Hi everybody,

I would like to hear some opinions (also negative ones!) about the following question: as a premise, I wish to remark that I am aware of the existence of the rainflow method, of the Goodman or Gerber diagram, and so on. But I am asking myself if, at the stage of early pre-dimensioning, it would be possible and meaningful, for the purpose of preliminary fatigue verification, to “condense” the verification in one single “point”, at a given “equivalent fatigue wind speed”.

My reasoning is the following:

given a (known) Weibull distribution with parameters [k,Vm], and assuming that the load on a part is proportional to the wind speed squared (a rough assumption, I know), this will probably generate something like a “distorted” Weibull distribution of load (of stresses) on the part. Thus it should be possible to take an appropriate “mean” or “equivalent” wind speed that represents the whole fatigue life of the part, for the purpose of fatigue verification. Is this reasoning, at least in first approximation, correct? And if yes, how would you choose this “mean” from the mathematical/numerical point of view?

thanks a lot

Best regards

Hi Claudio,

I’ve thought about doing this before, but never have been able figure it out. I haven’t found anything in the literature, although that doesn’t mean that it doesn’t exist or isn’t possible.

I think one large problem is that fatigue is very much design dependent: sometimes a design is driven by extreme events (or high amplitude cycles) and sometimes by more fatigue like (low-amplitude) cycles that are more frequent. Although usually the large cycles dominate in composite blades. The physics of these loads are often different, somtimes extreme loads scale with wind speed squared, but sometimes they occur from a combination of wind direction change and rotor

overspeed - which I’m not sure how that would scale. Often extreme winds occur when a turbine is parked, but that doesn’t mean that the loads are highest then either. For teetered rotors, teeter strikes are dominant…etc., etc. OK, since most turbines are 3 bladed upwind pitch-regulated these days, you might not have issues like this, but who knows. You may be able to get away with looking at the point of maxium operating load (rated wind speed for a variable speed trubine) and the 50 year extreme gust (parked) and the ETM (extreme turbulence model) from the IEC.

The only way to know is to test it out. I would take a simple turbine and run it using our RunNTM (or CondorNTM) and RunIEC codes, then look at the faitgue distribution binned by wind speed. Calculate the fatigue damage contributed in each of these bins and see how many bins you can neglect without significantly lowering the total fatigue damage. If all of them except one, then you’re in business. After you’ve this change the wind environment and repeat. So, yes this is a lot of work, but if you do it, we sure would be interested in the results. Sound like an excellent topic for a paper.

Anyway, those are my ideas at the moment. Cheers!

Hi Cladio, Pat,

‘Wind speed for fatigue preliminary dimensioning’ is a very interesting topic and initial ideas expressed by both of you are admirable. The work involved in this is a putting togather lot of load cases and developing a algorithm which would convert the mean wind speed individual scatter of speeds to represent the particular mean speed.

I am interested in this topic and working on this topic further. Is it possible for you to share the effort put till date so that I can try to put my experiene and efforts to march the progress of this subject project? [if not progressed further]

Presently I am involved in static and fatigue analysis and testing of the wind turbine blades.