Page 459 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 459

CURVED PIPES CONVEYING FLUID                     43 1

             pipes  of  uniform  curvature  as to  S- or  a-shaped pipes - without  having  to  determine
             the equivalent of new comparison functions for each case. According to this method, the
             continuum is subdivided into a number of elements, at the edges of which there are one
             or more nodes (one for a structure such as this, in which spatial variations involve only
             one  coordinate,  0. The  equations  of  motion  are  satisfied at  the  nodes  rigorously  and
             elsewhere  approximately  via  interpolation  functions.  A  variational  formulation  ensures
             a  systematic minimization  of  the error  for the  number  of  elements utilized.  The use  of
             the  finite element  method has now become routine,  and hence the  uninitiated  reader  is
             referred to one of several texts on the subject, e.g. Zienkiewicz & Cheung (1968), Desai
             & Abel (1972), Becker et al. (1984) or Zienkiewicz & Taylor (1989).
               The  details  of  the particular  form  of  finite element  analysis  employed  here  may  be
             found in Van  (1986). However, different forms may be employed, and this is partly the
             reason for not presenting the minutiae of the analysis.


             6.3.1  Analysis for inextensible pipes

             (a) In-plane motion
             The pipe is discretized  into  n  elements. The initial curvature of  a particular  element  is
             constant,  although  it can vary from element to  element. The variational  statement used
             for the finite element discretization is

                                                                                  (6.75)


             where 87;  is an arbitrary variational displacement, Ai(q;)  represents the left-hand side of
             equation (6.60) and   is the length of  the jth element. The subscript i  in Ai  stands for
             ‘in-plane’ .
               The longitudinal displacement  q;  may be expressed in the form

                                            11;  = [N31{qiIe,                     (6.76)
             where  [N3] is  a  matrix  of  interpolation  functions  at  the  space  coordinate  {, and  {si)“
             is  the  element  displacement  vector,  of  appropriate  dimension  and  dependent  only  on
             time; the superscript e stands for ‘element’. Hence, by using (6.75) and (6.76), integrating
             by  parts, and applying the boundary conditions - thereby eliminating the integrated-out
             components - one obtains the discretized equation

                                 [M,Ie{ijile + [DiIe{4iI“ + [KiIe(qiJe = (01,     (6.77)

             where
   454   455   456   457   458   459   460   461   462   463   464