cyclic loading beyond yield strength - spinning disk 2

[nanobot29]

Here's the problem: a disk made Inconel 625 (no heat treatment) was installed with an interference fit on a rotor. Since the disk was not heat treated, the hoop stress will reach the yield strength while its spinning. The yield strength is about 65 ksi and the hoop stress is expected to reach 70 ksi (von mises stress will reach about 80 ksi).

My assumption is that the disk will become permanently deformed at the inner surface and as a result, will loose some of its initial interference pressure when the rotor returns to a stand still.

my question is: what are the long term and short term effects of cycling a material to about 10% above its yield strength? does the disk continue to expand non linearly after every cycle? or is the loss of preload the only major consequence?

any insight would be greatly appreciated.

 

[MintJulep]

Take a paper clip.

Bend it.

Bend it some more.

Bend it even more.

Does it look the same as when it started?

 

[nanobot29]

also, what happens when the load is help constants over a long period of time beyond the yield limit of the material? is there continuous deformation?

 

[TheTick]

When circular saw blades are manufactured, they are pre-stressed in such a way to mitigate such stress. I can't remember exactly how, but it's as if the discs are stressed in compression toward the center.

I learned this from a family member who owns a business that makes blades for the lumber industry.

http://www.ctwsaws.com/

 

[MikeHalloran]

Our friends who make targets commonly stress steel beyond its yield point. ... but they only do it once. ... and it's not Inconel.

Maybe Special Metals has data that can help, but for right now, you are spinning expensive scrap, and that will probably lead to making more, more expensive scrap. ... and that's the best case scenario, assuming no one gets hurt.


Mike Halloran
Pembroke Pines, FL, USA

 

[nanobot29]

Thanks for the feedback,

I've been asked to analyze the situation. I did my analysis and I'm supposed to present the results tomorrow. I was hoping that my numbers should show that there is no way this would work and that the rotor assembly should be rebuilt. However, i feel that the managers will respond with "ok and so what if it expands a little, we can live with that." so now i'm trying to gather additional information on the long term effects of this scenario to help persuade them to do the rework. i really don't feel that this is safe

 

[drawoh]

nanobot29,

I did an analysis of a high speed laser chopper, many years ago. I tried all sorts of configurations and exotic materials, and I could not get stresses under the yield point. I suspect that carbon fibre would have worked, but I had no data or experience on it. The outside edges of the device would have been supersonic, so I needed to factor in shear due to air resistance.

I submitted a report stating that the device was a safety hazard, and that any design should be subcontracted to a firm of professional engineers. That was the last I heard of it.

JHG

 

[1gibson]

Is this a one-off, or going to be produced multiple times? Don't forget the difference between ASTM min yield strength, and actual yield strength from an ASTM.

Don't want to make a liar out of yourself if you say "this absolutely will fail" and they try it anyway.

 

[nanobot29]

This is a one-off thing going inside a 2 MW motor. We would like to get 15 years out of the motor.

my concerns right now are: what will happen long term? im sure the disk will be fine for a few cycles but will it continue to expand due to a constant load above the yield? is creep an issue at 150 degC?

 

[dhengr]

Nanobot29:
Ah... presentation of the results tomorrow. It’s a shame you didn’t wait a few more days to start asking questions, so nobody else would have to think about the problem. And, the engineering managers “ok, and so what if it expands a little, we can live with that.” would be all the blessing you needed, then. It’s obvious how they got to be managers, they couldn’t do engineering, so they naturally progressed to their level of total incompetence, just as the Peter Principal suggests. There is also another old saying... “ignorance is bliss.”

I need more time to think about your whole dilemma, and we need much more info. on the details of the problem. A sketch with dimensions and other pertinent operating conditions would be a good start, along with the mech. props. of the Inconel 625 and the stress/strain curve. With an Fy of 65ksi and a max. hoop stress of 70ksi you may not even get real much yielding. You should be working pretty low on the stress/strain curve, and should return back to zero stress (and zero rpm) on a line parallel to the slope of the elastic part of the curve, but with a small offset in strain, and a bunch of residual stresses. On restarting, the stress and stain will run back up this new curve (straight, sloped line) back to 70ksi, at the same rpm. If you increase the rpm, you’ll climb the stress/strain curve a bit higher, and then trace a new parallel sloped line back to zero stress, with a slightly greater strain offset. This is all in the lower reaches of the stress/strain curve, the start of the strain hardening part of the curve. And, is usually not a problem. At the moment the only reason we know anything about the Von Mises stresses is because your FE program told you so. And, it seems to confuse the hell out of everyone.

What happens at the shaft and hub and w.r.t. a shrink fit is yet another matter, with more need for the details of the situation. What shrink-fit, what dimensions, what shaft material, etc. etc.? Some loss of shrink-fit preload is possible, but may not be a killer either. Look in your Strength of Materials, Theory of Elasticity and Machine Design texts for more on both of these subjects. That’s what I’ll have to do to go much further. But, in the Fy=65 and hoop stress of 70ksi stress range you probably don’t have a failure mechanism likely to happen on start-up. Fatigue and creep are yet added issues with no quick answers. Obviously, this should be a nice clean disc to minimize fatigue. How often is this going to cycle?

Your’s is an interesting problem, but awfully short notice.

 

[GregLocock]

Shrugs, if you want decision made about a 15 year projection in life for a reasonably expensive machine then at the meeting you'd better say that you need more time, and resources.

Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

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

 

[Tmoose]

Are there any holes drilled in the flywheel/disk near the interference fitted center hole?

 

[nanobot29]

dhengr,

thank you for the information; i appreciate the feedback. I wish i would have posted this earlier. i really wasnt given much time. Here's some additional information.

Disk Stresses due to interference fit (0 rpm)
Radial stress: 28,900 psi
Hoop Stress: 49,500
(.011 inch diametral interfere on 5.20 inch diameter)

Disk Stresses at Max Speed (24,500 rpm)
Radial stress: 14,200 rpm
Hoop Stress: 69,600 rpm

On/off cycles: 250,000

A concern i have is in interpreting the stress-strain curve which i downloaded from ASM. It seems that i can end up with a strain offset of anywhere between .003 and .006 in/in which would reduce the interference pressure by 50% to 90% (assuming it does yield and strain offset is in the hoop direction. If this occurred,the second cycle would cause the disk and rotor to separate well below the operating speed.

I've attached a stress-strain curve for inconel 625. it shows a yield strength of 68.9 ksi, however, the material cert for the part call's out 65 ksi. Any help in interpreting the curve would be greatly appreciated.

thanks

 

[nanobot29]

Tmoose - no holes other than the center bore.

Greg - i agree... i'm gonna request some more time.

 

[MintJulep]

Yes, if stressed beyond yield continuously there will be continuous deformation.

But in your case the interface will slip, the disk will decelerate, the hole will shrink along the shifted elastic curve, accelerate again and repeat until failure.

There will likely be galling when the interface slips, which makes things more interesting.

Ultimately the failure should be rather spectacular.

 

[rb1957]

there are two issues, right ?

1) fatigue stresses above yield, the stress in the disc when it's up to speed, cycling on/off 250,000 times (with or without stress concentrations), and
2) the loss of interferences as the hub material cycles.

the probelm area is the hub (right?). this has a constant stress due to the interference ("y", easily calculatable) and a cyclic stress (0 to "x", also easily calculatable) when the disc is running. so the fatigue cycle is x to (x+y), a mean stress of x+y/2, a cyclic stress of y/2. but (thinking for a minute), isn't the constant stress in the circumferential direction (hoop around the hub) and the cyclic stress in the radial direction ?

when you say the inconel isn't heat treated, you mean it's annealled ?

 

[nanobot29]

Mintjulep,
that's interesting; i hadnt visualized it that way. that failure mode makes a lot of sense.

rb1957,
yes the problem area is in the hub. I have calculated that the radial stress goes from about 28 ksi down to 14 ksi while the hoop stress starts at 49ksi and rises to about 69 ksi at full speed. So the amplitude isnt very high but my concern is cycling into the yield strength. Yes the material is in the annealed condition.

 

[nanobot29]

i just realized something... if there's any plastic deformation, it would only occur at the center hub where the stresses exceed the yield limits. The radial and hoop stresses reduce at higher radii so i'm definitely over estimating the amount of deformation because not the entire disk reaches the yield limit. most of the disk would want to return to its original state with the exception of the inner surface. is there a way to determine this amount of deformation if only a portion is deformed? would this situation lead to residual stresses?

thanks again everyone for the valuable input

 

[dhengr]

Nanobot29:
Of course, this analysis should have been done before this machine was put together, when the shrink-fit or disk mat’l., and the like, could have been changed. Now, we are in the situation of trying to make something work which might be boarder line in some respects. All this to save face for some decision which was made without sufficient forethought or engineering study. I think that this should be part of your request for more time, politely phrased, of course.

Is this a gas fired turbine or a steam turbine? This machine doesn’t start or stop suddenly does it? While we want the disk to be stable on the shaft, the shrink-fit isn’t dealing with inertial starting and stopping forces of the disk too, is it?
What does the disk do? What are the i.d. and o.d. and thickness of this disk? What is the o.d. of the shaft, and shaft mat’l.? Was this a press fit or a shrink-fit?
RE: Disk stresses at 24,500rpm; where are the two stresses 14.2ksi and 69.6ksi located on the disk? Are these stresses on the disk only due to the rotation or are they combined stresses (shrink-fit stresses, plus rotational stresses, superimposed)? I’d like to see your shrink-fit calcs. and the stresses and stress patterns you got from your FEA. Put this in an attachment at a big enough scale so it can be read on the screen, I don’t have a color printer.

You do understand that the lower stress/strain curve is only the first (left most) 20% of the 0.0 to 0.05 column in the upper stress/strain curve. I would like to talk with a good materials guy or the material supplier on this stress/strain curve, it looks a little generic to me. I think you are reading these stress/strain curves correctly; that is a longer, flatter, plateau (plastic range) than I would have expected for this material; and you are working right at this stress level so you could get some larger strain excursions than I would have expected. But, this is not creep, and I don’t know how the rest of the system constrains any movement. I’m not sure about the creep problem, but at 150°C (300°F) that might be getting boarder line, and I would like to confirm that with a good materials guy, someone smarter than I am on this matter, or the material supplier. It depends upon where the max. stresses are, their orientation and the confinement around them.

Working on this type of equipment, don’t you have any senior engineers in your own office/shop who have a fair amount of experience on these matters. You are certainly stretching my off-the-top-of-my-head knowledge. I don’t work with high speed rotary equipment every day. I need to spend more time thinking and reading on this. You have the advantage of much more computer power than I have, and you’ve been studying this problem for a few day, so my thinking/calculating process is a little slower than yours.

 

[MikeHalloran]

The MBAs will certainly ask this question: "If an annealed part is good enough, why does the print specify heat treatment?"



Mike Halloran
Pembroke Pines, FL, USA