Page 412 - Biaxial Multiaxial Fatigue and Fracture
P. 412

3 96                         M. FILIPPINI ET AL.

              -         SONSIWS CRITERION - SAE 1015         NEWCRITERION--SAE  1015

              20-
              18  -
                -

             5 12
             3

               6
               4
               2
              30   -20   -10   0   10   20   30
                       Loganthmic error index [%]          Logarithmic error index [%]
            Fig.  10: Logarithmic error index for  SAE  1045 steel: Sonsino-Grubisic  criterion (left), new
            criterion (right)
              For  the  examined  materials,  the  new  proposal  gives  errors  between  predicted  and
            experimental fatigue lives that pile up in the classes around 0% with a rather small scatter. It is
            observed that,  by  employing the  Sonsino-Grubisic  method, the  majority  of the  logarithmic
            error values  concentrate on the positive side, showing a more  conservative prediction. This
            could be attributed to the fact that the beneficial effect of the stress gradients is not taken into
            account, thus predicting a fatigue life in torsion lower than the experimental one. It is noticed
            that  the  basic  material  curve  used  in  the  new  proposal  is  the  axial  straidlife  curve.  An
            improved version of the Sonsino criterion, including the effect of stress gradients [20], may be
            used instead, where the fatigue life of notched components could be  also assessed. Another
            source  of  the  slight  discrepancy observed  in  torsion  tests  might  be  the  anisotropy  in  the
            material, particularly in the case of the SAE 1045 steel [ 181.


            COMPARISON BETWEEN CRITERIA
            As mentioned previously, the von Mises approach, the Sonsin+Grubisic  method and the new
            proposal are, in principle, all applicable under in-phase  and out-of-phase  loading conditions.
            However, it is reminded that for two loads with the same strain amplitudes, the one being in-
            phase  the  other  being out-of-phase,  the  von  Mises approach produces  the  same equivalent
            strain,  leading  thus  to  the  same predicted  life.  This  contradicts  the  experimental  findings.
            Actually, several experimental results show a fatigue life reduction for the out-of-phase  case
            due to the variation of the principal strain directions and the interaction of the deformations
            acting along different directions. Both the Sonsino-Grubisic criterion and the new proposal are
            able to correctly model this behaviour, predicting a higher equivalent strain value and thus a
            fatigue life reduction for the out-of-phase  case in comparison to the in-phase  condition. For
            the  new  criterion  the  study of  the  interaction  of  the  deformation  in  different  direction  is
            extended to all possible material interference planes, obtaining thus an equivalent strain lower
            than the value predicted by the Sonsino-Grubisic criterion.
              In Fig. 11 a combined axial and shear strain loading is examined. On the left, the equivalent
            strains calculated according to the three approaches above are shown for the case where the
            axial and shear strains are in-phase.  It is noticed that the new proposal leads to the same value
            of equivalent strain as the von Mises method. On the right, the axial and shear strains are out-
   407   408   409   410   411   412   413   414   415   416   417