Page 204 - Introduction to Computational Fluid Dynamics
P. 204

P1: IWV
                                                                                   May 25, 2005
                           CB908/Date
            0521853265c06
                                        0 521 85326 5
                        6.3 UNSTRUCTURED MESHES
                           This discretised equation, however, is of little use because the values of variables 11:10 183
                        at fictitious points P 2 and E 2 and at vertices a and b are not known. We must therefore
                        relate values at these fictitious points to the values at nodes P and E. This matter is
                        developed in the next subsection.
                        6.3.7 Interpolation of Φ at P 2 ,E 2 , a, and b

                        If it is assumed that the   variation between P and P 2 is linear then to first-order
                        accuracy

                                                                       ·∇   P ,            (6.77)
                                             P 2  =   P +    P =   P + l PP 2
                        where
                                                                       − x 2,P ),          (6.78)
                                           l PP 2  = i (x 1,P 2  − x 1,P ) + j (x 2,P 2
                        and

                                                         ∂         ∂
                                                ∇   P = i       + j       .                (6.79)

                                                         ∂x 1 P    ∂x 2 P
                        Taking the dot product in Equation 6.77 therefore gives

                                                       2

                                                 P =     (x i,P 2  − x i,P )  ∂      ,     (6.80)
                                                                    ∂x i
                                                      i=1               P
                                  − x i,P mustbeevaluatedintermsofpointswhosecoordinatesareknown.
                        where x i,P 2
                        Thus
                                                            − x i,c + x i,c − x i,P .      (6.81)
                                            x i,P 2  − x i,P = x i,P 2
                        However, from the construction shown in Figure 6.11,
                                                                  − x i,e .                (6.82)
                                                 x i,P 2  − x i,c = x i,P 1
                        Therefore, Equation 6.81 is further reformulated as
                                                      − x i,e + x i,e − x i,P ] + x i,c − x i,e .  (6.83)
                                     x i,P 2  − x i,P = [x i,P 1
                        Now, the equation to the face-normal passing through e is given by

                                                                           1


                                                                    )    i β + j β 2
                                         i (x 1,e − x 1,P 1  ) + j (x 2,e − x 2,P 1  1  1
                                       n =                            =            ,       (6.84)
                                                      l P 1 e                A c
                        therefore
                                                                l P 1 e  i
                                                       − x i,e =−   β                      (6.85)
                                                                     1
                                                  x i,P 1
                                                                 A c
                           and Equation 6.83 can be written as
                                                      − x i,P = l xi + d xi ,              (6.86)
                                                  x i,P 2
   199   200   201   202   203   204   205   206   207   208   209