Page 187 - Basic Structured Grid Generation
P. 187

176  Basic Structured Grid Generation

                                                                                           1
                          A harmonic map will take M onto a unit square N in the (ξ, η) plane, with ξ = ξ,
                          2
                        ξ = η. Straight lines ξ = const., η = const.in N are what results from mapping ξ,
                        η curvilinear co-ordinate curves in M. The ‘induced’ covariant metric components are
                        given by
                                                                2
                                                         2
                                                g 11 = (x ξ ) + (y ξ ) + (z ξ ) 2
                                                g 12 = x ξ x η + y ξ y η + z ξ z η
                                                         2
                                                                       2
                                                                2
                                                g 22 = (x η ) + (y η ) + (z η ) .
                          However, since z = f (x, y), we have, using Chain Rules,
                                           z ξ = f x x ξ + f y y ξ and z η = f x x η + f y y η ,  (6.91)

                        giving
                                               2
                                                                           2
                                                    2
                                  g 11 =[1 + (f x ) ](x ξ ) + 2f x f y x ξ y ξ +[1 + (f y ) ](y ξ ) 2
                                               2                                 2
                                  g 12 =[1 + (f x ) ]x ξ x η + f x f y (x ξ y η + x η y ξ ) +[1 + (f y ) ]y ξ y η
                                               2    2                      2    2
                                  g 22 =[1 + (f x ) ](x η ) + 2f x f y x η y η +[1 + (f y ) ](y η ) .  (6.92)
                        Exercise 7. Verify that

                                                                2
                                                                                   2
                                                         2
                                                                             2
                                    g = det(g ij ) =[1 + (f x ) + (f y ) ](x ξ y η − x η y ξ ) = γJ ,  (6.93)
                                                                                         i
                                                                                i
                        where J is the Jacobian (x ξ y η − x η y ξ ) of the mapping from the ξ sto the x s.
                          We now have
                                         g 11  = g 22 /g,  g 12  =−g 12 /g,  g 22  = g 11 /g.
                          Substituting into eqns (6.87) gives

                                     L(x) = g 22 x ξξ − 2g 12 x ξη + g 22 x ηη

                                              g    ∂   1 + (f y ) 2  ∂  f x f y
                                            −√            √      −      √      = 0,
                                               γ   ∂x      γ       ∂y     γ

                                     L(y) = g 22 y ξξ − 2g 12 y ξη + g 22 y ηη
                                              g      ∂     f x f y     ∂    1 + (f x ) 2
                                            −√    −      √     +        √        = 0       (6.94)
                                               γ    ∂x     γ     ∂y       γ
                        with γ , g 11 , g 12 , g 22 ,and g given by eqns (6.89), (6.92), and (6.93).
                          To solve these equations numerically, the second-order accurate finite difference
                        approximations (4.3) for first derivatives and (4.4), (4.5), and (4.6) for second deriva-
                        tives may be used, taking uniform differences h, k in ξ and η, respectively, in the
                        computational domain N. To calculate f x and f y we need to invert eqns (6.91), so that
                                            1                       1
                                       f x =  (z ξ y η − z η y ξ ) and f y =  (z η x ξ − z ξ x η ).  (6.95)
                                            J                       J
   182   183   184   185   186   187   188   189   190   191   192