Institut für Geophysik, Westfälische Wilhelms-Universität Münster (Germany)
Local Project ID:
HPC Platform used:
JUQUEEN of JSC
Researchers at the University of Münster investigated precession driven flows in planets by direct numerical simulations on the JUQUEEN cluster. Precession of the rotation axis is an often neglected driving mechanism for flows in planetary cores, a field of research were other scientists mainly focus on the influence of thermal or chemical effects. As an additional complication that moves the models closer to the physical reality, the project considered the spheroidal shape of the planet, whereas previous research has been focused on the idealized case of a perfect sphere.
All the planets in our solar system (except mercury) are precessing, meaning that their axis of daily rotation is moving very slowly around another axis that is pointed at a right angle to the path of the planet around the sun. For example, such a motion takes about 26,000 years for the earth, with an angle of 23.5°. Precession is caused by the gravitational pull of a moon or the sun on the equatorial bulge of the planet.
The daily rotation of a planet leads to the well known Coriolis force, which changes the direction of the wind in regions of high or low pressure and also has a great influence on the flow of liquid iron in the earth's core. The effect of precession is similar, although more subtle: Due to the inertia of the flow, it aims to keep its old direction of rotation while the rotational direction of the planet itself changes continually, leading to complex flows. They have been proposed as a possible source for planetary magnetic field which are of great relevance for the existence of life.
Such a precessing flow is one of the cases were it is evident from theoretical work that the approximation of a perfectly spherical planet misses out on some essential physics. A spheroidal or ellipsoidal shape remains a challenge for the numerical simulation of geophysical flows, since certain simplifications applied in a sphere are not usable and therefore more computational power is necessary to reach the relevant parameters. The scientists in Münster therefore modified an existing code called Nek5000 to handle a spheroidal geometry, a problem that is much less established and researched than flows in spheres. The code has already been proven (e.g. on the Jülich Blue Gene/P Extreme Scaling Workshop 2010) to offer excellent parallel scaling capabilities to take full advantage of a modern high performance system for up to hundreds of thousands processing cores.
The computational capabilities of the JUQUEEN cluster at the Jülich Supercomputing Center allowed the researchers to study the basic dynamics of precessional flow when approaching more realistic geometries and parameters, where they found complex small-scale dynamics. One of the most interesting and far reaching findings from the simulations was the confirmation of a hysteresis effect that had already been predicted by theoretical considerations in the sixties. This means that the resulting flow in a planet depends not only on its current state, but also on its history, in this case on whether the time period of precession is increasing or decreasing – this is tantamount to an orbital companion moving closer to its neighboring planet or farther away from it.
Institut für Geophysik
Westfälische Wilhelms-Universität Münster
Corrensstraße 24, D-48149 Münster (Germany)
e-mail: jan.vormann [at] uni-muenster.de