Please advise on fatigue broken leaf spring

[Handig]

In one of my companies products, a steel leaf spring operates two plastic parts. The spring sometimes brakes before its required life. See attached photo's.

Description:
size between eyes is 15.0mm after production. After assembly, the size is 16.4mm. In one cycle, the spring is strained to 20.0mm, relaxed to 19.8mm, strained to 20.0 and relaxed to 16.4mm again.
Required life is 100k cycles. At this moment, about 10% of the springs break at 40k-60k cycles, all other reach 250k cycles without cracks.
Spring material is WS 1.4310 (AISI 301) with tensile strength 1700-1900MPa. After bending, the edges are rounded, spring is annealed and shot-peened.

Strange thing is that we have a similar spring in a similar product that seems to be heavier loaded, but does make the required number of cycles.
This size between eyes is 13.0mm after production. After assembly, the size is 16.4mm. In one cycle, the spring is strained to 19.8mm, relaxed to 19.0mm, strained to 19.8 and relaxed to 16.4mm again.
As far as I can figure out, mean stress in first spring is much lower than in first spring and amplitude of cyclic stress is only slightly higher. This should lead to a longer fatigue life of the first spring, but it doesn't.

I have consulted the spring manufacturer and other experts.
Options tested (all with negative result): changing shot-peening parameters, investigation of tooling on burrs and sharp edges, change in spring material.

Can anyone explain the shorter fatigue life of the first spring, or better: give advise on how to extend fatigue life?
Thanks in advance.

 

[GregLocock]

"mean stress in first spring is much lower than in first spring and amplitude of cyclic stress is only slightly higher. This should lead to a longer fatigue life"

Um, no. Fatigue life is strongly correlated with stress reversals, which are more likely as you drop the mean stress. I strongly suggest you get the SAE leaf spring design manual.

Your attachments don't work as you are effectively trying to link your hard drive directly to the whole internet, a good trick if you can do it but likely to end in tears.






Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[Handig]

Thanks.

Too bad the attachments do not work, I will try to upload the file again, but probably with the same result.

You are right in your reaction. If I review the Goodman or Soderberg curves, fatigue life is more related to alternating stress than to mean stress.
But still, here are the figures:
First (problem) spring: mean deflection is 3.2mm, amplitude of deflection is 1.8mm.
Second spring: mean deflection is 5.1mm, amplitude of deflection is 1.7mm.
Mean stress drops by over 30% from second to first spring, alternating stress increases only 6%, but the fatigue life is shortened.

If anyone else has an tips to improve the fatigue life, or an explanation, please respond.

 

[mcclain]

What was the other material considered?
What is your surface finish?
If possible, a clear zoom in on the cracked surfaces may help.
Thanks

 

[vpl]

Handig

What I found interesting was the statement that "about 10% of the springs break at 40k-60k cycles, all other reach 250k cycles without cracks."

If 90% of the springs are lasting well beyond your expected life cycle, then the design is probably good; you should consider looking to see what is different in the 10% that fail. Have you looked into the manufacturer's quality control program? Have you checked to see if all the problem springs came from the same batch?


Patricia Lougheed

Please see FAQ731-376: Eng-Tips.com Forum Policies for tips on how to make the best use of the Eng-Tips Forums.

 

[NickE]

Why are you annealing fully hard rolled 301? What is the cycle?

How are you rounding the edges, what is the radius? How consistent are you on this measurement/process?

How high are the residual stresses introduced by the peening? How consistent?

Is there a difference in operating environment of the failing 10%?

Difference in assembly?

You may have to chase this one for a while.

Nick

 

[btrueblood]

What Nick said. Annealing the material drops the fatigue strength in rough proportion to the reduction in yield strength.

 

[IRstuff]

If I understand your picture correctly, it's always breaking at that same spot? Which is, coincidentally, where the spring flattens out again.

That might seem to imply that there's something different in the actual physical behavior of the ones that break vs. the ones that don't. Hypothetically your arm curve is supposed to distribute the amount of bending to accommodate the displacements you've described. However, then, the ones that break are apparently not do that or not do not do it well enough, so there is a bending fatigue at the one spot.

This might suggest that the arms are somehow stiffer than they're supposed to be on the ones that break. Is there a way you can determine the bending behavior as a function of position along the arm? Perhaps the annealing is non-uniform along the arms.

TTFN

FAQ731-376

 

[TVP]

I think the use of "annealing" is a translation issue-- this is likely a low temperature heat treatment in the range of 200-400 C, not the full softening annealing treatment used for austenitic stainless steels.

 

[israelkk]

To be able to respond intelligently to your questions please upload the exact manufacturing drawings of the springs with all dimensions and tolerances.

 

[Kwan]

I think IR might be on the right path. I'd worry about manufacturing differences in the area of the bend. For example: during the bending operation the part ends up with a slight scratch. Anyway, this is all speculation.

Have you sent the broken spring to a material lab to perform a evaluation? Material tensile strength, striation count, crack initiation location, etc. This is the first step. By the way, you would want to also give them a good part within the same lifetime as a control. The material lab may be able to give you some enough insight to solve the problem.

 

[gwolf2]

It is as Greg Locock says - fatigue life is far more sensitive to alternating stress than mean stress.

In your case I did a quick calc and concluded that 30% mean stress reduction and 6% alternating stress increase would lead to a fatigue life "roughly similar" within the bounds of fatigue life scatter.

I think you are in that tricky area of fatigue where the good spring only just works and the bad spring only just fails. If you were unlucky in the future during manufacturing of the good springs you could end up with the same problem.

Solution: Make small design changes to both springs to reduce alternating stress. Detailed FE models would probably solve this - it's not too difficult a problem because the failed area looks like it's not in contact with anything, and you have real test data to correlate.

For a consultancy it looks like about two working weeks max to model and correlate both, try say 3 new profile variants, and report. IF you are a little bit lucky in that there is a simple solution, and the consultancy is reasonably competent, looks like a $3000-10000 problem.

 

[Handig]

Thank you all for your thoughtful answers.
Let me try to answer the questions in all the posts above:
Production tools are checked on burrs, sharp edges, etc. Same tools are used for the second spring, but with slightly different setup.
Spring are checked on burrs, sharp edges before shot-peening.
See picture of crack surface close up.
All problem springs are from the same (and only) batch.
All springs are tested on same machine, same conditions.
With annealing I mean 30min at 250C.
Crack appears at same position every time. We are currently adjusting the tooling such that we can see if it is always left, or always right.
Unfortunately, I cannot upload drawings or pictures of the assembly due to non-disclosure reasons. I know this information would help, but too bad.
Unfortunately, changing the shape of the spring is very complex: nearly every change would affect the function negatively.
Other materials tested is brandname Thenox, seems to be very similar in composition as AISI 301.

 

[israelkk]

Handig

If you can not upload a drawing at least give the following:

1. Strip thickness and width
2. Distance from the eye center to the surface where rhe crack started
3. Distance between the eyes
4. Bending radius adjacent to the crack.
5. Eye inner/outer diameter

 

[Handig]

1. Strip thickness: 0.9mm Strip width: 10mm
2. distance is 10.5mm
3. distance between eyes 15mm, 19.4mm centre-centre.
4. bending radius R4.5
5. eye outer diameter: 4.4mm (R2.2).

 

[mcclain]

It appears that your crack initiation site is at the valley of a shot peen indentation.
Things to consider:
Is the heat treatment sufficient to remove residual tensile stresses.
Polishing prior to shot peening may improve your life.
A different shot peening process may improve the resulting surface finish.

Does anyone know if polishing after peening is helpful?

 

[IRstuff]

Have you played with the other end of the broken ones? Do they behave the same as the unbroken ones, w.r.t. spring constant, etc.?

Perhaps it's the angle of the photo, but the unbroken one in your picture appears to have a more rounded transition from the straight part to the curved part, whereas the broken one seems to have a more abrupt transition. That might suggest that there may be a difference in how the two springs shown were formed.

TTFN

FAQ731-376

 

[Handig]

Spring constant of all springs is identical, at least not significantly different.
Of course there is difference in eye-eye distance due to manufacturing tolerances. But both springs with small and with large distances crack.

 

[israelkk]

According to my calculations the load to 20 mm is 120N and the stresses in the crack area are 1660 N/mm^2. Those stresses are basically in the range of the ultimate tensile strength. The shot peening raises the local ultimate stresses on the surface which improve the situation. Therefore, I assume without better surface roughness there is nothing more you can do with this alloy. You can try the Elgiloy alloy which has a minimum ultimate tensile strength of 2070 at that strip thickness.

 

[israelkk]

Wanted to add that the use of such spring for this application is a wrong approach. The residual stresses in the crack area after the spring forming are residual tensile stresses. Therefore, pulling the eyes away from each other adds stresses to those stresses. Usually such a spring should work to the opposite direction i.e. pushing the eyes closer to each other instead of pulling them away. The shot peening is not completely effeftive because most of its contribution goes to negate the tensile residual stresses instead of adding compressive residual stresses.