Page 52 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 52

CONCEPTS. DEFINITIONS AND METHODS                    35

               downstream station, and a selected group of  these integral equations is solved, often by
               a relatively simple numerical method (White 1974; Schetz 1993).]
                 Presuming now that v, has been determined, substitution of  (2.99) into (2.85) gives
                       au,
                      --+u  au,        1 ap, +(v,,+v,)-+- a2u,  av,  (av, -+- 2)   (2.106)
                                                                           ,
                       at      ax,      axi           ax;   ax,   ax,

               where pr = P + $pK is the turbulent ‘total pressure’. Equation (2.106) may be written in
               the usual, but perhaps less convenient, form
                    av                1
                    - +(V*V)V= --V~~+(V,,+V~)V~V+(VLJ,.V)V+(W).VV~, (2.107)
                     at               P

               where W is the so-called dyad, a vector (Wills 1958; Tai  1992). Examples of  the use
               of these equations andor the ideas summarized in this subsection are presented in Chap-
               ters 7- 10.

                (g) Empirical formulations
               As  intimated  in  the  foregoing,  mixed  analytical-empirical formulations  of  the  fluid-
               dynamic forces may be the only convenient way to analyse some fluid-structure interaction
               problems (e.g. provided that there is no large-scale flow separation, by analysing the flow
               as if  it were  inviscid, thereby  obtaining the pressure-related forces, and  adding empir-
               ical  expressions for the  viscous stresses acting on  the  body  surface). Indeed, in  many
               cases  involving complex flows, e.g. cross-flow of  heat-exchanger tube arrays, the  very
               foundation of the theoretical model may be empirical or quasi-empirical.
                 In analysing the empirical (experimental) data, it is convenient to express the unsteady
               fluid loading,  F(t), acting on  an  oscillating structure in  terms of  components in  phase
               with acceleration, velocity and displacement of  the structure, locally linearized; thus, for
               a one-degree-of-freedom system,

                                         F(t) = -m’z  - c’i - k’z.                 (2.108)

               When this is substituted in the equation of motion of the structure, mz + cz + kz = F(t),
               one obtains
                                    (rn + m’)z + (c + c’)z + (k + k’)z = 0,        (2.109)
               hence  the  appellation  of  m’,c’ and  k’  as  the  added  mass,  added  damping  and  added
               stiffness [e.g. Naudascher & Rockwell (1994, Chapter 3)].
                 For example, for a long cylinder of cross-sectional area A  and length L, oscillating in
               unconfined dense fluid of density p, the added mass per unit length is m* = m’/L = PA,
               if  end effects are negligible. If the cylinder is in a conduit of complex geometry, m’ may
               be determined analytically, numerically or experimentally, and the added mass per  unit
               length expressed by
                                                  m’
                                            m* = - = C,pA.                         (2.110)
                                                  L
               In  general,  C,  will  be  a  function of  geometry, viscosity  and  frequency (hence  of  the
               oscillatory Reynolds number), amplitude, and other factors as discussed in Sections 2.2.2
               and 2.2.3 and by others (Chen 1987; Gibert 1988; Naudascher & Rockwell 1994). In many
   47   48   49   50   51   52   53   54   55   56   57