Page 207 - Practical Design Ships and Floating Structures
P. 207
182
reliable than the short-term approach under the circumstances of same environmental information
available.
- __ -WIS6,JONSWAP . -.
N .c!
WIS6, Brestchneider
z300 ...... . Is -
O
d
- W391,JONSWAP 01
v,
W391,Bretschneider
P
? 200 ;lo 3
e ll.
ll.
2
p 1.00 5 :
k e
vl z
0 00 0
0 30 90 0 2 4 6 8 1 0
6o 2
Stress, (Kgflcrn ) WX)
Figure 7: Long-term probability density Figure 8: Long-term extremes of stress for
function&) of stress responses for deck plate deck plate with return period: 20, 50, 100 years
5 FATIGUE DAMAGE EVALUATION
The fatigue strength of welded joints and structural details in highly dynamically stressed areas need to
be assessed to ensure structural integrity and to optimize the inspection plan. The analyses of fatigue
strength should be based on the combined effects of loading, material properties, and flaw
characteristics. At the global scantling design level, the fatigue strength check for hull-girder members
can be conducted for screening purposes. At the final design level, analyses for structural notches,
cutouts, bracket toes, and abrupt changes of structural sections need to be performed. In addition, deck
facilities, pipelines, risers, mooring system and its interface structure with the FPSO hull also need to
be designed with adequate fatigue strength.
Stress types commonly used by fatigue analysis based on the S-N curve include nominal stress, hot-
spot stress, and notch stress. Each of these methods has specific applicable conditions. Although
nominal stress is used in the examples, the analysis approach is not limited to any stress type.
5.1 Spectral Fatigue Analysis
Spectral Fatigue Analysis (SFA) based on the S-N curve and Palmgren-Miner's cumulative damage
hypothesis has been widely applied in the fatigue damage assessment of offshore structures. However,
choosing a proper method is important to avoid unnecessary numerical errors. Figure 9 shows the
procedure for fatigue assessment, which applies the 'closed-form' integration to eliminate the error due
to numerical integration.
Denote the stress range by s (s = 2a, where a is the amplitude), then the short-term probability density
function p&) can be approximated by the Rayleigh distribution as
where ovk, is the variance of stress responses, i.e., a$, = m:'. The total damage in Z', (unit in years)
can be represented by