Page 110 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 110

92                SLENDER STRUCTURES AND AXIAL FLOW

                   in Figure 3.9 and u 2: 9.0+ in Figure 3.10), they leave the axis, indicating the onset of
                   Pafdoussis-type coupled-mode fluttert as defined in Section 3.2.3 and by  Figure 3.4(d).
                     The behaviour of  Figure 3.11 (j3 = 0.5) is different. The w = 0 solution for u 2: 8.99
                   does not correspond to a second divergence, but to restabilization of the system. This lasts
                   to u 2: 9.3, whereupon coupled-mode flutter occurs via a Hamiltonian Hopf bifurcation,
                   as defined in Figure 3.4(c).
                     What is  particularly interesting about this predicted coupled-mode flutter is its  orig-
                   ination.  As  discussed  in  Section 3.2.1  and  as  shown by  equations (3.5)  and  (3.6), for
                   periodic motions there is no energy transfer between the fluid and the pipe. Hence, since
                   the system is conservative, the question arises as to how the instability can be supported
                   whilst the total energy of  the system remains constant. As pointed out by  PaYdoussis &
                   Issid(1974), the  question is  not  quite  like this,  since the  critical point for the onset of
                   flutter, unlike  for  the  nonconservative (cantilevered) system, is  not  a point  of  neutral
                   stability; rather, it involves the coincidence of  two real frequencies, and hence growing
                   oscillations of the form q(c, t) = %e[f(c)(a + bt) exp(iwt)], with w real. The source of
                   energy is of course the flowing fluid, yet how  some energy is channelled to generate the
                   oscillatory state remains the question. A possible answer was provided, via an ingenious
                   set of arguments, by Done & Simpson (1977) for a pipe with supported ends but with the
                   downstream end free to slide axially [Figure 3.1(b)].
                     First, one may consider a two-mode Galerkin approximation of the system, namely




                   For clamped and pinned ends, b,,  = 0 and b,,  = -brs;  for pinned ends, csr = 0 for all
                   r # s,  while  the  same applies to  clamped ends  for  r + s odd, which  is  the  case here.
                   Hence, equation (3.91) may be written as

                                 0      -2/3112~b21] ir+  [A;'  +;'cI~
                         q + [ 2/3'/2~b21   0                      A; +  u2c22  I  q = 0.   (3.92)
                                                                      O
                   It  is  of  interest  to  remark  that  (i) the  damping matrix  is  skew  symmetric, which  is  a
                   characteristic  of  the  system being  gyroscopic  conservative,  as  already  remarked,  and
                   (ii) by setting det[K] = 0, [K] being the stiffness matrix, one retrieves the zeros for static
                   loss of stability Ucd  = 7t and u = 237,  exactly for simply-supported ends and approximately
                   for clamped ends (since in this case the matrix is not fully diagonal for N  > 2).
                     Then, solutions of the form q = qo exp(At) are considered, leading to the characteristic
                   equation
                                              p4h4 + p2h2 + po = 0,                     (3.93)
                   with  p4  = 1, p2 = [A;'  +A; - u2(cll + c22) - 4Bu2b&], po = (A:  - u2cll)(h;  - u2c22).
                   The  condition of  coalescence of  two  eigenfrequencies corresponds to  two  equal  roots
                   of (3.93), which occurs if  pi - 4p4po = 0. The results for clamped ends are shown in
                   Figure 3.12,  where  it  is  seen  that  all  the  critical  points  of  Figures 3.10  and  3.11  are

                     ?This flattering appellation, coined by  Done  & Simpson (1977), has been retained here for this particular
                   form of coupled-mode flutter. This phenomenon, however, although analytically intriguing, was shown to  be
                   physically doubtful with  he appearance of  Holmes'  (1977, 1978) work. This would have rendered  any claim
                   to fame by this book's  author rather ephemeral, were it not for the fact that, luckily, the physical reality of the
                   phenomenon is firmly established for another system (Chapter 8)!
   105   106   107   108   109   110   111   112   113   114   115