sandbox/M1EMN/Exemples/viscous_collapse_noSV.c

    collapse of a rectangular viscous column in 2D

    As a simple case we propose the Huppert collapse, this is the explicit resolution of the mass equation for a very viscous flow (viscous stress=pressure gradient) \displaystyle \frac{\partial h}{\partial t}+ \frac{\partial Q(h,\partial_xh)}{\partial x}=0, \text{ with } Q = -\frac{h^3}{3} \frac{\partial h}{\partial x}

    Code

    mandatory declarations:

    #include "grid/cartesian1D.h"
    #include "run.h"

    definition of the height of interface its O(\Delta) derivative and its O(\Delta^2) derivative, time step

    scalar h[];
    scalar Q[];
    double dt;

    Main with definition of parameters

    int main() {
      L0 = 10.;
      X0 = -L0/2;
      N = 200;
      DT = (L0/N)*(L0/N)/10 ;
      run();
    }

    initial elevation: a “double square” of surface 2

    event init (t = 0) {
      foreach(){
        h[] = (fabs(x)<1) ;
        Q[]=0;
        }
      boundary ({h,Q});
      }

    print data in stdout

    event printdata (t += 10; t <= 500) {
      foreach()
        fprintf (stdout, "%g %g %g %g \n", x, h[], Q[], t);
      fprintf (stdout, "\n");
    }

    integration

    event integration (i++) {
      double dt = DT;

    finding the good next time step

      dt = dtnext (dt);

    the flux Q = -\frac{h^3}{3} \frac{\partial h}{\partial x}

      foreach() //_face()
        Q[] = -1./3 * (( h[0,0] - h[-1,0] )/Delta) * pow((h[0,0] + h[-1,0])/2,3);
      boundary ({Q});

    update \frac{\partial h}{\partial t}= - \frac{\partial Q(h)}{\partial x}

      foreach(){
        h[] -=  dt*( Q[1,0] - Q[0,0] )/Delta;
      }
      boundary ({h});  
    }

    Run

    Then compile and run:

    qcc  -g -O2 -DTRASH=1 -Wall viscous_collapse_noSV.c -o viscous_collapse_noSV
    ./viscous_collapse_noSV > out

    To run

    make viscous_collapse_noSV.tst
    make viscous_collapse_noSV/plots
    make viscous_collapse_noSV.c.html
     
    source c2html.sh viscous_collapse_noSV

    Results

    The analytical solution is obtained in observing that a selfsimilar solution exists \displaystyle h(x,t) = t^{-1/5} (\frac{9}{10} (b^2 - {(xt^{-1/5})}))^{1/3} \text { with } \int_{-b}^{+b} hdx =2. with b^2=\frac{3\ 10^{2/5} \left(\Gamma \left(\frac{2}{3}\right) \Gamma \left(\frac{11}{6}\right)\right)^{6/5}}{\pi ^{9/5}}=1.28338 for mass conservation

    In gnuplot type

       p[0:2]'out' not w l 

    which gives h(x,t) plotted as a function of x at t=10..500

     set output 'plot1.png'
     set xlabel "x"
     set ylabel "h(x,t)  "
     p[:]'out' not w l
    (script)

    (script)

    The self similar solution

       p[0:2]'out' u ($1/($4**.2)):($2*($4**.2)),(9./10*(1.28338-x*x))**(1/3.)
        

    which gives h(x,t)t^{1/5} plotted as a function of (xt^{-1/5})

    h is zero for (xt^{-1/5}) > b =1.13286

     set output 'plot2.png'
     set xlabel "x/t^{1/5}"
     set ylabel "h(x,t) t^{1/5}"
     p[0:1.5]'out' u ($1/($4**.2)):($2*($4**.2)) t 'num' w l,(9./10*(1.28338-x*x))**(1/3.) t'Self Sim'
    (script)

    (script)

    Links

    Bibliographie

    OK v2