Roll of steel cleanliness in failure 1

[UCengno1]

We have experienced a rash of failures of a highly loaded part due to what is believed to be low cycle fatigue. We have identified the existence of small ridges in a root of a gear tooth and this was determined to be the primary failure initiation point. The mode of failure of the low alloy steel part, i.e. 4820 and 9310, is a rapid, explosive failure. There is little evidence of fatigue in the failure so we are assuming low cycle fatigue as the culprit.

Back to cleanliness. Our metallurgist believes that the existance of longitudinally aligned, in bar stock, non-metallic inclusions are the primary culprit for the explosive nature of the part.

We have addressed the stress risers but should we continue to be concerned about cleanliness? The material in use meets AMS requirements.

 

[unclesyd]

Have you verified the MOC?
Have you checked for Lead or Selenium?
What is the specified heat treatment for the gears?
What is the gear type?

 

[TVP]

Have you verified that the microstructure is correct? Case depth and hardness? NMIs are certainly an area of concern for fatigue, especially in highly loaded gears. What about residual stresses? How were the gears manufactured?

 

[UCengno1]

Need clarification on "MOC"

We were looking primarily for sulphide stringers in inclusions.

Carburized/hardened to 56-60 Rc , case depth .015-.02, core 32-46 Rc

6/8 pitch, 12 tooth

 

[CoryPad]

Material(s) Of Construction

Regards,

Cory

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

 

[UCengno1]

The gears are hobbed from 9310 or 4820. Vendor can use either.

 

[TVP]

UCengno1,

It is difficult to say whether or not the NMI's are strongly linked the failure. The poor surface condition was definitely a contributor, so it is good that you improved this. We would really need to have some additional information on the following:

1. Case microstructure - presence of any NMTPs (non-martensitic transformation products), intergranular oxidation, carbide networks, and retained austenite? If the retained austenite is in the normal range of 10-20% and there are none of the defects I specified, then this can be ruled out.

2. Residual stresses - the surface residual stress and the case-core transition are important items to understand. It is possible that the case-core transition area has very high residual tensile stresses, which can lead to catastrophic failure when coupled with a high stress intensity at the root (poor surface and high stresses).

3. Keep in mind that SAE AMS standards allow inclusion levels that are much too high for highly stresses gears. Do you know the sulfide and oxide inclusion ratings (ASTM E 45 or other method) for the steel bars? If not, do you know the sulfur and oxygen content? Oxygen levels should be < 50 ppm and sulfur below 0.025%, preferably below 0.010%. I would certainly specify cleanliness according to SAE AMS 2300K (Premium Aircraft Quality) instead of SAE AMS 2301J (Aircraft Quality) for highly stressed designs. Have you performed stress calculations for this design? Any numbers that you can share?

 

[redpicker]

As has been mentioned, you need to verify the chemical composition and microstructures (both case and core) to be sure there is not something out of whack there. With a chemical analysis of a failed part, you will also get the actual sulfur content in the material. As TVP states, the sulfur should be less than 0.025% and, particularly with highly stressed parts, perhaps lower limits are justified.

Inclusions are often blamed for failures (particularly when nothing else can be identified), but all steels contain inclusions. Unless these failed parts have excessively higher inclusions than is found in normal production, then the problem is likely elsewhere.

It should be noted that, particularly with the nickel-bearing low alloy steels (as both 4820 and 9310 are), "medium" sulfur levels of 0.015-0.025% are often given as "aims" to improve some machining operations (hobbing is an operation that the higher sulfur levels offer improvement, I believe). If low (less than 0.010%) sulfur levels are needed for you application, you will need to be sure it is specified since a lot of these steels are specifically made with higher sulfur levels.

ab

 

[UCengno1]

TVP and redpicker,

Sulphur content is ranging from .015-.025%. We are really only interested in attacking sulphur content if it buys us something tangible. It appears we are within specification but are pondering whether it is worth the effort to go beyond this.

We have been down the road of microstructure issues. We are watching the heat treatment closely and have verified fine tempered martensite in the failure region.

I am interested in how we could recognize residual stress in the failure region. Any thoughts here.

All of the guidance given so far has been very valuable. Thanks.

 

[redpicker]

If you try to put the pieces from a broken piece back together, do they fit? If they won't (and there isn't much yielding), that's a pretty good indication there were a lot of residual stresses in the part prior to failure. If they fit together nicely, except where obvious yielding has occurred, then likely not. You can get x-ray diffraction performed on new parts to quantify the residual stresses, if necessary.

As far as the sulfur content goes, that can be a bit more difficult. You mention you have experienced a "rash" of failures. Has this part performed satisfactorily in the past? What was the sulfur level in those parts? If the "good" parts have .003 Sulfur, that could be your answer.

rp

 

[NickE]

"If you try to put the pieces from a broken piece back together, do they fit?"

I was always told that the above is the WORST thing you can do with a fracture surface.

Nick
I love materials science!

 

[redpicker]

Apparently, NickE, you were always told wrong.

rp

 

[NickE]

Really I've never been taught or read anywhere that re-assembling a fracture surface is good.

It seems to me that the very fine details of the fracture surface are vital to examination under the SEM. These features can easily be destroyed by re-assembly.

Here is a quote from "Understanding How Components Fail" D.J. Wulpi; ASM 1983g.2

"Do not put the mating piecies of a fracture back together, except with considerable care and protection. -- Even in the best circumstances, fracture surfaces are extremely delicate and fragile and are damaged easily, from a microscopic standpoint.... ....Many such examinations have been frustrated by careless repositioning of the parts, by careless packaging and shipping, and by inadaquate protection from corrosion, including contact with fingers."

Also in my work it is common to have an inconclusive failure analysis due to damage of the fracture surface.

So in summary, I respectfully disagree, there is no need to re-assemble the fracture surface, and it can be highly destructive to the information contained in the surface.

Also in hard materials that undergo rapid brittle failure (as the OP states) there isnt going to be any yield zone.

I guess what Im saying is what is the concensus of the collected knowledge here? I would prefer to send a part out for XRD before I would re-assemble a fracture.

(I'm also trying to say I dont know everything and am interested in knowing whether redpicker's technique is valid.)

 

[aviat]

I have the same book that NickE has, and I have seen what he refers to. I have seen other books on Failure Analysis that repeat the same precaution as regarding mating the broken pieces together. Also the labs tell us not to assemble the pieces as in the site below that says under mechanical damage.
“DO NOT attempt to fit two mating fracture faces together.”

http://andersonlabs.com/Failure.html

I just phoned a company that does this kind of work and he said that when you put the two pieces together that you smear the metal and it makes it more difficult to determine problems. He also said 75% of people still put the parts together, and that it is only human nature that people want to see if it fits.

The following site has some failure problems associated with gears. From what I understand of the problem it would be case crushing or in Wulpi’s book and other ASM info it is called Subcase-Origin Fatigue. (fig 10)

http://www.elecon.com/gearworld/dat-gw-failure.html

 

[metengr]

Apparently, NickE, you were always told wrong.

I would have to believe that the statement above by redpicker was intended to be a sarcastic joke?

 

[redpicker]

Since we are quoting, here is an exerpt from ASTM (American Society for Testing and Materials) Standard A370, with regards to measurement of elongation on a tensile specimen:

13.1.4 Fit the ends of the fractured specimen together carefully and measure the distance between the gage marks to the nearest 0.01 in. (0.25 mm) for gage lengths of 2 in. and under, and to the nearest 0.5 % of the gage length for gage lengths over 2 in.

Why would they specify doing such a thing if it was the worst thing you could do?

My argument is not that it possible to damage delicate features of a fracture surface by fitting the surfaces together, but rather that such an action is not "the WORST thing you can do to a fracture surface". Frankly, there are much worse things you could do (consider what could be done with a hammer, an acetylene torch, or a mixture of hydrochloric and nitric acids, to name a few).

By assembling the broken pieces back together (obviously, you would not want to do this until after any SEM work was completed on the pieces in question, OK?), you might be able to tell if there was a significant amount of residual stress that was relieved by the fracture. That is the question and I know of no other way to answer it. XRD can detect residual stress patterns in new parts, but will not identify if they existed in the failed part or if they existed in the failed region. This might quickly answer the OP's question instead of having him scrach his head at XRD results or admire the fine details detecable with the SEM (but do very little to help him with his problem).

The SEM is an amazing instrument, but after more than 25 years of working in labs that have them (as Operator and as Lab Supervisor), I have found out that I can almost always tell as much (perhaps more) with a stereoscope and a metallograph, and a whole lot quicker. Knowing what to look for is often much more important that what you look with.

rp

 

[redpicker]

[quote}I would have to believe that the statement above by redpicker was intended to be a sarcastic joke?[/quote]
Well, yes, that, too...

rp

 

[unclesyd]

I may have missed it, is this a through hub failure or tooth failure?

Low cycle fatigue beach marks in hardened steels at times are hard to ascertain visually or optically and may require SEM to verify. Be very careful in preparing the surface for examination.


If you have several failures why not play with one.

PS:
Do you know how to find the origin of the failure?
Drop it and it will hit and mess up the origin.

 

[swall]

Maybe we should take a poll on this. Fracture fitting vs leaving it alone. Myself, I never fit fractures from failures back together and I have been known to scold other people handling the parts that try this (especially QC types). Tensile test bars--don't just fit them, rap them with a small hammer if you have to!