Page 149 - Basic Structured Grid Generation
P. 149

138  Basic Structured Grid Generation

                        where the effect will be to double the spacing between grid points in one part of the
                        physical domain.

                        Exercise 5. With ϕ(ξ) given by eqn (5.87), show by integrating eqn (5.82) in the two
                        intervals 0   ξ< ξ 0 and ξ 0 <ξ   1 and eliminating K that

                                                    (b − a)
                                              
                                               a +        ξ,         0   ξ< ξ 0
                                              
                                                    (2 − ξ 0 )
                                              
                                       x(ξ) =                                    .         (5.88)
                                               2a − bξ 0    (b − a)
                                                         + 2ξ      ,ξ 0 <ξ   1
                                              
                                              
                                                  2 − ξ 0    2 − ξ 0
                          In this example the grid spacing in the physical plane changes at x = x 0 ,where
                        x 0 = a + (b − a)ξ 0 /(2 − ξ 0 ).Thisgives
                                                          2(x 0 − a)
                                                     ξ 0 =          .
                                                         b + x 0 − 2a
                          For more complicated weight functions, it may be difficult to ensure that changes in
                        grid spacing in the physical plane occur precisely where they are wanted. It may be
                        more convenient to use weight functions which are functions of physical co-ordinates,
                        so that in the one-dimensional case we have ϕ(x), with


                                                           x i+1 + x i
                                           x i+1 − x i = Kϕ         (ξ i+1 − ξ i )         (5.89)
                                                              2
                        instead of eqn (5.81), with the limiting forms

                                                            dx
                                                               = Kϕ(x),                    (5.90)
                                                            dξ
                                                  d     1 dx
                                                               = 0,                        (5.91)
                                                  dξ  ϕ(x) dξ
                        in place of eqns (5.82) and (5.83), giving

                                                    2
                                                   d x    1 dϕ     dx    2
                                                       =             .                     (5.92)
                                                   dξ 2   ϕ dx  dξ
                          This is a non-linear second-order differential equation (with the same end-conditions
                        x(0) = a, x(1) = b), with no guaranteed solution in general. Integration of eqn (5.90)
                        yields
                                                            b  1
                                                     K =         dx.                       (5.93)
                                                          a ϕ(x)
                          Equation (5.92) is equivalent to eqn (5.80) if we express P as a function of x and put
                                                              1 dϕ
                                                    P(x) =−         .                      (5.94)
                                                                2
                                                             Kϕ dx
   144   145   146   147   148   149   150   151   152   153   154