Page 93 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 93

76                SLENDER STRUCTURES AND AXIAL FLOW

                  instance,+ for metals and certain types of rubber-like materials, and over frequency ranges
                  of  practical  interest, energy dissipation can  adequately be  accounted for by  hysteresis;
                  then, when a specimen of such a material is subjected to harmonic loading with a (real)
                  circular frequency  52,  the  energy dissipation per  cycle can be  calculated by  taking the
                  Young’s modulus to be complex, in  the form E(l + pi), where E  and p  are constants
                  independent of  52,  and p <<  1. This implies that the small stresses related to hysteresis
                  are in quadrature with the principal, linear-elastic stresses. This representation remains a
                  reasonable approximation for lightly damped oscillation - i.e. provided that sm(52) <<
                  %e  (52) when  52 = %e(52) + i9m(52); however, if  there is  another source of  damping
                  (e.g. flow-induced damping in  cantilevered pipes conveying fluid) such that the overall
                  damping is large, misleading results may be obtained. Nevertheless, within the limits of its
                  applicability [e.g. close to a flutter boundary or for lightly damped conservative systems
                  where sm(f.2) <<  %e(52)], the hysteretic model is very convenient. In that case, the first
                  term of  equation (3.38) may be replaced by

                                                                                       (3.39)


                    Finally, a variant of the equation of motion, first introduced by Gregory & Paidoussis
                  (1966a) for experimental convenience (Section 3.5.6)  will be discussed. For  simplicity,
                  consider the  horizontal  system with  dU/dt = 0 and  neglect dissipation. Then  suppose
                  that the downstream end of  the pipe is  fitted with a convergent nozzle, assumed to be
                  weightless and very short compared to the total length of the pipe. The discharge velocity
                  Uj is given by Uj = U(A/Aj), where A,  is the terminal cross-sectional area of the nozzle
                  flow passage. Equation (3.36) in this case simplifies to


                                          (T - pA)l   - (T - pA) = 0;                  (3.40)
                                                  x=L
                  consideration  of  momentum  at  x = L - cf.  the  second  and  third  terms  of  equation
                  (2.63) - gives
                                                  I
                                          (PA - T)/   = MU(Uj - U),                    (3.41)
                                                   x=L
                  which,  in  view  of  (3.40),  applies for  all  x.  Hence,  substituting into  (3.34),  simplified
                  according to the assumptions made here, yields the modified equation of motion
                                  a4          a2         a2w           a2
                                                             +
                                                  +
                               EI - +-MUU. - 2MU - (M +m)-                 = 0.        (3.42)
                                   ax4         ax2       axat          at2
                  3.3.3  Hamiltonian derivation

                  The difficulty in deriving an expression of Hamilton’s principle for this problem lies in the
                  fact that the system is open, with in-flow and out-flow of mass and momentum. Housner’s
                  (1952) derivation of the equation of motion for pipes with supported ends by means of

                    +See also Payne & Scott (1960), Snowdon (1968) and the workshop proceedings edited by Snowdon (1975)
                  and Rogers (1984).
   88   89   90   91   92   93   94   95   96   97   98