Page 85 - Essentials of applied mathematics for scientists and engineers
P. 85

book   Mobk070    March 22, 2007  11:7








                                                                                                     75




                                                 CHAPTER 5


                             Solutions Using Fourier Series


                                               and Integrals





                   We have already demonstrated solution of partial differential equations for some simple cases
                   in rectangular Cartesian coordinates in Chapter 2. We now consider some slightly more
                   complicated problems as well as solutions in spherical and cylindrical coordinate systems to
                   further demonstrate the Fourier method of separation of variables.

                   5.1    CONDUCTION (OR DIFFUSION) PROBLEMS
                   Example 5.1 (Double Fourier series in conduction). We now consider transient heat con-
                   duction in two dimensions. The problem is stated as follows:

                                            u t = α(u xx + u yy )

                                      u(t, 0, y) = u(t, a, y) = u(t, x, 0) = u(t, x, b) = u 0
                                     u(0, x, y) = f (x, y)                                       (5.1)

                   That is, the sides of a rectangular area with initial temperature f (x, y) are kept at a constant
                   temperature u 0 . We first attempt to scale and nondimensionalize the equation and boundary
                   conditions. Note that there are two length scales, a and b. We can choose either, but there will
                   remain an extra parameter, either a/b or b/a in the equation. If we take ξ = x/a and η = y/b
                   then (5.1) can be written as
                                                  a 2           a 2
                                                    u t = u ξξ +  u ηη                           (5.2)
                                                  α             b 2
                                                2
                                                                                      2
                   The time scale is now chosen as a /α and the dimensionless time is τ = αt/a . We also choose
                   a new dependent variable U(τ, ξ, η) = (u − u 0 )/( f max − u 0 ). The now nondimensionalized
                   system is
                                                       2
                                           U τ = U ξξ + r U ηη                                   (5.3)
                                    U(τ, 0,η) = U(τ, 1,η) = U(τ, ξ, 0) = U(τ, ξ, 1) = 0
                                    U(0,ξ,η) = ( f − u 0 )/( f max − u 0 ) = g(ξ, η)
   80   81   82   83   84   85   86   87   88   89   90