When the stochastic input loading acting on the structure is exceeding limits of the
material properties, damage and a failure of the structure can occur. These limits
are analyzed for ultimate and fatigue loading, and are described by the ultimate limit
state (ULS) and the fatigue limit state (FLS), respectively. In ULS-analyses, response
loads of the structure are investigated for the exceeding of elastic deformations,
16
while in FLS-analyses, the impact of cyclic loading on the lifetime of the structure is
calculated.
Stress calculation
The stress calculation for tubular beam elements is in general a simple task. Axial
stresses are calculated by the division of axial force and structural area; in-plane and
out-of-plane bending stresses are calculated by the division of bending stress and
moment of inertia, multiplied by the radius of the beam cross section. However,
due to the connection of several members in one tubular joint, the calculation of
stresses is more complex in the joints of lattice tower structures. 4 or 6 members are
typically connected in X- and K-joints (cf. Fig. 2.2), respectively. The interaction
of several members in one joint leads to stress concentrations. Therefore, stress
concentration factors (SCF) have to be calculated by dedicated formulas for tubular
joints, as for example described in DNV-RP-C203 [66]. SCFs are defined as the ratio
of hot spot stress (HSS) to local nominal stress.
The superposition of stresses in tubular joints is calculated at in total eight
different points around the circumference of the intersection. HSS σ
1..8are derived
by the summation of the single stress components from axial, in-plane and out-of-
plane action. Therefore, the structural analysis of a multi planar joint such as a
K-joint in lattice tower structures is a comprehensive task. More details about the
number of stresses to be evaluated in such structures are described in Paper 2 in
Appendix A.
Ultimate loads
Each HSS time series obtained by the stress calculation is analyzed on its maximum
absolute stress, representing the ultimate load of the limited time series. Expected
ultimate loads for longer periods, as for example the 50 year recurrence period,
are calculated based on statistical extrapolation of ultimate loads, adapted to the
procedure described in Annex F of IEC 61400-1 [67]. This method extracts a selection
of extreme loads from the response time series and calculates their probabilities
of occurrence. By the use of fitting functions, the expected ultimate load for the
probability of the 50 year recurrence period can be calculated.
Fatigue assessment
For cyclic loading with a constant stress range, fatigue damage occurs when the
number of applied load cycles is exceeding the maximum allowable number of cycles
before fatigue failure for this specific stress range. Fatigue properties of materials
are documented in standards and are typically given as S-N-curves (S-stress range,
N-number of cycles), as for example in DNV-RP-C203 [66]. The fatigue investigation
of stochastic responses is based on the same principle; however, each individual
stress range of the time series has to be analyzed. This is done by performing a
rainflow counting analysis [68]. This analysis extracts all load cycles with its stress
ranges from the response time series. Due to the large amount of data for longer
time series, results are normally collected in a histogram for stress range bins with
its summed up number of cycles.
A common method for fatigue damage calculation has been defined by Palmgren
[69] and Miner [70]. The method is based on the definition that a fatigue failure
occurs when the number of applied load cycles divided by the maximum allowable
number of cycles exceeds D = 1 (D-damage). This simple definition is valid for cyclic
loading with a constant stress range, and has to be extended for the application of
stochastic loading with various stress ranges (Eq. 2.1).
D =
k i=1n
iN
i(2.1)
where D is the accumulated fatigue damage, k is the number of stress range
bins, n
iis the number of stress cycles in stress range bin i and N
iis the number of
cycles to failure at constant stress range Δσ
i.
The above mentioned histogram data is used for the total fatigue damage
calculation based on the Palmgren-Miner rule. Therefore, each stress range bin
contributes with a fraction of fatigue damage to the total fatigue damage D of the
analyzed time series. The result obtained by the total fatigue damage D can be put
into context to the expected lifetime of the cyclic loaded material by dividing the
length of the analyzed time series by the fraction D. For results with D < 1, the
material can survive a corresponding loading longer than expected; for D≥ 1, the
lifetime is actually shorter than expected.
For analyses with focus on relative comparisons only, a simplified fatigue assess-
ment can be performed based on a so-called damage equivalent load (DEL) [71].
This method is not following the complete fatigue assessment with stress analyses
and lifetime calculations, but is only focusing on the fatigue characteristics of the
stochastic response load time series. The basis of this method follows the Palmgren-
Miner rule, too. However, results are treated differently in the DEL-analysis. For
a defined reference number of cycles, the fraction of total fatigue damage can be
transferred back to a specific load cycle range, assuming that this constant load cycle
range (=DEL) causes the same amount of fatigue damage as the stochastic response
18
load time series. The so found DEL can be used as a fatigue characteristic of the
response time series. By this, the DEL allows for example for direct comparisons of
several simulation results in a parameter study.
In document
Simulation and Optimization in Offshore Wind Turbine Structural Analysis
(Page 30-33)