Fatigue limit 2

[jacques29]

One of the product we design is a rotor can for an Electro Hydraulique Power steering for the automotive industry. To prove that our design is going to whistand its life expectancy our client ask us to perform a sort of fatigue test on the overall rotor can assembly. As all the parts involve use ferrous material we are planing a 10^6 cycles test with a load representing the worst condition. I would now want to back up this assumption with more robust data. My first idea is to look for the S/N curve of the different material and proove that at 10^6 cycles the material we use have reached their fatigue limit.


Do you know where I could find this information?

Thanks for your help

 

[TVP]

Before suggesting a resources for fatigue data, I wanted to ask a couple of questions:

Do you know what steel alloy is used?

Do you know how the manufacturing process (stamping, forging, etc.) affects the nominal strength of the material?

Do you have a good understanding of the surface characteristics of this part (surface roughness, presence of micro or macro-cracks, geometry changes that locally modify the stress field, etc.)?

Do you have a good understanding of the loading environment in terms of forces (min & max), cycles, temperature, etc.?

Is it appropriate to take the maximum stress and apply it for 10⁶ cycles as a means to validate this design? The effects of mean stress and variable amplitude loading would not be captured by this type of test.

Now, good places for fatigue data are the following:

ASM HANDBOOK Volume 19 Fatigue And Fracture

http://www.asminternational.org/Tem...=Ecommerce/ProductDisplay.cfm&ProductID=10494

Atlas of Fatigue Curves

http://www.asminternational.org/Tem...=Ecommerce/ProductDisplay.cfm&ProductID=10557

S/N Fatigue Life Predictions for Material Selection and Design

http://www.asminternational.org/Tem...=Ecommerce/ProductDisplay.cfm&ProductID=10332

Fatigue Data Handbook

http://www.ttp.net/web/search/b_detail.cfm?id=0-87849-803-6

Handbook of Fatigue Crack Propagation in Metallic Structures

http://www.elsevier.com/inca/publications/store/5/2/5/2/6/0/index.htt

MIL-HDBK-5H

http://www.grantadesign.com/cgi-bin/stat-tree.cgi?src=MIL5-H.xml&br=3

http://www.engineering-e.com/datamart/1319_DataMart2002_Datasheet.pdf

http://www.autosteel.org/bsfd.php3

http://www.swri.edu/4org/d18/mateng/matint/NASGRO/Overview/NASMAT.htm

 

[jacques29]

Thank you very much for your support.

Here is some answers to your questions:

We use 4 different materials:
-Stamping: Half hard mild steel cold rolled HV135/165 CS4 to BS 1449
-machined: Mild steel 230M07 (EN1A-UNLEADED) to BS 970
-deep draw: Hot rolled steel sheet/coil HR1 to BS 1449
-machined: 16Mn Cr5 casehardned

We don't know the impact of the manufacturing process on the material characteristic

Our understanding of the surface finish charcteristic is very limited.

Our loading condition are as follow: the torque apply to the shaft vary from nil to 4Nm depending the output required from the EHPS. It is unidirectional as the motor always rotates in the same direction. It is very difficult to capture the number of cycle as the torque vary all the time. So our idea for the test was to say, let test it by oscilating the torque from 0 the maximum torque (4Nm)for 10^6 cycles (where I understand most of the steel alloy reach their fatigue limit). We also did a destructive test where we measured the troque we need to apply to strip the shaft / rotor can assembly. Our results show that the torque to apply is around 40Nm.

I now would like to check that our test is going to capture the fatigue aspect of our application. I would also like to find some technical information to prove that 10^6 cycles is enough (ie: S/N curve of the material used).

Thanks for your help.

 

[Metalguy]

If you examine all the parts well after 1 million cycles and no fatigue cracks have initiated, you should be OK.

This way you don't even need the fatigue curves for the metals-they're all steel, and they all have a limit-unlike alum., for example. Just make sure your loads/stresses are realistic.

ASM has many refs. on fatigue.

 

[Metalguy]

Also make sure your test-piece surface finishes are the worst you'll produce, and the steels are on the low side of the hardness band.

 

[jacques29]

Yes this my approach. We oscilate the torque from 0 to 4Nm for 10^6 cycles and as at this number of cycles the materials reach their fatigue limit we expect not fatigue failure for the product even if we go above 10^6 cycles. But my issue is that I would like to have some proof that 10^6 is enough. So my idea was to have the S/N curve of the material and show that all these materials reach their fatigue limit before 10^6 cycles.

 

[Metalguy]

Jacques29,

I just looked this up on a CD-ROM I have from the ASM (Failure Analysis). I found only a few references to the 1 million cycle limit for "steel", and nothing which indicated any other number of cycles.

You may have to do a Google-type search on the subject and find hard refs, for what you require.

 

[TVP]

jacques29,

The answer to your question about the endurance limit of steel is yes, most steels under controlled fatigue testing reach an endurance limit by 10⁶ cycles. The value of stress (maximum stress, stress amplitude, etc.) is what varies at the endurance limit. The references that I gave in my previous post will support this.

 

[Merlinz]

I think 10^7 is normally considered infinite life for ferrrous materials. 0.45 of yield strenght is also a commonly quoted Endurance limit. If you were to calculate the real stresses and include stress raising factors etc, and all the results are well below 0.45 of yield (If that is correct) then you are likely to have infinite life anyway.

 

[Metalguy]

Merlinz,

Fatigue strengths/limits don't correlate well with YS, but they do with UTS. The 45% is wrt UTS.

 

[NickE]

Jacques29-- I usually double the critical test input parameter and then go for 10^6 cycles. IE in your case I would oscillate from 0 to 8Nm of torque. Depending on your geometry this doesnt seem like a huge amount of stress. You also may want to very closely examine any keyways, fillets, and other stress concentrators before and after cycling.

Another method for evaluating this may be to perform a stress governed test: (IIRC stairstep method?) Load your part such that it break in btw 1000 and 10,000 cycles. Mark the load and cycles to failure. Reduce the load by some factor. 1/8th 1/4 1/2 etc... record load vs lifetime. With a large number of repetitions this could be used as proof that your part has sufficient strength to survive usage.


Nick
I love materials science!