Page 444 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 444

416                SLENDER STRUCTURES AND AXIAL FLOW

                   configurations, Their equations  of  motion  have  a  significant difference  from  those  of
                   most of  the previous studies: they include the effect of the initial forces arising from the
                   centrifugal effect and the pressure of the fluid. They obtained the unexpected result that, if
                   the initial forces are taken into account, then pipes with both ends supported do not  lose
                    stability, no matter how  high the flow velocity may be!  Similar observations had been
                    made by Svetlitsky (Svetlitskii) in an earlier paper published in Russian (Svetlitskii 1969)
                    and  subsequently  in  English  (Svetlitsky  1977); see  also  Svetlitskii  (1982).  Svetlitsky
                    considered cantilevered pipes as well and noted that this system loses stability by flutter
                    at sufficiently high flow velocities, even when initial forces are taken into account.
                      Doll  & Mote  (1974,  1976) studied a  more  general case,  where  the  fluid-conveying
                   pipe is both curved and twisted. The equations of  motion were derived via Hamilton’s
                    principle and solutions were obtained using the finite element method. They considered
                    two  cases:  (i) the  ‘constant curvature’  case,  in  which  the  curvature  does  not  change
                    with flow velocity, and (ii) the  ‘variable curvature’ case in which variations in curvature
                    with changes in flow velocity are accounted for. The first case corresponds to the analyses
                    of Unny et al. and Chen, while the second is similar to Hill & Davis. Both Doll & Mote
                    and  Hill  & Davis  take  into  account  the  extensibility  of  the  centreline  of  the  curved
                    pipe. An important difference between the two studies is that in the latter the equilibrium
                    configuration and forces are calculated via a linearized set of equations, on the assumption
                    that  the initial and  the flow-deformed equilibrium configurations are close; in  the Doll
                    & Mote  study, on  the  other  hand,  a  cumulative application of  linearization is  utilized
                    for  small  flow  velocity  increments,  which  is  more  general.  The  main  conclusions  of
                    both  studies,  however, are  the  same:  the  eigenfrequencies of  pipes  supported at  both
                    ends  are  not  sensitive  to  flow  velocity,  and  hence  no  instabilities  should  arise.  Doll
                    & Mote  also compared their  curved-pipe theory  to Liu  & Mote’s  (1974) experimental
                    data  for  nominally  straight but  actually  slightly curved  pipes,  supported  at  both  ends
                    (see Section 3.4.4).  This comparison is discussed in Sections 6.4.5 and 6.6.2.
                      More recently, Dupuis & Rousselet (1985) have carried out a study on the dynamics
                    of fluid-conveying planar curved pipes modelled as Timoshenko beams. The extension of
                    the centreline was taken into account. This study used the transfer matrix method (Pestel
                    & Leche  1963), in  preference  to  either  analytical  or  finite  element  methods.  Once
                    more, flutter instabilities are predicted for cantilevered curved pipes, but results for pipes
                    supported at both ends are not presented.
                      Misra  et al. (1988a,b,c) re-examined  the  dynamics  and  stability  of  fluid-conveying
                    curved pipes ab initio. The main objective was to shed light onto the underlying reasons
                    for  the  fundamentally different dynamical behaviour for  pipes  with  supported ends as
                    predicted by the extensible theories of Hill & Davis, Doll & Mote and Svetlitsky, on the
                    one hand, and the inextensible theories of Chen and Unny et a1 ., on the other: the former
                    predicting no loss of  stability, while the latter predict divergence at high  enough flow.
                    For this purpose, three theories were formulated and their results were compared:
                       (i)  the  conventional  inextensible  theory,  in  which  the  centreline  of  the  pipe  is
                           assumed to be unstretched, and the steady (initial) flow-induced forces introduced
                           by the pressure and centrifugal forces are entirely neglected;
                      (ii)  the extensible theory, in which the shape of the pipe changes with flow velocity
                           under the action of the steady flow-induced forces;
                      (iii)  the mod$ed  inextensible theory, in which the assumption of inextensibility of the
                           centreline is retained, but the steady flow-related forces are taken into account.
   439   440   441   442   443   444   445   446   447   448   449