Page 133 - Introduction to Computational Fluid Dynamics
P. 133

P1: IWV
                                        0 521 85326 5
                                                                                   May 20, 2005
                                                                                                12:28
            0521853265c05
                           CB908/Date
                     112
                            5.2.3 Pressure-Correction Equation   2D CONVECTION – CARTESIAN GRIDS
                            In the collocated-grid SIMPLE algorithm, the nodal velocities are determined using
                            Equations 5.12 written for   = u 1 and u 2 . The pressure gradients appearing in the
                            source terms of these equations are of course evaluated by central difference [for
                                                                          l
                                                       l
                                                  l
                            example, ∂p/∂x 1 | P = (p − p )/(2 x 1 ), where p is the guessed pressure field
                                                       W
                                                  E
                                                                                         l
                                                                                   l
                            and l is the iteration number]. The task now is to correct the u and p fields such
                                                                                   i
                            that mass conservation over the control volume surrounding node P is satisfied. To
                            do this, and to remain consistent with the SIMPLE-staggered grid, we imagine that
                                                                                              l
                            the momentum equations are also being solved for the cell-face velocities u . The
                                                                                              fi
                            discretised versions of these imagined equations with underrelaxation will appear as

                                            α             l+1      ∂p l+1    l            l
                                     l+1
                                    u   =             A k u  −  V        + D    + (1 − α)u , (5.20)
                                     f1                   f1,k               u 1           f1
                                           AP  u f1                 ∂x 1
                                                   k

                                            α             l+1      ∂p l+1    l             l
                                     l+1
                                   u    =             A k u  −  V        + D    + (1 − α)u , (5.21)
                                     f2                   f2,k               u 2           f2
                                           AP  u f2                 ∂x 2
                                                   k
                                            l
                                    l
                            where D and D contain source terms (if any) other than the pressure gradient,
                                    u 1     u 2
                            α is the underrelaxation factor, and the summation symbol indicates summation
                            over all immediate neighbours of the cell-face location under consideration. Thus,
                            when Equation 5.20 is written for cell face e, for example, running counter k refers
                            to locations ee, Ne, w, and Se. Now, at iteration level l + 1, it is expected that
                                      ∂(ρ l+1 )  1 ∂      l+1  l+1    1 ∂     l+1  l+1
                                             +         r ρ  u f1  +        r ρ   u f2  = 0.    (5.22)
                                        ∂t      r ∂x 1              r ∂x 2
                            Substituting Equations 5.20 and 5.21 in Equation 5.22 we can show that
                              ∂(ρ l+1 )  1 ∂      l+1  l     1 ∂     l+1  l
                                     +         r ρ   u f1  +       r ρ  u f2
                                ∂t      r ∂x 1             r ∂x 2

                                  1 ∂     r ρ l+1  α                  l+1      ∂p l+1
                                                          l
                                =                  AP  u f1  u −  A k u f1,k  +  V   − D l
                                                          f1
                                  r ∂x 1   AP  u f1                             ∂x 1     u 1
                                                               k

                                    1 ∂     r ρ l+1  α      l           l+1       ∂p l+1   l
                                  +                   AP  u f2  u −  A k u f2,k  +  V  − D       .
                                                            f2
                                    r ∂x 2   AP u f2                              ∂x 2     u 2
                                                                  k
                                                                                               (5.23)
                               To develop the pressure-correction equation, we introduce the following
                            substitutions:
                                                                                    l
                                                                l
                                             l

                                     u l+1  = u + u ,   u l+1  = u + u ,    p l+1  = p + p ,   (5.24)


                                      f1     f1   f1      f2    f2    f2                 m
   128   129   130   131   132   133   134   135   136   137   138