Page 188 - Introduction to Computational Fluid Dynamics
P. 188

P1: IWV
                                        0 521 85326 5
                           CB908/Date
            0521853265c06
                        6.2 CURVILINEAR GRIDS
                           Similarly, it can be shown that                         May 25, 2005  11:10 167
                                                      %
                                                        
  1    2  
  2    2
                                                dA 2 =   β    + β    dξ 1 .                (6.26)
                                                          2       2
                           Comparison of the last two equations with Equations 6.21 shows that dA 1 and
                        dA 2 represent areas with dξ 1 = dξ 2 = 1.
                        Elemental Volume
                        The volume element in curvilinear coordinates is given by
                                                   r
                                                         r
                                                               r
                                                  ∂      ∂    ∂
                                           dV =      ·     ×       dξ i dξ j dξ k .        (6.27)
                                                 ∂ξ i   ∂ξ j  ∂ξ k
                        Thus, taking i = 1, j = 2, and k = 3, it follows that
                                             r
                                                  r
                                            ∂    ∂            
  1  2   2  1
                                     dV =      ·    dξ 1 dξ 2 = β β − β β 2  dξ 1 dξ 2 .   (6.28)
                                                                  2
                                                                1
                                                                       1
                                           ∂ξ 1  ∂ξ 2
                           Comparison of Equation 6.28 with Equation 6.14 shows that the Jacobian J is
                        nothing but element volume dV with dξ 1 = dξ 2 = 1.
                        The Normal Fluxes
                        Note that Equation 6.20 can be written in the following form:

                                           ∂(ρ )     ∂           ∂
                                        rJ        +     [rq ξ 1 ] +  [rq ξ 2 ] = rJ S,     (6.29)
                                             ∂t     ∂ξ 1        ∂ξ 2
                                        are given by
                        where q ξ 1  and q ξ 2

                                                       
 eff   2  ∂        ∂
                                           = ρ U f1   −     dA                  ,          (6.30)
                                        q ξ 1                  1    + dA 12
                                                        J       ∂ξ 1       ∂ξ 2

                                                       
 eff   2  ∂        ∂
                                           = ρ U f2   −     dA                  .          (6.31)
                                        q ξ 2                  2    + dA 12
                                                        J       ∂ξ 2       ∂ξ 1
                        With reference to Figure 6.4, these expressions represent total (convective + dif-
                        fusive) transport of   normal to the two curvilinear directions, respectively. The
                        convective transport ρ U fi   is thus directed normal to the constant-ξ i lines. In other
                                                                            i
                        words, U fi is directed along the contravariant base vector  a . Note that, in general,
                        lines of constant ξ 1 and ξ 2 do not intersect orthogonally. Thus, the total normal
                                                                                             2
                        diffusive contribution is made up of two components. The first, containing dA ,is
                                                                                             i
                                                                                    a
                        due to the property gradient along the covariant base vector direction   i , the second,
                        containing dA 12 , is due to the property gradient along the direction ξ j , j  = i.If
                        the intersection of coordinate lines were to be orthogonal, dA 12 = 0. Also, from
                        Equations 6.13, it is clear that dA 12 can be both positive as well as negative.
   183   184   185   186   187   188   189   190   191   192   193