Page 381 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 381

PIPES CONVEYING FLUID: NONLINEAR AND CHAOTIC DYNAMICS         351

              well not to rely on just  one measure  in deciding that motions are chaotic or otherwise.
              Professor F.C. Moon made the point eloquently in a course he gave at Cornell, through
              a parable, paraphrased here from memory.  ‘If you see something that Zooks  like a duck,
              it does  not  necessarily mean that it  is  a duck.  If, however, it also walks like a duck, it
              swims like a duck, and it quacks like a duck, then it is much more likely to be  one!’+
                The experimental system was also studied analytically, initially by the simplest possible
              model (PaYdoussis & Moon  1988; Paldoussis et aZ. 1989). As motions are relatively small
              because  of  the  constraining bars,  the linear  version of  equation (5.39) is used  as a first
              approximation - i.e. equation (3.70) - apart from the forces associated with impacting.
              A good model for the latter is a trilinear spring: zero stiffness when no contact is made, and
              a large stiffness once it is. For analytical convenience, this can be approximated by a cubic
              spring (Figure 5.33); hence, the following term is added to the dimensionless equation of
              motion:  KQ~S($ - &,),  where &,  is the  dimensionless  axial location of  the  constraints,  K
              the dimensionless cubic-spring stiffness, K  = kL5/EI, k  being the dimensional value, and
              S is Dirac’s delta function. Thus, the equation of  motion is






              The system is discretized by  Galerkin’s  method into a two-degree-of-freedom  (N = 2),
              four-dimensional  (4-D) model.  Solutions  are  obtained  by  numerical  integration  via  a
              fourth-order Runge- Kutta integration algorithm.


























              Figure 5.33  Diagrammatic view of the idealization of the trilinear (or ‘bilinear’) motion constraint
                            (-)   by a cubic spring (---)  (Pdidoussis & Moon 1988).


              threshold of chaos is provided by  the calculation of  the Lyapunov exponents, discussed later. For this problem,
              however, such calculations were confined to the analytical model, although they are also possible, but  not at
              all easy, for experimental data (Moon 1992).
                +The author, having recently discovered the excellent Belgian beer Kwak, feels that this argument is further
              reinforced, since thirsty humans may just as plausibly emit  ‘Kwak, Kwak’ as itinerant ducks.
   376   377   378   379   380   381   382   383   384   385   386