I agree with you and it is a good idea and the yield stress factor can be used when in the notch case..as in the case of crack. The notch root stress is limited by the yield stress. I can try my best to check with available data

Thanks Johan, I hope you can find some relevant data to check against. Regards, Clive.

Clive,

You can get some reasonable estimates by using fatigue crack growth data with high R ratio, as those should be loaded close to the yield stress of the material.  Alternatively, you can use fatigue data obtained directly from welded material which has not been stress relieved.  The applied load in those cases is always superimposed on the residual stresses, which in the case of welded steels would be very near the yield stress in tension.  In both cases, the base metal's yield stress is probably provided in the literature.  I know this is not exactly the data you are interested in, but it may serve as an adequate substitute.

Thanks Brian, I will have a look. Regards, Clive.

Clive,

The yield strength is often included in the AFGRO database with da/dN data, and for some specimens the load also, but it is rather time-consuming to extract it (and you have to be an AFGRO subscriber).   However, there is data available for many materials.

You can check out some models on my website (fracturelab.com). 

Rick

Dear Rick, an interesting website and the PN correlations are impressive. I have not been able to find the data I want, but would you have any comments on the approach? Checkout my website

 http://statact.squarespace.com

go to Fracture Mechanics and there is a relatively short power point presentation of the theory.

Regards, Clive.

I opted for the longer paper.  You are right that published crack growth data (even in the NASA databases) is often in terms of delta K and da/dN alone, with no applied stress (or yield stress, or crack length vs cycles) provided, and that makes the data far less useful.  US Air force data often sporadically preserved these other parameters, but the only working database where it can be readily accessed (though still quite arcane) is through AFGROW.  

Jim Newman's strip yield model (derived from FASTRAN center-cracked panel closure predictions) does include a yield ratio effect, though in practice it has been widely neglected (see NASGRO manual, which comes with the free demo version of NASGRO).  Part of the reason is that it is unclear how define the yield ratio for a complex structure being analyzed, or even for a compact-tension specimen.  Craig McClung has approached this using a K-analogy approach.  The nature of the stress-level sensitivity problem is different when you have compressive loading as I discuss in a .ppt on my "more info" page.  

Your assertion that practitioners miss the stress level effect because it is constant for given constant-amplitude test misses the fact that in practice tests are often purposely run at high and low stresses (smaller and larger cracks) to purposely evaluate stress level effects.  In the aerospace world, at least, crack growth data is typically run on material procured to company specifications, and all the parameters of the tests are known.  Models used (like the NASGRO equation, Walker-interpolated tabular models, proprietary models) are fit to the data, which often includes multiple stress levels.  

To be sure, fatigue crack growth practitioners have long been aware of stress-level sensitivity effects, though we often get away with neglecting them, perhaps choosing to err on the conservative side where necessary.  Your approach appears novel.  You are right to seek well-defined datasets to compare with your predictions.  I wish you luck in trying to shed more light on this issue.

Rick Pettit