Page 360 - Marine Structural Design

336 Part III Fatigue and Fracture

If p is the per cent of the test specimens that fall below the design S-N curve, the design S-N
curve may be defined as follows: en as:
logN=logK,, -'IPSd -mlogS (17.9)

where Sd is the standard deviation of logK,, . The relationship between the value of /2, and
the failure probability is (BV, 1998):
fail safe design: p=2.5%, dp = 2 (normally used to derived design S-N curves)

safe life design: p=O. 1%, 'Ip = 3 ( for special welded specimens that represent structural
details which can not be easily inspected and repaired.)

17.2 Estimation of the Stress Range
The fatigue analysis procedure is based on the ranges of cyclic principal stresses.
To determine the stress range, two approaches have been developed. The "nominal stress"
approach has been applied to plated structures and the "hotspot stress" approach has been
developed for tubular joints. Note that a "notch stress" approach is also suggested by some
design codes. In recent years, attempts are being made to apply the hot spot stress approach to
plated structures.
17.2.1 Nominal Stress Approach
In the nominal stress approach, the stress concentrations caused by the weld profile have been
included in the S-N curves.
The determination of stresses applied to fatigue analysis of structural details, is generally
undertaken by a global-local finite element analysis of the pertinent stress in accordance with
the chosen S-N curves. In other words, the calculated stress for the considered local hot spot
area of structural details should resemble the nominal stress of the test specimens from which
the S-N curves were established. Unfortunately, in most cases, structural details are more
complex than the test specimens, both in geometry and in applied loading. Consequently, a
relationship between the S-N data stress and calculated stress may not be easily established.
Another problem associated with the nominal stress approach is the classification of structural
details. The primary difference between UK DEn curves and recent European S-N curves is
that UK DEn curves do not have fatigue endurance limit. The fatigue endurance limit, which is
found in the constant cyclic loading test, usually does not exist for marine structural details
due to a variety of causes such as welding, corrosion, and load sequence effects of the random
loading.
The relevant fatigue stress for fatigue design would be the tensile stress c, for example, for the
weld shown in Figure 17.3a. For the weld shown in Figure 17.3b, the stress concentration
factor for the global geometry must be accounted for, using the relevant fatigue stress of
SCF.0, where SCF is the stress concentration factor due to the hole.
If a comer detail with zero radii is modeled, the calculated stress will approach infinity as the
element size is decreased to zero. The modeling of a relevant radius requires a very fine
element mesh, increasing the size of the finite element model. In addition, selection of the
proper radius to be used for the analysis will be a matter for discussion.