Numerical Simulations of Suction Vortices

page last modified on:
05:11 AM CDT, Thu 12 Mar 2009


This page shows results from numerical simulations of suction vortices, conducted in 2006. The simulations are configured as in the 1998 article by B. H. Fiedler in the Quarterly Journal of the Royal Meteorological Society. The computational domain could be called the "Fiedler Box": a 4x4x1 box that is a natural companion to the "Fiedler Can" used for the axisymmetric simulations. For 2006, the grid resolution is twice that used in the 1998 paper, in all directions (or 8 times the original number of points). Also, for 2006 the viscosity profile is simplified. The dimensionless viscosity (or inverse Reynolds number) is a constant below z=0.5 (here either .00005 or .0001). Above z=0.5 the viscosity linearly increases to 0.001 at the top boundary (z=1.).

The final sentence of the 1998 paper still stands: These results are exactly in line with the deduction of Fujita (1971), who estimated that a suction vortex would have a wind speed twice that of the parent vortex.

See a video of real suction vortices in the Ellis County tornado of May 4, 2007.

visc=.0001, swirl=.07

Single central vortex capped by a spiral breakdown.

visc=.0001, swirl=.10

Occasional twin vortices, but usually at most one intense vortex orbiting in the parent vortex.

visc=.0001, swirl=.15

Frequently twin vortices. Both can be intense.

visc=.00005, swirl=.07

Frequently two or more vortices. Two may be intense.
Simulations in a narrow range of the parameter space are shown in these four links. The viscosity is selected to be low enough to allow for suction vortices with a spiral vortex breakdown, but not so low as to lose confidence in the resolution provided by the grid. The swirl is selected to be large enough to allow for intense vortices, and to show the transition between single and multiple vortices. Larger values of swirl are not shown, because a larger parent vortex occurs, which tends to place the suction vortices outside the region of high resolution.

The transition to multiple vortices occurs with either increasing swirl or decreasing viscosity: either course provides a viscous boundary layer that is too thin to turn into the vertical and to provide the core of a single, central vortex with a central updraft that would be compatible with the available pressure deficit.

visc=.0001, swirl=.10
also contains a control experiment showing that friction is essential for making suction vortices.

This work was supported by National Science Foundation Grant ATM-0646914