Treatment of residual half-cycles in rainflow count./fatigue

[Bob.Prinsen]

Dear all,

I am currently researching approaches for reducing the cost-of-energy of the Upwind 5.0 MW turbine by implementation of (higher harmonic) Individual Pitch Control. An important part of the work is assessment of relative fatigue damage, calculated through equivalent load formulations (no structural data is available for the conceptual turbine, therefore conversion to stresses is not possible). Using a Matlab based rainflow counting algorithm, I transform load sequences to cycles. However, I noticed a lot of residual half-cycles remain, which make a significant difference in the final relative fatigue damage assessment. Looking through the documentation of MCrunch I noticed that the program has the ability to discard them, treat them as full cycles and anything in between those two options (weighing parameter from 0 to 1). What is a generally accepted view of dealing with these residuals? I hope someone can give me some pointers. Thank you for your consideration.

Kind regards,
Bob Prinsen

[Marshall.Buhl]

Bob,

I’m not an expert in fatigue, but Herb Sutherland is. Here’s what he said in Section 4.1.1.4 of his 1999 report (http://prod.sandia.gov/techlib/access-control.cgi/1999/990089.pdf):

4.1.1.4. Residual Cycles

The rainflow counting algorithm proceeds by matching peaks and valleys to form closed hysteresis loops. When the algorithm reaches the end of time-series data record, a series of unmatched peaks and valleys remains unclosed and, therefore, are not counted by the algorithm. These so-called “half-cycles” typically include the largest peak and valley in the record and they may also include other large events. Thus, the potentially most damaging events (the largest cycles) contained in the time series are not counted by the classical formulation of the rainflow algorithm. If desired, the rainflow counting algorithm can be easily modified to include these half–cycles in the cycle count [28, 29].

Various researchers have proposed several techniques for handling these half-cycles in wind turbine applications. Some ignore them, some count them as half of a complete cycle (hence, their name), and others count them as full cycles. The latter is the most conservative approach. However, the recommended practice, as stated in Madsen [25], is to treat all unclosed cycles as half cycles. This recommendation is currently being retained in the IEC standards [31].

[Bob.Prinsen]

Dear Marshall,

Thank you for taking the time in quoting this report. It answers my question perfectly and I will use this approach for my fatigue calculations.

Kind regards,
Bob