Page 530 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 530

500               SLENDER STRUCTURES AND AXIAL FLOW

                   i.e.  (I - P>f(C) = C cos(n.$),  where  C  is  a  constant to  be  determined. Therefore, the
                   projection P is defined by

                                            Pf(8 = f(C> - c cos(n0.                     (F.90)

                   The constant C may be calculated from the condition that each element in the range of
                   L is orthogonal to ker L* ,


                                                                                        (F.9 1 )

                   which leads to C = 2 Jd  f(() cos(?rc) de. Thus, the projection of any function f(4) onto
                   kerL* is given by

                                                                                        (F.92)

                     Now we decompose the solution according to (F.83), $ = $c  + qS, where

                              qC = q cos(7tC) E ker L   and   $s  = h(q, u, 6) E range L'.   (F.93)

                   In (F.93), q is the amplitude of the buckling mode. The bifurcation equation is obtained
                   by  projecting the original equation (F.88) onto the kernel of  L* through  (F.92), which
                   leads to

                                     I' [$" + u($  -    + . . .)] cos(rC)dt = 0.        (F.94)

                   Equation (F.94)  is  solved by  making  use  of  (F.93),  and  by  introducing the  unfolding
                   parameter ,u  = u - u,.  From the  symmetry of  the problem, it can be  seen that it is not
                   necessary to find h explicitly if one wants to find a bifurcation equation to the third-order
                   only. After some manipulations, this leads to


                                                                                        (F.95)

                   The  unfolding  parameter  p  represents  the  small deviation  from  the  bifurcation  point,
                   and from equation (F.95), it is obvious that  p = O(q2). Therefore the term pq3 can be
                   neglected in comparison to u,q3, so that the final bifurcation equation then becomes
                                                (8p/n2>q - q3 = 0.                      (F.96)


                   F.6.3  The method of alternate problems

                   Although  the  method  of  alternate  problems  is  usually  applied  to  finite  dimensional
                   problems,  it  is  introduced  very  briefly  here  because  of  its  similarity  with  the
                   Lyapunov-Schmidt  method presented in  Section F.6.2. A  detailed presentation may  be
                   found in Hale (1969) and Bajaj (1982). The method is particularly useful if the scaling
                   relationship between the small parameters and the amplitude is a priori  unknown, this
                   scaling  being  suggested  ultimately  by  the  bifurcation  equation,  as  in  the  example  of
                   the previous  section. In  spirit, the  method of  alternate problems is  very  similar to the
   525   526   527   528   529   530   531   532   533   534   535