Page 41 - Biaxial Multiaxial Fatigue and Fracture
P. 41

26                           G. SAVAIDIS ET AL.














                                                                  Detail of  weld modelling


            Fig. 5. Finite element mesh and simulation of the load configuration


            The accuracy of the FE calculation has been checked in [IO]  evaluating the strain response of
            the structure under monotonic loading with F, and/or Fy (load components acting on their own
            and  simultaneously). At  six different locations along the  length  of  the  tube  measured  (by
            means of  strain gauges) and calculated strains have been compared, showing an overall good
            agreement. Additionally, numerical studies with various element lengths reported in Ref. [7,
            101 showed  that  the  element  sizes  according to  the  IIW  guideline  [I]  as  used  here  yield
            reliable hot spot stress results.


            Hot spot stress and fatigue life calculation
            The  analysis of  fatigue  behaviour focuses on  determination and estimation  of  the  stress /
            strain state acting at the welding undercut, which is regarded as the failure-critical location of
            the component. This undercut has a distance of  12 mm to the surface of the forged component
            and corresponds to the fifth element ring starting from the end of the tube.
              For  the  nonproportional loading situation investigated  here, fatigue  life calculations are
            carried out for all shell elements of the fifth ring, separately for the outside and inside surface
            of  the shell. In general, calculations of this kind can be made in accordance with conventional
            approaches  (based  on  integral  criteria  such  as  energies,  equivalent  values  of  stresses  or
            strains) or  approaches  which  assess the  stress  and  strain  state  acting  in  certain  directions
            (critical plane  approach).  In  the  case  of  the  critical  plane  approach, it  is  assumed  that  a
            microcrack that forms at the surface propagates on a preferred plane on which the normal
            stress-strain or shear strain response or a combination of both reaches its maximum.
              Using wide-ranging experimental information, Sonsino [ 1 1,  121  reports that conventional
            hypotheses often fail in the case of nonproportional loading. For this reason the critical plane
            approach is used in this examination.
              Figure 6 shows the weld detail with the distribution of  stress o, at bending, where the y-
            direction corresponds to the tube axis.
              Starting from the applied nominal loads, the stress sequences o,(l) and g(t) and all the rest
            of  the stress components acting at each element along the ring are evaluated. The sequences
            odt) and  zdt) in  various planes  of  each  element  are calculated using  the  transformation
            equations based  on  the  equilibrium  of  an  infinitesimal material element. The subscript
            denotes the angle between the zy- and actual plane under consideration.
   36   37   38   39   40   41   42   43   44   45   46