Page 309 - Computational Fluid Dynamics for Engineers
P. 309

10.3  Shock  Capturing                                                299



           1.00
                                       Rankine-Hugoniot
                                       Isentropic
                                       Isentropic-TSD M 0.6
                                       Isentropic-TSD M 0.8
                                       isentropic-TSD M  1.2
















         Fig.  10.2.  Shock jump  relations  for  the  TSD,  full-potential  and  Euler  equations.


         Small-Disturbance  relation  is  losing  accuracy  as  the  freestream  Mach  number
         gets  away  from  unity,  and  for  near  sonic-flow  conditions,  all  relations  are  in
         agreement  with  the  correct  Rankine-Hugoniot  jump.  For  Mach  numbers  above
         1.3, the  TSD  and  full-potential  jumps  are  in disagreement  with  the  Euler  shock
        jump.  This  stems  from  the  use  of  the  isentropic  flow  relation  (10.2.12)  across
         the  shock,  which  results  in  a  momentum  imbalance  across  the  shock  wave.



         10.3  Shock   Capturing

         A numerical method to compute the correct  shock jump  as part  of the solution  to
         the TSD, Full-Potential  or Euler equations  is now discussed.  Such algorithms  are
         called  Shock  Capturing  schemes.  The  model  equation,  Eq.  (5.1.1)  is  considered
         and  was  derived  from  the  linear  wave  equation,

                                     du       .du
                                                                          (10.3.1)

         Here  u  represents  a  variable  to  be  conserved  across the  computational  domain,
         including  any  discontinuities,  and  c is  a function  of  u.  Note that  the  linear  wave
         equation  is recovered  if  c is  a constant.  For  simplicity,  Eq.  (10.3.1)  is  discretized
         using  a  forward  time,  central  space  procedure:

                          u^ 1             - c -
                           jn+1  _ n   W + l         i+1  XU)             (10.3.2)
                                  t
         The  application  of  the  discretization  of  Eq.  (10.3.2)  to  the  one-dimensional
         domain  D  (Fig.  10.3)  leads  to
   304   305   306   307   308   309   310   311   312   313   314