440C Stainless & Massive Chromium Carbide Particles

[nickjk]

Hello all,

I currently have a project consisting of a sleeve being used like a inner bearing race.
The material of the sleeve is 440C stainless and is vacuum furnace heat treated AMS 2759/5 gas quench (Rock "C" 58-61)

Recently I had several failures where the sleeve material yielded and the diameter closed in after approx. 400,000 cycles.
The failure was found on three of the 15 units tested, the other 12 seem fine. To be successful they must reach 1,000,000 cycles.

In the past I have had very good results with a product based on the same design but smaller in size.
I questioned if there was a problem with the material or heat treat.

Brought two sleeves to the lab, the one that failed and a smaller one that went 1,000,000 cycles with out failure.
The lab felt from past experience it would be best to start by comparing the microstructure on the two sleeves and correlate to mechanical properties.

The test results revealed the sleeve that failed has 2.93% massive chromium carbide particles while the one that went 1,000,000 cycles had 1.13%.
The metallurgist concluded that percentage of massive chromium carbide particles found would affect the mechanical and fatigue properties of the parts.
I was told the massive carbide particles cause stress risers.
I was also told this is difficult to control because it is not uniform through the material which could explain why the other 12 parts are fine.

I do not know what percentages are considered acceptable?
How can I prevent this from happening in the future, cannot find data from material suppliers?
Will these values vary from different suppliers?

If anyone has experience in this area I would appreciate your thoughts

Attached you will find pics of the Micro-Images

Thank you,
nickjk

 

[Maui]

The carbides you see in this microstructure are not "massive". They are primary carbides that are produced during solidification of the ingot when the steel is melted. They are always there to some extent. The relative size and distribution of the primary carbides is a function of the melting practice and the ingot mold geometry. Ingots that are physically larger and take longer to solidify tend to have more numerous and larger primary carbides. They also tend to have coarser grain structures.

I suggest having your lab tech measure the prior-austenitic grain sizes of both samples for comparison purposes. The Snyder-Graff grain size is what you should measure. The ASTM grain size will probably result in a measurement of "10 or finer" which will be useless for comparison purposes. The part with the finer grain size will likely have better fatigue properties, all other things being equal. But the presence of surface defects and discontinuities in the material may overshadow the impact of small grain size differences.

What heat treatment were they supposed to receive? If your aim hardness is 58 - 61 HRC, then your tempering temperature will be quite low, around 200 - 300F. Does the hardness you measure correlate with the appearance of the tempered microstructure? I would also recommend measuring the hardness of each part too. Let us know what you find.

Maui

 

[nickjk]

Maui,

The report from the tech states "the average equiaxed prior austenitic grain size was estimated to be 10.0 in the section examined as rated to ASTM E112-13 Plate 1A and Plate 1B see fig.3 & fig.4".
For what ever it is worth the Material Mill Cert states the grain size as 8.5.

My customer measured the sleeve Rock "C" at 58

Thank you
Nick

 

[Maui]

The ASTM grain size won't help. You'll need a Snyder-Graff grain size measurement to see any differences between the two samples. The mill cert grain size won't help here either.

What were the hardness values measured on each of the two samples?

 

[tbuelna]

Are you sure this metallurgical condition was the primary cause of the failure? You mentioned the thru hardened 440C sleeve was "used like an inner bearing race". If so, what type of bearing is it used for- rolling element or plain bearing? What are the dimensions of the sleeve? How is the sleeve installed? And what are the loading conditions for the 1M cycle fatigue test?

You also mentioned the "sleeve material yielded and the diameter closed in". Did you see any indications of fractures in the failed parts, or did the material simply yield during the testing?

 

[EdStainless]

And a few more questions.
What was the tempering temperature and process?
What is the calculated contact stress in the bearing (the old smaller one and the new larger one)?

My hunch is that this is related either to the parts being too soft, or the contact stresses being too high.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[tbuelna]

EdStainless-

Based on what the OP stated, I don't see contact stress being a problem. Subsurface material defects can definitely be a problem when it comes to hertzian contact fatigue, since these defects can initiate subsurface shear fractures. But this condition would usually result in surface spalling, rather than the buckling/yielding described. I agree with your hunch that the material was too soft or the part structural stiffness was inadequate for the loads applied in the fatigue test.

 

[nickjk]

I would like to thank everyone for replying

I will contact the lab to see if they can measure the prior-austenitic grain size of both samples (Snyder-Graff grain size).

When I was at the customers facility I had them measure hardness of the sleeve that failed, it was 58 RC. I do not have the hardness of the smaller sample and will have the lab measure both pieces for comparison.

I did not see any fractures in the failed part, the material appears to have yield during testing.

On the part print I specified AMS 2759/5 with a hardness of Rock "C" 58-61. That being said the tempering temp should have been 350F. During initial prototype testing I had tight communication with my heat treat source. At this time the heat treater is selected by the part supplier, I would hope they are following the AMS 2759/5 spec, currently all we check is hardness when parts are received.

This is not a typical ball bearing application. There is no rotation and axial movement is only .030"-.040".
From calculations using Roark's formula in table 14:1 condition 1a the max tensile stress of the failed part is 122,399 psi I believe the max tensile stress on the smaller part was 113,900 psi. There is also a bending stress of 26,640 psi applied.

Between the races there are .079" diameter 440C balls.

The sleeve measurements are approx. 1.031" O.D. x .120" wall x 1.732" lg.

Thanks again for your help

Nickjk

 

[mrfailure]

I think it is a mistake to first jump to the differences between sleeves. Instead, I would recommend performing a failure anlysis investigation to understand how the sleeve failed and under what mechanism. Then you can look at how material properties like microstructure might have contributed. It is quite possibl that your real root cause has nothing to do with maerial differences, but with stresses associated with scaling up the component size.

 

[tbuelna]

nickjk-

It appears my 6th edition of Roark's uses different table numbering than your version. Can you describe the analysis case you are using?

Based on your last post it sounds like this is an inner race for a linear ball bearing. You also mentioned the race is subject to some bending. It would be very helpful to see a sketch of how the inner race is constrained (fixed at one end, fixed at both ends, etc), the contact geometry between the ball complement and the inner race (how many rows of balls, how many balls per row, internal clearance of the bearing, axial spacing of the rows, etc), and how forces/moments are applied to the bearing system (thru the outer race?).

It seems possible that your inner race wall could have "collapsed inward" if not adequately supported with a moment applied that produced high local inward radial forces at opposing ends of the bearing. Depending on the number/spacing of balls in a row and the internal radial clearance in the bearing, the radial force at opposing ends of the bearing might only be transferred thru 2 or 3 adjacent balls at each end. So you might want to check your inner race wall for buckling including this radial load condition.

 

[nickjk]

tbuelna

I am using formulas for stress and strain due to pressure on or between elastic bodies Table 14.1 Roark's 7th edition
Condition Sphere on a flat plate.
The bending stress is because the race is slit. When load is applied the race closes in removing clearance, when load is removed the race opens up and clearance is restored.
Axially the race is constrained at the bottom with a light preload force pushing down on a locator surface. Radial the race is allowed to float, The balls centralize the race when load is applied. There is approximately 13 rows of balls with approx. 36 balls per row. The internal clearance is approx. .001-.0015
The Lab stated they did not see any stress fractures during inspection.

As described above, three units failed before 400,000 cycles. The remaining units on test are now at 800,000 cycles without any sign of failure.
I had the supplier measure the hardness of the two parts being studied. Hardness of the larger part that failed was 57 Rock "C", hardness of the smaller part that passed was Rock "C" 56. Part print specification is Rock "C" 58-62.

Moving forward per request of the lab, a sample of the material prior to heat treat will be kept, a sample part prior to heat treat will be kept and a sample part after heat treat will be kept. The micro-structure of these parts will be studied to determine if the problem with the carbides is in the original material or happening during heat treat. Note, currently I do not have a sample of the material prior to heat treat.

Any further suggestions would greatly be appreciated.

Thank you,
Nickjk

 

[metengr]

As mentioned earlier, you should have a failure analysis performed on one of the failed test pieces to characterize the fracture surface(s) and determine if the microstructure was to blame.

 

[nickjk]

Metengr

Specifically what type of failure analysis are you suggesting be performed on the test piece?
Exactly what would you test or look for?
The sample piece is not large enough to check for strength values and at this point I do not have a sample prior to heat treat.

Nickjk

 

[metengr]

Nickjk;
I would carefully examine the fracture surfaces in a metallurgical lab to determine the failure mode. Once the failure mode is identified, determine if it is associated with the presence of blocky carbides or other microstructural anomalies related to fatigue crack propagation. The effort would include fractography and metallographic examinations. At this point, I really don't believe you have any smoking gun related to premature failure of the sleeves from fatigue testing.