Functional vs. Process Driven GD&T

[pmarc]

Hello,
This time a little bit less specific question:
I am looking for some good real life examples showing differences between function and process driven GD&T, and somehow proving that function-based dimensioning and tolerancing is indeed beneficial in comparison to the latter approach.

Did anyone ever experience a situation when for instance manufacturing dept. insisted on adjusting GD&T scheme to their needs (tolerance values, datum structures, etc.), but at the end of the day, after doing some stacks, it occured that tolerances were unnecessarily tighten from functional point of view?

Any input will be really valuable.
Thanks a lot.

 

[fsincox]

pmarc,
Simple, I see it all the time. The shop wants to hold stuff like mounting screw holes for mounting simple electrical cover plates to main datums systems on large caastings. This is so they do not have to establish the actual (functional) cast electrical box datums and can machine in one set up. In the "good old days', before CNC, they would have had a jig to locate the holes for the cover and had a real person drill them on a press. The parts were actually functionally better then, but more capital and labor intensive.
Frank

 

[drawoh]

pmarc,

GD&T should be process oriented to the extent that your part can be fabricated by the process you think they are going to use. If the process cannot achieve the tolerances, either you must loosen them somehow, or you must go to another process.

Otherwise, GD&T should be based on functional requirements.

Designers must have some understanding of what fabrication processes can achieve, and apply tolerances accordingly. If the tolerance stack does not work, there is re-design to be done.

JHG

 

[pmarc]

drawoh,
Let's say that designer has sufficient understanding of what fabrication processes can achieve, but, as Frank nicely described, the shop demands other than functional set of datum features in order to make their life easier during fabrication. So in consequence the designer is forced to assign non-functional datum features on a print. What then?

 

[powerhound]

The only scenario that I've been involved in was regarding hole sizes and how they were reported. Several years ago we had some discrepant parts returned to us from our customer. The issue was with hole sizes. These parts were checked entirely on a CMM and the report generated from that was the law. The hole sizes on the report were within tolerance but the interfacing part would not fit. What happened was that the CMM was reporting average hole sizes (which was not functional in this case) rather than the smallest and largest size of the holes (which was functional). If memory serves me correctly, the inspector said that that was how ISO reported diameters, as an average. He said the same thing went for datums. A datum plane was always the average of the high and low points of a surface, not the highest points.

I don't know much about ISO things and I don't even know if what this inspector told me was correct but it seems that using averages are inherently flawed. I don't even know if I'm really addressing your question.

Powerhound, GDTP S-0731
Engineering Technician
Inventor 2010
Mastercam X6
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[fsincox]

drawoh,
Theory is fine but it is the fact it is not followed in practice or doesn't apply to real world applications that is the whole problem I see with many of my day-to-day GD&T issues. I keep trying to stress it here amongst (sorry!) the elites. This is also why I now think the ISO’s method may be the best way, just hold their feet to the fire!
Frank

 

[fsincox]

pmarc,
The worst part is, all the time and money we spend blaming the casting vendor because his casting is not accurate enough when it was just fine before, it is our processing that has changed.
How many do you want?
Frank

 

[KENAT]

Just last week I had an email chain (purchasing as middle man) on this topic.

I'd actually used a roll pin in a part to help position/self fixture part during assembly. I'd made the hole the pin wend into the tertiary datum for related holes etc.

Well the vendor had instead used an edge near the hole as the datum and all features were shifted by .15 accordingly.

I pointed out that I'd dimensioned based on function. They replied back with the old truism that it was bast to go from a corner of the part.

Initially I resisted making change but then lost the will to care and made the surface they'd actually used the datum as the tolerances etc. were well within process capability & in the application an argument could be made that that surface was a functional datum (though in my opinion if was a weaker argument than using the hole).

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[drawoh]

drawoh,
Let's say that designer has sufficient understanding of what fabrication processes can achieve, but, as Frank nicely described, the shop demands other than functional set of datum features in order to make their life easier during fabrication. So in consequence the designer is forced to assign non-functional datum features on a print. What then?

I have never been in this argument.

If you prepare your drawings carefully, your datums will work. For example, if you are designing a machined casting, you should specify datum targets, so that everyone works from the same set of datums.

When I was trained on ASME Y14.5M-1994, I was told that the datums specify fixturing for both fabrication and inspection. My final cop-out would be to inspect the part using the official datums. If it passes, it is okay, and I do not care how the fabricator did it.

Conceivably, a fabricator could use your fabrication drawings that specify the part, and create his own fabrication drawings. He could use his own datums that suit his fabrication process, tightening tolerances to make your official datums work. If your parts are good, you should be happy.

JHG

 

[fsincox]

Ken,
I am assuming you meant "best"?
I remember those days, "0"-lines on the edges, inspection determined the order of precedence. We are not required to go quite that far, I am sure there are many who still wish be did, though. Definately, no envelope principle here, though!
Frank

 

[KENAT]

Sorry, yes best.

Posting guidelines faq731-376

http://eng-tips.com/market.cfm?

(probably not aimed specifically at you)
What is Engineering anyway: faq1088-1484​

 

[powerhound]

drawoh,

Your last paragraph is a very common scenario. Your last sentence is spot on.

Powerhound, GDTP S-0731
Engineering Technician
Inventor 2010
Mastercam X6
Smartcam 11.1
SSG, U.S. Army
Taji, Iraq OIF II

 

[pmarc]

Thank you guys for all the comments given so far.
I was somehow hoping to hear that process driven selection of datum features is quite common practice in the industry. Your replies seem to confirm my experience with similar situations.

There is an example in A.Krulikowski's "Advanced Concepts of GD&T" (7th edition, pages 13-6 & 13-7) explaining a disadvantage of such datum structure conversion. Though I do not think the problem is always such black and white, in my opinion the example quite nicely emphasizes the issue. I attached these two pages below.

http://files.engineering.com/getfil...556fb61ee6b2&file=funtion_vs_process_GD&T.pdf

 

[CheckerHater]

What is it with “larger-than-life” questions?

In a perfect world design engineer should possess significant amount of manufacturing as well as quality control related knowledge. In today’s world of $10/hr CAD-monkeys – don’t get me started.
So in reality we end up with some sort of compromise.

Example one: small shop to small shop. GD&T is not used. Shop is trying hard with some amount of trial and error. Once they find “the sweet spot” they can run good parts forever. (Until new management shows up at either company and starts improving things.

Example two: manufacturing starts yelling “we cannot do that” even before the last print is unrolled. They are trying to make their life easier even before job started. Sometimes you may be surprised that after loosening all the tolerances your machine still works (which may prove that you didn’t really know what functional requirements are)

Example three: the product is really big, complicated and important. You will never get it right from the first time. On prototype stage, when trying to assemble it, you figure out REAL functional requirements. On preliminary production run shop starts realizing what is holding them back, and comes with some feedback. By the time of full-scale production your drawings start taking decent shape.

So, I don’t think you will find definitive answer in the real world.

 

[fsincox]

CH & Ken,
At least you guys work in the same world I do! I am glad to know I am not alone out here.
Frank

 

[fsincox]

How about this one, The Company I work for makes mostly small parts (fit in your hand). They also can apparently live with +/-.005 on three place dimensions for the most part (except for all the parts they bring to us for review EVERY DAY). I worked for a different division that made bigger parts and always specified tolerances of drilled holes based on the Metal Cutting Tool Institute recommended tolerances for drilled holes, which MCTI data shows tend to average oversize. ISO and ASME recommend something like a H13 tolerance or so (no minus). When someone came in and trained them in GD&T years ago they definitely got the message: “do not specify DRILL anymore”. However it seems they were content to live with +/-.005 for drilled holes as in the small sizes they use it covers it. Location tolerance calculations are a mess because they do not use realistic size tolerances on the hole size and therefore, waste location tolerance on size. What they really do is use the “+/- to position” chart they got when they were trained to convert from +/-.005 to .014 and ignore the rest, who need logic and math anyway, I have a chart that says..!
Frank

 

[CheckerHater]

I will try to throw in few more examples to support my case.

Look at this 30-something years old book intended for draftsmen, not even engineers.

The paragraph 10 “Shop processes” stretches from page 267 to page 297, that’s 30 pages.

The paragraph 12 “Tolerancing” goes from page 335 to page 355, that’s only 20 pages.

As you can see both are getting at least equal attention (at least back in time when drafting was respectable profession).

 

[CheckerHater]

Sorry, forgot the picture.
There is never time to do it right, but always time to do it over.

 

[fsincox]

CH,
Can you, please, give me that new ISO book title again?
Frank

 

[CheckerHater]

Frank,
The ISO Geometrical Product Specifications – Find your way in GPS (378 pages, B5 format, English only, ISBN 978-87-7310-721-8)
I am afraid you will wait forever for it to appear on Amazon.
I ordered copy directly from ISO. It is not difficult; all you need is a credit card. The shipping cost hurts (a little).
The more I read it, the more I think it was worth it.
By the way, the guy is not big fan of ISO 2768. Go figure…