Page 140 - Biaxial Multiaxial Fatigue and Fracture
P. 140

The Irtpuence of Static Mean Stresses Applied Normal to the Maximum Shear Planes in ...   125

          where  Aymax  is the largest shear strain range,onmax  is the maximum  stress normal to the
          shear plane and  ay is the yield stress of the material.

          Second Modified Kandil, Brown and Miller parameter [16]




          where  Aen is the normal strain range and  on, is the mean stress perpendicular to the maximum
          shear strain amplitude.

             A static tensile mean stress normal to one of the planes of maximum shear strain amplitude in
          a tension  torsion  machine has been  shown  by  a  number of  investigators to  reduce torsional
          fatigue strength [17,  181.  However, there is only a very small amount of  data to document an
          increase of  fatigue strength in shear due to static compressive stresses normal to the two planes
          of maximum shear stress range (if the compressive  stress is applied to only one maximum shear
          plane, failures continue to occur on the other maximum shear plane at the same shear stress
          range).  Sines and  Ohgi  [I91 in  an extensive 1981 review of fatigue under combined stresses
          show only three data points to support their conclusion that static compressive stresses normal to
          the planes of maximum shear are beneficial.  Two of these points come from a reanalysis of tests
          on notched specimens by Seeger [20] in which the notch constrained cracks from growing on the
          second maximum shear plane.  Other evidence of  the beneficial effect of  compressive normal
          stresses on shear fatigue that we have uncovered consists of crack growth experiments by Smith
          and  Smith [21] who reported that a compressive normal stress decreased shear crack growth
          rates, and work by  Bums and Parry [22] who showed that a superimposed hydrostatic pressure
          which  introduced compressive stresses on the planes of  maximum  alternating shear increased
          fatigue strength.
            Although it was assumed by Sines [6] and subsequent authors [23, 241 that for a given fatigue
          life, there is a linear relationship between  the maximum alternating shear stress and the static
          mean stress normal to the planes of maximum alternating shear for both tensile and compressive
          normal stresses, there is little data to support linearity in the compressive region.  Recent work
          by  Bonnen  [25] and Varvani-Farahani [26] which relates the effect of  tensile mean  stress to
          crack face interference suggests that, once a tensile mean stress opens the crack so that there is
          no longer interference, higher tensile mean stresses will not cause a further reduction in fatigue
          strength.  Both authors obtained interference free crack growth by combining the application of
          periodic Mode I overloads with an alternating shear stress.  Bonnen verified that the crack faces
          were  growing  free  of  interference with  each  other  by  performing  two  tests  with  different
          overloads for a given shear stress amplitude.  For  the second test, the Mode I overload was
          increased to almost double the overstrain level of the first test.  The equivalent fatigue life of the
          second test was the same as the first result and, consequently, crack face interaction was assumed
          to be at the lowest possible level.  Varvani-Farahani measured biaxial crack opening stress with
          the confocal scanning laser microscope (CSLM). Biaxial specimens were internally pressurized
          to separate the crack faces, and .increasing magnitudes of internal pressure were applied until the
          crack depth remained constant.  His CSLM imaging revealed that, with  the application of the
          Mode I overstrains (yield stress magnitude), the shear cracks were fully open at zero internal oil
          pressure. Theresearch completed by Bonnen and Varvani-Farahani is summarized in Fig. 1, where
   135   136   137   138   139   140   141   142   143   144   145