Page 398 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 398

374               SLENDER STRUCTURES AND AXIAL FLOW








































             Figure 5.46  Bifurcation  diagrams  of  the  dimensionless  pipe-end  displacement  versus  u  for
             the  system  of  Figure 5.43(a)  by  the  FDM  method  for  (a) p = 0  and  (b) p = 0.06;  -,
             N  = 3  in  the  Galerkin  discretization;  A, N  = 4.  (c) Time  trace  and  (d)  PSD  in  terms  of
             f = w/2n  for  p = 0.06, N  = 3  and  u = 8.5.  In  all  calculations the  parameters  are  as  in  the
             experimental system: B = 0.142, y  = 18.9, and the measured logarithmic decrements of damping:
               6, = 0.037, SZ = 0.108,63 = 0.161 and extrapolated 64 = 0.220 (Pdidoussis & Semler 1998).

               For p = 0.06, the dynamical behaviour is initially similar, as shown in Figure 5.46(b).
             For 8.2 < u < 8.625, however, there is a band of quasiperiodicity, involving two frequen-
             cies,  as  made  clear  in  the  time  trace and  PSD  in  (c) and  (d) of  the  figure. The  two
             dimensionless frequencies are fl  1: 0.5 and f2 1: 3, while the third peak in the PSD is at
             2f2 + fl,  and so on; f = w/2n, while w is as defined in equation (3.73). For u > 8.625,
             periodic oscillations resume, but at a smaller amplitude. This theoretical evolution bears
             some resemblance to the experimental behaviour, but there are some obvious discrepancies
             as well. Comparing the results of Figure 5.45 for p = 0.06 with those of Figure 5.46(b),
             it is seen that theory and experiment agree in  the following aspects: (a) the  values of
             UH for the first Hopf  bifurcation are similar (u~ 5.3 in  the experiments, compared to
                                                      1:
             UH = 5.35 in  theory); (b) the nonlinear model and the experiments both predict a qual-
             itative change in the behaviour of  the pipe at a higher flow; and (c) the values of  u for
             the second bifurcation are relatively close (uheory 1: 8.2 versus uexp 1: 7.8). On the other
             hand, only periodic solutions are predicted in the experiment (before the onset of chaos),
             while the motion is also found to be quasiperiodic in theory, prior to becoming periodic
   393   394   395   396   397   398   399   400   401   402   403