Page 62 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 62

CONCEPTS, DEFINITIONS AND METHODS                    45

              where  C,,,  is  given  by  equation (2.127), e = (smallest gap  between  the  cylinders)/R;,
               r = R,/R;. For larger values of r, the results are given in figure form. Results for a variety
              of  other systems may be found in the compilations of Chen (1987) and Blevins (1979).

               (ii)  Aspect  ratio effects.  As two-dimensionality of  the flow  is violated, the validity
              of the foregoing deteriorates. A particularly simple example illustrating this is a simply-
               supported cylindrical beam oscillating in a narrow  annulus [so that  the  approximations
              leading to (2.134) are valid], the ends of which are open to large cavities (Gibert 1988).
              It is found that
                                                                                  (2.139)

              where  C,,,  is  given  by  expression (2.134) for n = 1. Clearly, the  shorter the beam, the
              smaller is C:am,  as compared to an infinitely long one. The physical reason is that, near the
              ends, the fluid takes the easy way  around the beam, partly  in the third (axial) direction;
              hence,  less  than  the  total  force that  would be  obtained by  2-D  analysis is  realized. A
              more general analysis (PaYdoussis et al. 1984; Chen 1987), not making the assumption of
              a narrow annulus, gives

                                                                                  (2.140)

              where I1  and  K1  are, respectively, the  first-order modified Bessel  functions of  the  first
              and second kind; the primes denote derivatives with respect to the argument. The effect
              of R;/L  is strong for  1 < R,/R; < 2, but relatively weak for wider annuli.
              (iii)  Effects  of  compressibility  and  two-phase  flow.   If  the  flow  is  compressible,
              the wave equation, V2+ + k2+  = 0, k2 = o/c,  needs to be solved instead of the Laplace
              equation, V2@ = 0. Hence, the results are found to depend also on  an oscillatory Mach
              number, hfk  = wR;/c,  where c  is the  speed of  sound. The effect of  compressibility for
              Mk  5 0.2 is rather weak (Chen 1987).
                It has been found (Carlucci 1980; Carlucci & Brown 1983) that in gas-liquid  two-phase
              flows the  measured added mass is generally considerably lower than  that  predicted by
              homogeneous mixture theory  [in which average quantities are assumed for the mixture;
              e.g. if the void fraction is cr and the densities of the liquid and gaseous phases are p/ and
              pg, the mixture density is p  = (1 - cr)p, + upg]. Since the two-phase flow may be consid-
              ered as a flow with the density of the liquid phase and the compressibility of the gaseous
              one, it was supposed that the discrepancy may have been due to the neglected effects of
              compressibility (PaYdoussis & Ostoja-Starzewski 198 I).  Also, the effect of random varia-
              tions in the surrounding fluid density, inherent in two-phase flows was investigated (Klein
               1981). These effects, although qualitatively working  in  the  right  direction (Chapter 8),
              proved  incapable of  accounting fully  for the discrepancy quantitatively, and  the  search
              for more elaborate models continues.
              (iv)  Amplitude effects.  All  of  the foregoing apply to cases where the amplitude of
              oscillation is small enough for separation in the cross-flow not  to occur. This brings into
              play another dimensionless number, the Keulegan-Carpenter  number, KC = 2nV,/(oD),
              where V,  is the amplitude in velocity fluctuations. For a harmonically oscillating cylinder
              in quiescent flow, this  reduces to  KC = 2n(A/D), where A  is the amplitude of  motion
   57   58   59   60   61   62   63   64   65   66   67