Page 34 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 34

CONCEPTS, DEFINITIONS AND METHODS                   17
               a way  of  solving the problem correctly while utilizing boundary conditions (2.29), but
               the meaning of the domain 9 has to be expanded (Friedman 1956; Meirovitch 1967); an
               example is given in Chapter 4 (Section 4.6.2).



               2.1.5  Self-adjoint and positive definite continuous systems
               The eigenvalue problem of  equation (2.36), and thereby  the  system, is  said  to  be  se@-
               adjointt  if for any two comparison functions, u and u,
                                                   I            s,
                     s, u2?[u] d9 =   vZ[u] d9,       u&[u] d9 =   v&[u] d9,       (2.42)


               are satisfied. A consequence of self-adjointness is that the eigenvalues are real. Another
               consequence is that  a generalized or weighted orthogonality of  the eigenfunctions then
               holds true for nonrepeated eigenvalues; thus,





               Furthermore, if
                                r                         r
                               1% uZ[u]d9 > 0     and    J’, uA[u] d9 > 0

               for  all  nonzero  u, the  operators are positive definite,  and  hence so  is  the  system. The
               consequence  of  this  is  that  the  eigenvalues of  such  a  system  are  positive - refer  to
               Section 2.3 for the significance of this and back to Section 2.1.3 for further clarification
               of the different usage of  the word  ‘eigenvalue’. In cases where 2 is only positive, rather
               than positive definite, i.e. when the first integral (2.44) can be zero for some nonzero u,
               while A remains positive definite, the system is called positive seniide$wite, and  admits
               solutions with A  = 0.
                 Clearly, for the system of  equations (2.22) and (2.23), the problem is self-adjoint. To
               illustrate the case of  a non-self-adjoint system in  as simple a manner as possible while
               still keeping in the framework of  the examples already discussed, consider the system

                        Z[W]  = EI(d4/dr4) + P(d2/dr2),   A[w] = m;                (2.45)
                         ~(0) 0,      ~’(0) = 0,   EItv”(L) = 0,   EZw”’(L) = 0.   (2.46)
                             =
              This could represent a cantilevered beam, subjected to a compressive tangential ‘follower’
               force P, such that  the boundary  conditions remain  unaffected. A  follower force is  one
               retaining the same orientation to the structure in the course of  motions of the system, in
              this case remaining tangential to the free end.$ By applying the integrals (2.43) it is found
              that the integrated out parts do not vanish [since Pu(l)v‘(  1) and Pu’(l)u( 1) are not zero
              for the boundary conditions given].

                 ‘If  Y and A are complex operators, the equivalent property is for the eigenvalue problem to be Hennirion.
                 *In fact,  such a compressive follower  force could be  generated  by  a  light  rocket  engine  [M,/(nzL) z 01
              mounted on the free end of the cantilever, so that the force of reaction is always tangential to the  free end.
   29   30   31   32   33   34   35   36   37   38   39