Possible to Estimate Initial Fatigue Crack Size?

[KANN]

Is it possible to estimate whether the initial fatigue crack size exceeds 1 micron in Type 316L (A270) without a background in fracture mechanics or metallurgy?

In other words, is the science of crack initiation accessible enough that someone with a mechanical engineering background, not metallurgy, can find the material properties and fracture mechanics equations to estimate an initial crack size in austenitic stainless steel?

This is to determine whether there is a sufficient crevice size for bacteriological growth, not to avoid failure.

In reviewing the ASM Handbook, some fracture mechanics texts, and some NUREG publications, it seems that estimation of initial crack size is not described, but rather once a crack has occurred, there are methods to predict crack growth rate, and remaining life.

 

[metengr]

KANN;
Without getting into too much detail, fatigue crack propagation is well understood for most materials and can be predicted based on what you stated above. Once a crack is identified by one or more nondestructive test methods in service, the crack can be monitored until a critical crack size is reached in a component.

There is no formula or methodology that can predict when a fatigue crack occurs. One approach used for high cycle fatigue service is approximately 80% of the cycles to failure is spent in initiating a fatigue crack. Other than this, I do not know of any formula that can predict when a crack occurs in service.

 

[CoryPad]

A perfectly smooth specimen can develop a fatigue crack by formation of Persistent Slip Bands (PSB's) or some other mechanism. Or, fatigue can progress be propagation from some other flaw (scratch, crack, pore, pit, etc.).

One micrometer is a very small size - you can assume that any object will have some discontinuity that is at least one micrometer long. So, you can use equations to predict lifetime.

 

[unclesyd]

Here is an excellant site on fatigue calculations posted by metengr a long time ago. I don't know how it handles the initial crack size as most of the time one is looking for the critical crack size.

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

 

[KANN]

Metengr and CoryPad,
Thank you for confirming some of my understanding.

Rather than predicting when a fatigue crack will occur, I'd like to estimate the size of the initial crack, once a crack does begin to occur.

The subject piping surface has a 25 Ra surface finish (~0.64 micron). I'd like to know if and when a fatigue crack originates, whether it is likely to be larger in length (or depth) than the normal surface roughness.

One source states "During fatigue loading of smooth test specimens, surface cracks 10 microns or longer form quite early in life (< 10% of life) at surface irrigularities or discontinuities either already in existance or produced by slip bands, grain boundaries, second-phase particles, etc."

Is this to say that if cycled below the endurance strength of the material there are no fatigue cracks initiated, but then if the stress exceeds the the endurance strength of the material, for sufficient cycles, then the initial crack is infitisimally small or is it of a certain minimum size to be considered a fatigue "crack"?

I suppose part of my question is when is a relative material displacement change due to fatigue considered a "crack"?

 

[arunmrao]

With what means will you detect a discontinuity of 1 micron. Also 316 is a very ductile material and crack propagation will not be easy,unless there is a very severe corrosive attack.

If you need to avoid crevices for bacteria to grow,perhaps you could consider coating the surface.

I have not failed. I've just found 10,000 ways that won't work." — Thomas Edison
_____________________________________

 

[EdStainless]

Since you are using mechanically polished tubing I would say that cracks are the least of your worries. The polish creates at lot of crevices and traps. If you have sections of tubing that may be seeing fatigue you might want to replace those areas with Electropolished tubing.

= = = = = = = = = = = = = = = = = = = =
Plymouth Tube

 

[israelkk]

Initial crack size estimation is the minimum crack that the testing applicable facilities can detect. For example: if a material is processed acoording to a spec and the spec require crack detection test such as magnaflux or ultrasonic or any other method. The initial estimated crack size for fracture mechanics analysis will be the minimum crack that the detection method can detect.

 

[KANN]

This application is high stress, low cycle fatigue (under 10,000 cycles). The tubing is installed, and although new fittings could be coated or electropolished in future installations, if fatigue cracks do develop, there is no benefit with respect to the development of fatigue cracks, to performing those surface treatments. The better approach, and one that I am investigating, would be to reduce the stresses below the endurance strength, if cost effective.

What I'm gathering from the input here is that microstructure fatique cracks can develop early, but larger cracks, those that are detectable by say liquid penetrant, in the 10 micron depth or 10 micron width range will occur in the last quartile of cycles to failure (at least in high cycle fatigue, not sure if applicable to low cycle fatigue), and then progress rapidly in size to failure.

It seems that the answer to my initial question about estimating initial fatigue crack size is that there is no specific crack size that suddenly develops as a fatigue crack, but rather the crack begins at the microstructure scale in the slip planes and progresses at different rates through the fatigue life of the crack. The technical use of "initial crack size" is that crack size, that is detectable, and it is that detectable size, that is used as the starting point in the remaining life calculations to determine the critical crack size at point of failure. The Fatigue Calculator website by Prof. Socie was very helpful.

I'm certainly interested in any other input, but if none, thank you all for your time and input.