Predicting the Unpredictable
By Geoffrey Giller
Cornell Engineers turning complex phenomena into models
The universe is endlessly complex. But that has never stopped humans from trying to tame its complexity into laws, equations, and models to help us understand what happens all around us. While we may never achieve a grand “theory of everything,” humans have made great strides over the past few centuries towards describing the physical world.
Despite these advances, however, there is still a great deal that remains unknown and unpredictable about our world. At Cornell Engineering, researchers are continually seeking new ways turn those unknowns into knowns, to make the universe slightly less complex: they are breaking the rules to predict the unpredictable.
Some researchers are tackling problems that have obvious applications. Olivier Desjardins, for example, is seeking to predict and model the movement of individual fuel droplets in combustion engines. Others are doing more fundamental research, like Susan Daniel’s work to categorize the different shapes that water droplets form when vibrated at high frequencies.
While the application of some research is not readily apparent, that’s the beauty and mystery of science: sometimes, the most esoteric discoveries are the ones that lead to the greatest technological advancements. But even knowing more about how the universe works fulfills a deep human motivation to take the chaos that surrounds us and make it a little more predictable.
Susan Daniel’s research epitomizes the notion that if you look closely at anything, you’ll see that it’s never as simple as it appears. A drop of water on a table, for example, may seem utterly mundane—but not for Daniel, an associate professor and director of graduate studies at the Robert Frederick Smith School of Chemical and Biomolecular Engineering. She sees the contact area between the drop and the table, which tells her something about the chemical makeup of the table’s surface; the angle the edges of the droplet form; even the way gravity makes the droplet flatten slightly.
And what if you jiggle the table that the droplet is on? The area of contact with the tabletop may change, along with the line of contact at the droplet’s edge. While even Daniel’s trained eyes can’t pick up the movement of the drop itself, it is revealed when filmed with a super high-speed camera registering up to 5,000 frames per second. (For the record, the “tabletop” Daniel and her students use is really a high-tech platform where the vibration frequency can be fine-tuned and on which she and her students attach slides with various chemical coatings).
Daniel and her students have been compiling a set of images of the shapes that droplets make when vibrated at different frequencies. The shapes are surprisingly beautiful: some resemble sea stars; others look like stacked, offset crosses. Many look artificial, as though sculpted by human hands, although in fact these shapes are natural properties of water plus just the right input of energy.
Daniel is also interested in taking two drops of water and seeing what happens when the two of them come into contact. The two points of contact with a solid surface become one, the area of contact with the surface changes and there is even some perturbation in the liquid when the two drops touch. Understanding these movements and changes, says Daniel, could result in “a new way to characterize a liquid-solid interface.”
You might wonder: what is the point? While this is fundamental scientific research—meaning that the end value is usually many steps removed and probably unpredictable from this early stage—Daniel says that understanding the fundamental properties that explain and predict the movement of individual droplets could have wide-ranging applications. “You can imagine making better coatings based on understanding the fundamentals of drop spreading,” she says. “Things like paint, high-resolution printing, any place where you’re going to have a high-velocity movement of liquid that’s coating a surface.
Daniel’s next project will tackle the final frontier: space. Next year, she and colleagues at Cornell will be sending an experiment up to the International Space Station (ISS) to take advantage of the nearly zero-gravity environment. One thing that they’re hoping their work will impact is how to maximize the efficiency of heat exchangers in space. On Earth, when liquid droplets condense on a cool plate as part of a heat exchanger’s function, those droplets will eventually slide off due to gravity, keeping the heat exchanger functioning efficiently. “In space, you can’t rely on gravity to remove the liquid droplets. So you’ve got to come up with some other way that you can move and manage fluids on surfaces,” Daniel says.
Working with Daniel is Paul Steen, also a professor in the Smith School of Chemical and Biomolecular Engineering. While Daniel’s work is experimental, Steen provides the theoretical underpinnings and predictions for the work with water droplets. He and his students have developed predictions for what frequencies of vibration will produce which shapes, which Daniel can then test and fine-tune in the real world.
One of the advantages of working in near-zero gravity, Steen says, is that the droplets can be much larger than they are on Earth. “Droplets that take spherical shape on Earth have to be smaller than about 3 millimeters,” he says. Any larger than that and they start to deform and flatten out due to gravity. But on the ISS, he says, much larger droplets of water stay nearly spherical due to surface tension. Working with larger drops of water means that equivalent vibrations move more slowly, allowing Steen and Daniel to observe phenomena that happen too fast even for the high-speed cameras on Earth. “We get about a 30-fold magnification of time,” Steen says, which will increase the detail that they can observe.
Almost everything that moves people or products around the world—cars, airplanes, ships, even rockets—uses internal combustion engines. And nearly all of those engines use fuel injection. That entails a stream of liquid fuel—gasoline or diesel in cars and ships, rocket fuel in rockets—that breaks apart into tiny droplets before burning up, in the presence of oxygen, to form the heated gases that make the engine work.
Olivier Desjardins, an associate professor in Cornell’s Sibley School of Mechanical and Aerospace Engineering, wants to model how those droplets form—a process called atomization—and what happens to them inside the engine in the fractions of a second between when they’re injected as a liquid stream and when they combust. But, as he explains, this is no easy feat: each droplet is likely made up of multiple components, each of which is evaporating at different rates as it atomizes. Surface tension of each droplet changes as different components evaporate. “The complexity is enormous,” Desjardins says. And while he works with equations that do explain a lot of this behavior, ultimately “the solution is intractable,” he says.
For a given objective—say, creating a more efficient engine—the level of understanding required may differ from another objective, like creating an engine that is more clean-burning. Desjardins notes that we understand the behavior in combustion engines well enough to create engines that operate safely. “We fly in airplanes all the time,” he points out, not to mention driving around in cars. But a better understanding of the fuel injection process may lead to engines that are not only safe, but also produce less soot or fewer nitrogen-oxide emissions.
Desjardins plays a video of a stream of liquid transforming into droplets in slow motion. In black and white, the shimmering water shoots out, first as a solid cylinder, but quickly devolving into chaos with droplets breaking off and strands of water swirling in mesmerizing shapes. Although it looks real, it’s actually an animation—a visualization of one of his modeled systems.
One of the advantages of using simulations as opposed to conducting experiments with actual fuels, he says, is the ability to see inside the field of liquid droplets. “You lose optical access,” he says of real-world experiments with fuels. While it is possible to use X-rays and magnetic resonance imaging to look inside, those methods require expensive and massive machinery. “The main idea is that you can’t simply take a picture and get useful data... that’s the power of simulations,” says Desjardins.
Of course, the downside of a simulation is that if the parameters describing the fuel’s behavior are incorrect, then the simulation will also be wrong. So Desjardins collaborates with experimentalists to validate that what he sees in his simulations matches what actually happens in the real world.
William Blake may have seen a world in a grain of sand, but Michel Louge sees worlds in the sand dunes of Qatar’s deserts. These crescent-shaped mounds move interminably across the landscape and can bury villages, disrupt railways and wreak havoc on airports. But to combat these shifting, slow-moving disasters, scientists need to understand how exactly they function.
That’s where Louge’s work comes in. The professor of mechanical and aerospace engineering originally studied avalanches in Utah, Montana and the Swiss Alps, using probes that measured snow density. But he was drawn to the problem of sand dunes when he witnessed firsthand the capability of dunes to engulf entire houses in West Africa. He realized that, with a few tweaks, the probes he used in the snow could also tell him about the sand in the dunes, and that he could possibly help combat both desertification and the damage caused by these dunes.
The dunes that Louge studies are, as dunes go, relatively simple. They form on hard surfaces, moving across them as individual units. Known as barchan dunes, they are “the simplest dunes to try and understand,” says Louge. Such dunes also form in the Namib Desert and have even been observed on Mars.
Louge and his French and Qatari colleagues have discovered some strange things about these dunes. For example, they found a rapid shift in humidity that propagated rapidly just below the surface of the dunes. “We just don’t know where it comes from,” says Louge. “Whether it’s important or not, I don’t know.” They also found evidence for what Louge calls a “bistable signal,” where the humidity reading seemed to be two values simultaneously. “It really surprised us greatly,” he says. “There were things we never expected to see.”
One option that Louge and his colleagues are considering for dealing with impending dunes is stopping them by injecting a slurry of microbes that basically turns the sand surface into concrete. For this, he has teamed up with Anthony Hay, an associate professor of microbiology at Cornell. It’s an idea that was presented by architect Magnus Larsson at a TED talk in 2009. Louge isn’t sure if the plan will work, but he feels that their work helps inject “a tiny bit more realism” into Larsson’s vision.
After his work in Qatar concludes next year, Louge hopes to turn towards China, which has been spending massively on efforts to fight desertification. “They know the importance of it,” he says.
Seeing the Future
As computers get more powerful and modeling gets more sophisticated, our predictions about the future will get more and more accurate. Given a set of inputs and a contained system—be it water droplets, engines, sand dunes or almost anything else imaginable—we’re getting closer to seeing the future. But those predictions will likely always be constrained. We will never build a computer like the demon imagined by French scientist Pierre-Simon Laplace, who, given perfect knowledge of the state of every particle in the universe and the laws governing them, could predict all the things that will ever happen. But we can start to predict more than ever thought possible—and through that knowledge, improve the technology we use every day