Variable Amplitude Fatigue Analysis - Super Confused on how

[kepler456]

Hey everyone I'd be really grateful if anyone could help me out here, been going round in circles for two weeks now and I need to get going soon. Jumping right to the point, here's my problem:

We are currently working on a student project and designing a Wind Turbine for future opensource use, much like the NREL Wind Turbine project, but this time with a 150m Rotor diameter. I must say that it's a part of our course and each of us are looking into different aspects. I'm working on the FEM and Fatigue analysis of the Hub and Machine Bed.

Here's what I already know:

Use a SN curve, Modified Goodman diagram and Miners rule to predict failure. I've got a load time series and I wrote a rain-flow count script to give me the range and the number of occurrences of each Force range.

I've been reading a lot of papers and watching a lot of videos, what I fail to understand is this:

SN Curves are generated by plotting varying amplitudes with a mean stress of zero and joining the points on a log-plot. But what I've been seeing around is references that say you need to make sure you consider a SN Curve with the same "range" as your stresses in consideration. I know that the Range = (Minimum Stress/Maximum Stress). But how does this influence the outcome? Can I not just consider my amplitude and use a general SN curve as my first step? If not how do I get a SN curve based on my Range?

Once I figure out the range issue, I put in all my correction factors onto my SN curve and proceed with a Modified Goodman Diagram. Here I'm not sure what I need to take on the vertical axis. The horizontal is the ultimate stress (mean stress) and the vertical needs to be the amplitude. What I'm seeing in references is that they pick the amplitude based on the number of cycles they want the component to survive, then go to their SN curve and pick a corresponding Stress Amplitude. I'm going to have quite a few stress ranges and I was wondering if I need to pick a stress corresponding to 106 (which is what we usually design a component if we seek infinite life) or maybe 103 (in case I got a range that happens just 1000 times). As both the cases are going to give me very different slopes.

Once I got my Goodman line drawn, I put in the yield stress line at a 45 degree angle from the horizontal axes (mean stress) and get my enclosed area where the component would be safe. Let's say the component is safely under the curve. How do I count this into the damage criteria (Miners rule) since I do not have a relation between the stress and cycles to failure anymore. I need to somehow take my information from the modified Goodman diagram and implement it onto my SN Curve - how do I do this? Do I get some kind of constant from the modified Goodman Diagram that I need to multiply into my SN curve? Like some sort of correction factor?

Thanks a lot for those who went through this post and everyone who can help in any way. Answering or pointing me in a right direction regarding even a single of the three doubts would help me a lot.

P.S: If you look at this post's image, that's the process I need to follow. https://www.physicsforums.com/threads/fatigue-analysis-problem.416943/ I went through that thread, but it doesn't answer my doubts.

 

[kepler456]

Just for those of you who may need this some time in the future, the Goodman correction gives you a new stress amplitude that you need to consider. This takes into account the mean and amplitude that you put in. The new stress amplitude is equivalent to a R = -1 condition.