Strain Life Fatigue Data

[rickfischer51]

I have a strain life plot where the x-axis is labeled "Life (2Nf,reversals)." Never seen this before. What's 2Nf? Twice the number of cycles to failure? Second Normal Form?

Rick Fischer
Principal Engineer
Argonne National Laboratory

 

[Dhurjati Sen]

Dear rickfischer51,

2N_f is the Reversals to failure (log scale) from Basquin's equation.

Regards.

DHURJATI SEN

https://www.nace.org/people/dhurjatisen

 

[realtime2]

Cycles to failure used to be the standard for plots and equations
before the 1950s. During the days of Coffin and Manson, there was a lot of
argument about how to incorporate the tensile test into the plot. Is it 1 cycle?
or a half cycle(reversal)? Makes a difference to curve fitting I guess. As more and
more service histories came to be used, where it is quite difficult to determine
when a cycle begins or ends, Reversals,the turn-around points, became more
popular and the 2Nf plots and their fitting parameters are now in many fatigue
data bases.

 

[rickfischer51]

My fatigue experience has been with high cycle fatigue, but I am now faced with an extension spring that has failed after an estimated 5000 cycles. I've run fea and the max first principal strain occurs exactly where the failure occured. I have a reference that discusses Coffin-Manson and says that "2N is the number of strain reversals to failure, where one cycle is two reversals." Whats a reversal? R=-1? R=0? Ive read that low cycle fatigue is not sensitive to mean stress as is high cycle fatigue. My extension spring cycles for 0 to S1>0. Is that one cycle or one reversal?

Rick Fischer
Principal Engineer
Argonne National Laboratory

 

[realtime2]

A reversal refers to the change in direction of loading.
Thus there is a reversal at Smax and another at Smin,
so that in any given "cycle" there are two reversals.
If your estimated life is Nf= 5000 cycles, then that is equal to
2Nf = 10000 reversals.

A 0 to Smax loading may need a mean stress correction. It kind of depends on
how much plasticity your part is seeing. Given that it is a spring
material it is probably very hard, so not much plasticity even at 5000 Nf.
For tensile mean stresses its probably best to compute the Smith-Watson-Topper
parameter: SigmaMax x StrainAmpl. for your service condition and then compare
it to your fully reversed (R=-1) fatigue curve which is also translated into
SigmaMax x StrainAmpl. This does not work well with compressive mean stress
conditions however.

If your calculation for the hot spot stress is elastic, and your material is
actually going into plastic you may also need to do some sort of plasticity
correction such as the Neuber plasticity correction.

 

[rickfischer51]

Thanks for the reply. At this point metallurgy and hardness are unknown. Im trying to find a local source to do some hardness testing, so I can determine surface hardness and whether its case or through hardened. I've done some googling and have fatigue curves for induction hardened 1070. Its just a guess at this point, but the application stresses economy over life, so I'm guessing an inexpensive spring material. I have curves for the hardened surface and the ductile core. First principal strain is .0057, so I'm above yield in the core but well under for the 600 Brinell case hardness in the fatigue curve. Obviously I need hardness data. Another facet is that there is a possible flaw in the form of a pair of grooves on the wire surface. Cant tell yet if it is on the surface or a seam. Hope to section the wire and get it under the microscope soon.

Rick Fischer
Principal Engineer
Argonne National Laboratory