your process under control. Your Engineering and the Management are quite obviously too cautious about any negative feedback from customer,and also not prepared to learn, whereas you yourself are far ahead of them in matters related to building, ensuring and verifying quality.

As you must be aware from your own experience ingrained attitudes do not change overnight. They need some education;therefore patience will be needed for sometime. So let us see how the following approach works.

Firstly you already have the practice of dimensional check of the first off produced to confirm the correct machine/process setting. In case the first off does not meet the specs you make appropriate changes and repeat checks till the set up is OK.It shouldn't be a problem keeping a record of the first off piece dimensions. Now you run the process to produce the entire batch.

I do not know the size of a typical batch of your product, nor how long it takes for a single batch to be produced from a given work center. But suppose it takes 6 to 8 hour run for producing the batch after the set up, you could take a piece around the middle of the run and measure and record the dimensions. Do the same for one sample from the completed batch also.

Next you can plot these 3 sets of dimensions on a simple chart, something like how you do for preparing a histogram, with upper and lower limits duly marked for each dimension. For this exercise use of an excel sheet may be a good idea. If possible you may use some temporary identification mark for each of these samples. i.e.3 per batch. If you do it for a week of 5 day working @ one batch per day. you would have at least 15 sets of reading marked sequentially against each dimensional parameter on the chart.

Are all reading within the specified tolerance? Good. You are now ready to show it to Engineering and convince them about the process being stable and producing the parts within the specified limits. If you take a little more trouble, you can use the historical data on readings of dimensions, plot it same in the same fashion in a different sheet and show it to Engineering how good the process control has been working.

But wait a bit. Here is a crucial point. In order to succeed you would need to combine the techniques of Quality masters like Deming & Juran, with the art of Dale Carnegie. You know what I mean. Give all credit to Engineering for their excellent and realistic tolerancing of each drawing dimension that has been a cornerstone of achieving effective process control, etc...They will love you for it and themselves take up with the Management the case for dropping 100% measuring and recording of dimensions of each product.

I have assumed here that you have had no glitches in production (e.g. need for resetting the process often) or customer complaints in recent past.
Sorry for the long post. Best of luck. I shall be happy to know the progress you make in this journey.

Regards.