Page 175 - Computational Fluid Dynamics for Engineers
P. 175

162                        5.  Numerical  Methods  for  Model  Hyperbolic  Equations



         ej+ij  denotes  an average  value  of e for cell  i +  1, ,  etc.  Also  Eq.  (5.6.2) is
                                                         j
         treated  as  an  equation  for  solving  Uij  as  the  unknown.
            Equation  (5.6.2)  can  be applied  to the  continuity,  momentum  and  energy
         equations.  Since  for  the  continuity  equation,  Q s — 0,  Eq.  (5.6.2)  becomes
                d
                 Qij  ,  (gM)»+ij ~ (gtt)t-ij  ,  (ev)ij+i  ~ (g^kj-i  _  n  fKK<i*\
                                                                            - -^
                ~df  +         2k         +         2h          "  U      (5 b  j
         since  U = Q, e = QU and  /  =  gv.  Similarly  the  ^-component  of  the  momentum
         equation  can  be  written as
                      2
                                  2
          d,   x  ,  (su ) i+ij  -  (eu )j-ij  ,  (guv)jj +1  -  (euv^j-i  _  pj-ij  -pj+ij
         dt [0U)ij+          2k          +           2h          ~      2k
                                                                           (5.6.14)
         and  the  energy  equation as

                 9  , „ .  ,  (Etu)i+ij  -  (Etu)i- ltj  (Etv)ij+i  -  (E tv)ij-i
                  {Et)ij  +                      +
                 di                  2k                      2h
                            _  (Ptt)j-lj  -  {pu)j+lj  ,  (pv)i,j~l -  (pv)i,j+l  / C R I C X
                            ~         2k         +          2h            ( 5 b 1 5 j
                                                                            - '
            The  solution  of Eqs.  (5.6.12)  to  (5.6.14)  together  with  the  ^-component
         equation  can  be  obtained  conveniently  with  an  explicit  time-marching  solution
         based,  for example,  on a  standard  Runge-Kutta  numerical  integration  with
         respect  to  time.  As  in  time-marching  solution,  the  flowfield  is assumed  to  be at
                                                                                n
              n
         time  t . In Eqs.  (5.6.12) to (5.6.14),  the  right-hand  sides  are  evaluated  at t ,
        thus  they  are  known.  These  equations  are  of the  form
                                        d
                                         ^)"    = C                      ( 5 A M )

        with  C denoting  a constant  which  is different  for  each  conservation  equation.
         This equation  represents a system  of first-order  ordinary  differential  equations  in
        t  that  can  be integrated  conveniently  with a fourth-order  Runge-Kutta  method
         described  in Chapter  12 to calculate  the  values  of Uij  at the  next  time  step,
         n+l
         t   as  shall  be  discussed  in  some  detail in Chapter  12.

         Example  5.5.  Use the  finite  volume method  with  (a)  central  and  (b)  upwind  differencing
        to  solve the  one-dimensional  steady  convection  and  diffusion  equation
                                     « {u4>)= *( f)        0 <X<L
                                            v
                                    dx       ax \  ax  J                   (E5.5.1)
                                     0(0) =  1  «A(L) = 0

        where u =  2.5  and  v = 0.1. Compare  your  results  with  the  exact  solution  </>(#)
                                 4>(x)  — 4>(0)  exp(ux/u)  — 1            (E5.5.2)
                                 4>{L) -  <f>(0)  =  exp(uL/z/) - 1
   170   171   172   173   174   175   176   177   178   179   180