Page 60 - Principles of Applied Reservoir Simulation 2E
P. 60

Part I: Reservoir Engineering Primer  45


        vdC/dx.  When the diffusion  term is much larger than the convection term, the
        C-D equation behaves like the heat conduction  equation,  which is a parabolic
        partial differential  equation  (PDE). If  the diffusion  term is much smaller than
        the convection term, the C-D equation behaves like a first-order hyperbolic PDE.
             The C-D equation is especially valuable for studying numerical solutions
        of fluid flow equations because the C-D equation can be solved analytically and
        the  C-D  equation  may  be  used  to  examine two  important  classes  of  PDEs
        (parabolic and hyperbolic).  To solve the  C-D equation,  we must specify  two
        boundary conditions and an initial condition. The two boundary conditions are
        needed because the C-D equation is second-order  in the space derivative. The
        initial condition satisfies the need for a boundary condition  in time associated
        with the first-order derivative in time. The boundary conditions for the miscibie
        displacement process are that the initial concentration of displacing fluid is equal
        to one at the inlet (x = 0), and zero for all other values of x. The mathematical
        expressions  for these boundary conditions are concentration  C(0,  t)  =  I  at  the
        inlet, concentration  C(°°,  f)  = 0 at the edge of the  linear system  for all times t
        greater than the  initial time t = 0, and the  initial condition  C(x, 0)  = 0 for all
        values of x greater  than 0.
             The  propagation  of  the  miscibie  displacement  front  is  calculated  by
        solving the C-D equation. The analytical solution of the one-dimensional C-D
        equation is [Peaceman,  1977]


                                JC  -  Vt            X  +  Vt
              C(jc,  0  =  -U  erfc                                 (5.17)
                        2        2 ]/Dt
                          l
       where the complementary  error function  erfc(y)  is defined  as

                                        2   r  z 2
                         erfc(j)  =  1 -  —  je         rfz         (5,18)
                                       V*   o
       Abramowitz and Stegun [ 1972] have presented an accurate numerical  algorithm
       for  calculating the complementary  error function  erfcO/). A comparison  of the
       analytical  solution  of the  C-D equation  with numerical  solutions  is given in
       Fanchi [2000].
   55   56   57   58   59   60   61   62   63   64   65