Fatigue Design of Wind Turbines - Wind Turbine Technology: Fundamental Concepts of Wind Turbine Engineering - page 696

Fatigue Design Procedures

When fatigue load spectra have been defined for critical sections in a wind turbine ( e.g.,
welded and bolted joints, access holes, fillets, etc. ) the problem of designing wind turbine
components to resist these loads for the design lifetime becomes one for which accepted and
validated procedures are available. Standard stress analysis procedures are used to convert
load spectra to stress spectra, and allowable fatigue stresses are based on standard laboratory
tests of the materials of construction [see, e.g. , Manson 1965 and 1966, Mitchell 1979, Fuchs
and Stephens 1980, Dowling 1993, Mandel et al. 1993, 1994, 1997, 2002].
There is little that is unique about either fatigue theory or the fatigue design process as
applied to wind turbines, once the load spectra are specified. Figure 12-2 illustrates one of
the few differences between fatigue design of wind turbines and that of other structures. The
magnitudes of the allowable stresses in critical wind turbine components are significantly
lower because of their longer fatigue life requirements. Thus, structural integrity of wind
turbines requires very conservative yet standard fatigue design procedures.
One of the most important elements in the fatigue design process is the definition of al-
lowable fatigue stresses. This is a technical specialty of its own, in which engineers modify
the results of laboratory fatigue tests on material and joint specimens to account for the
expected spectrum of stress, size effects, cost-effective manufacturing and inspection pro-
cedures, environmental effects, and maintenance planned for the structural system during
its lifetime. Thus, the specification of fatigue allowable stresses for various materials in a
structure is an integral part of the manufacture and operation of that structure.
Verification of the total fatigue design process by field experience with prototype (or at
least similar) structures is necessary for confidence in the long-term structural integrity of
the design. As a result, different fatigue design procedures and different fatigue allowable
stresses will be utilized by different manufacturers, because each will have its own field ex-
perience and will have modified its design procedures accordingly.
Discussion of all of the factors that enter into the specification of a fatigue allowable
stress is well beyond the scope of this chapter. Instead, two general methods will be dis-
cussed that have been used successfully to account for the stress spectrum effect , which is one
of the first steps needed to modify laboratory test data into fatigue allowable stresses. These
will be referred to as the S-N linear damage and the fracture-mechanics methods.

S-N Linear Damage Method

Fatigue damage is both a physical process ( e.g. the initiation and propagation of defects in
the material) and a mathematical representation of that process. Here we shall deal only with
the latter. In general, fatigue damage and fatigue lifetime are inversely proportional. One of the
simplest models of the accumulation of fatigue damage during repeated cycles of stress is the
linear damage hypothesis , proposed by Palmgren [1924] and Miner [1945]. According to this
hypothesis, if the stress cycle remains constant throughout a fatigue lifetime equal to N , then
the fraction of that lifetime consumed on every cycle is constant and equal to 1/N . This fraction
is also defined as the damage per cycle , and it follows that the total damage at failure is equal
to unity. Furthermore, if the stress cycles change during the lifetime, the damage fractions
per cycle are linearly additive, and fatigue failure still occurs when the accumulated damage
reaches unity. If a stress spectrum is subdivided into groups of cycles or layers within which
the stress cycles are relatively uniform, then

I
å l

n i
N i

m = 1.0 at f at igue f ailure (12-9a)