Effect of compressive stress on fatigue?

[dculp1]

It is generally accepted that a compressive stress on the surface of a metal part (e.g., by shot peening or nitriding) will allow higher alternating (fatigue) stress. However, in constructing a Goodman or Gerber diagram (to consider the effect of static loading on fatigue), a compressive mean stress through the part's entire cross section apparently has no effect on the allowed alternating stress. (For example, see Juvinall, "Engineering Considerations of Stress, Strain, and Strength", p. 271, fig. 14.5)

So why do compressive surface stresses improve fatigue performance while mean compressive stresses have no effect?

Thanks,
Don Culp

 

[metengr]

Don Culp;
A general statement like

a compressive mean stress through the part's entire cross section apparently has no effect on the allowed alternating stress.

is not entirely correct. If you review the equation for constructing a Goodman diagram, you will notice that in the equation the term tensile mean stress/ultimate tensile strength is subtracted from 1. The reason is that the Goodman diagram like other diagrams evaluate the effects of tensile mean stress on fatigue life. This means that this ratio must be equal to or less than one for plotting the Goodman diagram. Compressive mean stresses (-) must be dealt with differently because the mean stress/ultimate tensile strength term being negative would result in fatigue strength values proportionally greater than values obtained under zero mean stress conditions (R=-1). I have seen similar statements where the effect of compressive mean stress is nil or can actually lead to longer life under the same stress amplitude. However, for calculation purposes, most use a mean value of zero for compressive mean stresses.

Now, regarding compressive stress on the surface of a component; typically, fatigue cracks will not initiate and propagate within a compressive stress region.

 

[CoryPad]

a compressive mean stress through the part's entire cross section apparently has no effect on the allowed alternating stress

This statement is false. In Mechanical Behavior of Materials, Dowling discusses how the Goodman model is not correct for negative mean stress. There are several figures that show a longer lifetime (or higher allowable alternating stress) for mean compressive stress.

Regards,

Cory

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

 

[dculp1]

Cory --

I would tend to think, as you indicated, that a compressive mean stress should give either a longer lifetime or higher allowable alternating stress. This makes sense, since the compressive mean stress would reduce the maximum stress that the part would experience. This is the reasoning that is usually cited for fatigue improvements due to nitriding or shot peening.

However, Juvinall, "Engineering Considerations of Stress, Strain, and Strength", p. 271, fig. 14.5 shows fatigue data points for steel and aluminum where a compressive mean stress has been applied. A number of these points fall at or just slightly above the fatigue strength Sn of the material (i.e., the fatigue strength where there is neither tensile or compressive static load). This occurs even where the compressive stress is considerable (in one case very near the yield strength). (Shigley, "Mechanical Engineering Design", p. 178, fig. 5-35 shows similar data.)

My question is why these data do not show improvement with a mean compressive stress, similar to the improvement when a part is nitrided or shot peened.

Note: the figures do show a number of data points where the alternating failure stress > Sn when a compressive mean stress has been applied. However, I would have expected that ALL of the data points would have shown such improvement. I don't have any additional information on how these tests were performed. Is it possible that the apparent inconsistency could be due to low-cycle versus high-cycle testing or the type of test (rotating, bending, or push-pull)?

By the way, the above Goodman diagrams show the mean stress (either tensile or compressive) on the horizontal axis and only the alternating stress amplitude (peak value) on the vertical axis.

Thanks,
Don Culp

 

[CoryPad]

Is it possible that the apparent inconsistency could be due to low-cycle versus high-cycle testing or the type of test (rotating, bending, or push-pull)

Yes

Regards,

Cory

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

 

[dculp1]

Cory --

Can you elaborate on why these factors might have an effect and which might be the most significant?

Don Culp