Page 152 - Basic Structured Grid Generation
P. 152

Differential models for grid generation  141

                           5.9 Surface-grid generation model


                        A natural extension of the differential model given by eqns (5.6) to grid generation on
                        a surface in three dimensions is obtained by replacing the Laplacian operator by the
                        Beltrami operator, as defined in Section 3.9. Thus if the surface is to be covered by a
                                                                                            2
                                                      α
                                                                                 1
                        curvilinear co-ordinate system of ξ curves, we could force ξ(= ξ ) and η(= ξ ) to
                        satisfy the equations
                                                    B ξ = P,    B η = Q,                  (5.104)
                        where P , Q are control functions. Note that the surface may already be effectively
                                                  α
                        covered by parametric curves u , α = 1, 2, but these may not give rise to a satisfactory
                        grid. We assume here that the surface has four edges (four known space-curve seg-
                                                                                   α
                        ments) which can be mapped onto the edges of a unit square in the ξ computational
                        plane.
                                                                               α
                          Substituting into eqn (3.165) with the ξ α  system instead of u immediately gives
                        the ‘inverse’ equation
                                                   2
                                                 ∂ r       ∂r     ∂r
                                             αβ
                                            a          + P    + Q    = 2κ m N,
                                                  α
                                                ∂ξ ∂ξ β    ∂ξ     ∂η
                        which after multiplying through by a may be written as

                                                         ∂r     ∂r
                                               Dr + a P     + Q     = RN,                 (5.105)
                                                         ∂ξ     ∂η
                        where the operator D is given by

                                                     ∂ 2       ∂ 2      ∂ 2
                                             D = a 22  2  − 2a 12  + a 11  2  ,           (5.106)
                                                    ∂ξ        ∂ξ∂η      ∂η
                        and
                                                    αβ
                                      R = 2aκ m = aa b αβ = a 22 b 11 − 2a 12 b 12 + a 11 b 22 .  (5.107)
                          Equation (5.105) comprises three scalar equations for the cartesian components
                        (x, y, z) of grid-points in physical space. These can be solved numerically when the
                        equation of the surface is given, either explicitly, implicitly, or in terms of parameters.
                        Specification of the two constraining eqns (5.104) provides the core of the method.
                          Here we shall just consider the case where P = Q = 0, and take the surface to be
                        defined by the explicit equation z = f (x, y). Then eqn (5.105) reduces to the three
                        equations

                                                          2
                                                                    2
                                                2
                                               ∂ x       ∂ x      ∂ x
                                            a 22  2  − 2a 12  + a 11  2  = RN x
                                               ∂ξ       ∂ξ∂η      ∂η
                                                          2
                                                                    2
                                                2
                                               ∂ y       ∂ y      ∂ y
                                            a 22  2  − 2a 12  + a 11  2  = RN y           (5.108)
                                               ∂ξ       ∂ξ∂η      ∂η
                                                2
                                                          2
                                                                    2
                                               ∂ z       ∂ z       ∂ z
                                            a 22  2  − 2a 12  + a 11  2  = RN z ,
                                               ∂ξ       ∂ξ∂η      ∂η
   147   148   149   150   151   152   153   154   155   156   157