Buttress thread fatigue failures

[scottlel]

Well the tile sums it up we are having a fatigue failure that involves a buttress thread. I have a cylinder of 6061-T6 Aluminum we have cut an internal buttress thread in. After 7-10,000 cycles to pressure, 3,000psi, the tube fails at the last thread of the sealing cap. I've been through two designs. One I inherited that had a very high axial stress in the wall- 26,000psi. I thickend up the wall and have reduced the axial stress to 9,000psi. This did nothing for the life of the tube. After about the same number of cycles I still end up with the same failure.

Any ideas on what else I should be doing to fix this problem?

 

[rb1957]

more details ...

what thread form ?

net thickness (at the root of the thread) ?

basic thickness (how much material is removed by the thread, how much local bending is happening) ?

how did you determine your stresses (hand calc, FEM) ?

did you predict a fatigue life ? what Kt ?

can we change material ?

 

[scottlel]

The thread form is- 3.50-10 PUSH-BUTT-2B, on the current design.

The wall thickness at it's thinnest is .220", The depth of cut for the thread is .058".

I just calculated the stress for the wall outside the thread quickly by hand.

Fatigue life was determined by testing actual parts.

Material choice could be changed. I'd like to stick with aluminum because of weight. We try to stick with 6061 on pressure vessels since when they fail its not as likely to fragment as something like 7075.

 

[rb1957]

thx for the info

i don't know your thread form but there isn't much info. out there on Kts ... is it a profile dsigned with fatigue in mind (generous root radius would be something to look for). thread Kts tend to be big, something like 8.

not surprised about the location of the failure. you could try unloading the last thread by thinning down (stepping) the end-cap wall (making it a little more flexible, encouraging the load to stay in the cyclinder); but i'd've thought this'd be a small effect (given your large change in nominal stress).

how did you accomplish this ... locally thickening the wall of the cyclinder ? did this extend somewhat beyond the thread (to make sure it was working as you expect) ?

 

[eromlignod]

Aluminum buttress threads??!

Don
Kansas City

 

[sreid]

Three things;

1)Threads do not equally share the load. The first thread at a bolt-nut interface can carry 1/3 to 1/2 the load.

2)The stress at the thread root usually exceeds all other stresses.

3)No matter how low the cyclic stress, aluminum will eventually fail from fatigue.

 

[desertfox]

Hi scottlel

What diameter is the cylinder o.d & i.d and I assume that the full axial load generated by the pressure is taken on the thread?
I may be able to help just need some more info first.

regards

desertfox

 

[MintJulep]

What is the failure direction? What is the combined stress? Cutting the thread how?

 

[scottlel]

The cylinder is a 4"ODx3.25" ID. The full axial load is taken by the threads. We added a radial seal before the cap that eliminates the hoop stress on the threads- this didnt help any with the fatigue life.

The failure originates at the root of the first thread then works it's way out to the surface. We then end up with a tear/crack that goes about 1/2 way around the OD.

The thread is being cut on a CNC lathe with a carbide insert made to the thread spec. I looked breifly into thread forming. The quick reading I did said it didnt work well with buttress threads.

One other point of interest I did some FEA work on it today after my 1st post. I found if I changed the thread pitch from 10 to 12 there was ~20% reduction in the stress at the thread root. For reference that reduced the thread height by .010" and made the root radius smaller by .002"(.007 to .005).

 

[MikeHalloran]

I'd try relieving the end plug internally so that it has a pretty thin wall at the inner end and a thick wall at the outer end. The idea is to get the plug threads to stretch enough to distribute the axial load more uniformly along the helix.

Or taper the pitch surface of the threads; similar reasoning.

If at all possible, I'd eliminate the joint by machining the cylinder and cap from one piece of billet, or make it an impact extrusion.



Mike Halloran
Pembroke Pines, FL, USA

 

[MikeHalloran]

Or, cut (round bottom) deep groove at the root of the male threads, so that each thread is at the crest of a fin... again to provide some flexibility.

Or, switch to titanium.



Mike Halloran
Pembroke Pines, FL, USA

 

[desertfox]

Hi scottlel

I was going to suggest you look at the stresses on the threads but you already have.
Is the tear your seeing going around the threaded portion like a helix?
If so you need to look at the combined stresses of the tension and shear as a helical crack can be formed by the
principle tensile stresses running at approx 45 degrees to
axis of the thread.
Using the Von Mises theory for failure of ductile materials might help.

regards

desertfox

 

[rb1957]

i'd like to return to the thread form given above "3.5 10 ...". it the tube is 4" OD and 3.25" ID and "10" is the tpi, is the "3.5" the mean thread dia ? this would add some secondary bending to the thread.

at 3000psi, the hoop stress is 3000*(7.25/4)/(.75/2) = 14.5ksi and the endload stress is 7.25ksi ... maybe in focusing on the endload on the thread you've forgotten the hoop stress component (seems higher than the stresses you quoted above).

i guess part of your answer is even though you reduced the nominal stress significantly, but didn't see an improvement in fatigue life, is that you're possibly still in the plastic range of 6061.

could you shrink a collar onto the outside of the part ... this would cause compression stresses in the cyclinder and help the cyclic fatigue ?

i'd be hesitant to reduce the thread root radius as this has a direct relationship to the Kt.

 

[rb1957]

a random thought ... is there nothing is the pressure vessel design code dealing with this ?

 

[btrueblood]

Because the hoop stress is relieved below the threads, the tube essentially necks down at the end, creating a reverse taper, and exacerbating the concentration of shear load on the 1st thread. If at all possible, try adding more axial distance from the seal location to the first thread. Try increasing the thread engagement length, too. Somehow increasing the thread pitch seems counter-intuitive, but it might work. I assume you are using a (US) National Buttress thread form with an 8-degree flank angle, could you use a form like the German 3-degree flank angle buttress thread? Doubt this will make much difference, but again, may be worth a try.

Also, look into using a spiral retaining ring (e.g.

http://www.smalley.com).

We use these a lot, and like the way they perform. Calculating the Kt for a single groove is a lot simpler, and could allow you to design in more fatigue capability. Clamping the ring down to hold it flat under load (using a cover plate that screws down to the cap) is a way to improve the spiral ring capability beyond that shown on their website (which gives the failure stress in the body for a cantilevered ring). Talk to their rep about your design, and ask them to give you some numbers.

 

[Cockroach]

I wouldn't be using Thin Wall Pressure Vessel equations for this analysis. This hoop stress equation is far too simplistic, errors exist even in the limiting case where wall thickness is less than one-tenth internal radius of the pressure vessel.

Rather, the triaxial stress should be used since radial and longitudinal stess add to elemental principle loads. Probably the Von Mises-Hencky model is better suited. (i.e. twice the gradient squared equals the cross product of the gradient with itself PLUS three times the shear stress. set shear to zero to get the principle stresses)

Using your numbers: P=3000 psig, D=4.0 in, d=3.25 in I would get a stress of 15,290 psi. This is 5.447% greater than that stated by earlier posts. The error is outside that of standard measurement, probably of interest in this case since we are dealing with failure analysis.

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

 

[hydtools]

Fatigue strength of 6061-T6 is about 14,000 psi.

The value of 15,290 psi calculated by Cockroach exceeds that.

It would seem a redesign is in order to reduce stress or use material with higher properties. Most aluminum alloys, whether cast or wrought, have similar fatigue strengths.

Ted

 

[rb1957]

actually, i'd've thought the von Mises stress would have been higher than that ... vM = sqrt(s1^2+s2^2+(s1-s2)^2) = s1*sqrt(1+.25+.25) = 1.225*s1 = 17.8ksi

if you changed to 2024T3, and guess a Kt=5 ('cause that the highest in the book), fatigue life = 100,000 cycles
(for 6061T6, life with Kt=1 is about 1E8 cycles)

 

[EdDanzer]

If you are trying to use aluminum in hydraulic cylinders you cannot have work induced load pressure or force spikes. We do hydraulic cylinder repair and have never seen threads in aluminum work well. Buttress threads don’t work well in anything I’ve seen. We have used Acme threads on internal thread cylinders with the best success. Any internal thread cylinder will see barrel swelling from internal pressure that reduces the thread strength.

Ed Danzer

http://www.danzcoinc.com

http://www.dehyds.com

 

[dimjim]

I think I would relieve the threads beyond the cap engagement and not mess with the top threads. By having a small groove beyond the threads might help to relieve the
stresses. Certainly a different approach.