Page 257 - Computational Fluid Dynamics for Engineers
P. 257

8.2  Solution  of the  Orr-Sommerfeld  Equation                       247



         8.2.1  Numerical  Formulation

         To formulate  the  numerical  scheme  employing the  box method,  we consider  Eq.
         (2.5.13)  and  its  boundary  conditions  given  by  (8.1.1)  and  reduce  them  to  an
         equivalent  first-order  system.  We  define

                                          4>' = f                         (8.2  !.la)
                                                2,
                                       f  = * + &                         (8.2  Mb)
                                          J  = g                           (8.2  i.lc)
         and  write  Eq.  (2.5.13)  as
                                      g'  =  &  - &<f>                    (8.2 .Id)

         With  these  variables,  the  boundary  conditions  become
                                  y  =  0,  0  =  0,  /  =  0             (8.2  !.2a)

                                                 6
                         y = 6;  s +  (£i +  &)/ + (6  + 6)0 = 0;
                                        g +  &s  =  0                     (8.2.2b)
        where  £2, defined  in  Eq.  (P2.7.6b)  attains  its  value  at  y  — 6.
           We  now consider  a nonuniform  mesh  with  y  =  0 represented  by  yo and  y  =  6
        by  yj  and  approximate  the  quantities  (/,  s,g,</))  at  points  yj  by  (fj,Sj,gj,<f>j).
        As  in  subsection  4.4.3,  we  write  the  finite-difference  approximations  of  Eqs.
         (8.2.1)  for  the  midpoint  y-\  using  centered-difference  derivatives  to  obtain
                                J  2
                           </>j  -  <t>j-i  -  c 3(fj  +  fj-i)  =  {n)j  =  0  (8.2.3a)

                             s
                                 s
                                                          r
                                         c
                          c
               fj  ~  fj-i  - 3( j  + j-i)  ~ i(0?  +  <l>j-i)  =  ( 3)j-i  =  0  (8.2.3b)
                            s   s                     r
                            j  ~ J-i  -  °s(9j  +  9j-i)  =  ( 2)j  =  0  (8.2.3c)
               9j  ~  9j-i  -  C4(sj  +  Sj_i)  -  c 2 (0j  +  0j-i)  =  (u)j-i  =  0  (8.2.3d)
        Here,  with  hj-i  denoting  yj  —  yj-i,
              C3 =  - ^ ,  Ci  =  ^ C 3 ,  C 2  =  - ( 6 ) , _ I C 3 ,  C 4  =  (ff , - ! < *  (8.2.4)
                                                                )
        As  in  the  procedure  for  solving  the  boundary-layer  equations,  Eqs.  (8.2.3)  are
         again  written  in  a  sequence  which  ensures  that  the  A-matrix  in  Eq.  (4.4.30)  is
        not  singular.
                                                2
           Equations  (8.2.3)  are imposed  for  j  =  1, , . . . ,  J  and  additional conditions  at
        . 7 = 0  and  j  =  J  are  obtained  from  the  appropriate  boundary  conditions  which,
        for  an  external  flow,  follow  from  Eqs.  (8.2.2)  and  can  be  written  as

                              00  =  (ri)o  =  0,  /o  =  (r 2 )o  =  0   (8.2.5a)
                fj  +  c 3sj  +  ci0j  =  (r 3)j  =  0,  gj  +  c 4sj  =  (r 4 ) j  =  0  (8.2.5b)
   252   253   254   255   256   257   258   259   260   261   262