Page 64 - Biaxial Multiaxial Fatigue and Fracture
P. 64

Multiaxial Fatigue Assessment of Welded Structures by Local Approach   49


         torsion and random uniaxial tension for 3 different random spectrums. In the case of random
         loading, the life is expressed as the number of blocks up to crack initiation (Nf). The results are
         plotted with two straight lines indicating a deviation interval band between N$3  and 3 Nf. This
         comparison makes it possible for the prediction to be assessed within an interval between Nf/4
         and 4 Nf.

              1E+06
                       Random tension, specmi Tens
              1 E+05   Random tension, spectrum Bend
                      . Random tensmn, spectrum Tors
           .cI  1E+04 -   0  Proporhonal tensiodtorsion
           e!
           .-         X Non-proportional tensiodto
              1E+03 -
           a
           ;E  1E+02 -
                  E-
              l E+O l

              1 E+OO
                 1E+00    1E+01     1E+02    1E+03    1E+04     1 E+05   1 E+06
                                         N, experimental

                     Fig. 6. Comparison of predicted lives against experimental ones


            The  highest  deviations  between  predictions  and  experiments  are  obtained  for  non-
         proportional loading and random amplitude. In the case of non-proportional loading, the most
         important part of the error comes from the fatigue criterion used. In fact, several authors [9,10]
         showed that,  for non-proportional loading, the critical plane  criterion taking  account of the
         maximum of the shear stress amplitude encountered during one cycle give inaccurate results.
         The average of the shear stress on the facet over the cycle or the integration of shear stresses
         over  all planes may  give better results.  This concept developed  in  some multiaxial  fatigue
         criteria is the so-called integral approach. Considerations on how to integrate such a criterion
         for random loading are necessary. In the case of random tension loading, the calculation is
          always optimistic, because the accelerating influence of previous small amplitude cycles onto
         the  damage  rate  due  to  large  amplitude cycles  is  not  taken  into  account.  To  improve the
         prediction, a more physical  damage parameter such as cumulative micro-plasticity on  each
         facet or micro-cracks nucleation on each facet should be used.


         APPLICATION OF THE LOCAL APPROACH TO WELDED COMPONENTS

         This section details all the steps of the local approach procedure which are successively run on
         in order to assess the fatigue lifetime of a welded mechanical assembly.
            Welded joints are particular structural details characterised by:
         - Stress concentrations induced by local shape of joint and manufacturing geometrical defects
         produced by the welding process (angular deviation and misalignment of the connected metal
         components),
   59   60   61   62   63   64   65   66   67   68   69