Fatigue life of non standard threads 1

[MichelM]

I am actually involved with fatigue of threaded hollow shaft used to tie a stack of rotating components. I am looking for information on how to establish the LCF life of the threads. The approach based on the ratio of nominal stress / ultimate strength (frequently used for fasteners) is not applicable. For cost reason, the threads are cut (instead of rolled) which reduces the LCF life. The size of the thread is around 3 inches diameter with a pitch of 14 threads per inch. In comparison with standard fasteners, the nut has a relatively small thickness and small outside diameter.

1- What is the recommended approach to establish the LCF life of the thread?
2- What is the debit of a cut thread versus rolled thread?

 

[unclesyd]

Here is a excellent place to start your project courtesy of [b[metengr[/b].

http://www.fatiguecalculator.com/index.htm

 

[TVP]

MichelM,

Do you have access to finite element analysis software? Do you have reliable data on the forces involved? Is this joint subjected to variable amplitude loading? How accurately is the fastener preload during tightening? Do you have good data on the material (composition, strength, fatigue crack growth vs. stress intensity, etc.)? Do you have an understanding of the residual stresses prior to rotation?

The answer to 1) is to conduct a strain-life fatigue analysis or possibly an analysis based on fracture mechanics. The link that unclesyd provided is a good place to start. The answer to 2) depends on the surface roughness, fracture toughness, and microstructure of the base material.

 

[MichelM]

Thanks to both of you for the feedback.
I visited the site

http://www.fatiguecalculator.com

that I found interesting. The approach I normally used to evaluate the Low Cycle Life of mechanical components (shaft, housing, etc.) is very similar to the stress-life method of the fatigue_calculator.

I have access to finite element analysis software. The model is 2D axisymetric with contact elements between the threads and between faces of adjacent parts. I actually used no friction in the contact elements. The loads are involved are constant amplitude and well known. The preload is applied with an hydraulic actuator with a controlled elongation of the shaft. I also have good data for the material (strength, endurance limit) but I did not try to perform crack growth analysis. The residual stresses are not known.

TVP,
You recommend to conduct a strain-life fatigue. What would be the advantage of a strain-life fatigue versus a stress-life fatigue? I would probably need to run a plastic analysis with multiple load cycles and evaluate the residual stresses? One of the difficulty with the stress-life fatigue is to establish a Kt at the root of the thread. Elastic concentrated stress is known from finite element analysis but the nominal stress can’t be the only P/A as for standard fasteners.

For the debit of cut versus rolled threads, I would expect results from technical paper with test data. As mentioned by unclesyd in the thread725-102464, the surface roughness in the root of a thread is difficult to measure. Is there available information between thread material microstructure and the fatigue strength? I will contact ASTM to obtain copy of STP 1236 & STP 1391.

 

[TVP]

Michel,

What is the chemical composition of the material you are using? How has it been processed? Quenched and tempered steel with a hardness of...? Annealed copper wire?? Something else? I have some fatigue data for rolled vs. cut threads in steel, so let me know if this applies to your situation.

Regarding the type of analysis, stress-life is used for high cycle fatigue, not low-cycle fatigue. The whole concept of an endurance limit is that is applies to the fatigue strength at > 10⁶ cycles. It is actually a flawed concept, but that is a different matter. Strain-life analysis is used to better predict the behavior in the low cycle fatigue regime. A quote from the fatiguecalculator website is "This method is used for finite fatigue lives where plasticity around stress concentrations is important." Cut threads on a shaft qualify as stress concentrations.

The first thing to understand is the stress-strain behavior at the thread root. I am not an FEA expert, so I can't really comment much on the suitability of an axisymmetric model using contact elements. You should be able to determine the stresses at the thread root after tightening. You should not need to include a separate Kt factor-- the analysis software will show the stress at the thread root. This stress will be the mean stress.

Next you need to understand what the stresses will be during cyclic loading. This will be complex due to the nature of the assembly-- a collection of parts that have been mechanically fastened together. FEA can be very helpful here as well, but it will require a more sophisticated model. The cyclic stress is then added or subtracted to the mean stress with each reversing cycle which can then be used to obtain the fatigue life.

I may be able to provide you with some cyclic material properties that you can use for a strain-life analysis. SAE J1099 has some properties for commonly used steels and a few other materials.

 

[unclesyd]

Bar Steel Fatigue Database
Has both stress and strain life charts on some common materials.

http://www.autosteel.org/AM/Templat...TaggedPageDisplay.cfm&TPLID=11&ContentID=5842

 

[MichelM]

TVP,

The principal materials that I used are:
Steel AISI 17-22A heat treated for a hardness 35-40 HRC,
Steel AISI 4340 heat treated for a hardness 40-45 HRC and
Nickel Inconel 718 solutioned and precipitation hardened for UTS= 180 ksi.

Regarding the low or high cycle fatigue, different configurations of threaded specimen were tested on load frame. The number of cycles to rupture was from 5x10e4 to 10e6 cycles. I want to design for more than 5x10e4 cycles. Being more familiar with the stress-life, I used this approach up to now. Even simpler approach exists for standard bolt and nut joint; lifing is frequently based on the bolt nominal stress (force fluctuation / stress area) in the bolt. Nonetheless, I will look at possible advantages of the strain-life method.

For the analysis, I agree, the FEA give me the concentrated stress at the root of the thread. For example, I get a concentrated stress of 200 ksi (elastic model) in the root of the thread and a nominal stress of 13.3 ksi across the shaft section. This concentrated stress is due to the axial load transmitted through the shaft and also to the bending of the thread flank as a short cantilever wide beam. For the Kt, assuming two extreme values:
- Kt =1.0, calculated life is 1 cycles which we know does not reflect the reality,
- Kt = 15.0 (ratio of the concentrated stress / nominal tension across shaft section), calculated life is >10e6 cycles which also does not reflect the reality.
So, for me the Kt in the stress life approach plays a very significant role.

Regards.

 

[unclesyd]

Here are two sites that I had marked for checking out that might be of interest.

The Aerade site has a lot of information, but the site is rather convoluted. I was able to get to several indexes that carried some interesting topics.

http://www.machinedesign.com/ASP/viewSelectedArticle.asp?strArticleId=57766&strSite=MDSite&catId=0

http://aerade.cranfield.ac.uk/subject-listing/esdu/ES171.html

Anecdotal:
I am going after the inclined nut paper in hope it will be another validation of debate that happened many years ago. In analysing bolt failures in highly loaded and high cyclic service my comments on the data presented at that time was "It ain't that simple"

 

[TVP]

MichelM,

Based on your test results, I would say that a stress-life model may be able to adequately address the fatigue situation you have described. The link from unclesyd is a good review of FEA and fatigue, and it references MIL-HDBK-5, which has some good fatigue data on 4340 and I believe Alloy 718 (Inconel 718). You can obtain MIL-HDBK-5J by using the following link (caution- huge file = 79 MB):

http://assist.daps.dla.mil/eAccess/...689EBD&s_key=39DF30932220F8CBFCD2A8BEBDD7B244

Check out the following thread for data from a journal article on fatigue life of rolled vs. cut threads:

thread725-88946

The steel was 4340 heat treated to 37-41 HRC. Overall I think you have a good understanding of the fatigue life, you just need to work through some of the specific issues related to cut vs. rolled. As previously mentioned, this becomes a difficult thing to predict, but the data from the above mentioned paper should be a good place to start for both of the steels you mentioned. Definitely review MIL-HDBK-5J for material properties and the discussion on fatigue.

 

[MichelM]

TVP & Unclesyd,

I have access to MIL-HDBK-5 at work, I will look into it. I just received STP 1391 (good data on fastener fatigue) and some papers from STP 1236. I used as reference the ESDU Papers 84037 ‘‘Fatigue strength of external and internal steel screw threads under axial loading.’’ and 68045 ‘’Fatigue strength of large steel screw threads under axial loading’’. I will do some reverse engineering around the standard bolt geometry (fastener with nut loading and cut/rolled threads) and I will apply the findings to my threaded shaft having non-standard threaded joint geometry.

Like you mentioned ‘‘It ain’t that simple’’.

Regards,
Michel

 

[unclesyd]

You might want to take a look around. The search engines sometimes haven't indexed all the good stuff.

http://www.nasatech.com/

 

[arto]

ASME B&PV Code Sec VIII Div 2 App 5 has a bit on Fatigue of bolts - Fatigue strength reduction factor >4; min root radius = 0.003" , etc.

 

[unclesyd]

You might want to check the papers coming out of this conference next month.

http://www.fatigue.org/Bolt-project/bolt.html