Page 100 - Computational Fluid Dynamics for Engineers
P. 100

86                                                     3.  Turbulence  Models



            In Navier-Stokes calculations, Baldwin and Lomax replace the absolute  value
         of the  velocity  gradient  du/dy  in Eqs.  (3.2.1)  and  (3.2.11)  by the  absolute  value
         of the  vorticity |CJ|,
                                      \  \  \du   dv\
                                      M  =                                ( 3 2 1 3 a )
                                           | ^ ~ ^ |                       - '
         and  the  intermittency  factory  7  in  Eq.  (3.2.4b)  is written  as

                                               csy
                                 7    1 +  5.5                           (3.2.13b)
                                               2/max
         with  C3 =  0.3.  The  studies  conducted  by  Stock  and  Haase  [14]  clearly  demon-
         strate  that  the  modified  algebraic  eddy  viscosity  formulation  of  Baldwin  and
         Lomax  is not  a true  representation  of the  CS model  since their  incorporation  of
         the  length  scale  in the  outer  eddy  viscosity  formula  is not  appropriate  for  flows
         with  strong  pressure  gradients.
            Stock  and  Haase  proposed  a  length  scale  based  on  the  properties  of  the
         mean  velocity  profile  calculated  by  a  Navier-Stokes  method.  They  recommend
         computing  the  boundary-layer  thickness  S  from
                                      6  =  1.936y max                     (3.2.14)

         where  2/ max  is the  distance  from  the  wall  for  which  y\du/dy\  or  F  in Eq.  (3.2.11)
         has  its  maximum.  With  6  known,  u e  in  the  outer  eddy  viscosity  formula,  Eq.
         (3.2.3),  is the  u  at  y  =  <5, and  7  is computed  from
                                                      -1
                                  7    1  +  5.5                          (3.2.15)

         based  on  Klebanoff's  measurements  on  a  flat  plate  flow  and  not  from  Eq.
         (3.2.13b).  The  displacement  thickness  6*  for  attached  flows  is  computed  from
         its  definition,  Eq.  (3.2.4a),  and,  for  separated  flows  from
                                                  u
                                  6*  =  f  l l -  —  )dy                  (3.2.16)
                                       /
         either  integrating  the  velocity  profile  from  y  =  0,  or  y  =  y u=o  to  <5, or  using  the
         Coles  velocity  profile,  see  Eq.  (3.5.6).  The  results  obtained  with  this  modifi-
         cation  to  the  length  scale  in  the  outer  CS  eddy  viscosity  formula  improve  the
         predictions  of  the  CS  model  in  Navier-Stokes  methods  as  discussed  in  Stock
         and  Haase  [14].
            A  proposal  which  led  to  Eq.  (3.2.14)  was  also  made  by  Johnson  [15].  He
         recommended  that  the  boundary-layer  thickness  6 is calculated  from
                                        6  =  1.2y 1/2                     (3.2.17)
         where
                                                     0
                                                       5
                                Vl/2=y    a t  -B   = - "                 (3.2.18)
         For  a  discussion  regarding  other  length  scales,  the  reader  is  referred  to  [1,3].
   95   96   97   98   99   100   101   102   103   104   105