Page 194 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 194

176               SLENDER STRUCTURES AND AXIAL FLOW

                    simply reflects that, in the absence of  any support, a lateral displacement of  the pipe is
                    not opposed by any restraint.
                      Now, if the analysis is conducted with  (T  = 0 from the start, it is easy to show that in
                    this case the condition of neutral stability, A1,2  = 0, requires
                                             u2(1 - p) - f + n < 2h.                  (3.144b)

                    Clearly, since /3  < 1, this result is nonconservative; in particular, criterion (3.144b) predicts
                    a system to be stable when, in fact, through (3.144a) it is unstable (Roth  1964)! This is
                    a good demonstration of  Bolotin & Zhinzher's  (1969) thesis (Section 3.5.5).
                      It is Stein & Tobriner (1970) who consider wave propagation per se. They use the same
                    equation  as Roth, but  with  r = 0. They obtain a  general solution to  initial conditions
                    q(c, 0) = f(6)  and c(c, 0) = g(6) by  means of  Laplace transforms in  time (denoted by
                    an overbar) and Fourier transforms in space (denoted by  an asterisk), of  the form




                    The Laplace transform over  t is applied first, and then the Fourier transform over c, on
                    the resultant equation. After inversion, the general solution is




                             +-     e-ur/2[2B1/2~a cos(4 - p) - (T  sin(@ - p) cos 81  sinh 62
                                24-
                             + [2/3%a  sin(@ - p) + CT cos(@ - p)] sin 81  cosh 621
                                                                                        1
                             +-     e-ur/2[sin(q5 - p) cos O1 cosh O2  - cos(@ - p) sin 81 cosh 1921  da,
                                4-
                                                                                        (3.145)

                    where
                               r = {[a2(a2 - (1 - B)u2 + n) + k - $72]2 + [B1/2u~a]2}1/2,

                                                       /3'/2UCT
                               p = - tan                                                (3.146)
                                                                +
                                   2       a2[a2 - (1 - B)u2 + n] k - io2
                                                                               -
                              01  = zfi  cos p,   02  = r,h  sin p,   4 = w(~'/~ur c).
                    Numerical results in the case of  (T  = 0 and q = exp[i(Kc - wt)] are then considered. The
                    characteristic equation in this case is

                                       w2 - 2B1'2UKW  - K4  f (u2 - n)K2 - k  = 0.      (3.147)
                    Thus,  for  each  wavenumber (wavelength) there  is  an  associated frequency and,  in  the
                    absence of end constraints, all wavenumbers are permissible. The dominant wavelengths
                    depend on the spatial distribution of  the initial disturbance and the propagation charac-
                    teristics of each of its Fourier components. Equation (3.147) may be solved for the phase
   189   190   191   192   193   194   195   196   197   198   199