Page 49 - Introduction to Computational Fluid Dynamics
P. 49

P1: IWV/ICD
                                        0 521 85326 5
                                                                                                10:49
                                                                                   May 25, 2005
            0521853265c02
                           CB908/Date
                     28
                            Table 2.2: Explicit procedure with ∆t = 20 s (unstable).  1D HEAT CONDUCTION
                            Time   0 mm   1 mm   3 mm    5 mm    7 mm    9 mm   10 mm
                              0    250     30     30      30      30      30    250
                             20    250    241.5   30      30      30     241.5  250
                             40    250    148.0  131.7    30     131.7   148.0  250
                             60    250    238.3   90.63  127.8    90.63  238.3  250
                             80    250    178.6  179.5    92.05  179.5   178.6  250
                            100    250    247.7  137.0   176.1   137.0   247.7  250



                            be pressed together can be determined by interpolation as (t − 90)/(100 − 90) =
                            (140 − 130.33)/(141.31 − 130.33) or at t = 98.8 s. This calculated time, of course,
                            need not be considered accurate. Its accuracy can be ensured by repeating calcu-
                            lations with increasingly smaller  x (increasingly greater number of nodes) and
                            by taking ever smaller values of  t. Further, note that the temperature distribu-
                            tions at any time t are symmetric about x = 5 mm. This is because of the sym-
                            metry of the boundary and the initial condition. Now, unlike in Table 2.1, the
                            results presented in Table 2.2 show zigzag or nonmonotonic evolution of temper-
                            ature. For example, at any x, the temperature first rises (as expected) and then
                            falls (against expectation). In fact, the reader is advised to carry the computations
                            well beyond 100 s or with larger values of  t. Then, it will be found that the
                            evolved temperatures will show even more unexpected trends. That is, the interior
                                                                             ◦
                                                                  ◦
                            temperatures will exceed the bounds of 30 C and 250 C. Clearly, this is in vio-
                            lation of the second law of thermodynamics. Results of Table 2.2 are, therefore,
                            unacceptable.


                            2.5.2 Partially Implicit Procedure 0 <ψ < 1

                                                                                     o
                            In this case, if the condition of positiveness of the coefficient of T is invoked then
                                                                                     P
                             t must obey the following constraint:
                                                             ρ V i C
                                                                    o
                                                 t <                i          .               (2.36)
                                                       (1 − ψ)(AE i + AW i )
                                                                            min
                            However, computations of the previous problem will show that stable (monotoni-
                            cally evolving) solutions can be obtained even with

                                                           o
                                                    ρ V i C
                                       t <                 i               for    ψ< 0.5,      (2.37)
                                             (1 − 2ψ)(AE i + AW i )
                                                                   min
                            and, for ψ ≥ 0.5,  t can be chosen without any restriction. Clearly, therefore,
                            condition (2.36), though valid, is too restrictive on the time step. The reader
                            will appreciate this matter by solving Exercise 29. The more rigorous proof
   44   45   46   47   48   49   50   51   52   53   54