Design vs quality/production department anyone? 2

[Feynman81]

Hello everybody,
I'm a mechanical design engineer in a aerospace company in Europe. First post here, hope everybody is well during this uncertain times.

Well, as the post subject suggest I'm having some issues in my company with other departments (WHHAAAAAAAT?!?!? REALLY??!?!? NEVER HEARD OF THAT!!!).

The problems are of course related to tolerance(and more), few examples:
- we receive some parts from a supplier and it is not up to spec. Quality/production in conjuction with the supplier of the part says that tolerances are too tight (drawing is wrong) and must be modified so we can accept the parts.
It usually goes "we need this part right now! It's in our interest to accept that part but you need to change the drawing for that".
Well the problem is that the supplier oversold its capacity at the beginning and he said that could keep those tolerances "no problem", had the drawings in his hands for months and now (when the parts are being checked) he's saying that we cannot pretend such tight tolerances. I trusted him and did not ask to put it in writing so there's no paper trail.
We already changed the drawings a couple of times (enlarged holes to relax hole location tolerance or changed DATUMS) and it didn't work, there's always something wrong with the holes pattern location.
The problem is not even this particular supplier but it is a constant even with internal production. It seems that this is the "modus operandi" in this company, we release the drawings, nobody says anything for months and then when QC check the parts and they are out of spec ----> "the drawings are wrong", "we must change the drawings so we can accept the parts".

- Quality doesn't seem to follow best practice or norms in checking the parts. Typical example: when checking hole location with the CMM they use the edges of the part as the zero. I told them multiple times that's not at all best practice because the edges might not be straight or perpendicular to each other and they have to use a metrology square or other metrology tools to "simulate" the DATUMS but they simply say "that's all theoretical, the edge is straight". , to me it doesn' make any sense, and the theoretical thing is that the edge is straight..... Am I wrong on this? We receive several custom made parts (heat exchangers, fans etc..) for hundreds of thousands of dollars but QC doesn't care about ASME/ISO standard clearly stated on the drawing by the manufacturer.

After a recent exchange with them where they tried to apply the general tolerance to a basic dimension (?) I've come to the conclusion that they have no idea about modern GD&T and all the nice CMMs and tools we have are just the proverbial lipstick on a pig. To compound this impression they always ask me to use the plus or minus method to locate holes and other feature instead of the GD&T way, recently they increasingly ask drawing modification to put as many dimensions "as reference" as possible because "you don't need to check them".

When I confront them trying to explain all the above they became aggressive and basically say that I cannot teach them how to work.

Has anybody had a similar experience? I'm thinking going to upper management to present the situation, I don't think they are fully aware. I've been in this company a couple of years now and my impression is that they are on a downward trajectory regarding all the technical aspects.

Thanks in advance.

 

[axym]

Feynman81,

Have I had a similar experience? Well, put it this way - my entire business is based on troubleshooting GD&T issues like these. I try to find the root cause in terms of problems with one (or more) of the following three things:
-The parts
-The measurement
-The drawing (how the GD&T was specified)

It's taken me many years to learn to do this in an effective and convincing way. It required very deep dives into GD&T definitions from the ASME standards, and how CMM inspection doesn't always follow those definitions. I had to assemble quite an arsenal of specialized software tools as well.

Over the years, I've found that it's usually not a problem with the parts ;^).

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[greenimi]

Feynman81,

I can ensure you that that's happening everywhere (should I say unfortunately).
Meaning here supplier- customer disputes....

Also axym (Evan) is one of the best members here to assess the drawing, if you are willing to post it (or at least some portion).

 

[3DDave]

Yes - I had a supplier ask to remove all tolerances from our interface control drawing.

They soon went bankrupt, if that's any sign. Best of luck on your job search.

 

[TheTick]

Why is there such a disparity between process/vendor capability and tolerance requirements?

Whatever the problem is, it started inside your walls. You can't perform neorosurgery with a pickaxe.

SHOP: "All we got is a hammer!"
ENGINEERING: "Wow! It certainly is going to be difficult drilling holes with that thing. Maybe if we put a tight tolerance on it that will help. PUT YOUR BEST HAMMER GUY ON THE JOB!"

Why are you making tolerances without consideration of process capability? Precision costs money. Don't ask for more than you need. If you need more, you must pay for it.

 

[MintJulep]

I'm thinking going to upper management to present the situation, I don't think they are fully aware.

I suggest that you do not do this.

If you do, all that will happen is that you will become "the problem" in management's eyes.

Put your effort into making your resume better, and learning all that you can in your remaining time with this failed company.

Good luck!

 

[TheTick]

I'm thinking going to upper management to present the situation, I don't think they are fully aware.

You don't think so? Sounds like you are about to get "educated".

Tick's Rule #0 of Business Dynamics
Bull$#!+ persists because it is allowed.
[sub](It's Rule #0 because it is in effect before you even consider what Rule #1 might be.)[/sub]

 

[Belanger]

Instead of telling management that there is a problem, simply ask for some training to be brought in on the topic of GD&T. That way you can obliquely get your concerns addressed.
I've conducted countless classes where we go through the non-GD&T way of doing things and show why that leads to problems. Hopefully all involved will listen to the instructor as an impartial voice. If not, then instructor doesn't stand to lose anything other than a receptive audience.

 

[jassco]

Hi, Feynman81:

Not a good idea to approach your upper management. I agree with MintJulep's comment.

Do you have simple example that you can post to show your issue?

Best regards,

Alex

 

[CWB1]

Yes, I have seen some very poor quality practices but I have also seen some very poor engineering practices including unnecessarily tight tolerancing so don't discount the source of any possible issue.

If quality is insisting that your tolerances are overly tight, I would do them the courtesy of checking your work first to ensure that they aren't correct, same as any other colleague who questions my work. Compare the part tolerances in question to other legacy parts' that are similar. If the old and new part tolerances are vastly different then ask yourself whether/not they really need to be. Next review your parts' tolerance stacks to see what the tols need to be for the assemblies to function. With that information you can easily prove or disprove whether/not the prints are correctly toleranced. Without that information you're arguing for the sake of argument, which usually leads to an angry manager or HR's involvement.

Your CMM exammple doesnt makes sense btw. CMMs dont rely on the actual feature for measurement. They'll measure 2-2M points along an edge, create an average theoretical datum in the control software, and measure other points based off that to establish the relationship of datums to other theoretical features. Even the position of the part on the table is irrelevant bc you can virtually rotate it however you want to normalize your XYZ coordinate system. I normally find it easiest to simply overlay the CMM's point cloud with the part's CAD model.

 

[drawoh]

Feynman81,

Are you calling up manufacturable tolerances?

--
JHG

 

[3DDave]

If the tolerances were a problem the supplier should not have accepted the contract. If the tolerances were the problem QC/QA should not have signed off on the drawings.

I've seen this sort of extortionate behavior - take on a task, knowing full well there is zero intention to deliver, then wait for the last moment when there is no other "out" than to twist engineering's arm and convert an assembly that likely was self-fixturing to one that will require a lot of assembly skill to get parts to align and function. Then claim it is all the fault of engineering. Not procurement that didn't vet the supplier; not QA/QC that didn't ensure the supplier was vetted; not the supplier who plainly lied; not management - who let those other groups fail at their jobs because "engineering" could be blamed (gotta have a reason to hold back the bonus/raises.)

 

[Feynman81]

Hi all,
thanks all for your answers,even for those angry with the design department for applying too tight tolerances ;-)

Let me provide an example:

On this component (dimensions in inches sorry) we assemble a carbon fiber duct which basically is a duct (woooaaaAA) with a flange that is identical to the one you see. I kept the same tolerances for the holes (actually I only used A on the second row) plus a generous profile tolerance on the external perimeter and the internal perimeter (3 mm all along so +-1.5 mm deviation from the perfect form).

It was like this from the start and the supplier said ok until now when we are checking the first batch. The holes are machined by the supplier afterwards using a custom made drilling jig. I don't think the tolerances are impossible but some of those holes keep coming out of whack by 1 mm or more.

The problem, I discovered today, is that they use the jig to scribe the flange, then they drill a pilot hole and work around the pilot with a dremel tool (???) up until they reach the previously scribed trace.
This explained why the holes are not at all circular too (some can't even be said to be oval). I have no idea if this is standard practice in the carbon fiber world but it seems strange.

Second problem is QC, they keep pushing me to use the +- method to locate holes, I keep saying them that it's not accurate but we don't agree on this. Anyway when they check the part they use the edges of the part itself as the zero, they touch a point at the start and at the end of the edge to establish B and C. I don't think this is correct at all, to me the right procedure is to simulate the datums B and C (A is the mating face of course) with a metrology square and use the square's edge as the zero as the part edges are not at all guaranteed to be straight or perpendicular to each other.

I mean something like this:

https://www.youtube.com/watch?v=fAKSqOQmVi0

Am I wrong on this last point?

Thanks

 

[jassco]

Hi, Feynman81:

You have to consider accuracies of datum features B and C. If they are not accurate, then you can't ask for more accurate positions of the holes.

You don't need a metrology square to establish datums. You can measure all the features and establish datum.

Also, your part/assembly is symmetrical in vertical direction. But your choice of datums does not indicate that.

Best regards,

Alex

 

[axym]

Feynman81,

I would say that you're correct on the last point. The metrology square would establish the B and C datums according to how they are specified on the drawing (provided that the part is fully aligned to the two high points of the B edge first, and then translated to contact the high point of the C edge). Taking CMM points at the start and end of the edges won't achieve the same thing, if B and C are not flat and perpendicular.

I agree with Belanger that an instructor or consultant is often listened to more than the internal employee, even if they're saying the exact same thing. It's because they're an outside "expert", and assumed to be objective and impartial. I experience this quite often, where someone comes to me after a class or consulting session and says to me "I've been telling them that for a long time, but now they believe it because it's coming from you".

Evan Janeshewski

Axymetrix Quality Engineering Inc.

http://www.axymetrix.ca

 

[3DDave]

They are making a product in such a sloppy manner they would not pay for it themselves. They would take that out of the box, say it's crap, and go elsewhere for that part. They know it and you know it and QA knows it. As long as they can make it someone else's name on that package they don't care and just want to be paid.

Tough to be you.

 

[TheTick]

You really do have a quality problem. You could land a 747 in that tolerance zone.

 

[Burunduk]

Feynman81,
Out of curiosity, what explanation does QA provide for their request to locate holes by plus/minus toleranced dimensions?

 

[BiPolarMoment]

put as many dimensions "as reference" as possible because "you don't need to check them"

Translation: put "as reference" so we can justify accepting parts that do not meet specifications. (Whether or not you check the dimension has nothing to do with that)

 

[Feynman81]

@Burunduk
No real justification. Just "we've always done it like that". I tried to tell them gd&t is far better and more actual. Is there an obligation for a company to use gd&t that you know of? Or is it just considered best practice?