Fatigue question

[Neubaten]

Hello everyone,


I have a part that is made from a steel containing 0,35 C and 1,5 Mn. It is quenched & tempered to obtain a hardness of 47 HRC throughout all its section, and then it's shot-peened.

The parts go then to fatigue testing, flexion and torsion combined. There are two test ranges. One with higher loads and less number of cycles specified and another one with lesser loads and more N of cycles. The parts that go to the higher load test do very well (vs. the specification), but the parts that do less loads break under the number of cycles specified.

Any of you have any speculative idea, given the facts I've told, why a part does well in high loads and not that well in less loads??.

 

[davefitz]

Not sure, but the ratio between the stress per cycle and the number of cycles to failure follows a fairly predictable curve- roughly a doubling of the alternating stress will reduce the fatigue life by a factor of 20, up unitl the fatigue curve meets a point of inflection at the hi cycle - limit cycle transition.

So, if the high cycle had 4 times less stress, one might expect 400 times more cycles to failure.

What was the ratio of stresses to ration of cycles in your test case?

 

[davefitz]

also, the effective alternating stress needs to be adjusted for nonlinear-plastic yielding in the high load-low cycle case, so if this factor was not adjusted in the test it might explain the departure from expected results.

 

[NickE]

Fatigue at High Stress Low Cycle is generally controlled by surface effects. (stress concentrators, etc...)

High Cycle Low Stress fatigue life is generally controlled by internal material defects.

 

[Maui]

Neubaten, more detailed information will have to be supplied to the experts in this forum in order to provide you with an answer that is useful to you. What specifically is the grade of steel and the application? What are the maximum applied stresses in each test, the stress range, and the number of cycles? One application that could fit the description you provided would be an alloy steel bandsaw blade. These blades are subjected to bending, torsional, and tensile stresses as they are cycled on a typical industrial cutting machine, and a routine test that is performed to determine if the blades were manufactured properly is a fatigue test. There are several factors that influence the behavior of a component that is subjected to cyclic loading including maximum applied stress per cycle, stress range, number of cycles, material geometry, surface finish; orientation, size, and geometry of existing flaws; exposure to corrosive agents, etc. One previous thread that covered this topic is thread330-128281, and I strongly recommend that you review it to gain some additional insight. If this is not sufficient to answer your question, supply us with more detailed information.

Maui

 

[Neubaten]

First of all, thank everyone for your help.


The part we are talking about is this one:

http://img264.imageshack.us/my.php?image=bar1ew1.jpg

It's a bended tube with 3.5mm of wall thickness.

The steel is quenched at 815ºC and tempered at 270ºC.

The ultimate strength is around 1.500 MPa and the yield strenght is around 1330 Mpa, with an elongation of 11%.

The loads are applyed on both ends, alternativelly. The arms go up and down a certain distance.

The part is shot peened with an Almen factor of 12A.

The parts have to do 50.000 cycles with +50/-50 mm displacement, and 239.000 at +40/-40.

The parts do more than O.K. at the +-50 mm. requirement, but they break at around 140.000 in the other test.

The test are not one after another, they are different tests with different parts.

 

[NickE]

Have you performed failure analysis to determine the origins of the failure?

Nick
I love materials science!

 

[Neubaten]

Also, the force, read at the fatigue testing machine is about 2300 N.

The tubes are bended before the heat treatment.

 

[TVP]

The short answer to your original question is the addition of 10 mm of displacement (from + 40 to + 50 mm) does not necessarily add 20% to the stress on the component. I would suppose that the complex loading conditions of bending + torsion result in a similar stress when the displacement is + 40 mm or when the displacement is + 50 mm.

As NickE mentioned, you need to determine whether or not the failure mode is the same, i.e., are the fatigue cracks initiated at the surface for both load cases, or are some of the cracks initiated from internal defects like inclusions, etc. If all of the cracks initiate from the surface, then I would increase the shot peening intensity, maybe in the range of 0.18-0.24 mm.

 

[Neubaten]

Hi again, this is an old post, but I feel in debt because of your help.

A failure analysis and some micrographs showed a decarburized layer, formed by a 0.03mm layer of total decarburization, and a 0.05mm layer of partially decarburized steel. The crack initiated in the totally decarburized layer (the most external).

So, I made the following decission: I increased the hardness of the steel to 50 HRC, by downgrading the temper to 220ºC, instead of the 270ºC we were using before. This hardness is also allowed by our customer.

I did not touch the Almen intensity, but I shortened the saturation time, and then I exposed the part to a longer peening time (double the initial time, more or less).


And then it happened: Problem solved.

The part has improved a massive 800% in the number of cycles at the "problematic" test.

The funny thing is that I can't clearly explain why I tried that, apart from following TVR and NickE advices. I think I was reading at the time some papers by J.O. Almen and other classics of shot-peening, and somewhere it appeared the sentence "shot-peening better for harder steels".

Also I figured out that maybe I could "peel" the external decarburized layer using shot-blasting. Indeed, the ferrite layer dissapears after extended peening (not the 0,05 layer, though).

 

[dbooker630]

The .03mm total decarb can be eliminated. Combined with your .05mm partial decarb, that is significant enough that is can be addressed with proper atmosphere control. I would not depend on a shot peener or blast cleaner to remove heat treat irregularities beyond obvious loose scale.

 

[Neubaten]

I know, dbooker630. We are investing in a furnace with protective atmosphere for the final manufacturing line.

But what we had then were tubes that had the decarburized layer BEFORE heat treatment.

An engineer of my department (very experienced and with several diplomas in his pocket, by the way) asked for the tubes allowing the supplier 0.10 mm of decarb, when we have a specification of just 0.05mm partial decarb MAX.

She did a peculiar move, no doubt.

 

[dbooker630]

Maybe there was a price break. Got to watch those pseudo-metallurgists