Fatigue failure of low stressed shaft 8

[GregPerry]

We have a number of gearboxes installed in 1991, since then 10 shafts have failed by means of fatigue.

The failures happen at two positions. I have calculated the following stresses based on the material and shaft geometry:

Position 1: Stress = 67 MPa, Endurance Limit = 82 MPa @ 10^6 cycles
Position 2: Stress = 65 MPa, Endurance Limit = 137 MPa @ 10^6 cycles

According to the calculations the shaft should not fail.

The material used is BS970 817M40 condition T. Testing reveals the following:
UTS = 1089 MPa specification 850-1000MPa
Yield = 932 MPa specification 650 MPa min.
Elongation = 16 specification 13 min
impact resistance= 26 specification 35 min (under)
cleanliness ASTM E45 A=2, B=0, C=2-3, D=1
Grain size ASTM E112 size 5-6

My questions are:

1. Could the material condition reduce the endurance limit of the material to such an extent that fatigue failure would be garanteed?

2. Is there a graph, index or reference literature that shows some correlation between material cleanliness and fatige life and/or grain size and fatigue life.

Thanks,

Greg

 

[butelja]

Surface finish and corrosive environments can both have dramatic effects on the endurance limit.

Is this an internal shaft where the loads are known? If not, could excessive overhung loads be overstressing the shaft? Have you taken into account stress raisers at transitions (fillets, keyways, splines, etc) in the shaft?

 

[CoryPad]

Answer to question 1:

Yes, the material condition could reduce the fatigue resistance. The plastic zone size scales with the inverse square of yield strength, so your steel being very strong could greatly reduce its fatigue crack resistance. Also, the low impact strength is an indication that fracture resistance is lower than it should be. And lastly, the grain size seems to be high (low ASTM number), which can reduce resistance.

Answer to question 2:

There is significant literature on this subject. For example, these two articles appear to be very appropriate for your application. They both have "steel", "inclusion", and "fatigue" in the Title, Abstract, or Keywords.

Inclusion engineering for improved fatigue response in forged AISI 4140 steel, International Journal of Fatigue, Volume 19, Issue 1, January 1997, Page 95
S. R. Collins and G. M. Michal

Morphological and analytical characterization of inclusions: Relation with processing parameters and properties of use, Materials Characterization, Volume 36, Issues 4-5, 6 April 1996, Pages 321-326


For general trends, you could look at ASM Handbook Volume 19, Fatigue and Fracture. It shows trends such as increasing fatigue resistance with decreasing grain size and decreasing inclusion content.

Cory
cpadfield@omnimetalslab.com

 

[TVP]

Greg,

Cory Padfield provided some good answers to your questions, so I am going to offer some opinions that may not exactly answer your questions.

1. The stresses that you listed seem to be way too low to cause fatigue failure in this type of steel (Cr-Ni-Mo, similar to AISI/SAE 4340). Are you sure these are the correct stresses? The endurance limit for this type of steel should be much higher-- more like hundreds of MPa, even with Kt = 3.3, fully reversed loading, etc.

2. This alloy has one of the largest, if not THE largest, fatigue databases in the world. Cory mentioned Volume 19, Fatigue and Fracture, of the ASM Handbook series, which has some good data on this alloy. Another book from ASM, Atlas of Fatigue Curves, has some excellent data, including data on how inclusion size affects fatigue life. MIL-HDBK-5 has a huge amount of data on this alloy, but I didn't see anything at strength levels this low-- most of it was at tensile strengths of 190-220 ksi (1300-1500 MPa).

3. The low impact values, coupled with the high inclusion ratings, indicate some steel problems. The relatively high grain size of ASTM 5-6 is used for high hardenability, but this alloy, coupled with the low overall yield and tensile strength spec, means that you don't necessarily need it-- 6-8 would improve fatigue and fracture resistance. However, I do not know if any steel producers regularly produce this steel to an ASTM grain size of 6-8.

4. Back to the inclusion ratings: a typical ASTM E 45 worst field rating for a steel of this composition would look like this:
A B C D
thin 2.0 1.5 1.5 2.0
heavy 1.0 1.0 1.0 1.5


Your C ratings are much too high. The A rating is a function of sulfur content, so if it requires machinability, and the spec allows for anything more than 0.015, then you won't be able to reduce these below the 1.5/2.0 level. B's & D's depend on the deoxidation & grain refinement practice, as well as the level of sophistication used in fluxing, mold powders, electromagnetic mold stirring, etc. Since fatigue life is a consideration, you should try to keep everything below 1.5.

5. If you can give me a more complete loading history, I can give you some more definite numbers on fatigue life. Is this constant amplitude loading? What is the max stress, min stress, and mean stress. What is the R ratio (ratio of min stress to max stress)?

6. Last point is about processing. Decarburization will have a significant effect on fatigue strength. Have you examined any of the shafts for this? Are the shafts through hardened or only surface hardened? Hardness profile on induction-hardened components has a significant effect on fatigue life.

 

[rustbuster1]

Wow! I hope you can digest all this excellent fatigue theory. My experience with shafts that fail consistently is that significant loads are often dynamically induced, especially with drive gear. Starts and stops can be brutal, depending on how they are controlled - i.e. how the torque is applied, etc. If the fatigue breaks are torsional you can suspect this as a contributing factor. Fatigue also starts at small surface features, so you should find out what has been the initiating feature for these shafts. There is a lot of useful informationin the fractographic evidence - number of intiation sites, shape of crack propagation front, etc.

Finally, have you considered shot peening to upgrade the fatigue resistance of the shafts. This can do wonders at minimal cost - assuming the surface can be made accessible for peening.

 

[GregPerry]

TVP

Thanks for the input, herewith some more information:

The gear unit drives a conveyor and has a fluid coupling so there appears to be no shock loading. I unfortunately do not have a comprehensive loading history so I have calculated my design on the maximum design loading 344 kW and an overhung load of 10 tons. I have assumed it is constant loading but this may be incorrect.

The actual operating condition I have witnessed are approximately 30% below the design loading.

For the endurace limit I have used the following equation:

Sn=Sn'.CL.CS.CD.CT.CR

Where:

Sn'= 850.05 (Specification is 850-1000 MPa)
CL = 1 (loading factor for bending fatigue)
CS = 0.91 (surface finish factor)
CD = 0.704 (size effect - diameter 220 shaft)
CT = 1 (temperature effect)
CR = 0.753 (reliability factor for 99.9%)

Thus Sn = 205 MPa

Position 1: Keyway Kf = 2.5 - Endurance limit = 82 MPa
Position 2: Taper lock rigid coupling Kf = 1.5 assumed - Endurace limit = 137 MPa

I have revised our material specification to include cleanliness and grain size. Cleanliness has been limited to 2 for each index, you sugessted 1.5 I will consider revising this but our materila is not of the best quality here. Grain size I have limited to a minimum allowable index of 7.

Greg

 

[TVP]

Greg,

Thanks for the clarification. I am not trying to bash your profession, but I sometimes have difficulties with Mechanicals using "Endurance Limit" to describe a "design" instead of a "material", hence my confusion with the very low endurance limit of 82 MPa. Now that I understand your derivation, it seems quite plausible that 82 MPa is the limit.

I think you will find that revising your specification limits should improve the quality of your steel. The impact value of the incoming bar is a great indication that the previous steel was deficient-- hopefully you will see this number improve with the more stringent specification of inclusion ratings and grain size.

Having read the other members contributions, I think that failure analysis of the shaft(s) can definitely help you further. A full metallurgical study using Scanning Electron Microscopy can yield a lot of information on material condition, failure origin, effect of inclusions, etc. You may want to consider an outside source for this type of testing-- Cory Padfield of Omni Metals Lab does this kind of thing, and he is already familiar with your problem.

My last thought is regarding Finite Element Analysis. Do you have access to this? Have you considered performing this type of analysis? You can gain a lot of insight into how small details like radii, blend transitions, etc. can effect the stress distribution. Perhaps you can reduce the stress intensity around the keyway, and thus improve the fatigue life of the parts. Good luck with everything!

 

[metman]

Greg
You have been given a lot of good input but I agree most strongly with TVP's last reply. The primary thing you need to learn is where and why is the crack(s) initiating. If it starts at the keyway, is there any possibility of redesigning to elilminate the sharp corners (stress risers). This can drive your stress up to 3 times nomimnal stress. If it is starting from inclusions, cleaning up the chemistry and finer grain size will help but this approach seems to me as marginal design. If you cannot improve the design by changing the geometry (blunt or eliminate stress risers or increase section size), then I would be looking at changing the material. Alloys high in Nickel content have higher fracture toughness and therefore are great at blunting crack propagation such as 3310, 9310, or 4800 series. If one these alloys do not have high enough YS without carburizing you mightr consider a Maraging Steel. Yes the material cost can be prohibitive but if you do not have a relatively large mass (one big part or multiple little ones), the material cost becomes insignificant compared to all the ramifications of downtime and you might even circumvent lab testing if macro analysis can pinpoint the keyseat as the culprit, in which case you probably need a hefty solution. Jesus is the WAY

 

[tedg]

Greg:

Thank you for starting a discussion which has brought out all kinds of excellent input. I agree that it is important to determine the cause of the fatigue failure, but wonder how much time you have to study and investigate. For a FAST CURE of your problem, I would suggest you switch from an alloy similar to AISI type 440 to one of higher fracture toughness eg a precipitation hardening or - if necessary -even a maraging steel. These alloys cost more but can get you out of trouble while you develop a permanent solution.

If you need info on fracture toughness (which is different from toughness determined by IZOD or CVN) or specs or sources for the suggested grades, please advise.

ted gerson
ftg@gerson.ca

 

[GregPerry]

Thank you all for the input.

I think that the combination of a poor material quality, underestimated stress concentration factors and additional freting of the saft has eroded the endurace limit to such an extent that failure was iminent.

I will be looking at the following options to see which will be more cost effective:

1. Select a different material for the shaft
2. Increase the size of the shaft
3. Mount the gear unit on the floor with a flexible coupling

Ted I may take you up on those specs at a late date.

Thanks
Greg

 

[corus]

Fatigue of rotating shafts is a complex problem that involves not only the stress concentration features of the shaft, keyways, transitions etc., but also surface finish, size effects, reverse torque applications, and mean stress effects. At Corus TTC we developed software for assessing shafts of general shapes, which both evaluates stresses and assesses these against fatigue limits, for a range of materials, taking into account the various factors. If you require further details then contact me.

 

[tedg]

Greg:

This interesting thread seems to live on forever.

Have the various suggestions helped you to solve the shaft problem, or develop a program designed to solve it?

A quick update would likely be appreciated by all of us.

Ted Gerson

 

[GregPerry]

Ted,

We have inmplemented a improved material specification based on some of the information supplied in this forum and believe this will be suffiecent as the fatigue calculations show the stresses are acceptable.

We included:
1. A high temperature normalise, not done previously
2. A cleanliness specification
3. A grain size specification
4. And require a microstructure determination to ensure a Tempered Martensite grain structure.

Some of these shafts only failed after a number of years so it is still early days.

Here is another question for all of you out there:

What stess concentration factor would you use for a rigid flange coupling which locks onto a solid shaft by means of a taper locking element? I have assumed Kt=1.5 but I need to validate this assumption. Does anyone have some reference where I could do this please.

Greg