Professors Philip Varghese and David Goldstein with Bill McDoniel
From left to right: Billy McDoniel, Philip Varghese and David Goldstein. The computer image in the background shows the modeling of Pele, an active volcano on one of Jupiter's moons, Io.

Investigating the hypersonic flow of gas molecules and dust particles as they spew hundreds of miles into the atmosphere of a distant moon has a group of UT aerospace engineers hooked on space exploration.

Billy McDoniel, graduate student of aerospace engineering, is modeling the plumes that erupt from Pele, an active volcano on Io, one of Jupiter's four Galilean moons. He simulates the complex processes that cause the colorful patterns on the surface of the celestial body on Stampede, one of the fastest supercomputers in the world, at the Texas Advanced Computing Center.

"The amount of physics in his simulation code is unusually large," said David Goldstein, professor of the Department of Aerospace Engineering and Engineering Mechanics at The University of Texas at Austin. "Billy commonly uses hundreds of processors at one time."

Goldstein co-advises a team of graduate, post-doctoral and undergraduate students with Philip Varghese, director of the Center for Aeromechanics Research, and Laurence Trafton, senior research scientist with UT's Department of Astronomy.

The team has modeled planetary atmospheres since it first received funding from NASA in the mid-1990s. Their code, developed over time with the continuous addition of features required by different projects, is one of the most physics-complete for this type of research, Goldstein said.

"Certainly most other people doing work like this don't have access to the computers to do these massively paralleled simulations," Goldstein said.

In 1979, NASA's Voyager spacecraft captured the first images of the sulphur-yellow moon mottled with white frost, black lava lakes and orange-red rings. Io is the most volcanically active body in the solar system.

Perturbed by the other satellites, it stretches and squishes as it orbits Jupiter. Tides ripple through its rocky body and build heat that escapes through hot surface fissures.

"The tides move the solid ground 100 feet, which is a tremendous flexing of the material of Io," Varghese said. "And that's what keeps the core liquid and makes it such a volcanically and geologically active body."



In this video, aerospace engineering graduate student Billy McDoniel simulates a gas and dust plume of Pele, an extraterrestrial volcano located on the Galilean Moon of Jupiter, Io. Sulfur dioxide is shown in teal and three different sizes of dust are shown in other colors. Heavier dust does not rise as high as lighter dust or gas and is visible from the side as a dense plume core.

The volcanoes emit plumes of gas molecules, dust particles and sulphur dioxide droplets hundreds of miles into the tenuous Ionian atmosphere. Some particles are ionized and then trapped by Jupiter's magnetic field, the largest in the solar system, while others crash to the moon's surface in various patterns.

McDoniel has replaced the simple perfectly round holes that produced the plumes in decade-old modeling with multiple, irregular shaped surface cracks discovered in close-up infrared images produced by NASA's Galileo spacecraft in the mid-1990s.

"The structure of the internal flow and the irregular vents explain why the plumes do not leave clean circles on the surface," McDoniel said.

In his models, McDoniel writes substantial amounts of code to simulate the statistically accurate physics and motion of gas molecules and dust particles erupting from fissures and colliding with each other like ping pong balls.

"For example, I compute the drag that gas molecules have on dust particles," McDoniel said. "And I apply that force in my simulations."

The gas molecules and dust particles collide as they rise. Their motion can be described by continuum equations while the gas is dense, but as they move up and the density of the gas drops, the flow must be described by equations that describe molecular motion and the relatively long distances molecules travel between collisions. The dust and most of the gas eventually fall to the moon's surface under the influence of gravity. (A small amount of gas escapes Io and supplies the plasma torus orbiting around Jupiter.)

"We're the only team doing this sort of simulation of Pele," McDoniel said. "No one else is trying to do high fidelity, 3-D simulations that capture the physics of what is going on inside of the plume."

The dust, once separated from the gas, tends to move outward from the center of the simulated flow. The fallen particles are thought to create the dark "butterfly wing" spray on the moon's surface. The gas also expands outward, and the molecules are believed to fall in the orange-red oblong shape that encircles the volcano.

The group's implementation of the direct simulation Monte Carlo method is as novel as its goal to link models of Pele's plumes with those of Io's entire atmosphere.

"I don't think we have lot of competition in terms of doing a complete simulation of a planet's tenuous atmosphere while capturing a lot of the relevant physics in such a way that it is believable," McDoniel said.

Complete global simulations involve thousands of computer processors that run for days on the supercomputer, Goldstein said.

To optimize the computer's efficiency, the team dynamically adjusts the region of the moon assigned to each processor so that each one does nearly the same amount of work. A few processors work on large regions on the night side of Io, where there is nearly a vacuum, while many processors each work on a small region of the dayside where the gas is relatively dense and the computational load is high.

The computations are synchronized because they all depend on each other, McDoniel said. For example, the processors must run at the same speed and communicate with each other to simulate a global scale wind blowing across the surface.

"I'm interested in the programming and problem-solving side of the research," McDoniel said. "There is something to be learned about terrestrial volcanoes from those on Io, and there is value in developing methods like these to apply to other problems of more immediate engineering interest."