Page 42 - Fluid-Structure Interactions Slender Structure and Axial Flow (Volume 1)
P. 42

CONCEPTS, DEFINITIONS AND METHODS                    25

               (b) Potential flow theory
               Many  interesting inviscid  flows (e.g. a  uniform  flow approaching a  body) are  initially
               irrotational, i.e. the vorticity, w, is everywhere zero: o = V x V = 0. Hence, by  Kelvin’s
               theorem, such flows remain irrotational;’ the flow is then referred to as potential flow and
               is associated with the velocity potential, 4, where V = V4. Euler’s equations in this case
               simplify to the well known unsteady Bernoulli, or Bernoulli-Lagrange,  equation

                                            a4         P
                                            - + $2  + - = 0,                       (2.67a)
                                            at         P
               where p is measured relative to the stagnation pressure of the free stream.$ This form of
               the equation applies if  there are no body forces. If  there are, for example due to gravity,
               the following form may be more useful:

                                          a4
                                         - + ;v2+  - +gz =o,                       (2.67b)
                                                     P
                                          at         P
               where z  is the vertical height. There exists a highly developed mathematical treatment of
               potential  flow - see, e.g. Lamb (1957), Streeter (1948), Milne-Thomson (1949,  1958),
               Karamcheti (1966), Batchelor (1 967).
               (el Very low Reynolds number flows

               In  this  case,  when  Re + 0, inertial effects become negligible, and  the  Navier-Stokes
               equations reduce to the equations of creeping flow,


                                               vp = pv2v.                           (2.68)

               A  number  of  well  known  solutions  exist,  e.g.  for  the  plane  Couette  and  Poiseuille
               flows, classical lubrication theory (Lamb 1957), Stokes flow past a sphere and constant
               pressure-gradient laminar flow through pipes; but,  surprisingly perhaps, not  for low-Re
               two-dimensional cross-flow over a cylinder (Stokes’ paradox).
               Id) Linearized flows

               In some problems there is one dominant steady flow-velocity component, while all others
               are perturbations thereof, say induced by structural motion, e.g. V = Ui + v, where llvll  <<
                  i
               U; is the unit vector in the x-direction. In such cases, the Navier-Stokes  equations may
               be linearized and simplified considerably. Thus, if  U is steady, i.e. not  time-dependent,
               and spatially uniform, the Navier-Stokes  equations reduce to
                                       av     av     1
                                       - + u-  = -- v p + v v2v.                    (2.69)
                                        at    ax     P
                 ‘Interestingly,  this is not so if there is a density gradient to  the  fluid!
                 iThus,  the integration constant that  would  otherwise  appear on the  right-hand  side  reduces  to  zero.  This
               constant, C(t), is generally  a function of time if, unusually, the  hydrostatic pressure  varies with time.
   37   38   39   40   41   42   43   44   45   46   47