Deflection during machining causing out-of-tolerance parts? 4

[geesamand]

My company purchases machined cast iron housings that hold bearings and shafts. Like an automotive transmission, we rely on a pair of dowel pins to align the two halves accurately.

I suspect that some of our newer designs are adequately rigid in operation as an assembly, but I'm concerned that the halves are moving around during machining and causing lack of flatness and out-of-location bores and dowel pins.

Any suggestions how to determine if this is part of the problem? Measure deflection of the part on the milling machine? Estimate machining forces and check in FEA?

David

 

[1gibson]

Machine the last pass twice. If you are removing material the second time, the part is deflecting.

 

[tygerdawg]

??? Your question is very curious, perhaps I am misunderstanding something. Do you not have engineering drawings to measure parts against? And therefore definitive knowledge that the parts do / do not meet the drawings?

TygerDawg
Blue Technik LLC
Virtuoso Robotics Engineering

http://www.bluetechnik.com

 

[BrianE22]

Measure the part in the machine after the final machining operation. Then unclamp it and re-measure.

 

[Twoballcane]

Not an expert in machining, but cast iron is a brittle / non-ductile material. Any slight deflection should crack the martial. If you see no cracking, there should not be any gross deflections (well to through off the machining process by +/-). So maybe there is something else going on? My $0.02 anyway…

Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

Perception is reality: Your reality is how others perceive you, not how you think of yourself.

 

[geesamand]

We have drawings. I would argue they are drawn well as we've taken our lessons learned and revised them as we go. We use CMM to measure a sample of the parts after they arrive at our shop. The problem is that vendors struggle to meet the tolerances for some of the parts.

I think the challenge for me is that this is multiple parts being machined at multiple vendors. It's not that we can't control the output, it's that vendors seem to struggle with certain parts. They claim their machines are .000x accurate, and that the parts measure the same on the machine whether clamped or not, but they measure out of tolerance here. I want to be sure on some level that the parts are rigid and stable enough to hold their tolerances when machined with a milling machine and handled in a typical way. I want to know if thicker wall sections in the castings will help. An analytical approach that would guide us in the design phase is ideal.

As far as cast iron goes, yes it is brittle but it's not weak or glassy. In these parts it could deflect many thousandths of an inch before cracking. It would take extremely poor cast iron material or much thinner sections to crack during machining.

 

[IRstuff]

"the parts measure the same on the machine whether clamped or not, but they measure out of tolerance here"

OK, so before anyone starts designing a fix, shouldn't you verify WHY the measurements are disagreeing? Is it a procedural problem or an equipment problem? Is the error within the test accuracy ratios (TARs) of the test equipment?

TTFN
faq731-376

 

[geesamand]

Sorry, the term TAR is not used by me or our CMM department. What I do know is that our CMM repeats measurements of these types of parts/features within .0001", where tolerances go no tighter than .001" on location. 10:1 is a pretty good accuracy vs. tolerance range ratio IME. We have repeated the measurements using artifact parts on outside CMM's in the past. If there is a hole in the process, it's that few vendors have CMM capability as good as ours and most of our production parts are not CMM'd prior to shipping to us.

I believe the issue is absolute accuracy of the machining equipment (repeatable numbers on the digital readout are not the same thing as correct numbers), occasionally clamping error, and in some cases stress relief. These have all be pursued in some manner in the past.

Judging by how this discussion immediately sweeps away from part design aspects and toward process issues, am I getting the answer to the question?

 

[dhengr]

Geesamand:
These types of problems lend themselves much more to experience and well earned knowledge and judgement than they do to some cutesy FEA program. Although, I’m sure there is some software vendor who professes to solve that problem too. Are the poor tolerances always of the same magnitude + or -, in the same location, etc. Does that point a finger at a weak shell or heavy, slow cooling stiffener or fillet radius, or some such? Can you change them by changing your exact machining process, or by using a different machine? The final pass should certainly be a light cut. Have one machine shop do both parts and guarantee the fit-up at their plant. Talk with your casting supplier, I don’t think it’s his fault, but he may offer some experienced guidance. I’d have to do some digging to be sure where cast iron fits in this scenario, but maybe the rough parts should be stress relieved before machining, due to built-up casting cooling stresses. How old are the castings before they are machined, or after machining but before fit-up, how have they been stored, etc? Because, some parts made of some materials tend to stress relieve by aging, temperature cycling, and vibration in transit.

 

[mfgenggear]

any time castings are involved, it becomes a complicated matter.

castings can also deviate, causing work holding issues.
as mention above casting could be stress relieving (dimensionally moving) after machining.
or it's just bad workmanship.

#1) are casting to print? is the quality correct. Inspect the casting supplier for review of their procedures.
#2) are the vendor rough machining then stress relieving, then finish machining.
#3) how tight are the tolerances? is it resonable.
#4) do inprocess verification at the vendors facility, have them do a (R & R) repeatbility or reproducibility.
#5) is the inhouse receiving inspection & the vendor using the same proceedure. verify both.
#6) 99% of the time the vendor quality is bad because of bad procedures. make them give a cause & corrective action.
maybe limit the amount of vendors that can machine this part.
#7) part of # 4, does the vendor have the right equipment for machining & inspection. do a vendor survey. verify their claibration & machine maintance procedures. Is SPC necessary.

Mfgenggear

 

[IRstuff]

OK, so if it's been determined that the supplier's test capability is subpar, then why are you still going to them? If it's critical enough that you contemplate changing the design, then maybe your company should just buy them the correct equipment, or go elsewhere.

The bottom line is that if they cannot accurately measure now, how is changing the design going to improve their measurement capabilities later? So long as they have substandard test equipment, their measurements will never be as accurate as yours and you will continue to have these issues with measurement consistency.

Now, one thing that can change the apparent accuracy is the way the drawings are actually converted into cutting instructions, i.e., the order of operations can have a negative impact on the ultimate machining repeatability.

TTFN
faq731-376

 

[geesamand]

1: Yes, we don't see dimensional issues with the castings. The material hardness and strength are confirmed.
2: Currently the housing is stress relieved after casting only. Machining vendor rough machines, then finish machines.
3: That's the crux of the question, in a sense. We have been making these types of parts with these types of tolerances for decades. However we had our own machines and back then the fight between inspection and machining was internal.
4: I'm working toward this. It's the fact that some parts cause more problems that suggests I might consider other machinability factors (rigidity) as well.
5: I think our price target precludes heavy CMM cost per piece. Like most job shops, our vendors have more capbility in machining than inspection.
6: We have an approved suppliers list as well as a root cause / corrective action requirement when parts are found out of tolerance. It's when we see parts repeatedly miss tolerance from different vendors where I am taking a closer look. Still that does not relieve our need to find usable parts when a vendor struggles.
7: SPC would be nice, but I don't have a department of quality engineers to follow through with that.

All of this does jog my thought process: perhaps we need an artifact part or two that resembles this type of work. We send it to a CMM shop for 3rd party verification. Then we send it to a potential vendor and ask them to inspect it (by CMM or their machine center, or both) and ask them to detail the errors in the part. That could be a very telling exercise.

 

[MikeHalloran]

ISTR that Henry Ford stress relieved and aged his engine block castings by the simple expedient of storing them outside. ... for a year.

I don't think Ford goes to that extreme these days, but you might wish to set aside a few castings, machine some minimal targets on them, and reinspect them with your CMM over a period of time, just to gauge how long it takes for the castings to stabilize.



Mike Halloran
Pembroke Pines, FL, USA

 

[1gibson]

Send defective parts to vendors (don't mention that they are defective) and ask them to inspect to print. Stop using the ones who can't find the deficiencies. Give more work to the ones who can tell you why the deficiencies are there.

 

[geesamand]

"ISTR that Henry Ford stress relieved and aged his engine block castings by the simple expedient of storing them outside. ... for a year.

I don't think Ford goes to that extreme these days, but you might wish to set aside a few castings, machine some minimal targets on them, and reinspect them with your CMM over a period of time, just to gauge how long it takes for the castings to stabilize."

Ha, yeah we've played around with these things before. The folklore is that 'seasoned' castings were the best for machining. Stress relief is important, that's for sure.

Setting the parts outside for a year will certainly be a simple experiment, but not my definition of expedient. I myself get my mords wixed too.

 

[stanier]

When I was an apprentice forty years ago we aged all of our castings by leaving them on a pallet for over 6 months for the very reasons you annunciate. Perhaps this aging process could be accelerated by alloying the material or heat treatment.

These cast iron components were for gears, rollers, cams and pulleys used on envelope folding machines. This was for a product where paper, a few thou thick, was cut, printed, folded and gummed at 1000 per minute. Out of roundness and vibration could not be tolerated.

“The beautiful thing about learning is that no one can take it away from you.”
---B.B. King

http://waterhammer.hopout.com.au/

 

[KENAT]

Have you eliminated differences in inspection set up/process per mfgenggear #5?

Are you working to ASME stds, if so how are you and/or your vendors deriving the initial datum planes etc? Is one of you pushing the part up against known 'flat' 'perpendicular' surfaces and interpolating from that, while the other tries to interpolate direct from component surfaces? (Or equivalent)

Even without 100% inspection, just how you and vendor are applying datums etc. (not just in inspection but even in mounting on tool) could impact matters.

Related to this, do the parts allow for consistent fixturing on the machines, e.g. well defined datum targets etc.

Posting guidelines faq731-376

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

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