Heavy Shrink Fit 4

[nanobot29]

Are there any risks in using a heavy shrink fit on a 4340 steel disk? i heard of rules of thumb that suggest 2 to 3 thou per inch but my application requires a shrink fit of 0.035 inch interference on a 7.000 inch diameter rotor. The reason i need such a large shrink fit is because the rotor is spinning at 18000 rpm which is causin the disk to expand by over 0.030 inches.

i calculated all the stresses and fatigue life. streses are ok as long as i heat treat to a min yield of 200 ksi.

Thes disk is 1 inch thick and has a 15 in OD.

is there anything else i need to consider? is it possible to achieve a 0.035 inch diametral interference in a 7 inch rotor? thanks

 

[TVP]

nanobot,

I'm not sure that a casting is exactly what you need, but it should not be ruled out because of general concerns like "they are too brittle". There is a huge difference between cast iron and casting 4340 steel using an investment casting process coupled with vacuum or counter gravity pressure that is subsequently quenched and tempered to very high strength. The CLA process from Hitchiner is one example:

http://www.hitchiner.com/countergravity-casting.html

Regarding interference fits, 0.8% diametral interference is routinely used in heavily loaded forging tools. As long as you have a press big enough, there should not be a problem with your indicated interference fit.

 

[Cockroach]

Okay NanoBot29, I think you need to step back and think this one through. There are too many experienced and prominent forum contributors here that are questioning your application.

You have a steel disk, 15.0 OD X 7.0 ID X 1.0 inch thick dimensions, presumably SAE 4340 Alloy Steel. The requirement is to use this "flywheel" at 18,000 RPM. You have a concern about fastening to the shaft, your believe is that the dynamic forces will expand the rotating device by 1/32 inch. In order to counter this effect, you wish to apply an interference fit of 0.035 inches at the shaft, flywheel interface.

Let's get the material straight. SAE 4340 is an alloy steel and not some form of cast. It is typically made as a forging (1800 - 2250F) were the result, in an annealed state, make this material heat treatable. Heat treat can be performed at 1525F followed by an oil quench AFTER normalizing at 1650F (i.e. tempering). Tempering CANNOT be performed for material strengths greater than 220 ksi for fear of degragation in impact resistance, i.e. Charpy Notch Toughness. Typically you can temper at 950F and obtain a strenght in the range of (125,200)F.

My first concern is your specification for this choice of material. Noting the correction that the material is a forged steel and not cast, use is acceptable for gears and associated machinery but I highly doubt 18,000 RPM. The reason is that with a good chance in lack of tempering control and the wish that you get close to the limits of notch toughness for a given stength of 200 ksi, I would seriously question whether the device would hold together at 18,000 RPM. Any chance of an imperfection, surface or internal, the disk would simply initiate a fracture from a zone of very high stress concentration. Given the rotational energy, you would cause serious surrounding environmental damage, probably put personnel safety at risk. So I question the usage of this material for your application, and then wonder if you have done the research on the material.

My second concern is the intent to apply an interference fit of 0.035 inches at the mating surface. Presumably you would have this done as a heat shrink, mechanical application would severely challenge the hydraulic equipment. Given you could press fit this, you would be beyond the material strength anyway and be working with plastic deformation along the inner core of the disk and shaft. So that would leave thermal application. I don't see your computation, but your quotation to Kenat for 700F would push the hardened material back to the annealed state. I note the annealing process of 600F air cooled at a rate of 50F per hour. So you have lost your desired material properties of 200 ksi and would have about 135 ksi at best. Again, the material selection as a function of your assembly methods makes me wonder if command over the process would make your intent even remotely achievable.

Let's talk about the connection method, assuming you could adequately get 0.035 inches out of 200 ksi alloy steel. What are the contact pressures along the interface of disk to shaft? Again, I am inclined to think that given the deformation of 1/32 inch nominal, you would have a contact stress nearing the yield strength of the material, clearly an undesired consequence.

So I think RB1957 is correct with mentioning keyway or cross bolting as a mechanical interference for taking shear and moment (i.e. torque). You would need to address your shaft setup for bearing stresses, deflections, dynamic loading and accelerations causing unwanted critical loading. You seem not to mention the energy input mechanism, what about frequency interference computations as a result of starting the system up from zero and accelerating to 18,000 RPM.

I wish to point out that your proposal is really, really scary. Please get some solid engineering consultation and controls in your process.

Kenneth J Hueston, PEng
Principal
Sturni-Hueston Engineering Inc
Edmonton, Alberta Canada

 

[nanobot29]

Kenneth,

Thank you for the detailed information that you provided. I'm questing the application too... which is why i turned to experts like yourself and the kind people of this forum that have offered their expert opinions. I agree, this project is really really scary and I'm taking the feedback that I'm receiving from the contributors into consideration. I also intend to seek a consultant to come in and look over my work.

To answer your questions... yes im intending to use 4340 steel (and i know its an alloy). The disk and shaft will be machined as two different pieces or forged into one. The tempering temperature for 200 ksi steel is specified to be 800 deg F. Assuming that the disk can be heated to 700 degF without degrading its properties, the contact stress on the interface would be about 55ksi and the hoop stress would be around 90 ksi once assembled. The equivalent Von Mises is below 150 ksi taking into account the stress concentration at the edge of the interference. The contact stress reduces to about 5ksi at full speed while the hoop stress and von mises stresses increase to 160 ksi. Even though the stresses are high, the stress amplitude is relatively low, which is why the calculated fatigue life is over 500,000 cycles. The lowest safety factor on the yield strength is calculated to be about 1.3 at the edge of the disk.

I would love to be able to use a key or a bolted disk but, as i mentioned in an early post, the stress concentrations created from the key or bolt holes makes the design even more dangerous than a shrink fit.

If any one knows a good consultant in Los Angeles, I'd appreciate contact info.

 

[RossABQ]

What are the thrust loads and where are they applied? Since this disk is rotating, I assume it is also attached to the rotating coils (or magnets)? How?

 

[nanobot29]

Thrust loads are relatively low. no more that 4,000 lbs distributed over the entire face. My stress and fatigue calcs take the axial loads into account. They seemed very low compared to the centrifugal effects.

 

[Cockroach]

Okay my friend, okay. I found the problem, your application was really grating on my brain.

Taking a fresh look at your situation and noting the flywheel of dimensions 15.0 X 7.0 X 1.0 inches, 18000 RPM would generate only 3569 psi (maximum at the inner radius) of hoop stress and 437 psi maximum radial stress at an offset distance of 5.123 inches from the centre of the shaft, i.e. 10.246 inch diameter.

Computing your energy storage content for this flywheel of SAE 4340 HRc 18-22 alloy steel, 1.08 kg gives about 48,619 J or 35,850 ft lbf energy. Noting this is the braking torque for the flywheel, I get a press fit of 10,928 lbf between your shaft and flywheel, which would be capable of holding 38,249 ft lbf torque.

Working backwards, you would need a maximum interference of 0.011 inches, about one-third of what you originally stated. This is most likely the source of your error, I would check that computation.

Final dimensions for your flywheel are: 15.0 X 6.980 X 1.0 inches and the shaft is of 7.000 OD. Material specification is SAE 4340 machined originally in the annealed state and then hardened to HRc 28-32. I would recommend case hardening by liquid nitration for the shaft only.

Flywheel critical dimension is the hoop stress at the interior wall with the shaft. In your case the contact mating pressue of 3,313 psi would be sufficient to hold 38,249 ft lbf of torque, slightly above your requirement of 48,619 J (35,860 ft lbf).

To me, 18,000 RPM is really quite high, but the mathematics does show that your application will work. The earlier conversation(s) really began taking the tangent, my apologies for adding to that deviation.

Kenneth J Hueston, PEng
Principal
Sturni-Hueston Engineering Inc
Edmonton, Alberta Canada

 

[nanobot29]

Kenneth,

I greatly appreciate you taking the time to calculate the stresses. But i respectfully disagree with the numbers that you came up with. I'm not sure i understand the meaning of the 5.123" offset. My calculations where verified by two other engineers and I came within 4% of FEA results. A spinning disk of dimension 15x7x1 will reach a hoop stress of almost 150 ksi at the inner surface and will have a diametral expansion of 0.030 inches.

I apologize for not providing much information. This is not a flywheel. Its simply a piece of a thrust bearing. I need to maintaing the surface area at that particular diameter.

 

[unclesyd]

I personally don't see getting 200 ksi tensile out of a 7" bar 0f 4340 steel due to the mass effect. You could get this strength on a .5" round but not on 7" round.
As stated above a shaft and hub is the way to go but if you have go with a shaft and wheel of any material would get the wheel made by ring rolling. I definately get all the price and delivery for all the options.
Your calculations should include stresses from axial, thermal, radial, and spin. Both Shigley and Hamrock cover this this type project.
The following excerpt from a machine design book has a very good treatment of problems similar to yours.

if the link doesn't work type the second line in the search box of IE.

https://highered.mcgraw-hill.com/sites/dl/free/0073121932/365766/chapter07.pdf

bud21932_ch07_346-394

as a matterof curosity what type of bearings are you proposing.

 

[Cockroach]

Thanks NanoBot29.

I concur with UncleSyd, my computations do not show 200 ksi hoop stress at the inner diameter due to mass affects and dynamic loading. Maximum radial stress occurs at a radial displacement of 5.123 inches. Your loading situation is not sufficient to generate the 1/32 inch nominal deformation.

Good luck with you application.

Kenneth J Hueston, PEng
Principal
Sturni-Hueston Engineering Inc
Edmonton, Alberta Canada

 

[nanobot29]

Unclesyd,
thanks for the info. I'll look deeper into the expected strengths of large bars. My calcs did take into account thermal, axial, and centrifugal effects. Thank you for the link. there alot of good tips in there.

Kenneth,
I'll double check my analysis and assumptions. The one thing i feel sure of is that the disk will expand 1/32 (1/64 radially).

 

[nanobot29]

**Update** incase anyone is interested...

We're forging the shaft and disk as one piece. I appreciate everyone's feedback, all the red flags was an eye opener.

i also did a thermal analysis of the assembly process. a disk at 700 degF mounted on a -80 degF shaft results in thermal shock. the stresses are beyond the material's limits.

 

[unclesyd]

Appreciate the feedback and it's nice to get good results.

I think you are on the right course. I would start looking around for a balance machine that can handle your part. You may have to go to a turbine service center.

 

[gerhardl]

Interesting reading.

It might be that I have not understood the problems properly, but although I am no expert: is it possible or an idea (practical and moneywise) to use a shaft of highgrade stainless steel and shrink it in liquid nitrogen ( minus 196 deg C) to make the proper connection? (I am just curious)

 

[unclesyd]

gerhardl,
We do this all the time on SS shafts , the majority of them are used in centrifuges, The real problem with this type assembly ly you are not going to able to unstack the parts. As you know with CS you heat the disk and push and it's off or drop the assembly and let inertia do the job.

 

[moon161]

How bout balance and then spin weld?