Barrel Spring Unusual Failure

[sbozy25]

This could be a tough question, but any input would be much appreciated.

I have a barrel spring that had some strange failures recently. We had 2 out of 8 springs fail at about a quarter of the test time at the OD of the part. Typically if we see an early failure it will be on the ID and is usually due to high stresses and or material imperfections. Well these springs are low stress, have good life characteristics, and the wire was 100% eddy current tested for imperfections.

So my question is this.... What could cause stresses to shift to the OD of a barrel spring to cause this sort of failure? If you need any more information, please let me know and I will see what I can provide.

Thanks in advance...

 

[btrueblood]

a "barrel spring" being a helical coiled spring with a double-taper? Some ideas, or guesses really:

Damage to the o.d. (either during winding, heat treating, subsequent handling/installation or during loading/use) causing stress risers?

Twisting/torsion loading of the spring (developing bending stresses on the coils)?

 

[sbozy25]

Twisting and torsion are possible.... we are trying to look into this. However, it is not very likely as there were no smoking guns on the broken part to show torsional forces....


No there wasn't any damage to the wire. It was as perfect as perfect can get.... if that is possible... lol...


Yeah, allow me to crudely show the barrel shape, for any clarification to others....



_____
___________
_______________
___________
_____

Very crude, but you get the point...

 

[btrueblood]

Didn't think there'd be much evidence of reason #1 or you'd have said something. But, that type of spring does have the drawback that the coils can rub the i.d./o.d. surfaces against each other as the spring is deflected (depending on the geometry and deflection of course), resulting in stress risers due to the rubbing/erosion. Is there a really polished-looking area near the fractures? Evidence of "flattening" of the cross section?

Isrealkk will likely be along soon and have more ideas.

 

[israelkk]

btrueblood

I was watching this thread and had the same suspicion however, I wanted to see if sbozy25 is going to give more information such as the actual data of the spring such as sized, forces, number of coils, deflections, which coil was failed, etc.

If this spring has a constant pitch then your suspicion has a basis because the largest coil will bottom first.

This is basically two conical spring in series. To receive a linear behavior the coils pitch has to be different from coil to coil to assure that all coils will touch each other at the same time. Otherwise, the spring load deflection will be non-linear with a progressive increasing force vs. deflection.

sbozy25 didn't say which coil failed. Your suspicion will be justified if it didn't happen in the largest coil.

 

[sbozy25]

More infomation....

Coil clash is not an issue in this design. When the coils to make contact they make blunt contact and do not slip bye, which is the perfered method of clash. This only really becomes an issue once these parts are coated and shipped on to the customer and are in a corrosive environment....

As far as more info on the part...

.521 wire diameter with a free state OD of 5.5" It has a 2.4" stroke which keeps coil clash to a minimum, only the 1st and 2nd coil make contact, and it is minimal. HRC is at 53, with a maximum stress on the part of 240 Ksi which gives a predicted life of 900,000 cycles.

The part broke 1 1/3 coil from the end on a 45° plane on the OD of the part. There was no evidence of material imperfections, and decarb was almost nil.

We have seen this more and more lately....

 

[israelkk]

sbozy25

Just for curiosity can you provide the spring drawing with the load and deflection requirements?

When you say "coated" what is the coating? Is this an electropalating coat?

 

[sbozy25]

Sorry, can't provide a drawing. Propriatary information....


No, it is epoxy coated. Hydrogen embrittlement was not present.


I can tell you this is an extremely high loaded part. At maximum deflection it can reach as high as 2,500 Lbs.

 

[israelkk]

sbozy25

In your first post you mentioned and I quote "the wire was 100% eddy current tested for imperfections". Was the Eddy current test before or after coiling?

As far as I know any NDI test has a limitation on the crack or imperfection size and shape it can detects. Therefore, there is always the chance that the wire has some minor defects the NDI can not find. Therefore, if you can tell the minimum defect detection capability you can try to do a crack propagation analysis using fracture mechanics analysis and predict the life cycle of the spring. When the loading is combined with corrosive environment a stress corrosion cracking process can lower the life cycle significantly.

Did you have a metallurgy lab test the broken parts for type of failure. They can do wire section test samples and look for cracks and defects through the entire wire section.

 

[MikeHalloran]

That's a very high stress level for a wire that big.



Mike Halloran
Pembroke Pines, FL, USA

 

[TVP]

I have to agree with Mr. Halloran, that seems like a huge stress. In order to get 900,000 cycles I am assuming that the peening intensity is quite high. Any chance that there is surface initiation due to overpeening?

 

[israelkk]

I believe the 240ksi is the wire tensile strength not the stresses in the spring which are shear stresses. I have a doubt that a spring material that thick can carry 240ksi shear stresses even with heavy preset. The life cycle estimation seams odd. Even if he analysis gives 900000 cycles it is customary to commit only to 1/4 of the calculated life cycle.

sbozy25 doesn't supply enough data of the spring which makes it difficult to reach intelligent conclusions or suggestions for possible reasons for the failure.

 

[sbozy25]

Well the good news is I have figured out what has caused this failure to occur. I will get to that in a minute...

Yes isrealkk is correct that is the tensile of the material not the stress the part see's. Yes I know these estimates seem odd, but they are actually quite accurate. This particular part can reach well over a million cycles in optimum conditions. My design program I have written takes into account true spring formulas as well as first hand data which is then merged together to predict my stresses and life. Also, yes the wire is eddy current tested prior to coiling...

I'm sorry about the fact that I can't give to much information out. But, unfortunately my company feels it is important to keep as much of this hush hush as possible.

Now for why the parts failed... Over the past 9 months I have noticed parts were sporadically failing with no given cause. There were no wire imperfections, no handeling marks, and no chemical reasons. Well I happened to be in the plant when they operators turned in their daily peening check sheets. One of them caught my eye that it had over the 10% allowable bad shot mix recorded. So I started digging through papers and plotted my data. I found that when our % of bad shot grew over the allowable 10% mark, our saturation dropped and thus our peening was not driving the residual stresses deep enough into the part. as a result we would have parts that were not processed properly for fatigue life. When I compared this graph to the dates of irregular failures, they were almost a perfect match. So after a nice barking session with manufacturing, I think the problem is fixed....

Thanks everyone for attempting to give a hand, and throwing ideas out there. It helped me get some ideas started....

 

[sbozy25]

Uhoh... oops, I lied....

The 240 ksi, is the stress seen by this part... the tensile is 290 ksi... I know this sounds bad, and trust me on paper these parts don't work, but in real life they do... that is why I have my program that takes test data into account.

 

[israelkk]

sbozy25

Something is not quite right. If the tensile is 290ksi and the wire is loaded by shear stresses (in compression springs the wire is in torsion) the ultimate shear stress is 0.577*290ksi=167ksi. Even with perfect theoretical preset operation where the whole wire section gone to plasticity during preset you can gain 33% more therefore, equivalent ultimate shear strength theoretically will comes to 167*1.33=222ksi.

The only way you can live with that is if the actual wire you have is stronger than 290ksi and is in lower range of mechanical properties (manufacturing tolerances). When you will have a batch of wire with 290ksi as the spec allows the spring may not hold.

Just a week ago I encountered such a case where a spring design that was manufacrured in the past could not be manufactured today because the designer didn't take the lower range of mechanical properties into consideration. The newer wire batch tensile properties ere in the lower range of the spec.

 

[sbozy25]

Really? wow... I never thought of that....

See, this wire is spcialty wire "Suzuki Wire" I believe from Japan. Our corporate technical guru was the person that invented our process and was the person that first designed this type of part for it's application. He is the one that tells me that it looks bad on paper, but showed me how to fudge it to replicate true life.

I think I might perhaps start some investigation on my own based off what you said and see if I can discover what is truly going on. This has me concerned because it is my butt if things go bad...

Thanks

 

[israelkk]

Correction.

The second paragraph should be :" The only way you can live with that is if the actual wire you have is stronger than 290ksi and is in UPPER range of mechanical properties (manufacturing tolerances). When you will have a batch of wire with 290ksi as the spec allows the spring may not hold."