Page 313 - Wind Energy Handbook

BLADE FATIGUE STRESSES 287

A real instance of incipient aeroelastic instability was the development of an
edgewise blade resonance under stalled conditions on some larger three-bladed
machines. A negative rate of change of lift coefficient with angle of attack is
believed to have been the prime cause – see Section 7.1.9.
Another potential instance of aeroelastic instability is classical flutter, encoun-
tered in the design of helicopter rotors, in which the blade structure is such that
out-of-plane flexure in the downwind direction results in blade twisting, causing an
increase in the angle of attack. During the development of some of the early large
machines, the dangers of aeroelastic instability were considered to be a real concern,
and much analysis work was directed to demonstrating that individual turbine
designs would not be susceptible to it. However, partly no doubt because of the
high torsional rigidity of the closed cell hollow structure adopted for most wind
turbine blades, aeroelastic instability has not yet been found to be critical in
practice, and stability analyses are no longer regarded as an essential part of the
design process. This may change, however, if designs become more flexible.

5.9 Blade Fatigue Stresses

5.9.1 Methodology for blade fatigue design

The verification of the adequacy of a blade design in fatigue requires knowledge of
the fatigue loading cycles expected over the lifetime of the machine at different
radii, derivation of the resultant stress cycles and calculation of the corresponding
fatigue damage number in relation to known fatigue properties of the material. The
procedure is less or more complicated, depending upon whether blade loading in
one or two planes is taken into account. If bending about only the weaker principal
axis is taken into account, considering only aerodynamic lift forces, the steps
involved are as follows.

(1) Derive the individual fatigue load spectra for each mean wind speed and for
each radius. This is a non-trivial task because, unless wind simulation is used,
the information on the periodic and stochastic load components is available in
different forms, i.e., as a time history and a power spectrum respectively.
Sections 5.9.2 and 5.9.3 consider methods of addressing this difficulty.
(2) Synthesize the complete fatigue load spectrum at each radius from the separate
load spectra for each mean wind speed, including start-ups and shutdowns (see
Section 5.5.1).

(3) Convert the fatigue load cycles (expressed as bending moments) to fatigue
stresses by dividing by the appropriate section modulus. (The section modulus
with respect to a particular principal axis is defined as Second Moment of Area
of the cross section about that axis divided by the distance of the point under
consideration from the axis.)