I want to respond to this before Dale deems this discussion too far off topic. I am a naval architect (glorified ME) and I specialize in fatigue. Most of my work, however, is on very low grade steel that is at least 1/2" thick. The test as it is presented is adequate in comparing the relative fatigue life because it gives us a point on the S-N curve (stress vs. number of cycles to failure) for each frame for this loading condition. There are many other loading conditions that are not accounted for here that may drive the fatigue life rather than this sprinting case. In my work, small and medium sized waves do much more fatigue damage to ships than large waves.
The results are not very surprising because it is finding that the frames with poor fatigue details are failing first. On the De Rosa, I bet that it was a large step from the lug to the tube that caused the failure, rather than overheating as mentioned in the article. A little filing on the lug may have doubled the fatigue life of that location.
I think that this test may be overly conservative because it is assuming that all the force on the pedals is absorbed by the frame. When I ride, I would like to think that the majority of the force I put on the pedals turns into forward motion. I am also concerned that the head tube support may be too rigid and is causing the head tube failures on the lugged bike.
What lessons do I take away from the article? If I want a lugged bike to last a long time, I will file the lugs down very smooth. Other than that, I wouldn't worry about fatigue any more than I do now.
Sean Smith
Huntington Beach, CA