08 solution heat equation explicit

Upload: pondok-huda

Post on 08-Apr-2018

216 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/6/2019 08 Solution Heat Equation Explicit

    1/3

    Heat EquationSolution sheet6. week

    8 Heat Equation - Explicit

    Aim of this exercise is the numerical solution of the heat equation (seeexercise 7) in explicit form.

    a) Simplify the heat equation from exercise 7 to the one dimensional case.

    Solution:t k

    2x = 0 k > 0

    b) Discretize the time derivative operator as usual (Euler-method).

    Solution:

    (x)n+1 = (x)n + t k 2x(x)

    c) Discretize the spatial derivative in explicit form. The second derivativeis discretized in two steps:

    First discretize the outer derivative as the divergence of theflux. The resulting scalar quantity is therefore stored at the cellcenter.

    Secondly discretize the inner spatial derivative x at cell i.The result is assumed to be the above mentioned flux (in 3D thegradient is indeed a vector quantity) and should therefore bestored at the cell interfaces (staggered mesh).

    Solution:

    The outer derivative of F(x) = x yields to

    (xF(x))i Fi+1/2 Fi1/2

    x

    The inner derivative yields to

    Fi1/2 = (x)i1/2 i i1

    x

    1

  • 8/6/2019 08 Solution Heat Equation Explicit

    2/3

    Combining these

    2x(x) = x(x)

    Fi+1/2 Fi1/2

    x

    (i+1 i) (i i1)

    x2

    =i+1 2i + i1

    x2

    The discretized heat equation in explicit form is

    n+1i = ni + t kn

    i+1 2n

    i+ n

    i1x2

    d) Every explicit method yields to an upper limit of the timestep tmax.Derive this upper limit analytically (and test your result later on).

    Solution:

    The CFL condition leads to tmax = minxi|vi|

    . In this particular

    case the grid spacing x is constant in time and space and the velocityis given by the diffusion velocity:

    F = vdiff = k

    vdiff = kx(x)

    (x)

    The resulting upper limit for the timestep is

    tmax = min

    (x)x

    k|x(x)|

    e) Solve the explicit form of the heat equation numerically:

    Use 100 grid points, x = 1.

    Initial condition (for the density distribution) should be a simpleparabolic function (ax2 + b) with (x = xmin, t = 0) = 1 and

    (x = xmax, t = 0) = 0.

    The boundary of the domain is assumed to be an empty reservoirwith infinite capacity (x = ghost, t) = 0.

    2

  • 8/6/2019 08 Solution Heat Equation Explicit

    3/3

    Determine the upper limit of the timestep for this specific problem.

    Solution:

    With x(x) (x)x the upper limit of the timestep for this specific

    problem is given by

    tmax = min

    (x)x2

    k|(x)|

    The highest jump in density is given at the left border. The density atthis innermost interface is 1/2 = 0.5(ghost + 1) = 0.5 and the jumpin density 1/2 = 1 ghost = 1. Therefore this scheme is unstable

    for t > 0.5x2

    k .Download the corresponding Mathematica file.

    3