Page 366 - Marine Structural Design

342 Part III Fatigue and Fracture

The analysis method should be tested against a well-known detail, prior to using it for fatigue
assessment. There are numerous types of elements that can be used, and the SCF obtained,
depends on the elements chosen. Therefore, a consistent stress recovery procedure must be
calibrated when assessing data from finite element analysis.
Finite element analysis programs, such as NASTRAN, ABAQUS and ANSYS, use structural
elements such as thin, thick plate, or shell element. When modeling fabricated tubular joints,
the welds may not be properly modeled by thin plate or shell elements. Consequently, the
model does not account for any notch effects due to the presence of the weld and micro effects
due to the weld shape.
The stresses in thin shell plates are calculated from a membrane stress and a moment at the
mid-surface of element. The total free surface stresses are determined by superposition. At a
plate intersection, the peak stresses will be predicted at positions that lie inside the actual joint.
Comparisons between these values and experimental measurements have indicated that thin
shell analysis overestimates the actual surface stresses or SCF present in the real structure.
Most finite element elements are based on a displacement formulation. This means that
displacements or deformation will be continuous throughout the mesh, but stresses will be
discontinuous between elements. Thus, the nodal average stresses may be recommended.
However, limited comparison between these values and experimental measurements indicate
that this will generally over-predict hotspot stress or SCF especially on the brace side.
As opposed to shell elements, a model using solid elements may include the welded region,
see Figure 17.5. In such models, the SCFs may be derived by extrapolating stress components
to relevant weld toes. The extrapolation direction should be normal to the weld toes. However,
there is still considerable uncertainty associated with the modeling of weld region and weld
shape.
Fricke (2002) recommended hot-spot analysis procedures for structural details of ships and
FPSOs based on round-robin FE analysis. Some of his findings are:
If hot-spot stress is evaluated by linear extrapolation from stresses at 0.5t and lSt, the
fatigue strength may be assessed using a usual design S-N curve based on hot-spot stress
(e.g. Hobbacher, 1996 and Maddox, 2001).
If hot-spot stress is defined at 03 without stress extrapolation, the design S-N curve
should be downgraded by 1 fatigue class.
If the hot-spot stress is evaluated from strain measurements or from refined models with
improved finite elements, a stress extrapolation over reference points at distance 0.4t and
1 .Ot or a quadratic extrapolation is recommended (Hobbacher, 1996).
It should be pointed out the determination of hot spot stress based on finite element analysis is
still a very active field of on-going research since the accuracy and efficiency of the stress
determination are of importance. Other known research work includes Niemi (1993, 1994).