This article is the seventh in a series on NASA High-Performance Computing and Communications (HPCC) research teams.
Late 1998 witnessed the launch of the first two International Space Station modules into Earth orbit. A prime concern for station inhabitants is safety; at the same time surface dwellers may benefit greatly from experiments carried out in space. To promote safety and productivity, planners first need to know how the near absence of gravity changes the physical behavior of matter, particularly fluids.
"If you are on Earth, with full gravity applying, how do you design experiments to study how systems will behave in microgravity?" asks Graham Carey, professor of aerospace engineering and engineering mechanics at the University of Texas at Austin (UT).
"You can design drop towers, and there are studies going on at NASA's Lewis Research Center, where you monitor objects as they fall for a few seconds," he said."You can also fly an aircraft for a minute or so on a parabolic arc and reduce the effect of gravity that way." In either case, it is difficult to design and carry out measurable experiments.
To address this dilemma, Carey has marshaled two groups of UT researchers to compare results from supercomputer simulations and inventive laboratory experiments in which effects due to gravity are no longer dominant. An example is working with sufficiently thin fluid layers. The researchers' NASA HPCC-funded efforts are uncovering fundamental details about how fluids behave in space and how thin fluid layers behave on Earth.
Surface tension takes over
Many fluid applications involve temperature changes, either naturally or by design. For example, when boiling water on a stove, heat from below causes the water to expand. Warmer, lighter fluid wants to rise to the air-cooled surface due to the vertical force of buoyancy. Scientists dub this process buoyancy-driven convection. "Since buoyancy is associated with gravity, if you go to a microgravity environment, that effect is no longer present," Carey explained. "Other phenomena normally masked by gravity become important."
Pressure and temperature differences at a fluid's surface, where it meets another liquid or gas, set up a surface tension gradient. "As a result of that gradient, the surface tension pulls the fluid across the surface, and that in turn tends to cause the fluid to rotate below," Carey said.
On Earth, such behavior can be mimicked with thin layers of fluid, where little buoyancy is possible. Experiments at UT's Center for Nonlinear Dynamics aim "to understand the instabilities and ultimately how heat is transported," said Harry Swinney, Center director and physics professor. One investigation uses a thin layer of purified silicone oil about 0.5 millimeters thick, equal to stacking five sheets of paper. Researchers spread the oil on a heated metal mirror and place a cooled sapphire window over it.
As heating increases, a honeycomb-like pattern of hexagons appears in the oil. The hexagons "spontaneously form when a certain value of the temperature difference is exceeded," Swinney said. As captured by an infrared camera, "the convection goes through the whole of the fluid, but it is being driven by forces at the surface." (See this issue's Experiment .)
French scientist Henri Bénard first observed the hexagonal pattern in 1900. Later theory attributed it to surface-tension-driven convection, but attempts to compare observations with theory did not succeed until 1995 with the Center's carefully controlled experiments on thin fluid layers. A NASA microgravity grant supported the research undertaken by graduate student Michael Schatz, now on the Georgia Tech faculty.
From fires to computer chips
Computation enables "non-invasive measurement" and visualization of the behavior inside the fluid. "If you have an accurate, reliable model that captures the physics, it will give you the behavior at any arbitrary point inside," Carey said. The team's MGFLO (MicroGravity FLOw) computer code incorporates ideas learned over 20 years in Carey's Computational Fluid Dynamics Laboratory.
This long-term experience guided Laboratory manager Robert McLay, MGFLO's chief designer, in including "all the necessary properties and features in a fluid mechanics code." Heeding experiments by Swinney's group and others, the code also supports surface tension studies and allows properties to change as a function of temperature. Finally, it is a simple matter to have gravity off or low, McLay said.
MGFLO simulations show in detail why temperature differences on a free surface are paramount in moving fluids under microgravity conditions. Spencer Swift, currently at Silicon Graphics, Inc., is extending the code to include movement of chemicals through fluids and chemical reactions as part of his doctoral work. Understanding these dynamics may have important implications for space station safety, notably in fire prevention.
Combustion drives most fires, and flames move at a speed related to how fast they consume material. Not so with fluid fires in microgravity. "The radiation from the flame will heat the surface, and the resulting surface tension will pull the flame across the surface," Carey said. Safety engineers should then consider using surface barriers, rather than chemicals, to stop flames from spreading. Related research focuses on the fundamental structure of chemical concentration patterns arising in reacting systems. Experiments by the Center's Anna Lin and simulation studies by the Laboratory's Alexandre Ardelea and Anand Pardhanani have identified highly structured spiral patterns, as well as chaotic labyrinthine patterns.
Thin-layer convection also can lead to formation of a dry spot where fluid has drained away. This phenomenon was predicted 30 years ago but was not seen until Stephen Van Hook, a NASA graduate student research fellow in the Center, devised an experiment that revealed dry spot formation in liquid layers less than 0.03 centimeter deep. While difficult to observe on Earth, dry spots are expected to be the dominant instability in space for liquid layers as thick as 1 centimeter.
To fashion computing devices, chip manufacturers diffuse chemicals into ever thinner heated silicon films, which behave like molten glass. The dry spot phenomenon or similar nonlinear effects may impact Earth-based production as device-size shrinks. However, "things that we learn about thin liquid layers and how they behave are also relevant to thick layers in microgravity," Carey said. Working with easily observed thick materials, station-based experimenters could further probe dry spot formation and other phenomena to advise U.S. manufacturers on improving terrestrial processes.
Robert McLay, of the University of Texas, describes how the MGFLO microgravity software's portability allows use of the optimal computer for different size problems.
"Likewise, the goal with a simulator such as MGFLO is to be able to guide the design of experiments and predict the behavior of such complex systems," Carey added. "This can reduce significantly the time to production and market."
Moving back a step, chip manufacturers slice the thin films from silicon crystals. Over the years they have found ways to make larger and purer crystals, which unfortunately is time-consuming and expensive. Higher-temperature gradients in crystal fabrication processes mean faster production, but an imperfection-producing temperature wave will arise if the temperature gradient becomes too large. "We've developed a technique for creating some feedback into this temperature profile that inhibits this wave," Swinney said. Apply this method, and rapid growth of exceptionally large and pure crystals is conceivable.
Exploring with confidence
Space shows promise as an effective manufacturing environment for pharmaceutical and other industries. For NASA, microelectromechanical systems similar to those in newer cars and trucks might successfully dampen station vibrations. Studying such problems requires adding to MGFLO capabilities to simulate more sophisticated geometries, the effects of suspending solid particles in fluids and the ways electrical fields change fluid properties.
"When you have all these coming together in a multiphysics model, you end up with quite a complex code," Carey said. "At the same time, you'd like to resolve things at the fine scale. It becomes a computationally intensive problem that you can't solve on a workstation."
"Hence, it's important that large-scale computer facilities exist, like the one at NASA Goddard Space Flight Center, so we can explore phenomena with confidence that we're adequately modeling the physics," he stated. [See sidebar article "New Models of Computing"]
More information on this UT research team is available at the following URL: www.cfdlab.ae.utexas.edu/nasa_hpcc