Interference Fits on Tapered Shafts

[cfordyce05]

Would someone please point me towards some equations for calculating the required interference for transmitting torque through a 8620 case hardened tapered (11.5deg) steel shaft to a 6061-T6 aluminum hub? Can I just use the equations for cylindrical interference fits and pick either the large or small end for the dimensional input values? The shaft has a M6x0.75 external thread and a nut is used to press and hold the hub to the shaft. My end goal is to calculate the required torque on the nut to provide the calculated required interference fit.

Things I am fuzzy on:
How does the case hardening affect how I calculate the strength of the 8620 threaded portion of the shaft? I want to find a max torque value that won't exceed the strength of the threaded portion of the shaft.
Do you use the strength of the core or the case? I have HRC values for the both the core and the case hardened portions of the shaft. Is that enough information to be able to find a rough tensile strength?

Thank you,

Casey

 

[hydtools]

I hope you did not caseharden the threads. Thread roots, stess risers in casehardened material, are a failure waiting to happen.

Ted

 

[cfordyce05]

The shaft is from an off the shelf 3W engine. The whole shaft is case hardened, including the threads, and then the bearing seats and taper are ground.

 

[JJPellin]

I would strongly caution against using the torque on the retaining nut to determine the interference on the taper. There are too many variables. Torque, in general is a poor way to apply a precise clamping force. And even if you knew the force precisely, it would be very difficult to estimate the pull-up which is the true indicator of interference. I suggest you measure the pull-up directly which gives you an exact calculation of the resulting interference. Place the hub on the taper cold and measure the standoff. Make an appropriate stop which will result in the calculated pull-up. Heat the hub and mount it up to the hard stop. Once it cools, re-measure the standoff and verify that the proper pull-up was achieved.

I am not the best person to address your questions regarding the determination of the required interference. But once you determine the interference that you require, use pull-up to achieve it, not torque on the retaining nut. And, it if it is possible, I always prefer to use heat rather than brute force to achieve the pull-up.

Johnny Pellin

 

[dhengr]

Casey:
11.5° is an awfully steep taper isn’t it? That’s certainly not a ‘self-holding’ taper. The retaining nut will be needed to hold the hub in place, on that steep a taper, once it is where you want it. But, I do agree with JJPellin that that nut is not the way to try to get the press fit you want. Why don’t you ask the people who make that off the shelf 3W engine, what ever that is, how to deal with their equipment. A good sketch with dimensions and some working forces and loads would help you explain your problem, shaft size, hub torque, etc., if you really want some constructive answers. We can’t see what you are imagining from here. You also have to deal with two dissimilar materials in that hub/shaft press fit, and that can be tricky. There have been a couple threads on that issue in the past on this mech. forum. I don’t remember the titles of the threads, but a fellow by the name of Desertfox and I participated in one of them about six-eight months ago. So, do a search and see if you can find it, or look through his or my postings on the mech. forum for that thread.

 

[Cockroach]

Look through this forum on threads. I posted the papers on this some time ago. Two papers were from API on premium threads, and the second from an engineering conference circa 1962. I would need to find and repost them here if necessary.

Regards,
Cockroach

 

[Tmoose]

I know (only after Googling) this is a tiny engine for RC planes and the like.
11.5 degrees included looks mighty close to the 1;10 used to mount the clutch on many snowmobile cranks. No key, just friction from the interference fit. I don't have any numbers, but I expect the clutch moves way up the crank taper. They are installed with big torque on the center bolt, and removed with BIG torque on the puller bolt. The clutch hub is kind of thin, as I recall, at least on an old Salisbury.

 

[Cockroach]

Clinedinst was the author for both the papers I refer to above. Attached are those papers. You may also wish to reference API Specification 7 and 7G in addition to API 5B and 5CT for additional information, particularly premium oilfield threading.

Regards,
Cockroach

 

[Cockroach]

The link only loads one file at a time. Here is the second Clinedinst paper.

I would like to point out that I do not subscribe to the believe that case hardening threads is a bad thing. This is contrary to the commentary above. For over twenty years I have been case hardening threads by liquid nitration without any problem whatsoever. I strongly feel that, and have argued against this supposidly high stress concentrations and cracking at the thread arguments. Cracking and propagation of cracks is a Charpy phenonema and typically originate from sites of imperfections such as notches or dislocations within the material. Often this is due to incorrect manufacturing and poor quality thread cuts. Material selection and strength via notch toughness mitigate these effects.

I say mitigate because every thread is prone to failure, not necessarily by the root cracking phenonema. Line pipe threads made from stainless steel comes to mind. These are typically poor candidates for transverse loading and are prone to metal pickup off the flanks of the threads where the faces are high load bearing surfaces.

But to allude to case hardening as a poor practice as a post machining process, cart blanche, is simply nieve and a wrong statement. On the contrary, I fully recommend hardening of the threads by liquid nitration in order to reduce galling and oxidation in the joint as a result of galvanic corrosion as a secular phenonema.

Hope this helps you out somewhat.

Regards,
Cockroach

 

[cfordyce05]

Thank you for your responses. I have attached a picture of the shaft. There is a woodruff key on the other side that we have found the hard way should only be used for alignment of the hub. We had several shafts break from torsional stress cracks originating in the bottom corners of the key slot. We determined the cause of the cracking was because the hub taper did not match the shaft taper resulting in torque being transmitted through the key instead of the interference fit on the taper. This was due to the customer incorrectly tolerancing the taper on the hub drawing. The parts were machined to print, but the print allowed the taper to be +/-2deg from nominal! We have since corrected their print to hold the taper to +/-.1deg and have no longer had shaft breaking issues. The "bad" parts were only .5deg over nominal, but it resulted in about 1/3 the contact area on the shaft. I did confirm that the taper is 1:10 on a 10mm shaft. The taper is about .5in long, just using calipers.

The issue I am looking into now is periodic thread breakage when the hub is installed. The customer currently has us torque the hub nut to 200in-lb with a small application of Loctite 271. I believe I found the threads dhengr was referring to, but I have not had a chance to review them thoroughly. I also need to do some "light" reading in the papers Cockroach provided.

 

[desertfox]

I found this article on tapered interference fits:-

http://artec-machine.com/_images/documents/hydraulic_removal_coupling_hubs-keyed_keyless.pdf

but it assumes material for shaft and hub are the same.
Why are you using aluminium for the hub?

 

[cfordyce05]

Weight. The hub has a generator rotor pressed onto it so it already has a pretty good rotational mass. I have thought about seeing if the customer would be willing to possibly put a steel insert into the aluminum hub for the female taper. We have seen some pretty impressive examples of galvanic corrosion coupled with fretting corrosion on these parts.

 

[desertfox]

Well while the steel insert might improve the friction between that and the shaft,you still have the corrosion problem and the torque capacity of the sleeve interface with the aluminium, as it as just moved outboard slightly

 

[hydtools]

This may be useful:

http://www.mdesign.ftn.uns.ac.rs/pdf/2008/091-094_for_web.pdf

I would use the core strength in the calculations, not the case strength. The core supports the case, if the core gives way the will fail.

Ted

 

[desertfox]

Hi

I think I can shed some light on your thread failure, firstly using a very approximate formula for axial load derived from torque applied to a screw:-

F = T/0.2*d where T= torque
d= bolt/screw dia
F= Axial load
0.2= friction factor

convert the torque quoted in your earlier post to Nmm and this gives 200lbin equal to 22652Nmm.

F = 22652/0.2*6 = 18876.8N

Now if you take the minor diameter of an M6 screw (fine) its 5.08mm

Next divide the tensile area into F above

stress = f/A = 18876.8/(3.142*5.08^2/4)= 931Mpa
The yield for the material in the hardened condition is about 690Mpa according to the site:-

http://www.tatasteelnz.com/downloads/CaseHard_AISI8620.pdf

for upto 25mm section.
All that said my initial thoughts are your tightening the shaft thread to tight, however the threads don't fail every time, so there as to be reasons for that,I would put these down to an inaccurate friction coefficient in my formula above because it will change depending on thread geometry and surface finish of threads and mating parts and the yield stress quoted for the material is usually a minimum or guidance only.
The above fomula for axial load related to torque also shows how inaccurate using a torque setting is, for example just change the friction coefficient to 0.1 and for the same torque setting you can get double the axial load.
desertfox

 

[desertfox]

I would also like to add that if the M6 nut bears directly against the aluminium hub then the compressive load of 18876.8N would exceed the yield stress of the 6061-T6 material based on the matrial value found here:-

http://asm.matweb.com/search/SpecificMaterial.asp?bassnum=MA6061t6

My calculation shows the area under the nut to be 58.32mm^2 and dividing that into the axial force in my earlier post gives a stress of 323.6Mpa compared with a yield stress of 276Mpa for 6061-T6 material, if you use a plain washer between the hub and nut then the stress in the hub doesn't exceed 216Mpa based on the same axial load.
It is also prudent to note that the tolerence on preloading the shaft screw by torque wrench is about +/- 25% so even with a plain washer between hub and nut its possible to get to 270Mpa of stress on the hub face just on this tolerence alone.
Can you tell us what the interference on the taper is?
Also can you tell us more about the screw failure, is the screw breaking off as I suspect or the threads shearing?

desertfox

 

[cfordyce05]

Desertfox,

Attached is a picture of one of the failures. I don't have enough experience to be able to tell what type it is.

The nut that is used is a custom nut made from 4140. It has an enlarged bearing surface with an area of 88.89mm^2.
Something that I need to look into is the nut running out of threads and shearing off during tightening. After taking a second look at the break, it appears that may have happened. In that case, the 5 or 6 failures we have had may be attributed to dimensional inconsistencies in the hub that allows it to run out of threads. It really is a poorly dimensioned drawing and doesn't control the critical features very well. What I don't get is that the production guys just grabbed another core and the same hub that the threads failed with the first time worked with the new core. I haven't been able to find the other cores that have broken to make a comparison.

I don't know what the actual interference is with the current tightening procedure. I may try and measure before and after torquing dimensions of the hub relative to the end of the crank shaft on our CMM.

 

[cfordyce05]

Picture of the nut side of the break.

 

[desertfox]

Hi

I am on site so my getting on eng-tips is limited but weekend I can get a better signal.
The point of my post is to indicate to you that at the torque setting you're using can over stress
the male thread tensile area, the reason there not all failing is because the yield of the material used in the calculations is a minimum and in practice it may be higher, also the friction factor may vary as told in my post above.
The failure to me seems to be very brittle and I can see indications of multiple cracks around the screw thread tensile area.
Did this failure occur on tightening the nut at the assembly stage?