Endurance limits of new material

[nitinfem1]

Hello Everyone,
We make the castings using our own material.we performed RR moore rotating beam test and found the endurance limits through curve fitting of S-N curve.
Do we have to use modifying factors to further reduce the endurance values?
What is the general procedure to accept a new material values?

I appreciate if someone can throw some light on it.
Thanks

 

[cbrn]

Hello,

The effect of the sollicitation type is one of the several which do can influence fatigue strength of material.

The data you are collecting is relative to rotating flexure and thus can not directly be used to planar flexure or axial traction.

K-factors (named Kv) have to be used to translate the data from a sollicitation type to the other.
Taking your test case as the reference, generally used factors are:

- planar flexure: Kv = 0.8
- rotating flexure: Kv = 1 (obvious!)
- axial traction: Kv = 1.1

These factors are only relative to uniaxial stress cases.

This means that fatigue strength for planar flexure will be 1/0.8 times greater than your values. This is not the correct way to "see" the phenomenon: it's better to say that the sigma calculated for a planar flexure sollicitation case must be multiplied by Kv = 0.8 before being compared to your fatigue strength curve.

Hope this can help.

Regards.

 

[gurmeet2003]

nitinfem,

The fatigue test specimens are polished and have a very nice finish. The actual parts will usually have a rougher finish. Therefore your test results have to be modified to account for this. Also the test specimens are of small size. The actual parts may be much larger. The distribution of material defects may be different in the actual parts. Usually larger parts have a smaller fatigue strength. Therefore you need to apply modifying factors for these two effects;i.e. surface finish, size.

In absence of any other information , I would suggest that you consult the Fatigue Handbook of ASM (American Society of Metals) to get these factors.

In rotating bending fatigue testing the stresses are high on the surface. While in axial fatigue testing the stresses are uniform throughout the cross-section Therefore the axial fatigue tests would give slightly lower results, because they are affected equally by all the defects in the whole cross section.

Thanks,
Gurmeet

 

[gwolf]

Some more factors to consider:

1) You will have to apply some statistical analysis to your results to define minimum performance. This is done on the basis of standard deviations based on the scatter which you see from specimen to specimen. This also depends on how many samples you took and the target audience for your products.

As an example, for civil aerospace applications, the JAR (joint air regs) are quite specific about relating the number of test specimens to the statistical methods which should be used for castings. It is also worth looking at MIL-spec equivalents. I'm sorry but I don't have details to hand.

As a simple rule of thumb if you are not tied down by any particular legislation then do the following. For each S-N point on your test curves, determine the statistical mean stress and standard deviation stress (sigma). Then subtract 3xsigma from your mean stress. This means that roughly 99.9% of your components will have this fatigue strength.

2) On top of this there is another factor. Fatigue tests are usually done to failure (or close to failure). A tiny crack will have started well before this number of cycles. A widely used rule of thumb is that 2/3 of the cycles are used to initiate the crack and 1/3 of the remaining cycles to grow the crack to failure. This, once you have factored for stress as in 1) you also need to factor the number of cycles down by 2/3.

3) I'm not familiar with the RR moore test. If it has no stress concentration feature (KT=1) then all is well because this is the most commonly used type. You can get into fatiguing stress concentration features like holes (KT=3) but these components will show greater life (based on hole concentrated stress not bulk stress) than plain components stressed to the same level.

This approach is called "Safe Life" design. You should make it clear in any product literature that the curves which you produce are design curves not raw test curves.

Again, it very much depends on your target audience but check out the JAR's, they are very good at detailing what you need to do for castings.

Hope this adds useful food for thought.

Regards,

gwolf.

 

[feajob]

gwolf,

Regarding your second comment, I would like to add that:
Based on SAE course (Fundamentals of Metal Fatigue Analysis), crack initiation (strain life) is defined for the cycle when we notice 20% load drop.

AAY

http://www.geocities.com/fea_tek/

 

[gwolf]

feajob,

Yes, assuming the poster was looking for that during the fatigue test. If not then the 2/3 rule is a reasonable rule of thumb. Also the % load drop is material dependent.

gwolf.

 

[nitinfem1]

Hi all,
Thank you very much for the input.As gwolf suggested we reduced our mean values by 2 sigma and applied modifying factors.

We were able to resolve the issue.
Thanks once again