Page 104 - Introduction to Computational Fluid Dynamics
P. 104

P1: IWV
                                        0 521 85326 5
                           CB908/Date
            0521853265c04
                                                                                   May 25, 2005
                        4.6 BOUNDARY CONDITIONS
                        with the notion that the variation in   in the transverse direction asymptotically 11:7  83
                        approaches a value   ∞ (say) there. Thus, the fictitious notion of a boundary layer
                        thickness is associated with
                                                        −   I
                                                               = A,                        (4.62)
                                                       ∞ −   I
                        where suffix I refers to the inner boundary (wall or symmetry) and A is typically
                        taken to be 0.99 by convention. Note, however, that this boundary layer thickness
                        willbedifferentfordifferentmeaningsof andthemagnitudeofthicknesstypically
                                                    2
                        depends on the Prandtl number Pr   defined as
                                                               ν
                                                       Pr   ≡    .                         (4.63)

                        The Prandtl number is a property of the fluid. In fact, in Table 4.1, we may replace
                                                         .
                        k/C p by µ/Pr T and ρ D k by µ/Pr ω k
                           There is one further notion associated with the free stream. If we assume the
                        E boundary to be the free boundary (see Figure 4.1), the flow region above the
                        boundary can be taken to be a region in which there is no transverse convection or
                        diffusion and
                                                        b =   ∞ (x),                       (4.64)

                        where   ∞ (x) is specified. However, the physical location where this boundary
                        condition is to be applied is not a priori known because of the asymptotic nature
                        of variation of   in the vicinity of this boundary. To circumvent this problem,
                        Patankar and Spalding [50] relied on estimating the entrainment rate (− ˙ m E ) into
                        the boundary layer that occurs from the fluid above the E-boundary.
                           Thus, as previously mentioned, since there is no net flux of   in the transverse
                        direction, from Equation 4.17, it follows that
                                                                ∂    ∂

                                                       ∂
                                             (a + b ω) E     =      c      .               (4.65)
                                                       ∂ω      ∂ω     ∂ω
                                                           E              E
                           However, at the E boundary, ω = 1. Therefore,
                                                                    −1
                                             (a + b ω) E = a + b =−ψ EI  ∂ψ E  .           (4.66)
                                                                        ∂x
                        Thus, Equation 4.65 can be written as
                                                      −1
                                  ∂ψ E          ∂        ∂    ∂             ∂     ∂
                                       =−ψ EI                c     =−ψ EI       c
                                   ∂x            ∂ω     ∂ω     ∂ω          ∂      ∂ω
                                            ∂        ∂
                                       =−       r 
       =−r E ˙ m E .                    (4.67)
                                           ∂         ∂y
                        2  The term Prandtl number applies to variables T and h. When   = ω k , the appropriate dimensionless
                          number is called the Schmidt number (Sc). For velocity variables, of course, Pr   = 1. We thus use
                          Pr   generically to cover all  s.
   99   100   101   102   103   104   105   106   107   108   109