Running the program 'nsevol'
On this page a brief overview of the input for and output from the program
'nsevol', which solves the 2D Navier-Stokes equation in the
vorticity-streamfunction formulation,
as explained on the page about
Vortex Evolution with a Finite Difference
Method.
Contents of this page:
- General input data
- Initial vorticity distribution
- Initial tracer distribution
- Analysis of the results
The first part of the input file gives some general input data on the
discretisation of the equations, the time integration and the output,
i.e. about running the program 'nsevol'.
- Physical dimensions of the domain in the x,y-plane (in
dimensionless units).
- Number of grid points in the x and y direction.
- Boundary conditions for the left and right boundary, and for the
bottom and top boundary (the conditions on opposite boundaries are
always the same);
if a background flow (e.g. a strain flow) is desired, its
charactertics are given.
- Kind of background rotation (uniform, beta-plane, gamma-plane), if
any.
- Bottom topography, if any; this can be read from a file, generated
by an auxillary program by the name of 'nstopo'.
- Which solver to use for the streamfunction (FACR or NAG routine).
- Include nonlinear terms?
- Reynolds number, or no viscosity.
- Criteria to stop the computation: exceed some time limit or number
of time steps.
- Compute with constant or variable time step;
-- for constant time step: what is the time step, and
-- for variable time step: what is the (constant) Courant number.
- When or how often output is to be given and what kind of output
(there is a variety of ways to follow what is going on during and
after the evolution).
- The name of the run.
As initial vorticity distribution, a number of possibilities is
available, ranging from Bessel type
monopolar or dipolar vortices, to uniform vortices (Rankine, Kirchoff)
and Gaussian vortices. For these vortices, the location, size and
strength or velocity need to be given.
Also possible are: a uniform or random initial distribution, and other
distributions throughout the entire domain.
Furthermore, it is possible to read the initial vorticity distribution
from file, which comes in handy if one wants to continue a computation
from a previous run.
Tracers can be placed along lines (straight, circular/elliptic or
ractangular), as a blob (circular/elliptic or ractangular),
throughout the entire domain (either uniform or random), or as single
tracer points. It is also possible to put tracers on streamlines of
a Lamb dipole.
Furthermore, it is possible to read the initial tracer distribution
from file, which comes in handy if one wants to continue a computation
from a previous run.
What also needs to be specified is how often the tracer position is
written to a file and in what form.
===> Example input file
During the computation it is possible to follow what is going on in
a number of ways, e.g. by looking at the position and movement of
the tracers (thanks to a program from Chritiene Aarts and Pauline Vosbeek),
or by looking at a rough 'film' of the vorticity distribution.
Once the runs has finished, the flow field at certain moments in time
-- which is written to files by 'nsevol' -- can be analysed using an
auxillary program called 'nsfield', with which for instance plots and
profiles of the vorticity and/or streamfunction can be made (using a
suitable plot program, of course).
Also, the position of tracers can be plotted at specific moments in time,
or the time evolution of a specific tracer can be plotted, with the help of
the auxillary program called 'nstrs'.
Other output of 'nsevol' involves the maximum and minimum of vorticity and
their location, the circulation, energy and enstrophy, the Courant number
and time step, etc., as a function of time.
<=== Vortex Evolution with a Finite Difference
Method.
<=== Numerical simulations of 2D vortex
evolution with a Finite Difference Method.
Jos van Geffen --
Home |
Site Map |
Contact Me
created: 10 December 1996
last modified: 26 May 2001