Previous Up Next

4.1  PASS: Estimation of the numerical viscosity

Author
Stéphane Popinet
Command
sh reynolds.sh reynolds.gfs 1
Version
0.6.4
Required files
reynolds.gfs (view) (download)
reynolds.sh div5.ref div6.ref div7.ref reynolds.ref
Running time
3 minutes 11 seconds

The velocity field is initialised with an exact stationary solution of the Euler equations in a periodic 2D domain. An exact Euler solver would not change this field, however any numerical solver will introduce numerical dissipation which will slowly dissipate the kinetic energy of the initial solution. By monitoring the evolution of the kinetic energy, the dissipative properties of the numerical scheme can be measured (see [20] for details).

Figures 35 and figure 36 illustrate the evolution of the divergence of the velocity field with time. This is a check of the stability of the approximate projection and should remain bounded.

Figures 37 and 38 illustrates the evolution of the kinetic energy and the corresponding equivalent Reynolds number as a function of resolution. The higher the Reynolds number, the less dissipative the scheme.


Figure 35: Evolution of the maximum divergence.


Figure 36: Evolution of the L2 norm of the divergence.


Figure 37: Evolution of the kinetic energy.


Figure 38: Equivalent Reynolds number as a function of resolution.

4.1.1  PASS: Estimation of the numerical viscosity with refined box

Author
Stéphane Popinet
Command
sh ../reynolds.sh box.gfs 4
Version
0.6.4
Required files
box.gfs (view) (download)
../reynolds.sh div5.ref div6.ref div7.ref reynolds.ref
Running time
17 minutes 3 seconds

Same as the previous test but with a refined box in the middle and four modes of the exact Euler solution.


Figure 39: Evolution of the maximum divergence.


Figure 40: Evolution of the L2 norm of the divergence.


Figure 41: Evolution of the kinetic energy.


Figure 42: Equivalent Reynolds number as a function of resolution.


Previous Up Next