Page 411 - T. Anderson-Fracture Mechanics - Fundamentals and Applns.-CRC (2005)
P. 411

1656_C009.fm  Page 391  Monday, May 23, 2005  3:58 PM





                       Application to Structures                                                   391


                       the stress field. This latter method is necessary when the stress distribution does not have a closed-
                       form solution, such as when nodal stresses from a finite element analysis are used.
                          When computing K  for the internal surface flaw, we must also take account of the pressure
                                          I
                       loading on the crack faces. Superimposing p on Equation (9.9), and applying Equation (9.7) to
                       each term in the series leads to the following expression for K  [7,12]:
                                                                         I
                                    pR 2         a      a   2    a  3    a   4    π a
                              K =  R  2 o  o R −  i  2  2 G − 2    R   G + 3    R    G − 4    R    G + 5    R    G     Q  (9.10)
                                            o
                                I
                                                     1
                                                  i 
                                                               2
                                                                                   4
                                                                         3
                                         
                                                                     i
                                                           i
                                                                               i
                       Applying a similar approach to an external surface flaw leads to
                                     pR 2          a      a  2     a  3     a   4    π a
                               K =  R  o  2  i R −  i  2  2 G + 2    R   G + 3    R    G + 4    R    G + 5    R    G     Q  (9.11)
                                              o
                                I
                                                       1
                                                                             3
                                                    o 
                                                                  2
                                                                                        4
                                           
                                                                                   o
                                                                         o
                                                              o
                       The origin in this case was defined at the outer surface of the cylinder, and a series expansion was
                       performed as before.  Thus  K  for a surface flaw in a pressurized cylinder can be obtained by
                                               I
                       substituting the appropriate influence coefficients into Equation (9.10) or Equation (9.11).
                          The influence coefficient approach is useful for estimating K  values for cracks that emanate
                                                                            I
                       from stress concentrations. Figure 9.7 schematically illustrates a surface crack at the toe of a fillet
                       weld. This geometry produces local stress gradients that affect the K  for the crack. Performing a
                                                                               I
                       finite element analysis of this structural detail with a crack is generally preferable, but the influence
                       coefficient method can give a reasonable approximation. If the stress distribution at the weld toe
                       is known for the uncracked case, these stresses can be fit to a polynomial (Equation (9.6)), and
                       K  can be estimated by substituting the influence coefficients and polynomial coefficients  into
                        I
                       Equation (9.7).
                          The methodology in the previous example is only approximate, however. If the influence
                       coefficients were obtained from an analysis of a flat plate, there may be slight errors if these G n
                       values are applied to the fillet weld geometry. The actual weld geometry has a relatively modest
                       effect on the G  values. As long as the stress gradient emanating from the weld toe is taken into
                                   n
                       account, computed K  values will usually be within 10% of values obtained from a more rigorous
                                        I
                       analysis.













                       FIGURE 9.7 Application of the influence coefficient approach to a complex structural detail such as a fillet
                       weld.
   406   407   408   409   410   411   412   413   414   415   416