Page 249 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 249

230                SLENDER STRUCTURES AND AXIAL FLOW

                    is  concerned,  because  the  fluid discharging  from  the  downstream  end  is  assumed  to
                    enter a rigid pipe which experiences no deflection [Figure 4.16(a)]. Therefore, in expres-
                    sions (4.67), I  = 1 or Gn(.$) = 0.








                                               Flexible tube







                                              Rigid pipe








                    Figure 4.16  The physical form of the ‘collector pipe’ for a clamped-clamped pipe and the form
                               of  the free jet emerging from a cantilevered pipe (no collector pipe).


                      The  two  inner  integrals  of  the  expressions  in  (4.67) may  be  evaluated  analytically
                    (Luu  1983) or numerically, but the three outer integrals, which involve an infinite range
                    of  integration  over  Z,  have  to  be  evaluated  numerically; this  is  done  by  a  two-point
                    Gaussian numerical integration method.  Based on  a  check  on  convergence for  a  long
                    clamped-clamped  pipe,  calculations  (throughout  this  work)  of  the  generalized  fluid-
                    dynamic  forces  for  a  clamped-clamped  pipe,  either  long  or  short,  are done  with  the
                    integration range  -100  5 a! 5 100 and  the  integration  step  SZ = 2;  they  approximate
                    the  result  for  a  larger  range  of  5 (and  hence  -m  5 a! 5 m) and  a  finer  83 very
                     well.
                      The next  step is  to  undertake  a  comparison between  the  results  of  the  generalized
                    hydrodynamic forces Qfi, i = 1,2,3, for a long clamped-clamped  pipe conveying fluid
                     (A = 1012, corresponding to  E = L/2a = 8.5 x lo5)+ obtained by  (i) simple plug  flow
                    and  (ii) refined  fluid  mechanics,  where  the  $(t) used  are  the  eigenfunctions  of  a
                    clamped-clamped  Timoshenko beam without internal flow, as given in Appendix E. For
                    the  first  (lowest)  three  modes  of  the  system  (k, n = 1, 2,3), the  results  are  virtually
                     identical: the  largest discrepancy, associated with the  Q:;) term, is only 0.023%. This,
                    to  some  extent,  validates  the  refined  fluid  mechanics  model,  which  may  now  fairly
                    confidently be used for short pipes clamped at both ends.

                       +This value of  E  is clearly nonphysical, but  has been dictated by  the  desire to  obtain virtually identical
                    results to those of the Euler-Bernoulli  theory, to many significant figures. Pmctically  identical results may be
                    obtained for E - S(l@).
   244   245   246   247   248   249   250   251   252   253   254