Page 315 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 315

296                SLENDER STRUCTURES AND AXIAL FLOW

                   equations derived for cantilevered pipes, those by Lundgren et al. (1979) and Bajaj et al.
                   (1980) are found to be absolutely correct, while those by Rousselet & Herrmann (1981)
                   are found to be correct except for a small order-of-magnitude inconsistency. Furthermore,
                   both sets contain a distinct refinement vis-&vis  those derived here: the flow velocity is
                   not assumed to be constant; instead, the upstream pressure is taken to be constant, while
                   the flow velocity generally varies with deformation. Of  the equations derived for pipes
                   withJixed ends, the set derived here is considered to be the only one available, correct to
                   the same order as that for the cantilevered pipes. On the other hand, the simple equation
                   derived originally by Holmes (1977) is correct as far as it goes and may be preferred in
                    some cases because of  its simplicity. It is of interest that the origin of the terms in the
                   equations - even some of the linear terms - as well as the structure of the equations are
                   distinctly different for pipes with both ends fixed as compared to cantilevered ones.


                    5.3  EQUATIONS FOR ARTICULATED SYSTEMS
                    Traditionally, articulated models of columns subjected to axial loading have been widely
                    used  as  an  aid  in  the  study  of  their  continuous,  distributed  parameter  counterparts
                    (Herrmann  1967). The  same  has  occurred  with  pipes  conveying  fluid.  For  nonlinear
                    dynamics this is particularly attractive, since many of the methods of nonlinear dynamics
                    are best suited to low-dimensional discrete systems; with articulated systems, no questions
                    need arise as to the adequacy of the Galerkin discretization of a continuous system: the
                   physical  system is discrete and may be low-dimensional to start with.
                      Most of  the interesting dynamics is associated with cantilevered systems, and hence
                    most of  the research has been devoted to such systems. Furthermore, virtually all of that
                    work has been confined to two-degree-of-freedom (N = 2) systems (Figure 5.2), and this
                    despite the caveat  (Section 3.8) that the dynamical behaviour of the N  = 2 system is not
                    generic with respect to N  (Paidoussis & Deksnis 1970).
                      Two representative sets of  equations are given here, both for cantilevered two-degree-
                    of-freedom  systems.  The  first  set  was  derived  by  Rousselet  & Herrmann  (1977)  by
                    straightforward Newtonian  methods  via  free-body  diagrams  and  moment  balances  on
                    the two segments of the pipe, yielding









                                                                                        (5.64)









                    where M and m  are the  mass of  the  fluid and of  the  pipe per  unit  length, and  U the
                    flow velocity; ZI and  12  are the  lengths  of  the  upper  and  lower  segments of  the  pipe
                    (Figure 5.2), and kl  and kz are the stiffnesses of  the interconnecting springs. The flow
   310   311   312   313   314   315   316   317   318   319   320