What, if any, is today's "Industry standard" and/or "best practice" for creating 15

[Kjroonie]

Hello Colleagues ~ I work for a fairly large and diverse International company and have been “tasked” with implementing a “GLOBAL” 3D NX design standard to include the logic that “ALL parts are to be modeled to mid-range of tolerance”. Not a problem here in the USA, that’s how I was taught. However, it seems my European colleagues are NOT in agreement with this at all. Please help me to better understand their “rational” behind this. Thank you.

 

[TheTick]

What good is nominal? The designer wants everything in the middle. The mold maker wants everything steel safe (LMC). The machinist wants MMC. Sometimes the optimal is not the median, so the middle is meaningless.

RTFD, people!

 

[TWJR]

As one who programs CNC, sure, I'd like to see everything at the mean. However, I can't expect everyone to deviate from "normal" to make my life easier, so when I see unidirectional tolerances, I often have to "cheat" my tool path to achieve mean. That's just the way it is.

 

[3DDave]

Here's an interesting twist -

http://model-based-enterprise.org

which seems like just the place to get information, but it looks like something (or nothing) has happened to them. They had a fulls schedule for 2014, but no events at all on their 2015 calendar. The latest presentations were from mid-2014.

Things like this indicate to me that the cradle-to-grave concept is not yet a general purpose solution. The only people I've barely gotten to discuss their no-drawing operations seem to depend on previously negotiated expectations for results based on information that is usually depicted on the drawing, such as tolerances. This seems to happen between companies that have done similar work before, such as for mold suppliers.

I expect that unless it is a high volume exchange of very similar designs, the process has a lot of e-mails and group phone calls.

 

[capnhook]

Looks like model-based-enterprise.org gave up.

Stinkin' PHD's....

Proud Member of the Reality-Based Community..

[green]To the Toolmaker, your nice little cartoon drawing of your glass looks cool, but your solid model sucks. Do you want me to fix it, or are you going to take all week to get it back to me so I can get some work done?[/green]

 

[JNieman]

If I were to guess why they gave up, it would be something like the failure-loop described in an xkcd comic (sorry I am not linking xkcd directly, they are firewall blocked, here)

If there isn't already, there should be a law stating: "The longer a discussion continues, the probability of an xkcd comic becoming relevant approaches 1"

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[3DDave]

Thumbs up for XKCD, a comic and cartoonist who has a great track record; the optimistic version of Dilbert.

The model-based-enterprise group wasn't even working on a new standard; just trying to put other standards into practice - That's how bad the situation seems to be. It's as if, in general, MBE is toxic even to motivated people who are just trying to eliminate drawings, much less automate conversions based on context.

Ever see how an ant-lion works>? MBE is like that, and engineers are the ants. Management looks and sees that drawings cost money and so to save money says - let's get rid of drawings. Then they see that downstream users struggle with models and say - let's make engineers create universal models. Then costs go up, so they get rid of engineers to trim costs and bring in cheaper (newgrad or contactors) engineers who aren't as good, but may seem fast. Then no one can use the models. Meanwhile, although these models are the core of operation for the company, none of the managers can be bothered to learn the software to examine the models nor are they interested in taking several hours every week to review the models because that costs money.

 

[KENAT]

By the way Kjroonie Model Based Definition has been touched on many tmes before.

A quick search of 'Model Based Definition' in just this forum shows up numerous posts that may touch on this e.g.

thread1103-316370
thread1103-239768
thread1103-262305
thread1103-182500

http://www.eng-tips.com/search.cfm?...ITION&t=&h=&searchSite=1&searchArea=1&d=1&s=1

Posting guidelines faq731-376

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

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

 

[3DDave]

Kenat,

On "Current State of Model Based Definition," how did it turn out? It was sadly expected to see management inserting themselves in to a technical decision with little technical background. At this point it seems like a murder mystery with the last pages torn out.

The "Pulls hair out" is also interesting. If anyone wishes to follow this spark, please start a new thread, but whenever I gave a nominal value it was for the purpose of easiy comparison between features on other parts and not having to do with producing some 'target.' Essentially, the expected distribution was not Gaussian with the peak sitting on the nominal, but constant; every value in the range had the same probability of occurring.

So the +/+ and -/- make sense when it is more important to quickly realize the relationship between features on mating parts and not as a guide to a desired size distribution. For the purposes of MBD that would continue to be the case. While I didn't take it to an extreme, one might put dia 0.375 +.030/+.015 and make it clear that this is a clearance value of a dia 0.375 fastener rather than depending on knowing what the next larger nominal fractional drill size that was typically chosen to clear a 0.375 fastener. However in MBD that could be a problem as the acceptable range for the hole would not encompass the size of the surface representing that hole in the CAD model.

 

[JNieman]

There is no problem with a modeled feature size not being within the acceptable range (such as the hole measuring .3750 but the tolerance being +.032/+.016) - we do it all day every day and so do many of our customers.

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[KENAT]

3DDave, well the MBD thing went quiet except for complex castings and moldings etc.. It then reared its head again more recently. On our big glamor project (read resource sink) to speed things up they started sending lots of parts out without a drawing (or a very very minimal drawing + model) for prototype parts. As we move into production this is causing all kinds of issues at inspection...

I think I whined about this in another recent thread but maybe didn't accentuate the MBD portion.

http://www.eng-tips.com/viewthread.cfm?qid=385291

JNieman, I'm sorry but in some circumstances not having the model suite going straight into CAD/CAM can be perceived as a problem. For those of us in sectors where management is focused on chasing this quarters $ and maximizing share value....

Posting guidelines faq731-376

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

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

 

[JNieman]

We program our CNC machines straight from models. I reiterate - it isn't a problem. Possibly it should come with a disclaimer that it isn't a problem for people who can read a print.

However, there at least 3 places off the top of my head where a programmer or machinist can compensate/offset the cutting tool or path to account for a modeled feature outside the acceptable range. This isn't even an exceptional situation. This is just every day stuff for CAM software and machine programming. There's a reason "offsets" are a pretty bog-standard thing.

Even if your model is 100% to "median" you will still use offsets in your day-to-day programming when you have situations where you have a different toolpath for 'roughing' than 'finishing' as your roughing pass will want to stop short of the modeled position.

If someone is really looking into changing the whole way /intelligent/ models are being made, to try and make them idiot proof, then they need to stop hiring idiots. There is no gain I can think of in requiring models to be made to the -median-. The "problem" that policy "solves" is more elegantly, efficiently, and intelligently handled using settings in the CAM software. Whether or not the model is outside acceptable range or to the media, the programmer still has to go through the same number of steps to generate their outputs. The tolerance can dictate operation settings: tool selection, tool settings, cutting strategy, number of finishing passes, etc.

I don't know - I've been on every side of it. Design, Drafting, Programming, Quality... it's just never been a problem or a hurdle that caused preventable labor. I'm not saying it's a topic immune to "seagull manager input" but it's not a /real/ problem in my mind.

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[ctopher]

JNieman,
A star for you.

Chris, CSWA
SolidWorks '15
SolidWorks Legion

 

[JohnRBaker]

I always model to the 'nominal', or what I would expect if I could get exactly what I wanted. In that way the tolerances assigned represented what I was willing to accept, and for which I've accounted for, as the potential 'slop' in my design, so as to allow manufacturing enough leeway so that the parts can be produced at a reasonable cost. As mentioned in a previous response, modeling to any sort of 'mid-' or 'median' value, at least in my humble opinion, is shooting for what's potentially a 'moving target'.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.

 

[JNieman]

As a designer, you specify a tolerance zone and accept the fact that you cannot rightfully reject any part that comes within any of the in-tolerance numbers. But you will always be looking at the numbers and thinking "Dang it'd sure be nice if we held this one to the tighter end," but you know if you detailed it to that tight of a dimension, it'll drive the price of the part up, so you loosen it up within acceptable means, and deal with what you get.

As an inspector, no number is any better than any other number. There is good and bad, go or no-go, black and white.

As a button-pusher, they may shoot for the median to make their life easier, or shoot toward LMC so they don't have to change their tools as they wear, as often, so they can boost their production numbers.

As a machinist worth their salt, they will look at the purpose of the part and do their best to make the best part they can, including consideration for fits, datums, and other functional surfaces/components/holes.

Depends on what seat you're in.

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[JNieman]

To add, those are obviously simplifications of reality. I don't mean to marginalize or compartmentalize anyone's role beyond reason - it's purely a demonstrative description to convey my idea.

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[KENAT]

Enginerd9, acceptable tolerance range is acceptable tolerance range. This idea of specifying your ideal value with non symetric tolerances somehow guiding the machinist to be as close to your ideal value as possible has no traction in the standards, and little to none in reality of production as JNieman illustrates.

Posting guidelines faq731-376

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

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

 

[drawoh]

What about this interpretation:

You specify a non-symmetrical tolerance on a moving part so you can achieve a functional fit of some kind. Would you want the machinist to make it to median spec, and have a potential for premature wear on that surface? Or would it not be better to machine it to one side of the tolerance band which favors wear, thus longer service life?

Just throwing that out there. Maybe someone else in the thread beat me to it.

Your tolerance range is the range of dimensions you will accept from the fabricator. The shop's job is to do this as cheaply as possible. They will exploit your tolerances to accomplish this.

--
JHG

 

[drawoh]

Oops! I hit "Submit Post" way too soon. The message continues.

There are a bunch of scenarios here.

Possibly, the fabricator's process is marginally able to achieve your tolerance. When he gets inside it, he is done, and he is darn glad! The as fabricated dimension winds up where it winds up.

Example

Routinely, I specify holes as something like Ø5/4.2mm at zero positional error at MMC. Note how there is no possibility that I will get a Ø4.2mm[ ]hole. The fabricator will open the hole up enough to give himself a manageable positional tolerance. Let's pretend that the fabricator can do a Ø0.2 positional tolerance. That allows for Ø4.4mm hole. Is that perfect for me? Do I want the largest possible contact area for my fastener head, or do I want a bit of allowance in case the mating part is not to specification? Maybe the best thing is for neither of us to push any envelopes.

The fabricator's job is to manufacture pieces to my drawings. I do not send them my assembly drawings. Often, I don't want them to know how the parts will be used.

--
JHG

 

[JNieman]

@Enginerd9, are you talking about accepting parts out of tolerance on the MMC side knowing they'll wear within tolerance after a little run time?

If so.. yea, I've done that when they were close enough and knew we making the parts work as-is was cheaper/faster than remaking the part. There's the option for the manufacturer to ask engineering for a one-time allowable non-conformance or deviation or whatever you wish to call it.

_________________________________________
NX8.0, Solidworks 2014, AutoCAD, Enovia V5

 

[JohnRBaker]

Oops! I hit "Submit Post" way too soon. The message continues.

Please note that you can now edit (as well as delete), for a period of 24 hours, any of your OWN posts, so you could have gone back and added the additional text. After posting a response and you discover something that you wish to correct, just look for the 'Edit' and 'Delete' options in the lower Right corner of your post, as shown below:

Note that this is a new feature added to Eng-Tips in the past week or so. But note that there will be a 'paper trail' showing what it is that you edited so you can't use this tool to 'rewrite history' as it were.

John R. Baker, P.E.
Product 'Evangelist'
Product Engineering Software
Siemens PLM Software Inc.
Digital Factory
Cypress, CA
Siemens PLM:
UG/NX Museum:

To an Engineer, the glass is twice as big as it needs to be.