Page 133 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 133

PIPES CONVEYING FLUID: LINEAR DYNAMICS I              115

























                                     0      0.2      0.4    0.6
                                                   B
               Figure 3.31  Comparison  between  urf  and  wCf  obtained  by  the  exact  solution  (-),   cf.
               Figure 3.30. and Galerkin  approximations:  0, N  = 2; +, N  = 3, A, N  = 4  (Gregory & Paidoussis
                                                 1966a).


               of  Section 3.3.6(a), some obtained by  the Galerkin method of  Section 3.3.6(b) are also
               presented, for N  = 2, 3 and 4, N  being the number of beam modes utilized. It is obvious
               that, although N  = 3  and 4 may be  adequate for predicting ucf  (see also Figure 3.27),
               the  two-beam-mode  approximation  (N = 2)  is  not,  failing  to  reproduce  the  S-shaped
               behaviour, as  will  be  discussed further in  Section 3.5.4; on  the  other hand,  the N = 2
               approximation is  quite reasonable for B 5 0.2, or  even B = 0.25. In  general, higher-N
               approximations become necessary to adequately represent the dynamics of the system as
               u  and B  are increased. This contrasts sharply to the inherently conservative system [cf.
               equation (3.92) of  Section 3.4.1 and  the  attendant discussion], where N  = 2  and  even
               N  = 1 Galerkin approximations can predict u,d  very well.


               3.5.2  The effect of gravity

               The  motivation for  investigating the  effect of  gravity  (y # 0) on  the  dynamics of  the
               system comes from two sources. The first is to obtain theoretical results for comparison
               against measurements from experiments with pipes oscillating in  a vertical  rather than
               a horizontal plane, the former being easier to conduct. In this regard, recalling that  y  =
               (M + m)gL3/EZ, turns out that for metal pipes conveying fluid, unless L is very large,
                             it
               y  is small and its effect on the dynamics may well be negligible; for rubber or elastomer
               pipes, however, with which the majority of  the experiments are conducted, because E  is
               considerably lower, gravity effects should normally be accounted for. The second source
               of  impetus was provided by  Benjamin’s (1961a,b) findings with articulated cantilevered
               pipes conveying fluid: that horizontal systems lose stability exclusively by flutter, whereas
               vertical ones can do so by  divergence also (Section 3.8). Hence, since the continuously
   128   129   130   131   132   133   134   135   136   137   138