Page 100 - Introduction to Computational Fluid Dynamics
P. 100

P1: IWV
                           CB908/Date
                                        0 521 85326 5
            0521853265c04
                                                                                   May 25, 2005
                        4.5 DETERMINATION OF ω, y, AND r
                        for j = 2, 3,..., JN − 1. Note that superscript d is now dropped for    11:7  79
                        convenience.
                        4.5 Determination of ω, y, and r
                        Equation 4.43 represents a set of algebraic equations at a streamwise location x d .
                                                                            u
                        These equations can be solved by TDMA when values of   at x u are known along
                                                                            j
                        with the two boundary conditions at x d (i.e., at j = 1 and j = JN ). Thus, starting
                        with x = x 0 (say), one can execute a marching procedure taking step  x. This
                        situation is very much like the unsteady conduction problem in which the marching
                        procedure is executed with time step  t.
                           Thus, at x = x 0 , the u j ∼ y j relationship is assumed to have been prescribed
                        either from experimental data or from an analytical solution. One can use this
                        prescription to set ω j once and for all. Let

                                      ω j = ω P ,  ω c, j = ω s ,  ψ j = ψ P ,  ψ c, j = ψ s ,
                                       y j = y P , y c, j = y s , r j = r P ,  r c, j = r s ,  (4.44)


                        where, at x = x 0 , y j ( j = 1, 2,..., JN ) are known. Thus, one can set y c,1 = y c,2 =
                        y 1 where y 1 refers to the I boundary and y JN to the E boundary. Now, from the
                        geometry of Figure 4.1, it follows that r j and r c, j can be evaluated from the formula

                                                    r = r I + y cos (α),                   (4.45)

                        where α is function of x. This completes the grid specification at x = x 0 .
                           For evaluation of ω j , we first calulate ψ j . Thus, setting ψ 1 = ψ c,1 = ψ I (say),
                        where ψ I is arbitrarily chosen, one can use Equation 4.5 to set all other ψ j . The
                        relevant discretised equations are

                               ψ c, j = ψ c, j−1 + (ρ ru) j−1 (y c, j − y c, j−1 ),  j = 2, 3,..., JN, (4.46)



                         ψ j = ψ j−1 + 0.5 (ρ ru) j + (ρ ru) j−1 (y j − y j−1 ),  j = 2, 3,..., JN.
                                                                                           (4.47)

                           It is now a simple matter to evaluate ω j and ω c, j using definition (4.2). Thus,
                        ω j at y j represents the ratio of streamwise mass flow rate from y 1 = y I to y j to the
                        total mass flow rate from y I to y E at x = x 0 . It is now assumed that this ratio remains
                        intact at all values of x and thus the ω j distribution does not change throughout the
                        domain in the x direction.
                           Note, however, that the physical distance y (and therefore r) must go on changing
                        at different values of x as the boundary layer grows or shrinks. We thus seek the
                        y j ∼ ω j relationship applicable to every x.
   95   96   97   98   99   100   101   102   103   104   105