Page 369 - Computational Fluid Dynamics for Engineers
P. 369

360                                  12.  Compressible  Navier-Stokes  Equations


            As  in  Section  5.4,  where  central  differences  were  used  for  the  spatial  deriva-
         tives  in  Eq.  (5.4.12),  central  differences  are  again  used  in  Eqs.  (12.4.17)  and
         (12.4.18),  and  the  resulting  system  is  written  in  block  tridiagonal  form  and
         solved  with  the  block  elimination  method  described  in  Eqs.  (4.4.32)  through
         (4.4.34).
            As  discussed  in  Section  5.7,  in  some  implicit  methods  for  hyperbolic  equa-
         tions, damping terms are necessary to suppress high  frequency  oscillations  in  the
         solution.  This  can  be  accomplished  by  adding  constant  coefficient  implicit  and
         explicit  dissipation  terms  to the  solution  algorithm  and,  in the  Beam-Warming
         method,  this  is accomplished  by adding an explicit  fourth-order  dissipation  term
         to the  right-hand  side  of Eq.  (12.4.17).  If the  steady-state  is  of interest,  implicit
         second  order  dissipation  terms  are added  to the  left-hand  sides  of Eqs.  (12.4.17)
         and  (12.4.18)  and  the  resulting  equations  become

         Step  1:

                        d_
            m  + I  9 At   dx   [A]   [P] ~  [A*  Redx 2   [R] n  -  SiS x[I]  AQ  n - l / 2
                                                                2
                                     Re
                   +
                    Z
              At              Vi  +  V 2^ n
                   dx           Re         dy           Re
                6At  1                          l
                         l AV ^)        l (AWr )
             +  l  +  £Re  {  2     +    y
                       n
                    (AQ -')-e e(6Z  +  6*)Q n
               1  +  C
                                                                         (12.4.19)
         Step  2:

                   9 At   d_                                      2
                              [B]                       2  M    eid y[I)
                         dy            Re          Redy



         Here  8 2  and  6 4  denote  central  difference  operators  defined  by
                                6%u =  Ui+ij  -  2uij  +  Ui-ij

         and
                   6 xu  =  Ui+2,  Au.  «+ij   +  6u 1,3   4u  i - i j  +  u •i-2,3   (12.4.22)

            On mesh points adjacent  to the boundaries, Eq.  (12.4.22)  is approximated  by
        second-order  four  point  finite  differences  rather  than  five  point  finite  differences
        to  avoid  requesting  information  from  outside  the  computational  domain.  For
        example,  near  the  bottom  boundary,  Eq.  (12.4.22)  becomes

                              6yU «  ^,4 -  4^,3  +  5ui t2  -  2i^ 5i   (12.4.23)
   364   365   366   367   368   369   370   371   372   373   374