Page 290 - Computational Fluid Dynamics for Engineers
P. 290

280                                                      9.  Grid  Generation


         such that  the  values  of the  inner  boundary  control  functions  become  vanishingly
         small  at  the  outer  boundary,  and  the  outer  boundary  control  functions  become
         vanishingly  small  at  the  inner  boundary.
            Thus,  we  have:
                                     ^,o)= (0
                                             P
                                     P&r knax)=r(Z)
                                                                            9.5.7)

                                     <5(£,%iax)  =  *(0
         These components  of the control functions  are derived  as follows. Two  conditions
         are  imposed  at  the  boundaries,  i.e.  orthogonality  and  specified  grid  spacing.
         Since conditions  are imposed  at  the  r]  =  0 and  rj =  rjmax boundaries, the  spacing
         is along the  r\ coordinate.  If  S^  is the  desired  spacing  along  this  coordinate,  the
         condition  can  be  written  as:

                                        =  xl  +  y% =  S%                  (9.5.8)
                                   r v.r 7l
         The  orthogonality  condition  is expressed  as  follows:

                                       =  x ^  +     =  0                   (9.5.9)
                                  ^,f v         y^y v
         On  the  rj =  0  and  rj — rimax  boundaries,  all  the  derivatives  with  respect  to  £
         are  known,  since  they  involve  only  boundary  points.  Hence,  in  the  equations
         (9.5.8)  and  (9.5.9)  above, the  x^  and  y^  terms  are  known,  and  we thus  have  two
        equations  in  two unknowns,  yielding the  following  expressions  for  the  x^  and  y^
        terms:
                                    = S riy i/(xl  +  y ^  2             (9.5.10a)
                                 x v
                                                     1 2
                                  Vr,  =  S r}x i/{x\  +  yl) l           (9.5.10b)
         and
                                         =  Xrji +  y^j                   (9.5.10c)
                                      r v
        Thus,  r v  is known  on the  boundary.  Furthermore,  it  can  be  shown that  if  equa-
        tions  (9.5.5)  and  (9.5.9)  are  combined  with  equation  (9.5.4),  the  latter  reduces
        to  the  following  on  the  boundary:

                  ( V ^ ) ( %  +  Pf^)  +  (f^f^ifrfrj  +  Q ^ ) ! boundary  =  0  (9.5.11)

         Multiplying  the  above  equation  by  f^  or  by  r^  and  again  using  the  condition  of
        orthogonality,  yields  the  following  two  equations  for  the  control  functions:

               ^boundary  =  - ( ^ . r ^ ) / ( f f  .7^)  -  ( ^ . f ^ ) / ^ . ^ )  [boundary  (9.5.12a)

              Qlboundary  =  - ( V ^ / ^ . T ^ )  -  ( r ^ . f ^ ) / ^ . ^ )  [boundary  (9.5.12b)
        By  evaluating  the  above  equations  on  the  boundaries,  the  p,  q,  r,  and  s  terms
        described  in  equations  (5.5.7)  can  be  computed.  All  quantities  in  equations
   285   286   287   288   289   290   291   292   293   294   295